&EPA
United States
Envirofunmlal Protection
Agnncy
Policy Assessment for the Review of the
Ozone National Ambient Air Quality
Standards


First External Review Draft

-------
                                    DISCLAIMER
This draft document has been prepared by staff from the Ambient Standards Group, Office of Air
Quality Planning and Standards, U.S. Environmental Protection Agency (EPA). Any opinions,
findings, conclusions, or recommendations are those of the authors and do not necessarily reflect
the views of the Agency. This draft document is being circulated for informational purposes and
to facilitate discussion with the Clean Air Scientific Advisory Committee (CASAC) to inform
the EPA's consideration of the ozone National Ambient Air Quality Standards. Questions
related to this draft document should be addressed to Susan Stone, U.S. Environmental
Protection Agency, Office of Air Quality Planning and Standards, C504-02, Research Triangle
Park, North Carolina 27711 (email: stone.susan@epa.gov).

-------
                                                          EPA-452/P-12-002
                                                                August 2012
Policy Assessment for the Review of the Ozone National Ambient Air
                           Quality Standards
                     First External Review Draft
                     U.S. Environmental Protection Agency
                         Office of Air and Radiation
                  Office of Air Quality Planning and Standards
                   Health and Environmental Impacts Division
                          Ambient Standards Group
                  Research Triangle Park, North Carolina 27711

-------
This page left intentionally blank.

-------
                              TABLE OF CONTENTS
1.     INTRODUCTION	1-1
   1.1    PURPOSE	1-1
   1.2    BACKGROUND	1-2
      1.2.1    Legislative Requirements	1-2
      1.2.2    Previous O3 NAAQS Reviews	1-4
      1.2.3    Litigation and Reconsideration of the 2008 O3 NAAQS Final Rule	1-7
      1.2.4    Current O3 NAAQS Review	1-7
   1.3    OVERVIEW OF O3 MONITORING AND AIR QUALITY	1-9
      1.3.1    O3 Monitoring Network	1-9
      1.3.2.   Emissions and Atmospheric Chemistry	1-10
      1.3.3    Air Quality Concentrations	1-11
      1.3.4    Background O3	1-13
          1.3.4.1  Average Background O3  in the U.S	1-16
          1.3.4.2  Distributions of Background O3 in the U.S	1-17
          1.3.4.3  Summary	1-22
   1.4.    GENERAL APPROACH AND ORGANIZATION OF THIS DOCUMENT	1-22
   1.5    REFERENCES	1-23

2.     CONSIDERATION OF THE HEALTH EVIDENCE	2-1
   2.1    WEIGHT-OF-EVIDENCE CHARACTERIZATION IN THE ISA	2-1
      2.1.1    Approach to Characterizing the Weight-of-Evidence	2-1
      2.1.2    Weight-of-Evidence Conclusions for O3	2-3
   2.2    HEALTH EFFECTS FOLLOWING SHORT-TERM EXPOSURES TO O3	2-4
      2.2.1    Respiratory Effects	2-4
          2.2.1.1   Lung function	2-6
          2.2.1.2   Pulmonary inflammation and injury	2-12
          2.2.1.3   Airways hyperresponsiveness	2-14
          2.2.1.4   Respiratory symptoms and medication use	2-15
          2.2.1.5   Lung host defense	2-18
          2.2.1.6   Hospital admissions and  emergency department visits	2-19
          2.2.1.7   Respiratory mortality	2-25
          2.2.1.8   Summary of respiratory effects	2-25
      2.2.2    Total Mortality	2-26
      2.2.3    Other Effects	2-35
          2.2.3.1   Cardiovascular effects	2-35
          2.2.3.2   CNS effects	2-36
   2.3    HEALTH EFFECTS FOLLOWING LONG-TERM EXPOSURES TO O3	2-36
      2.3.1    Respiratory Effects	2-37
          2.3.1.1   New-onset asthma	2-38
          2.3.1.2   Asthma hospital admissions and other morbidity effects	2-42
          2.3.1.3   Pulmonary structure and function	2-45
          23.1.4   Respiratory mortality	2-47
          2.3.1.5   Summary of respiratory effects	2-47
      2.3.2    Other Effects	2-48

-------
          2.3.2.1   Cardiovascular effects	2-48
          2.3.2.2   Reproductive and developmental effects	2-49
          2.3.2.3   CNS effects	2-50
   2.4    PUBLIC HEALTH IMPLICATIONS	2-50
      2.4.1    Adversity of Responses	2-50
      2.4.2    At-Risk Populations	2-57
          2.4.2.1   Approach to classifying factors that increase risk	2-58
          2.4.2.2   Factors that increase risk	2-59
          2.2.4.3   Factors that potentially increase risk	2-68
          2.4.2.4   Summary of factors that increase or potentially increase risk	2-69
      2.4.3    Averting Behavior	2-70
      2.4.4    Size of At-Risk Populations in the United States	2-71
   2.5    INTEGRATED DISCUS SIGN OF THE EVIDENCE	2-72
   2.6    REFERENCES	2-75

3.       ASSESSMENT OF O3-RELATED EXPOSURES AND RISKS	3-1
   3.1     EXPOSURE ASSESSMENT	3-2
      3.1.1    Approach to Assessing O3 Exposures	3-4
      3.1.2    Key Observations: Exposures to O3 Concentrations Above Health
              Benchmarks	3-6
   3.2      RISK ASSESSMENT	3-8
      3.2.1    Risk Assessment Based on Epidemiologic Studies	3-9
          3.2.1.1   All-cause mortality	3-10
          3.2.1.2   Respiratory morbidity	3-12
          3.2.1.3   Key sources of variability and uncertainty	3-14
      3.2.2    Risk Assessment Based on Controlled Human Exposure Studies	3-14
   3.3      REFERENCES	3-14

   4.     PRELIMINARY STAFF CONCLUSIONS REGARDING THE PRIMARY O3
        NAAQS	4-1
   4.1      APPROACH TO REVIEWING THE PRIMARY O3 NAAQS	4-2
      4.1.1     Approach Taken in 2008 Review	4-2
      4.1.2     Approach in the Current Review	4-3
   4.2       PRELIMINARY STAFF CONCLUSIONS REGARDING THE ADEQUACY
             OF THE CURRENT PRIMARY O3 STANDARD	4-6
      4.2.1      Evidence-Based Considerations	4-6
          4.2.1.1   Lung function, respiratory symptoms, airway inflammation, and other
                  respiratory effects	4-7
          4.2.1.2  Respiratory-related hospital admissions, emergency department visits, and
                  mortality	4-14
          4.2.1.3   Total mortality	4-23
          4.2.1.4  Other health endpoints	4-27
          4.2.1.5   Integrated consideration of health evidence	4-27
      4.2.2      Exposure- and Risk-Based Considerations	4-30
          4.2.2.1   Exposures above health benchmark concentrations	4-30
          4.2.2.2   Estimates of Os-induced lung function impairment	4-33
                                         11

-------
         4.2.2.3   Estimates of O3-associated mortality and morbidity	4-34
         4.2.2.4   Integrated consideration of the exposure and risk information	4-39
     4.2.3      CASAC Advice from Previous Review	4-41
     4.2.4      Preliminary Staff Conclusions on the Adequacy of the Current O3
               Standard	4-43
  4.3       ADDITIONAL ANALYSES TO INFORM SECOND DRAFT PA	4-46
     4.3.1      Additional Exposure and Risk Analyses	4-46
         4.3.1.1   Averaging time	4-47
         4.3.1.2   Alternative Levels	4-50
     4.3.2      Approaches to Translating Epidemiologic Evidence	4-53
  4.4        SUMMARY OF STAFF CONCLUSIONS ON THE PRIMARY O3
             STANDARD	4-58
  4.5        KEY AREAS FOR FUTURE RESEARCH AND DATA COLLECTION ... 4-58
  4.6        REFERENCES	4-58

5.      CONSIDERATION OF THE WELFARE EFFECTS EVIDENCE	5-1
  5.1        WEIGHT-OF-EVIDENCE CHARACTERIZATION IN THE ISA	5-1
     5.1.1       Approach to Characterizing the Weight of Evidence	5-1
     5.1.2       Weight-of-Evidence Conclusions for O3	5-3
  5.2        MECHANISMS GOVERNING PLANT RESPONSE TO O3
             EXPOSURES	5-5
     5.2.1       Ozone Uptake into the Leaf	5-6
     5.2.2       Cellular to Systemic Response	5-7
     5.2.3       Detoxification	5-8
     5.2.4       Effects on Plant Metabolism	5-9
  5.3        EFFECTS ON VEGETATION 	5-10
     5.3.1        Visible Foliar Injury and Biomonitoring	5-10
     5.3.2        Growth, Productivity and Carbon Storage	5-13
     5.3.3        Factors that Modify Plant Response to O3	5-18
  5.4         BIOLOGICALLY RELEVANT EXPOSURE INDICES	5-20
     5.4.1        Exposure Factors that Influence Plant Response	5-20
         5.4.1.1     Relevant exposure periods: diurnal and seasonal	5-21
         5.4.1.2     Relevant exposure concentrations: differential weighting	5-23
     5.4.2.       Biologically Relevant Exposure Metrics	5-24
     5.4.3        Conclusions	5-27
  5.5         ECOSYSTEM-LEVEL IMPACTS	5-27
     5.5.1        Overview: Ecosystem Scale, Structure, Function, and Services	5-28
     5.5.2        Productivity and Carbon Sequestration	5-30
     5.5.3        Water Cycling	5-33
     5.5.4        Below-ground Processes	5-34
     5.5.5        Community Composition	5-35
     5.5.6        Insects and Other Wildlife	5-36
  5.6        CONSIDERATIONS REGARDING ADVERSITY IN PUBLIC WELFARE
              CONTEXT	5-36
  5.7         CONSIDERATIONS REGARDING OTHER WELFARE EFFECTS	5-38
     5.7.1        Ozone Effects on Climate Change	5-38
                                       in

-------
      5.7.2.       UV-B Radiation Effects on Ecosystems and Materials	5-39
   5.8         REFERENCES	 5-40

6.      CONSIDERATION OF EXPOSURE AND RISK ASSESSMENTS	6-1
   6.1          EXPOSURE	6-3
      6.1.1        Monitoring and Air Quality	6-3
      6.1.2        Approach to Estimating Exposures	6-4
      6.1.3        Exposure Surface Generation	6-5
   6.2          CONSIDERATION OF O3 RISK ESTIMATES	6-7
      6.2.1        Approaches to Assessing Risks	6-7
      6.2.2        Risk of Biomass Loss in Individual Sensitive Tree Species	6-8
      6.2.3        Risk of Foliar Injury Incidence	6-13
   6.3          ECOSYSTEM SERVICES	6-13
   6.4          UNCERTAINTIES IN RISK ASSESSMENTS	6-14
   6.5          REFERENCES	6-14

7.       PRELIMINARY STAFF CONCLUSIONS REGARDING THE SECONDARY O3
        NAAQS	7-1
   7.1          APPROACH TO REVIEWING THE SECONDARY O3 NAAQS	7-2
      7.1.1       Approach Taken in 2008 Review	7-2
      7.1.2       Approach in the Current Review	7-5
   7.2          PRELIMINARY STAFF CONCLUSIONS REGARDING THE ADEQUACY
              OF THE CURRENT SECONDARY O3 NAAQS	7-8
      7.2.1      Vegetation Evidence-, Exposure-, and Risk-based Considerations	7-8
         7.2.1.1    Visible foliar injury	7-8
         7.2.1.2   Reduced growth and productivity effects	7-10
      7.2.2      Biologically Relevant Exposure Indices	7-12
      7.2.3      CASAC Views on the Adequacy of the Current Standard	7-15
      7.2.4      Preliminary Staff Conclusions on the Current Secondary O3 Standard	7-17
   7.3         POTENTIAL ALTERNATIVE STANDARDS FOR FURTHER
              ANALYSIS	7-19
      7.3.1      Form	7-20
      7.3.2      Averaging Times	7-21
         7.3.2.1    Diurnal exposure period	7-21
         7.3.2.2   Seasonal exposure period	7-22
         7.3.2.3    Annual or 3-year average	7-23
      7.3.3       Level	7-24
   7.4         SUMMARY OF STAFF CONCLUSIONS ON THE SECONDARY O3
              STANDARD	7-28
   7.5        KEY AREAS FOR FUTURE RESEARCH AND DATA COLLECTION .... 7-28
   7.6        REFERENCES	7-28
                                       IV

-------
                                      TABLES
Table 1-1. Summary of Primary and Secondary O3 NAAQS Promulgated During the Period
         from 1971 to 2008	1-4
Table 1-2. Modeled median and 75th percentile 2006 MDA8 NAB concentrations at monitoring
         site locations (as presented in Henderson et al. (2012) and adapted from GEOS-Chem
         and CAMx modeling applications) when simulated base case Os MDA8 values were
         greater than 55 ppb	  1-20
Table 2-1. Gradation of Individual Responses to Short-Term Ozone Exposure in Healthy People
	2-55
Table 2-2. Gradation of Individual Responses to Short-Term Ozone Exposure in People with
         Respiratory	2-56
Table 2-3. Prevalence of asthma by age in the U.S	2-60
Table 4-1. Average Ozone Design Values in U.S. Cities where Epidemiologic Studies have
         Evaluated Associations between Short-Term Os and Respiratory-Related Hospital
         Admissions (HA), Emergency Department (ED) Visits, and Mortality	4-18
Table 4-2. Ozone Design Values in U.S. Cities where Epidemiologic Studies have Evaluated
         Associations between Long-Term Os and Respiratory Effects	4-21
Table 4-3. Ozone Design Values in U.S. Cities where Epidemiologic Studies have Evaluated
         Associations between Short-Term Os and Non-Accidental Mortality	4-24
Table 4-4. Average Estimated Percent Decrease in Maximum and Seasonal Os Concentrations
         with Simulation of Just Meeting the Current Standard (2006-2008) Using Quadratic
         Rollback	4-49
Table 6-1  Summary of Tree Species Biomass Loss Analyses	6-12

-------
                                       FIGURES

Figure 2-1. Summary of mortality risk estimates for short-term Os and all-cause (nonaccidental)
          mortality	2-28
Figure 4-1. Overview of Approach to Reviewing Primary Os NAAQS	4-5
Figure 4-2. Average Versus Highest 8-hour Daily Maximum Os Concentrations in New York City,
          1999-2006	4-20
Figure 5-1. An illustrative diagram of the major pathway through which O3 enters plants and the
          major endpoints that O3 may affect in plants and ecosystems. (Figure 9-1
          ISA)	5-5
Figure 6-1 U.S. Rural NCore, CASTNET and NFS POMS sites in 2010 (U.S. EPA, 2012a, Figure
          3-22)	6-4
Figure 6-2 National Ozone Exposure Surface of Recent (2006 - 2008) Ambient W126	6-6
Figure 6-3 National Ozone Exposure Surface Simulating Just Meeting the Current Standard . .6-7
Figure 6-4 Relative Biomass Loss Concentration-Response Functions for 11 Tree species. The
          vertical dashed lines represent the maximum W126 values for (A) recent (2006 to
          2008) conditions, and (B) just meeting the current standard	6-9
Figure 6-5 "Blow Up" View of Relative Biomass Loss Concentration-Response Functions for
          Tree species	6-11
Figure 7-1 Overview of Approach to Reviewing the Secondary Os NAAQS	7-7
                                           VI

-------
                                    1.      INTRODUCTION
      1.1    PURPOSE
 3          The U.S. Environmental Protection Agency (EPA) is presently conducting a review of
 4    the national ambient air quality standards (NAAQS) for ozone (O^). An overview of the
 5    approach to reviewing the Os NAAQS is presented in the Integrated Review Plan for the Os
 6    National Ambient Air Quality Standards (TRP, US EPA, 201 la). The IRP discusses the schedule
 7    for the review; the approaches to be taken in developing key scientific, technical, and policy
 8    documents; and the key policy-relevant issues that will frame EPA's consideration of whether
 9    the current NAAQS for Os should be retained or revised.
10          As part of the review process, a Policy Assessment (PA) is prepared by staff in the EPA's
11    Office of Air Quality Planning and Standards (OAQPS).  The PA is intended to help bridge the
12    gap between the relevant scientific information and assessments and the judgments required of
13    the EPA Administrator in determining whether, and if so how, it is appropriate to revise the
14    primary (health-based) and secondary (welfare-based) NAAQS for Os.  The final PA will seek to
15    provide EPA staff conclusions related to the broadest range of policy options that could be
16    supported by the currently available scientific evidence and technical information for
17    consideration by the Administrator. In so doing, we recognize that the selection of a specific
18    approach to reaching final decisions on the primary and secondary Os standards will reflect the
19    judgments of the Administrator.
20          In this first draft of the PA, we take into account the available scientific and technical
21    information as assessed in the third draft of the Integrated Science Assessment for Os and
22    Related Photochemical Oxidants (ISA, US EPA, 2012a) and the first drafts of the exposure and
23    risk assessment documents: Health Risk and Exposure Assessment for Ozone: First External
24    Review Draft (Health RE A, US EPA, 2012b) and Welfare Risk and Exposure Assessment for
25    Ozone: First External Review Draft (Welfare REA, US EPA, 2012c). In so doing, we focus on
26    information that is most pertinent to evaluating the basic elements of NAAQS: indicator1,
27    averaging time, form2, and level. These elements, which together serve to define each standard,
28    must be considered collectively in evaluating the health and welfare protection afforded by the
29    Os standards.  Although this first draft PA should be  of use to all parties interested in this Os
30    NAAQS review, it is written with an expectation that the reader has familiarity with the scientific
      lrThe "indicator" of a standard defines the chemical species or mixture that is to be measured in determining whether
      an area attains the standard.
      2The "form" of a standard defines the air quality statistic that is to be compared to the level of the standard in
      determining whether an area attains the standard.
                                                 1-1

-------
 1    and technical discussions contained in the third draft of the ISA and the first drafts of the Health
 2    and Welfare REAs.
 3           Following this introductory chapter, this draft PA is organized into two main parts.
 4    Chapters 2 through 4 focus on the review of the primary  Os NAAQS while chapters 5 through 7
 5    focus on the review of the secondary Os NAAQS. The remainder of this chapter provides
 6    background information on the NAAQS program and on the Os NAAQS in particular (section
 7    1.2); an overview of the Os ambient monitoring network, precursor emissions, and Os air quality
 8    (section 1.3); and an overview of the approach to reviewing the 63 NAAQS and of the
 9    organization of the remainder of this draft PA (section 1.4).

10    1.2    BACKGROUND

11    1.2.1   Legislative Requirements
12           Two sections of the Clean Air Act (CAA) govern the establishment and revision of the
13    NAAQS.  Section 108 (42 U.S.C. 7408) directs the Administrator to identify and list "air
14    pollutants" that in her "judgment, cause or contribute to air pollution which may reasonably be
15    anticipated to endanger public health or welfare" and satisfy two other criteria, including "whose
16    presence ... in the ambient air results from numerous or diverse mobile or stationary sources"
17    and to issue air quality criteria for those that are listed. Air quality criteria are intended to
18    "accurately reflect the latest scientific knowledge useful in indicating the kind and extent of all
19    identifiable effects on public health or welfare which may be expected from the presence of [a]
20    pollutant in the ambient air ... ." (42 U.S.C. 7408).  Section 109 (42 U.S.C. 7409) directs the
21    Administrator to propose and promulgate "primary" and "secondary" NAAQS for pollutants for
22    which air quality criteria are issued. Section 109(b)(l) defines a primary standard as one "the
23    attainment and maintenance of which in the judgment of the Administrator, based on such
24    criteria and allowing an adequate margin of safety, are requisite to protect the public health."3  A
25    secondary standard, as defined in section 109(b)(2), must "specify a level of air quality the
26    attainment and maintenance of which,  in the judgment of the Administrator, based on such
      3The legislative history of section 109 indicates that a primary standard is to be set at "the maximum permissible
      ambient air level.. . which will protect the health of any [sensitive] group of the population," and that for this
      purpose "reference should be made to a representative sample of persons comprising the sensitive group rather than
      to a single person in such a group" [S. Rep. No. 91-1196, 91st Cong., 2d Sess. 10 (1970)].

                                                 1-2

-------
 1    criteria, is requisite to protect the public welfare from any known or anticipated adverse effects
 2    associated with the presence of such air pollutant in the ambient air." 4
 3           The requirement that primary standards include an adequate margin of safety was
 4    intended to address uncertainties associated with inconclusive scientific and technical
 5    information available at the time of standard setting. It was also intended to provide a reasonable
 6    degree of protection against hazards that research has not yet identified. Lead Industries
 1    Association v. EPA, 647 F.2d 1130, 1154 (D.C. Cir 1980), cert, denied, 449 U.S. 1042 (1980);
 8    American Petroleum Institute v. Costle, 665 F.2d 1176, 1186 (D.C. Cir. 1981), cert, denied, 455
 9    U.S. 1034(1982).  Both kinds of uncertainties are components of the risk associated with
10    pollution at levels below those at which human health effects can be said to occur with
11    reasonable scientific certainty. Thus, in selecting primary standards that include an adequate
12    margin of safety, the Administrator is seeking not only to prevent pollution levels that have been
13    demonstrated to be harmful but also to  prevent lower pollutant levels that may pose an
14    unacceptable risk of harm, even if the risk is not precisely identified as to nature or degree. The
15    CAA does not require the Administrator to establish a primary NAAQS at a zero-risk level or at
16    background concentration levels, see Lead Industries Association v. EPA, 647 F.2d at 1156 n. 51,
17    but rather at a level that reduces risk sufficiently so as to protect public health with an adequate
18    margin of safety.
19           In addressing the requirement for an adequate margin of safety, EPA considers such
20    factors as the nature and severity of the health effects involved, the size of the population(s) at
21    risk, and the kind and degree of the uncertainties that must be addressed.  The selection of any
22    particular approach to providing an adequate margin of safety is a policy choice left specifically
23    to the Administrator's judgment. Lead Industries Association v. EPA, 647 F.2d at 1161-62;
24    Whitman v. American Trucking Associations, 531  U.S. 457, 495 (2001).
25           In setting primary and secondary standards that  are "requisite" to protect public health
26    and welfare, respectively, as provided in section 109(b), EPA's task is to establish standards that
27    are neither more nor less stringent than necessary for these purposes. In so doing, EPA may not
28    consider the costs of implementing the  standards.  See generally, Whitman v. America Trucking
29    Associations, 531 U.S. 457, 465-472, 475-76  (2001). Likewise, "[attainability and
30    technological feasibility are not relevant considerations in the promulgation of national ambient
31    air quality standards." American Petroleum Institute v. Costle, 665 F. 2d at 1185.
      4Welfare effects as defined in section 302(h) (42 U.S.C. 7602(h)) include, but are not limited to, "effects on soils,
      water, crops, vegetation, man-made materials, animals, wildlife, weather, visibility, and climate, damage to and
      deterioration of property, and hazards to transportation, as well as effects on economic values and on personal
      comfort and well-being."

                                                  1-3

-------
 1
 2
 3
 4
 5
 6
 7
 8
 9
10
11
12
13
14
15
16
17
18
            Section 109(d)(l) requires that "not later than December 31, 1980, and at 5-year intervals
     thereafter, the Administrator shall complete a thorough review of the criteria published under
     section 108 and the national ambient air quality standards . . . and shall make such revisions in
     such criteria and standards and promulgate such new standards as may be appropriate
     Section 109(d)(2) requires that an independent scientific review committee  "shall complete a
     review of the criteria . . . and the national primary and secondary ambient air quality standards . .
     .  and shall recommend to the Administrator any new . . . standards and revisions of existing
     criteria and standards as may be appropriate . .  . ." This independent review function is
     performed by the Clean Air Scientific Advisory Committee (CAS AC) of EPA's Science
     Advisory Board.5
     1.2.2  Previous O3 NAAQS Reviews
            Table 1-1  summarizes the 63 NAAQS that have been promulgated to date. In each
     review, the  secondary standard has been set to be identical to the primary standard. These
     reviews are briefly described below.
     Table 1-1. Summary of Primary and Secondary O3 NAAQS Promulgated During the Period
        from 1971 to 2008
Final Rule
1971
(36 FR 8 186)
1979
(44 FR 8202)
1993
(58 FR 13008)
1997
(62 FR 38856)
2008
(73 FR 16483)
Indicator
Total
photochemical
oxidants
0;,
Averaging
Time
I-hr
1-hr
Level
(ppm)
(108
0.12
Form
Not to be exceeded more than one
hr per year
Attainment is defined when the
expected number of days per
calendar year, with maximum
hourly average concentration
greater than 0, 1 2 ppm, is equal to
or less than 1
EPA decided that revisions to the standards were not warranted at the time.
0?
Ql
8-hr
8-hr
0.08
0,075
Annual fourth-highest daily
maximum 8-hr concentration,
averaged over 3 years
Form of the standards remained
unchanged relative to the 1 997
standard
     5 Lists of CAS AC members and of members of the CAS AC O3 Review Panel are available at:
     http://vosemite.epa.gov/sab/sabpeople.nsfAVebExternalCommitteeRosters?OpenView&committee=CASAC&secon
     dname=Clean%20Air%20Scientific%20Advisory%20Committeeand
     http://vosemite.epa.gov/sab/sabpeople.nsfAVebExternalSubCommitteeRosters?OpenView&committee=CASAC&su
     bcommittee=Ozone%20Review%20Panel respectively.
                                                 1-4

-------
 1
 2          The EPA first established primary and secondary NAAQS for photochemical oxidants in
 3    1971 (36 FR 8186, April 30, 1971). Both primary and secondary standards were set at a level of
 4    0.08 parts per million (ppm), 1-hr average, total photochemical oxidants, not to be exceeded
 5    more than one hour per year.  The standards were based on scientific information contained in
 6    the 1970 Air Quality Criteria for Photochemical Oxidants (U.S. DHEW, 1970).  The first
 7    periodic review of the NAAQS for photochemical oxidants was initiated in 1977. Based on the
 8    1978 Air Quality Criteria for Ozone and Other Photochemical Oxidants (U.S. EPA, 1978), EPA
 9    published proposed revisions to the original NAAQS in 1978 (43 FR 16962) and final revisions
10    in 1979 (44 FR 8202). The level of the primary and secondary standards was revised from 0.08
11    to 0.12 ppm; the indicator was revised from photochemical oxidants to O3; and the form of the
12    standards was revised from a deterministic to a statistical form, which defined attainment of the
13    standards as occurring when the expected number of days per calendar year with maximum
14    hourly average concentration greater than 0.12 ppm is equal to or less than one.
15          In 1982, EPA announced plans to revise the  1978 Air Quality Criteria document (47 FR
16    11561), and in 1983 EPA initiated the second periodic review of the O3 NAAQS (48 FR 38009).
17    EPA subsequently published the 1986 Air Quality Criteria for Ozone and Other Photochemical
18    Oxidants (U.S. EPA, 1986) and 1989 Staff Paper  (U.S. EPA, 1989).  Following publication of
19    the 1986 Air Quality Criteria Document (AQCD), a number of scientific abstracts and articles
20    were published that appeared to be of sufficient importance concerning potential health and
21    welfare effects of O3 to warrant preparation of a Supplement (U.S. EPA, 1992).  Under the terms
22    of a court order, on August 10,  1992 EPA published a proposed decision stating that revisions to
23    the existing primary and secondary standards were not appropriate at the time (57 FR 35542).
24    The notice explained that the proposed decision would complete EPA's review of information on
25    health and welfare effects of O3 assembled over a 7-year period and contained in the 1986
26    AQCD and its 1992 Supplement.  The proposal also announced EPA's intention to proceed as
27    rapidly as possible with the next review of the air quality criteria and standards for O3 in light of
28    emerging evidence of health effects related to  6- to 8-hour O3 exposures.  On March 9, 1993,
29    EPA concluded the review by deciding that revisions to the standards were not warranted at that
30    time (58 FR 13008).
31          In August 1992 EPA announced plans  to initiate the third periodic review of the air
32    quality criteria and O3 NAAQS (57 FR 35542). On the basis of the scientific evidence contained
33    in the 1996 Air Quality Criteria for Ozone and related Photochemical Oxidants (U.S. EPA,
34    1996a), the 1996 Staff Paper (U.S. EPA, 1996b), and related technical support documents,
35    linking exposures to ambient O3 to adverse health and welfare effects at levels allowed by the

                                               1-5

-------
 1    then existing standards, EPA proposed to revise the primary and secondary 63 standards on
 2    December 13, 1996 (61 FR 65716).  The EPA proposed to replace the then existing 1-hour
 3    primary and secondary standards with 8-hour average O3 standards set at a level of 0.08 ppm
 4    (equivalent to 0.084 ppm using standard rounding conventions). The EPA also proposed to
 5    establish a new distinct secondary standard using a biologically based cumulative, seasonal form.
 6    The EPA completed the review on July 18, 1997 (62 FR 38856) by setting the primary standard
 7    at a level  of 0.08 ppm, based on the annual fourth-highest daily maximum 8-hr average
 8    concentration, averaged over three years, and setting the secondary standard identical to the
 9    revised primary standard.
10          On May 14, 1999, in response to challenges by industry and others to EPA's 1997
11    decision, the U. S. Court of Appeals for the District of Columbia Circuit (D.C. Circuit Court)
12    remanded the 63 NAAQS to EPA, finding that section 109 of the Act, as interpreted by EPA,
13    effected an unconstitutional delegation of legislative authority.  In addition, the D.C. Circuit
14    Court directed that, in responding to the remand, EPA should consider the potential beneficial
15    health effects of O3 pollution in shielding the public from the effects of solar ultraviolet (UV)
16    radiation, as well as adverse health effects. On January 27, 2000, EPA petitioned the U.S.
17    Supreme Court for certiorari on the constitutional issue (and two other issues) but did not request
18    review of the D.C. Circuit Court ruling regarding the  potential beneficial health effects of 63. On
19    February  27,  2001, the U.S. Supreme Court unanimously reversed the judgment of the  D.C.
20    Circuit Court on the constitutional issue, holding that section 109 of the CAA does not delegate
21    legislative power to the EPA in contravention of the Constitution, and remanded the case to the
22    D.C.  Circuit Court to consider challenges to the O3 NAAQS that had not been addressed by that
23    Court's earlier decisions. On March 26, 2002, the D.C. Circuit Court issued its final decision,
24    finding the 1997 O3 NAAQS to be "neither arbitrary nor capricious," and denied the remaining
25    petitions for review. In response to the D.C. Circuit Court remand to consider the potential
26    beneficial health effects of O3 pollution in shielding the public from effects of solar (ultraviolet
27    or UV) radiation, on November 14, 2001,  EPA proposed to leave the 1997 8-hour NAAQS
28    unchanged (66 FR 52768).  After considering public comment on the proposed decision, EPA
29    published its final response to this remand on January 6, 2003, reaffirming the 8-hour O3
30    NAAQS set in 1997 (68 FR 614). Finally, on April 30, 2004, EPA announced the decision to
31    make the  1-hour O3 NAAQS no longer applicable to areas one year after the effective date of the
32    designation of those areas for the 8-hour NAAQS (69 FR 23966). For most areas, the date that
33    the 1-hour NAAQS no longer applied was June 15, 2005.
34          The EPA initiated the next periodic review of the air quality criteria and O3 standards in
35    September 2000 with a call for information (65 FR 57810). The schedule for completion of that

                                                1-6

-------
 1    rulemaking later became governed by a consent decree resolving a lawsuit filed in March 2003
 2    by a group of plaintiffs representing national environmental and public health organizations.
 3    Based on the Air Quality Criteria for Ozone and Other Photochemical Oxidants (US EPA, 2006)
 4    published in March 2006 and the Staff Paper (U.S EPA, 2007a) and related technical support
 5    documents published in July 2007, the proposed decision was published in the Federal Register
 6    on July 11, 2007 (72 FR 37818).  The EPA proposed to revise the level of the primary standard
 7    to a level within the range of 0.075 to 0.070 ppm.  Two options were  proposed for the secondary
 8    standard: (1) replacing the current standard with a cumulative, seasonal standard, expressed as an
 9    index of the annual sum of weighted hourly concentrations cumulated over 12 daylight hours
10    during the consecutive 3-month period within the Os season with the  maximum index value, set
11    at a level within the range of 7 to 21 ppm-hrs, and (2) setting the secondary standard identical to
12    the revised primary standard. The EPA completed the review with publication of a final decision
13    on March 27, 2008 (73 FR 16436), revising the level of the 8-hour primary Os standard from
14    0.08 ppm to 0.075  ppm and revising the secondary standard to be identical to the revised primary
15    standard.
16    1.2.3  Litigation  and Reconsideration of the 2008 O3 NAAQS Final Rule
17          In May 2008, state, public health, environmental, and industry petitioners filed suit
18    against EPA regarding that final decision.  At EPA's request the consolidated cases were held in
19    abeyance pending EPA's reconsideration of the 2008 decision. A notice of proposed rulemaking
20    to reconsider the 2008 final decision was issued by the Administrator on January 6, 2010. Three
21    public hearings were held. The Agency solicited CAS AC review of the proposed rule on
22    January 25, 2010 and additional CAS AC advice on January 26, 2011. On September 2, 2011,
23    the Office of Management and Budget returned the draft final rule on reconsideration to EPA for
24    further consideration.  EPA decided to coordinate further proceedings on its voluntary
25    rulemaking on reconsideration with the ongoing periodic review, by deferring the completion of
26    its voluntary  rulemaking on reconsideration until it completes its statutorily-required periodic
27    review.  In light of that, the litigation on the 2008 final decision is no  longer being held in
28    abeyance and is proceeding. The 2008 ozone standards remain in effect.
29    1.2.4  Current O3 NAAQS Review
30          On September 29, 2008, the EPA's NCEA-RTP announced the initiation of a new
31    periodic review of the air quality criteria for Os and related photochemical oxidants and issued a
32    call for information in the Federal Register (73 FR 56581, Sept. 29, 2008). A wide range of
33    external  experts as well as EPA staff, representing a variety of areas of expertise (e.g.,
34    epidemiology, human and animal toxicology, statistics, risk/exposure analysis, atmospheric
35    science,  ecology, biology, plant science, benefits analysis) participated in a workshop, held by
                                                1-7

-------
 1   EPA on October 28-29, 2008 in Research Triangle Park, NC. The workshop provided an
 2   opportunity for a public discussion of the key policy-relevant issues around which EPA would
 3   structure this O?, NAAQS review and the most meaningful new science that would be available
 4   to inform our understanding of these issues.
 5          Based in part on the workshop discussions, EPA developed a draft IRP outlining the
 6   schedule, process, and key policy-relevant questions that would guide the evaluation of the air
 7   quality criteria for O?, and the review of the primary and secondary  O?, NAAQS.  A draft of the
 8   integrated review plan was released for public review and comment in September 2009 and was
 9   the subject of a consultation with the Clean Air Scientific Advisory Committee (CASAC) on
10   November 13, 2009 (74 FR 54562; October 22, 2009).6 Comments received from that
11   consultation and from the public were considered in finalizing the plan and in beginning the
12   review of the air quality criteria. The EPA's overall plan and schedule for this review is
13   presented in the Integrated Review Plan for the Ozone National Ambient Air Quality Standards.7
14          As part of the process of preparing the Os ISA, NCEA hosted a peer  review workshop in
15   October 29-30, 2008 (73 FR 56581, September 29, 2008) on preliminary drafts of key ISA
16   chapters. The first external review draft ISA (US EPA, 201 la; 76 FR 10893, February 28, 2011)
17   was reviewed by CAS AC and the public at a meeting held in May 19-20, 2011 (76 FR 23809;
18   April 28, 2011). Based on CASAC and public comments, NCEA prepared a second draft ISA
19   (US EPA, 201 Ib; 76 FR 60820, September 30, 2011), which was reviewed by CASAC and the
20   public at a meeting held on January 9-10, 2012 (76 FR 236, December 8, 2011).  Based on
21   CASAC  and public comments, NCEA prepared a third draft ISA (US EPA 2012a; 77 FR 36534;
22   June 19,  2012), which will be reviewed at a CASAC meeting in September 2012.
23          The EPA's plans for conducting the Risk and Exposure Assessment (REA) documents
24   that build on the scientific evidence presented in the ISA, se assessments, including  the proposed
25   scope and methods of the analyses, were presented in two planning documents titled, Ozone
26   National Ambient Air Quality Standards: Scope and Methods Plan for Health Risk and
27   Exposure Assessment and Ozone National Ambient Air Quality Standards: Scope and Methods
28   Plan for  Welfare Risk and Exposure Assessment (henceforth, Scope and Methods Plans).8 These
29   planning documents outlined the scope and approaches that staff planned to  use in conducting
30   quantitative assessments as well as key issues that would be addressed as part of the assessments.
      6 See http ://vosemite .epa. gov/sab/sabproduct. nsfAVebProj ectsbvTopicC AS AC! OpenView for more information on
      CASAC activities related to the current O3 NAAQS review.
      7 EPA 452/R-l 1-006; April 2011; Available:
      http://www.epa.gov/ttn/naaqs/standards/ozone/data/2011 04 OzoneIRP.pdf
      8 EPA-452/P-11-001 and -002; April 2011; Available:
      http://www.epa.gOv/ttn/naaqs/standards/ozone/s o3 2008_pd.html
                                               1-8

-------
 1    The documents were released for public comment in April 2011, and were the subject of a
 2    consultation with the CASAC on May 19-20, 2011 (76 FR 23809; April 28, 2011).  In designing
 3    and conducting the initial health risk and visibility impact assessments, we considered CASAC
 4    comments (Samet 2011) on the Scope and Methods Plans as well as public comments.  In May
 5    2012, a memo titled, Updates to information presented in the Scope and Methods Plans for the
 6    Ozone NAAQS Health and Welfare Risk and Exposure Assessments., was made available that
 7    described changes to elements of the scope and methods plans and provided a brief explanation
 8    of each change and the reason for it.
 9    On July 16, 2012, the EPA made available for CASAC review and public comment two draft
10    assessment documents titled, Health Risk and Exposure Assessment for Ozone, First External
11    Review Draft and Welfare Risk and Exposure Assessment for Ozone, First External Review Draft
12    (77 FR 42495, July  19, 3023).  These two draft assessment documents describe the quantitative
13    analyses the EPA is conducting as part of the review of Os NAAQS. Along with the third draft
14    ISA and this PA, these documents will be reviewed at a CASAC meeting in September 2012.

15    1.3     OVERVIEW OF O3 MONITORING AND AIR QUALITY

16           This section provides overviews of the ambient monitoring network for Os (section
17    1.3.1); Os precursor emissions and atmospheric chemistry (section 1.3.2); of ambient
18    concentrations (section 1.3.3); and the available evidence and information related to background
19    63 (section 1.3.4).  These issues are also discussed in detail in the chapter 3 of the ISA (US EPA,
20    2012a).
21    1.3.1   O3 Monitoring Network
22           To monitor compliance with the NAAQS, state and local monitoring agencies operate O3
23    monitoring sites at various locations, depending on the size of the area and typical peak Os
24    concentrations (US EPA, 2012a, sections 3.5.6.1, 3.7.4). In 2010, there were 1,250 State and
25    Local Os monitors reporting concentrations to EPA (US EPA, Figures 3-21 and 3-22).  The
26    minimum number of Os monitors required in a Metropolitan Statistical Area (MSA) ranges from
27    zero, for areas with a population under 350,000 and with no recent history of an 63  design value
28    greater than 85% of the level of the NAAQS, to four, for areas with  a population greater than 10
29    million and an 63 design value greater than 85% of the NAAQS.9 For areas with required 63
30    monitors,  at least one site must be designed to record the maximum  concentration for that
31    particular metropolitan area. The spatial scales for Os sites are neighborhood, urban, and
      9The current monitor and probe siting requirements have an urban focus and do not address siting in non-urban, rural
      areas. States may operate O3 monitors in non-urban or rural areas to meet other objectives (e.g., support for research
      studies of atmospheric chemistry or ecosystem impacts).
                                                1-9

-------
 1    regional.10  Since 63 concentrations decrease significantly in the colder parts of the year in many
 2    areas, Os is required to be monitored only during the "ozone season," which varies by state (US
 3    EPA, 2012a, section 3.5.6 and Figure 3-20).11
 4    1.3.2.  Emissions and Atmospheric Chemistry
 5           Os is formed by photochemical reactions of precursor gases and is not directly emitted
 6    from specific sources.  In the stratosphere, ozone occurs naturally and provides protection
 7    against harmful solar ultraviolet radiation.  In the troposphere, near ground level, O3 forms
 8    through atmospheric reactions involving two main classes of precursor pollutants: volatile
 9    organic compounds (VOCs) and nitrogen oxides (NOx). Carbon monoxide (CO) and methane
10    (CH/t) are also important for Os formation in some areas (US EPA, 2012a, section 3.2.2).
11           Emissions of O3 precursor compounds can be divided into anthropogenic and natural
12    source categories, with natural sources further divided into biogenic emissions (from vegetation,
13    microbes, and animals) and abiotic emissions (from biomass burning, lightning, and geogenic
14    sources). Anthropogenic sources, including mobile sources and power plants, account for the
15    majority of NOx and CO emissions. Anthropogenic sources are also important for VOC
16    emissions, though in some locations and at certain times of the year (e.g., southern states during
17    summer) the majority of VOC emissions come from vegetation (US EPA, 2012a, section 3.2.1).
18    In practice, the distinction between natural and anthropogenic sources is often unclear, as human
19    activities directly or indirectly affect emissions from what would have been considered natural
20    sources during the preindustrial era.  Thus, emissions from plants, animals, and wildfires could
21    be considered either natural or anthropogenic, depending on whether emissions result from
22    agricultural practices, forest management practices,  lightning strikes, or other types of events.
23    (US EPA, 2012a, sections 3.2 and 3.7.1).
24           Rather than varying directly with emissions of its precursors, Os changes in a nonlinear
25    fashion with the concentrations of its precursors. NOX emissions lead to both the formation and
26    destruction of O3, depending on the local quantities  of NOX, VOC, and radicals. In areas
27    dominated by fresh  emissions of NOX, these radicals are removed, which lowers the Os formation
28    rate.  In addition, the scavenging of Os by reaction with NO is called "titration," and is often
29    found in downtown metropolitan areas, especially near busy streets and roads, and in power plant
30    plumes. This short-lived titration results in local valleys in which O?, concentrations are low
      Neighborhood scale represents concentrations within some extended area of the city that has relatively uniform
      land use with dimensions in the 0.5-4.0 km range. Urban scale represents concentrations within an area of city-like
      dimensions, on the order of 4-50 km. Regional scale usually defines a rural area of reasonably homogeneous
      geography without large sources, and extends from tens to hundreds of kilometers.
      11 Some States and Territories operate O3 monitors year-round, including Arizona, California, Hawaii, Louisiana,
      Nevada, New Mexico, Puerto Rico, Texas, American Samoa, Guam and the Virgin Islands.
                                                 1-10

-------
 1    compared to surrounding areas, but produces NC>2 that contributes to 63 formation later and
 2    further downwind. Consequently, Os response to reductions in NOX emissions is complex and
 3    may include ozone decreases at some times and locations and, in others, increases of ozone to fill
 4    in the local valleys of low ozone. In areas with low NOX concentrations, such as those found in
 5    remote continental areas to rural and suburban areas downwind of urban centers, Os production
 6    typically varies directly with NOX concentrations (e.g. increases with increasing NOX emissions).
 7          The formation of Os from precursor emissions is also affected by the intensity and
 8    spectral distribution of sunlight and atmospheric mixing. Major episodes of high ground level 63
 9    concentrations in the eastern United States are associated with slow moving high pressure
10    systems.  High pressure systems during the warmer seasons are associated with the sinking of
11    air, resulting in warm, generally cloudless skies, with light winds. The sinking of air results in
12    the development of stable conditions near the surface which inhibit or reduce the vertical mixing
13    of Os precursors. The combination of inhibited vertical mixing and light winds minimizes the
14    dispersal  of emitted pollutants emitted in urban areas, allowing their concentrations to build up.
15    In addition, in some parts of the United States (e.g., in Los Angeles), mountain barriers limit
16    mixing and result in a higher frequency and duration of days with high 63 concentrations.
17    Photochemical activity involving precursors is enhanced during warmer seasons because of
18    higher temperatures and the availability of sunlight (US EPA, 2012a, section 3.2).
19          Ozone concentrations in a region are affected both by local formation and by transport of
20    63 and its precursors from surrounding areas. Ozone transport occurs on many spatial scales
21    including local transport between cities, regional transport  over large regions of the U.S. and
22    international/long-range transport.  In addition, Os is transfered into the troposphere from the
23    stratosphere, which is rich in 63, through stratosphere-troposphere exchange (STE). These
24    inversions or "foldings" usually occur behind cold fronts, bringing stratospheric air with them
25    (U.S. EPA, 2012, section 3.4.1.1).
26    1.3.3  Air Quality Concentrations
27          Because Os is a secondary pollutant formed in the atmosphere from precursor emissions,
28    concentrations are generally more regionally homogeneous than concentrations of primary
29    pollutants emitted directly from stationary and mobile sources (US EPA, 2012a, section 3.6.2.1).
30    However, variation in local emissions characteristics, meteorological conditions, and topography
31    can result in daily and seasonal temporal variability in ambient 63 concentrations, as well as
32    local and national-scale spatial variability.
33          Temporal variation in ambient Os concentrations results largely from daily and seasonal
34    patterns in temperature, sunlight, precursor emissions, and  meteorological conditions (US EPA,

                                                1-11

-------
 1    2012a, section 3.7.5).  On average, ambient Os concentrations follow well-recognized daily and
 2    seasonal patterns, particularly in urban areas. Specifically, daily maximum Os concentrations in
 3    urban areas tend to occur in mid-afternoon, with more pronounced peaks in the warm months of
 4    the O3 season than in the colder months (US EPA, 2012a, Figures 3-54, 3-156 to 3-157).  Rural
 5    sites also followed this general  pattern, though it is less pronounced in colder months (US EPA,
 6    2012a, Figure 3-55). With regard to seasonal variability, median maximum daily average 8-hour
 7    (MDA8)  Os concentrations in U.S. cities in 2007-2009 were approximately 47 ppb, with typical
 8    ranges between 35 to 60 ppb and the highest MDA8 concentrations above 100 ppb in several
 9    U.S. cities.
10           In addition to temporal variability, there is also considerable spatial variability in ambient
11    O3 concentrations within cities  and across different cities in the United States. With regard to
12    spatial variability within a city, local emissions characteristics, geography, and topography can
13    have important impacts.  For example, fresh NO emissions from motor vehicles titrate Os present
14    in the urban background air, resulting in an Os gradient around roadways with Os concentrations
15    increasing as distance from the road increases (US EPA, 2012a, section 3.6.2.1). In comparing
16    urban areas, the ISA notes that  measured Os concentrations are relatively uniform and well-
17    correlated across some cities (e.g., Atlanta) while they are more variable in others (e.g., Los
18    Angeles) (US EPA, 2012a, section 3.6.2.1 and Figures 3-28 to 3-36).  In addition to differences
19    in local emissions characteristics, such differences in the uniformity of ambient Os
20    concentrations across urban areas can also result from differences in local geography and
21    topography (US EPA,  2012a, section 3.6.2.1).
22           With regard to spatial variability across cities, when the ISA evaluated the distributions
23    of 8-hour Os concentrations for the years 2007 to 2009 in 20 cities, the highest concentrations
24    were reported in Los Angeles, with high concentrations also reported in several eastern and
25    southern  cities.  The maximum recorded MDA8 was 137 ppb in Los Angeles, and was near or
26    above 120 ppb in Atlanta, Baltimore, Dallas, New York City, Philadelphia, and St. Louis (US
27    EPA, 2012a, Table 3-10). The pattern was similar for the 98th percentile of the distribution of
28    MDA8 concentrations12, with Los Angeles recording the highest 98th percentile concentration
29    (91 ppb) and many eastern and southern cities reporting 98th percentile concentrations near or
30    above 75 ppb.  In contrast, somewhat lower 98th percentile Os concentrations were recorded in
31    cities in the western United States outside of California (US EPA, 2012a, Table 3-10).
32           Although rural monitoring sites tend to be less directly affected by anthropogenic
33    pollution sources than urban sites, rural sites can be affected by transport of Os or Os precursors
      12 Table 3-10 in the ISA analyzes the warm season. Therefore, the 98th percentile values would be an approximation
      of the 4th highest value.
                                                1-12

-------
 1    from upwind urban areas and by local anthropogenic sources such as motor vehicles, power
 2    generation, biomass combustion, or oil and gas operations (US EPA, 2012a, section 3.6.2.2).  In
 3    addition, O?,  tends to persist longer in rural than in urban areas due to lower rates of chemical
 4    scavenging in non-urban environments. At higher elevations, increased 63 concentrations can
 5    also result from stratospheric intrusions (US EPA, 2012a, sections 3.4, 3.6.2.2). As a result, Os
 6    concentrations measured in some rural sites can be higher than those measured in nearby urban
 7    areas (US EPA, 2012a, section 3.6.2.2) and the ISA concludes that cumulative exposures for
 8    humans and  vegetation in rural areas can be substantial, and often higher than cumulative
 9    exposures in urban areas (US EPA, 2012a, section 3.7.5).
10    1.3.4   Background O3
11           As discussed above, and in more detail in the ISA (US EPA, 2012a, Chapter 3), ambient
12    concentrations of Os in a given location can be influenced by emissions from both anthropogenic
13    and natural sources, by long-range transport from within and outside the United States, and by
14    intermixing of stratospheric and tropospheric air masses.  In the last review of the 63 NAAQS,
15    EPA distinguished between ambient O3 that could be controlled through U.S. regulations or
16    through international agreements with neighboring countries and ambient Os not generally
17    controllable  in this manner (US EPA, 2007, section 2.7).  This distinction was judged appropriate
18    because it had the effect of focusing policy considerations on health risks that would be
19    controllable  through U.S. regulations and/or policies.  To facilitate such a distinction, EPA
20    defined policy  relevant background (PRB), referred to as North American Background (NAB) in
21    the current draft ISA (US EPA, 2012a, section 3.4) and in this draft PA, as the distribution of O3
22    concentrations that would be observed in the U.S. in the absence of North American  (i.e., U.S.,
23    Canada, and Mexico) anthropogenic emissions. The primary implication of this distinction was
24    that Os-related health risks were characterized for ambient Os concentrations above PRB (NAB
25    in current review).13 In this section, we discuss sources and contributions of background 63, as
26    well as estimated concentrations of background O3 across the U.S., with a focus on how
27    background  Os varies spatially and temporally across the U.S. and with respect to measured or
28    simulated total 63 concentrations (US EPA, 2012a,  section 3.4).
29           In this first draft PA, we discuss three definitions of background 63 concentrations:  (1)
30    NAB, which is simulated O3 concentrations that would exist in the absence of anthropogenic
      13Inthis review, the first draft RE A (US EPA, 2012b) focuses on estimation of risks down to zero concentrations of
      O3 and down to lowest measured levels (LML) of O3, as reflected in the epidemiology studies used in the REA. This
      is in agreement with CAS AC members' recommendation that EPA move away from using PRB in calculating risks
      (Henderson, 2007). In simulating air quality that just meets the current O3 NAAQS, the first draft REA uses
      modeled U.S. background concentrations as lowest values for the rollback (i.e., O3 concentrations are not rolled
      back below U.S. background concentrations) (chapter 3, below).
                                                 1-13

-------
 1    emissions from the U.S., Canada and Mexico; (2) U.S. background (USB), which is simulated 63
 2    concentrations that would exist in the absence of anthropogenic emissions from the U.S.; and (3)
 3    natural background (NB), which is simulated O?, concentrations in the absence of all
 4    anthropogenic emissions globally. All of these definitions include contributions from natural
 5    sources including the STE of Os, Os resulting from photochemical reactions of emissions from
 6    natural sources (e.g., wildfires, lightning, soil, biogenic), and global methane emissions, although
 7    approximately 60% of global methane emissions are anthropogenic (Olivier et al, 2005). In
 8    addition, both NAB and USB include international transport of Os and Os precursor emissions
 9    from outside of North America into the U.S.14
10           While some of these sources contribute to background 63 in a more consistent manner,
11    with limited day-to-day variability (e.g. biogenic and soil emissions), other sources contribute to
12    background Os more episodically (e.g. stratospheric intrusions, international transport events,
13    wildfires). These episodic events usually occur in relation to a specific event, such as a  strong
14    cold front or a wildfire, and occur more often in specific geographical locations, such as at high
15    elevations and in wildfire prone areas during the local dry season. In addition, these episodic
16    sources of background have been found to be the primary drivers of high background
17    concentrations (US EPA, 2012a, section 3.7.3).  It should also be noted that EPA has policies for
18    treatment of air quality monitoring data affected by these types of events. For example, EPA's
19    2007 Treatment of Data Influenced by Exceptional Events Rule allows  exclusion of air quality
20    monitoring data from regulatory determinations related to exceedances  or violations of the
21    NAAQS and to avoid designating an area as nonattainment if a State adequately demonstrates
22    that an exceptional event have caused an exceedance  or violation of a NAAQS.15 In addition,
23    Section 179B of the CAA also provides for  treatment of air quality data from international
24    transport when emissions emanating from outside of the United States have caused an
25    exceedance or violation of a NAAQS.16
26           Historically, two approaches have been used to estimate background O3 concentrations.
27    In the 1996 Os AQCD, and in earlier reviews, measurements from remote monitoring sites were
28    used to estimate background concentrations. However, this approach has the disadvantage of not
      14 USB also includes the transport of O3 and O3 precursor emissions from Canada and Mexico into the U.S.
      15 EPA's 2007 Treatment of Data Influenced by Exceptional Events Rule section 319(b)(3)(B) and section 107(d)(3)
      of the CAA: Exceptional events are unusual or naturally occurring events that can affect air quality but are not
      reasonably controllable using techniques that tribal, state, or local air agencies may implement
      (http://www.epa.gov/ttn/analysis/exevents.htm). Additional guidance related to this rule is currently under
      development.
      16 Section 179B states: "Notwithstanding any other provision of law, any State that establishes to the satisfaction of
      the Administrator that, with respect to an ozone nonattainment area in such State, such State would have attained the
      national ambient air quality standard for ozone by the applicable attainment date, but for emissions emanating from
      outside of the United States, shall not be subject to the provisions of section 181(a)(2) or (5) or section 185."
                                                  1-14

-------
 1    allowing unambiguous attribution of ambient Os concentrations to background sources. As
 2    noted in the 2006 O3 AQCD (U.S. EPA, 2006, section 3.9), given long-range transport of O3 and
 3    Os precursors from anthropogenic source regions, estimates of background concentrations in the
 4    U.S. cannot be obtained directly from measurements of ambient Os, even measurements obtained
 5    at relatively remote monitoring sites.  In support of this conclusion, recent analyses by Parrish et
 6    al. (2009) indicated that measured Os concentrations at Trinidad Head, CA, a site that has
 7    historically been characterized as reflecting NAB, only reflected background concentrations
 8    about 30% of the time during spring, with local anthropogenic source influences present on most
 9    days (US EPA, 2012a, section 3.4.2).
10          Since background Os concentrations as defined above are a construct that cannot be
11    directly measured, the 2006 AQCD adopted the use of chemical transport models (CTMs) to
12    estimate NAB (referred to as PRB in the 2006 AQCD).  An advantage of using CTMs is that
13    they are able to provide a broad range of 63 concentrations, spatially and temporally, from
14    various different environments. Another advantage of using these models to estimate background
15    Os concentrations is that specific emissions sources can be turned on or off in the model,
16    providing insight into contributions to ambient Os concentrations in the U.S. from natural
17    sources and international transport, when compared to measured or simulated base case
18    concentrations of 63 concentrations photochemically produced from all emissions sources.
19    However, it should be noted that modeled concentrations of O3 background are an estimate of O3
20    concentrations in the absence of specific anthropogenic emissions, and because of the nonlinear
21    nature of 63 chemistry, are only an approximation of how much of the 63 measured or simulated
22    in a given area is due to background contributions. In this way, there are important limitations to
23    consider when interpreting the modeling results.
24          Recent modeling efforts from Zhang et al. (2011) and Emery et al. (2012) have sought to
25    improve the spatial and temporal resolution of background  estimates and to better characterize
26    important sources of background O3 such as fires and international transport. These applications
27    have produced the latest estimates for background Os concentrations documented in the recent
28    literature, and the results of these modeling efforts are discussed in more detail in the ISA (US
29    EPA, 2012a, section 3.4) and below. The analyses provided by Zhang et al. (2011) (hereafter
30    referred to as GEOS-Chem) utilized the GEOS-Chem model at a grid spacing of 0.5° x 0.667°
31    (-50 km) over North America for modeling NAB, USB, NA, and base case Os. The analyses
32    provided in Emery et al. (2012)  (hereafter referred to as CAMx) employed the CAMx model at
33    an even finer grid spacing of 12km x 12km to model NAB and base case Os, with boundary
34    conditions being provided by a GEOS-Chem model run at a grid spacing of 2° x 2.5°. The most
35    readily discernible differences in the two modeling applications are in the model grid spacing

                                               1-15

-------
 1    and the treatment of wildfires. The ISA discusses in detail the model performance of the GEOS-
 2    Chem and CAMx base case model runs, as well as the estimates of background Os (US EPA,
 3    2012a, section 3.4). In addition, the information from the 2006 GEOS-Chem and CAMx model
 4    runs are analyzed in Henderson et al. (2012) to provide additional information about variations in
 5    the Os background concentrations spatially, temporally, and with respect to the distribution of
 6    simulated base case 63.
 7          In the remainder of this section, we discuss the estimated average background Os values
 8    across the U.S. (section 1.3.4.1), compare how the estimated MDA8 background 63 contribution
 9    varies with simulated Os concentrations (section 1.3.4.2) based on the GEOS-Chem and CAMx
10    results for 2006,  and provide a summary of this information (section 1.3.4.3).
11          1.3.4.1 Average Background Os in the U.S.
12          The ISA finds that both GEOS-Chem and CAMx were capable of simulating measured
13    seasonal or monthly mean MDA8 Os within a few parts per billion throughout the U.S., with less
14    acceptable model performance in California (US EPA, 2012a, section 3.4.3.2).  Both modeling
15    applications showed that background concentrations vary spatially and temporally, and that
16    simulated mean background concentrations are highest in the Intermountain West (i.e. at high
17    altitudes) in spring and lowest in the Northeast during the summer. Table 3-1 in the ISA (US
18    EPA, 2012a) provides the seasonal mean modeled MDA8 NAB Os concentrations for five
19    regions of the U.S. for 39 CASTNET site  locations.  Spring mean MDA8 Oj concentrations were
20    simulated by GEOS-Chem and CAMx to be 38-42 ppb in the West (including California) and
21    30-33 ppb in the  rest of the country (i.e. North Central, Northeast, and Southeast).  Summer
22    mean MDA8 O3  concentrations were simulated by the two modeling approaches to be 37-42 ppb
23    in the West and 27-33 ppb in the rest of the country (US EPA, 2012a, Table 3-1).
24          GEOS-Chem estimated March-August 2006 mean MDA8 NAB and NB concentrations at
25    CASTNET site locations to be approximately 29 ppb and 18 ppb, respectively,  at low  elevations
26    (<1,500 m) and approximately 40 ppb and 27 ppb, at high elevations (>1,500 m).  These data
27    suggest that intercontinental pollution contributes approximately 9 ppb at low-elevation sites and
28    13 ppb at sites in the Intercontinental West to seasonal average MDA8 Os concentrations. These
29    results reflect the increased importance of background sources such as stratospheric intrusions
30    and intercontinental transport with altitude.
31          GEOS-Chem results for seasonal mean MDA8 USB and NAB concentrations suggest
32    that USB concentrations are on average 1-3 ppb higher than NAB background,  reflecting the
33    influence of anthropogenic sources in Canada and Mexico. Very little variability was found  in
34    these concentrations, except in areas in the U.S. that bordered Canada and Mexico, where

                                               1-16

-------
 1    international transport from these two countries plays a greater role in contributing to 63
 2    background concentrations (US EPA, 2012a, Figure 3-9). These results were similar to those
 3    reported by Wang et al. (2009).
 4          Biomass burning has been found to be a significant contributor to background Os,
 5    particularly in the West. For a small number of western locations, generally not near major
 6    urban centers, NAB concentrations were estimated by Emery et al. (2012) to be considerably
 7    higher than the NAB concentrations estimated for the majority of the region. When Emery et al.
 8    (2012) removed biomass burning from the model runs, these high concentrations were
 9    dramatically reduced, suggesting that the highest estimates of NAB concentrations in the West
10    are driven largely by wildfires. This result is consistent with other research indicating that
11    biomass burning, including wildfires and the intentional burning of vegetation, can make
12    important contributions to background Os concentrations. Biomass burning exhibits strong
13    seasonality, with most fires occurring during the local dry season (US EPA,  2012a, section
14    3.4.1.2). Several studies have reported effects of U.S. or international wildfires on Os
15    concentrations in the U.S.  (Wang et al., 2006; Generoso et al., 2007; Jaffe et al., 2008; Mathur et
16    al., 2008).  Specifically, Jaffe et al.  (2008) reported a strong correlation between the area burned
17    in the western U.S. and the summer mean Os concentrations measured at various national park
18    and CASTNET sites in the surrounding regions.  The authors estimated that,  on average, 63
19    concentrations increased by 3.5 ppb during mean fire years and 8.8 ppb during maximum fire
20    years (US EPA, 2012a, section 3.4.1.2).
21          1.3.4.2  Distributions of Background O3 in the U.S.
22          In addition to discussing average background concentrations, it is also important to
23    understand how estimated background concentrations vary with measured or simulated Os
24    concentrations, as well as to understand the contributing sources. The remainder of this section
25    focuses on the results detailed by Henderson et al. (2012), which compares distributions of
26    background O?, concentrations with simulated total Os, as adapted from the GEOS-Chem and
27    CAMx model simulations performed by Zhang et al. (2011) and Emery et al. (2012). In this way,
28    we provide a more robust analysis by focusing on better understanding the variability of
29    concentrations across a region for many days rather than just one day of predicted 4th highest
30    MDA8 Os concentrations in a grid cell.
31          The analysis presented by Henderson et al. (2012) divides the country into 5 regions:
32    Northeast, Southeast, North Central, West (except for California), and California. It presents
33    NAB contributions at AQS, CASTNET, and  other special study monitoring  sites by season to
34    concentrations simulated by  both GEOS-Chem and CAMx. The analysis also uses Combined
35    Statistical Areas (CSA) to distinguish between "urban" and "non-urban"  sites, and presents
                                                1-17

-------
 1    results for CSA versus non-CSA areas in each region. In addition, results are presented for the
 2    location of high elevation versus low elevation monitoring sites in the West. The following
 3    discussion focuses on these regional results. Overall, NAB concentrations are found to be lowest
 4    in the Northeast, and highest in the West, particularly at high elevation sites.
 5          Little or no differences were found in estimated NAB concentrations for the location of
 6    "urban" and "non-urban" monitoring sites, and so the results discussed below are for the location
 7    all monitoring sites within each region. In addition, because Os concentrations are usually
 8    highest in the spring and summer, we focus our discussion on these two seasons. The analysis
 9    presented by Henderson et al. (2012) compares the contribution of modeled NAB to modeled
10    base case 63 and the results are broken into 5 ppb increment bins from >= 20 ppb to >=135 ppb,
11    based on MDA8 concentration values.
12          The median, 75th , and 95th percentile MDA8 values for the five regions of the U.S.,
13    discussed below and shown in Table 1-2, are 2006 MDA8 NAB concentrations estimated by
14    GEOS-Chem and CAMx (as presented in Henderson et al. (2012)) when simulated base case 63
15    MDA8 values were greater than 55 ppb. Median values provide a more robust indicator than
16    extreme values of the distribution for the purpose of examining and comparing the regional and
17    seasonal variability in NAB.  The NAB predictions toward the higher end of the distribution (e.g.
18    75th and 95th percentiles) are more reflective of infrequent or atypical events. Due to the overall
19    uncertainties and assumptions in the inputs to the two modeling systems, the higher percentile
20    NAB predictions are likely to have a greater degree of uncertainty than the median values.
21          As shown in Table 1-2,  similar concentration values for NAB were also found in North
22    Central and  Southeast. On days when simulated base case spring-time O3 concentrations were
23    above 55 ppb, median MDA8 NAB concentrations for GEOS-Chem and CAMx were 28 ppb and
24    33 ppb for North Central and 30 ppb and 34 ppb for the Southeast. The 75th percentile MDA8
25    NAB concentrations were 33 ppb and 37 ppb for North Central and 34 ppb and 38 ppb for the
26    Southeast, as estimated by GEOS-Chem and CAMx.  The 95th percentile MDA8 NAB
27    concentrations were estimated to be 40 ppb and 42 ppb for North Central, and 41 ppb and 45 ppb
28    for the Southeast. The NAB concentrations were lower in the summer. Summertime median
29    MDA8 NAB concentrations for the two models were 24 ppb and 33 ppb for North Central, and
30    29 ppb and 31 ppb for the Southeast. The 75th percentile NAB concentrations were 28 ppb and
31    36 ppb for North Central and 36 ppb and 34 ppb for the Southeast, as predicted by GEOS-Chem
32    and CAMx.  The 95th percentile NAB concentrations were estimated to be 39 ppb and 41 ppb for
33    North Central and 44 ppb and 41 ppb for the Southeast.
34          Estimated NAB background concentrations were lowest in the Northeast region. On days
35    when predicted spring-time Os concentrations were above 55 ppb, median MDA8 NAB
                                               1-18

-------
 1    concentrations in the spring in the Northeast were 23 ppb and 31 ppb for the two models, with
 2    estimated 75th percentile NAB concentrations of 26 ppb and 34 ppb. In the spring, 95th percentile
 3    NAB concentrations were estimated to be 33 ppb and 38 ppb. Os background concentrations
 4    were lower in the summer. Summertime median MDA8 values estimated by GEOS-Chem and
 5    CAMx were 18 ppb and 29 ppb, and 75th percentile NAB concentrations were 23 ppb and 32
 6    ppb, when simulated base case 63 concentrations were above 55 ppb. Ninety-fifth percentile
 7    NAB concentrations were estimated to be 34 ppb and 36 ppb
 8          Compared to the eastern U.S., NAB 63 concentrations in the West are more variable
 9    across locations. As compared to the rest of the western U.S., relatively lower background Os
10    concentrations have been estimated for California, where total ambient 63 concentrations are
11    among the highest in the nation. In addition, higher background concentrations have also been
12    found at higher elevation sites, compared to lower elevation sites (US EPA, 2012a, Figure 3-12).
13    In light of this spatial variability, we discuss separately background 63 in California. Since most
14    of the high  elevation sites are in the West, we also discuss NAB background concentrations at
15    monitoring sites in the West at high (>1500 m) versus low (<1500 m) elevation sites.
16          Spring and summer median MDA8 NAB concentrations in California were similar, when
17    simulated base case 63 concentrations were above 55 ppb. Median MDA8 NAB concentrations
18    for the two models were 34 ppb and 35 ppb in the spring, and 30 ppb and 36 ppb in the summer.
19    Seventy-fifth percentile MDA8 NAB concentrations estimated to be 40 ppb in the spring, and 36
20    ppb and 40 ppb in the summer, while 95th percentile MDA8 NAB concentrations were estimated
21    to be 48 ppb in the spring, and 45 ppb and 47 ppb  in the summer.
                                              1-19

-------
 1   Table 1-2. Modeled median and 75th percentile 2006 MDA8 NAB concentrations at
 2   monitoring site locations (as presented in Henderson et al. (2012) and adapted from
 3   GEOS-Chem and CAMx modeling applications) when simulated base case Os MDA8
 4   values were greater than 55 ppb.
Region
California
North Central
Northeast
Southeast
West
All sites
Low-
elevation
sites
High-
elevation
sites
Spring (GEOS-Chem/CAMx)
Median
(ppb)
34/35
28/33
23/31
30/34
44/43
43/41
45/48
75th
percentile
(ppb)
40/40
33/37
26/34
34/38
47/48
46/44
48/52
95th
percentil
e (ppb)
48/48
40/42
33/38
41/45
52/55
51/51
53/57
Summer (GEOS-Chem/CAMx)
Median
(ppb)
30/36
24/33
18/29
29/31
41/41
40/39
42/43
75th
percentile
(ppb)
36/40
28/36
23/32
36/34
46/46
45/44
47/48
95th
percentile
(ppb)
45/47
39/41
34/36
44/41
54/52
52/52
54/53
 6          For the rest of the West, on days when spring-time 63 concentrations were estimated to
 7   be above 55 ppb, median MDA8 NAB Os concentrations for the two models were 44 ppb and 43
 8   ppb, with 75th percentile MDA8 NAB concentrations of 47 ppb and 48 ppb and 95th percentile
 9   MDA8 NAB concentrations of 52 ppb and 55 ppb. Lower values of background were estimated
10   in the summer, with median MDA8 NAB concentrations of 41 ppb. Seventy-fifth percentile
11   NAB concentrations were estimated to be 46 ppb for both models, and 95th percentile NAB
12   concentrations were estimated to be 54 ppb and 52 ppb.
13          Higher NAB concentrations were estimated at high-elevation site locations, compared to
14   low-elevation site locations, in the West. On days when spring-time Os concentrations were
15   estimated to be above 55 ppb, median MDA8 NAB Os concentrations estimated by the two
16   models at high-elevation site locations were 45 ppb and 48 ppb, compared to low-elevation site
17   locations with median MDA8 NAB concentrations of 43 ppb  and 41 ppb. The 75th percentile
18   MDA8 NAB concentrations at high-elevation site locations were estimated by the two models to
                                              1-20

-------
 1    be 48 ppb and 52 ppb, compared to low elevation sites with 75th percentile MDA8 NAB
 2    concentrations of 46 ppb and 44 ppb, when simulated base case O3 concentrations were above 55
 3    ppb. The 95th percentile MDA8 NAB concentrations at high-elevation site locations were
 4    estimated to be 53 ppb and 57 ppb, compared to low elevation sites with 95th percentile MDA8
 5    NAB concentrations of 51 ppb. A similar pattern, with somewhat lower concentrations, was also
 6    estimated in the summertime.
 7          In general, for all areas of the country, background concentrations increase with
 8    increasing total O3 up to MDA8 values of approximately 55-60 ppb. Above MDA8
 9    concentrations of 55-60 ppb, the NAB O3 concentrations do not vary significantly across the 5
10    ppb bins between 55-135 ppb (Henderson et al, 2012). This result was also illustrated in
11    decreasing percent contribution of NAB to O3 concentrations with increasing O3 concentrations
12    above MDA8 concentrations of approximately 55-60 ppb (Henderson et al, 2012). In addition,
13    the median values of NAB are not dramatically higher than the typical seasonal average
14    background concentrations for the region. These results suggest that anthropogenic sources
15    within the U.S. are largely responsible for the highest MDA8 O3 concentrations.
16          It is important to understand source contributions to the background concentrations
17    discussed above and how they relate to U.S. EPA polices for data treatment. The GEOS-Chem
18    and CAMx results from the applications used for this analysis demonstrate that NAB O3
19    concentrations are quite variable  across U.S., with lower background concentrations in the East
20    and higher concentrations in the West. As discussed above, and in the ISA (US EPA, 2012a,
21    section 3.4.3,2), high values of background O3 are believed to be mainly due to some
22    combination of wildfires, stratospheric intrusions, and Eurasian emissions, with stronger
23    influences of Canadian and Mexico emissions also affecting areas bordering those countries. For
24    example, in the western U.S., fire emissions have been found to be a significant contributor to
25    background O3, particularly in the West.  Emery et al. (2012) found that removing fire emissions
26    in the West resulted in reductions of MDA8 NAB O3 concentrations of  10-50 ppb, with smaller
27    reductions elsewhere. Emery et al. (2012) found the highest values of background in Idaho,
28    Oregon,  Washington, and in the upper northwestern corner of California from wildfires. This
29    result is consistent with  other research indicating that biomass burning, including wildfires and
30    the intentional burning of vegetation, can make important contributions  to background O3
31    concentrations (Wang et al., 2006; Generoso  et al., 2007; Jaffe et al., 2008; Mathur et al., 2008).
32          In addition to wildfires and as discussed above, STE of O3 can contribute to surface level
33    O3 concentrations, especially at the high elevations found in the mountainous regions of the
34    western U.S. These events usually contribute episodically to high background O3. Lin et al.
35    (2012) estimated that contributions from strong stratospheric intrusions  to surface O3 could range

                                               1-21

-------
 1    up to approximately 55 ppb in the western U.S. While it is challenging to accurately estimate the
 2    exact contribution to surface-level Os from relatively small direct or indirect (i.e., resulting from
 3    shallow intrusions into the mid and upper troposphere that are then mixed downward into the
 4    planetary boundary layer) stratospheric intrusion events, some strong events have been identified
 5    and their contributions to surface level Os concentrations have been characterized (Langford et
 6    al., 2009, Hocking et al., 2007, US EPA, 2006).
 7           1.3.4.3 Summary
 8           While numerous large urban areas in the U.S. experience high ambient Os concentrations
 9    during the warm season, recent modeling efforts indicate that anthropogenic emission sources
10    are the  dominant contributor to these ambient concentrations (US EPA, 2012a, section 3.4.3  and
11    Henderson et al., 2012). In the Southeast, Northeast, and North Central, background
12    concentrations were lower in the summer (than in the spring) when measured Os concentrations
13    are usually the highest and the 4th highest MDA8 values usually occur. In addition, the GEOS-
14    Chem and CAMx model results suggest that background concentrations on the days with the
15    highest total O3 concentrations are not dramatically higher than typical seasonal average
16    background concentrations and, therefore, that anthropogenic sources within the U.S. are largely
17    responsible for 4th highest 8-hour daily maximum Os concentrations.  In areas where background
18    63 is highest, such as the western U.S. and at higher elevation sites, the sources contributing to
19    high background concentrations have been identified as wildfires, stratospheric intrusions, and
20    intercontinental transport (US EPA, 2012a, section 3.4.3). As noted above, EPA has policies that
21    allow exclusion of air quality monitoring data affected by these types of events.

22    1.4.    GENERAL APPROACH AND ORGANIZATION OF THIS DOCUMENT

23           The final PA will identify as broad an array of options for consideration as is supportable
24    by the available scientific evidence and exposure and risk information, recognizing that the final
25    decisions on the Os NAAQS will reflect the judgments of the Administrator.  In developing that
26    range options for consideration, staffs approach is framed by the series of policy-relevant issues
27    identified in the IRP (US EPA, 201 la, Chapter 3),  and as discussed in subsequent chapters of
28    this document.  Consistent with the approaches used in previous reviews, the current approach to
29    reviewing the Os NAAQS is based most fundamentally on considering the scientific evidence
30    and the results of exposure and risk analyses to inform staff conclusions on the range of policy
31    options that could be supported.  The approaches used in previous reviews of the 63 NAAQS,
32    and the preliminary approaches used in the current review, are discussed in more detail in
33    chapters 4 (primary standard) and 7  (secondary standard), below.
                                                1-22

-------
 1          Following this introductory chapter, this first draft PA is organized into two main parts.
 2    Chapters 2 through 4 focus on the review of the Os primary NAAQS while chapters 5 through 7
 3    focus on the review of the Os secondary NAAQS.  Staffs preliminary consideration of the
 4    scientific evidence and exposure/risk information related to the primary standard are discussed in
 5    chapters 2 and 3, respectively. Staffs preliminary  conclusions on the adequacy of the current
 6    primary 63 standard are discussed in chapter 4. Staffs preliminary consideration of the
 7    scientific evidence and exposure/risk information related to the secondary standard are discussed
 8    in chapters 5 and 6, respectively. Staffs preliminary conclusions on the adequacy of the current
 9    secondary 63 standard are discussed in chapter 7.

10    1.5    REFERENCES

11    Emery, C; Jung, J; Downey,  N; Johnson, J; Jimenez, M; Yarvvood, G; Morris, R. (2012).
12          Regional and global modeling estimates of policy relevant background ozone over the
13          United States. Atmos Environ 47: 206-217.
14          http://dx.doi.0rg/10.1016/j.atmosenv.2011.ll.012.

15    Generoso, S; Bey, I; Artie, JL; Breon, FM. (2007).  A satellite- and model-based assessment of
16          the 2003 Russian fires: Impact on the arctic region. J Geophys Res 112: 5302.

17    Henderson, R. 2007. Clean Air Scientific Advisory Committee's (CASAC) Review of the
18          Agency's Final O3 Staff Paper. EPA-CASAC-07-002. March 26.

19    Henderson, B.H., Possiel, N., Akhtar, F., Simon, H.A. (2012). Regional and Seasonal  Analysis
20          of North American Background Ozone Estimates from Two Studies. Available on the
21          Internet at:  http://www.epa.gOv/ttn/naaqs/standards/ozone/s o3td.html

22    Hocking, WK; Carey-Smith, T; Tarasick, DW; Argall, PS; Strong, K; Rochon, Y;  Zawadzki,  I;
23          Taylor, PA. (2007). Detection of stratospheric ozone intrusions by windprofiler radars.
24          Nature 450: 281-284. http://dx.doi.org/10.1038/nature06312

25    Jaffe, D; Chand, D; Hafner, W; Westerling, A; Spracklen, D. (2008). Influence of fires on O-3
26          concentrations in the western US. Environ Sci Technol 42: 5885-5891.
27          http://dx.doi.org/10.1021/es800084k.

28    Langford, AO; Aikin, KC; Eubank, CS; Williams, EJ. (2009). Stratospheric contribution to high
29          surface ozone in Colorado during springtime.  Geophys Res Lett 36: L12801.
30          http://dx.doi.org/10.1029/2009gl038367

31    Mathur, R. (2008). Estimating the impact of the 2004 Alaskan forest fires on episodic particulate
32          matter pollution over the eastern United States through assimilation of satellite-derived
33          aerosol optical depths in a regional air quality model.  J Geophys Res 113: D17302.
34          http://dx.doi.org/10.1029/2007JD009767.
                                               1-23

-------
 1    Olivier, J.G.J., Van Aardenne, J.A., Dentener, F., Pagliari, V., Ganzeveld, L.N. and J.A.H.W.
 2          Peters (2005). Recent trends in global greenhouse gas emissions: regional trends 1970-
 3          2000 and spatial distribution of key sources in 2000. Env. Sc., 2 (2-3), 81-99. DOT:
 4          10.1080/15693430500400345.

 5    U.S. EPA (U.S. Environmental Protection Agency). (2006). Air quality criteria for ozone and
 6          related photochemical oxidants [EPA Report]. (EPA/600/R-05/004AF). Research
 7          Triangle Park, NC. http://cfpub.epa.gov/ncea/cfm/recordisplay.cfm?deid= 149923

 8    U.S. Environmental Protection Agency. (2012a). Integrated Science Assessment for Ozone and
 9          Related Photochemical  Oxidants: Third External Review Draft, U.S. Environmental
10          Protection Agency, Research Triangle Park, NC. EPA/600/R-10/076C

11    U.S. Environmental Protection Agency. (2012b). Health Risk and Exposure Assessment for
12          Ozone, First External Review Draft, U.S. Environmental Protection Agency, Research
13          Triangle Park, NC. EPA 452/P-12-001.

14    Wang, HQ; Jacob, DJ; Le Sager, P; Streets, DG; Park, RJ; Gilliland, AB; van Donkelaar, A.
15          (2009). Surface ozone background in the United States: Canadian and Mexican pollution
16          influences. Atmos Environ 43: 1310-1319.
17          http://dx.doi.0rg/10.1016/i.atmosenv.2008.ll.036.

18    Zhang, L; Jacob, DJ; Downey, NV; Wood, DA; Blewitt, D; Carouge,  CC; Van donkelaar, A;
19          Jones, DBA; Murray, LT; Wang, Y. (2011). Improved estimate of the policy-relevant
20          background ozone in the United States using the GEOS-Chem global model with 1/2° x
21          2/3° horizontal resolution over North America. Atmos Environ 45: 6769-6776.
22          http://dx.doi.0rg/10.1016/j.atmosenv.2011.07.054.

23
                                               1-24

-------
 1              2.     CONSIDERATION OF THE HEALTH EVIDENCE

 2          In this chapter, we pose the following overarching question:
 3       •  To what extent has scientific information become available that alters or
 4          substantiates our understanding of the health effects that occur following short-term
 5          or long-term exposures to Os, and our understanding of the Os concentrations at
 6          which such effects occur?
 7    To inform our consideration of this issue, we consider the weight-of-evidence conclusions from
 8    the ISA (section 2.1); the scientific evidence linking short-term Os exposures to morbidity and
 9    mortality (section 2.2); and the scientific evidence linking long-term Os exposures to morbidity
10    and mortality (section 2.3).  The public health implications of Os exposures are discussed in
11    section 2.4 and an integrated discussion of the evidence is provided in section 2.5.

12    2.1    WEIGHT-OF-EVIDENCE CHARACTERIZATION IN THE ISA

13          Since the conclusion of the last review, the Agency has developed a more formal
14    framework for reaching causal inferences from the body of scientific evidence.  This framework
15    provides the basis for a robust, consistent, and transparent process  for evaluating the scientific
16    evidence, including uncertainties in the evidence, and drawing conclusions and causal judgments
17    regarding air pollution-related health effects.  The causality framework and the approach to
18    characterizing the weight-of-evidence is discussed briefly below (section 2.1.1)  and is described
19    in more detail in the third draft ISA (US EPA, 2012a, Preamble). The ISA weight-of-evidence
20    conclusions for Os are summarized in section 2.1.2.
21    2.1.1  Approach to Characterizing the Weight-of-Evidence
22          Characterization of the weight-of-evidence in the third draft ISA is based on the
23    evaluation and synthesis of evidence from across scientific disciplines.  Three general types of
24    studies inform consideration of weight-of-evidence conclusions for human health effects:
25    controlled human exposure, epidemiologic, and toxicological studies (US EPA 2012a,
26    Preamble). Each of these types of studies has strengths and limitations, as discussed briefly
27    below and in more detail in the third draft ISA (US EPA, 2012a, Preamble).
28          The most direct evidence in support of a causal relationship between pollutant exposures
29    and human health effects comes from controlled human exposure studies.  These experimental
30    studies evaluate the health effects of administered exposures in human volunteers under highly
31    controlled laboratory conditions. Limitations of controlled human exposure  studies include
32    generally small sample sizes and short exposure periods; the potential for responses to be
                                                2-1

-------
 1    influenced by uncharacterized pollutant exposures in the hours and days preceding the study;1
 2    and the use of relatively healthy individuals, not representing the most sensitive individuals in
 3    the population.  Given these limitations, the ISA notes that not observing an effect in controlled
 4    human exposure studies does not necessarily mean that a causal relationship does not exist and
 5    further concludes that the effects reported in controlled human exposure studies may
 6    underestimate the response in certain at-risk populations (US EPA, 2012a, Preamble).
 7          Epidemiologic studies also provide important information on the links between adverse
 8    health effects and exposures of human populations to ambient air pollution. These studies have
 9    the advantages of allowing the evaluation of more severe health effects than can be studied in
10    controlled human exposure studies (e.g., hospital admissions, premature mortality); the
11    assessment of pollutant-associated effects across a broad population, including particularly at-
12    risk groups within that population; and providing information on real world impacts of pollution
13    exposures, which occur within the context of a complex multi-pollutant atmosphere.  Key
14    uncertainties that affect interpretation of epidemiologic studies include the extent to which a
15    particular health outcome is due to the pollutant of interest, as opposed to one or more co-
16    occurring pollutants, and the extent to which  exposure measurement error influences associations
17    (US EPA, 2012a, Preamble). Confounding by co-pollutants can result in overestimates in the
18    contribution of a given pollutant to a health effect while exposure error can make it more
19    difficult to detect pollutant-associated effects, even when such effects are present.
20          The third main type of health effects evidence, animal toxicological studies, provides
21    information on the pollutant's biological action under controlled and monitored exposure
22    circumstances.  Understanding the biological mechanisms underlying various health outcomes
23    can prove crucial in establishing or negating causality, particularly in the absence of data from
24    controlled human exposure studies.  However, given species differences, there are important
25    uncertainties associated with quantitative extrapolations of reported pollutant-induced
26    pathophysiological alterations between laboratory animals and humans (US EPA, 2012a,
27    Preamble).
28          In considering available evidence from each of these types of studies, the Os ISA draws
29    conclusions within the context of a causality framework with a five-level hierarchy.  This
30    framework  is used to classify the overall weight-of-evidence into one of the following
31    categories:  causal relationship, likely to be a causal relationship, suggestive of a causal
32    relationship, inadequate to infer a causal relationship,  and not likely to be a causal relationship
      1 Though in a number of the available O3 controlled human exposure studies, researchers have addressed this issue
      by using a randomized, crossover design with each subject serving as their own control (US EPA, 2012a, section
      6.2.1.1).
                                                 2-2

-------
 1    (US EPA 2012a, Preamble Table II). In making such judgments regarding causality, the ISA
 2    evaluates several aspects of the evidence including the consistency of effects across studies, the
 3    coherence of the evidence across different types of studies, the strength of reported associations,2
 4    and the biological plausibility of a causal relationship (US EPA, 2012a, Preamble, Table I).
 5    Confidence increases that Os exposures cause a given health effect as the number of consistently
 6    supportive studies increases, as the coherence of the evidence across different types of studies
 7    increases,  as the strength of the relationship with O3 increases, and as the support for biological
 8    plausibility increases. The ISA also evaluates evidence related to concentration-response and
 9    exposure-response relationships in order to inform conclusions on the concentrations at which
10    effects are present.  Considerations related to weight-of-evidence conclusions and concentration-
11    and exposure-response relationships are discussed in more detail in the Preamble to the third
12    draft of the O3 IS A (US EPA, 2012a).
13    2.1.2   Weight-of-Evidence Conclusions for Os
14           In the last review of the 63 NAAQS, the strongest evidence was for respiratory effects
15    following  short-term exposures. The 2006 AQCD concluded that there was clear, consistent evidence
16    of a causal  relationship between short-term exposures to OB and respiratory health effects (US EPA,
17    2006).  In the current review, in applying the updated causality framework, the ISA draws the
18    following weight-of-evidence conclusions (US EPA 2012a, section 2.5.2):
19        1.  "there is a causal relationship between short-term 63 exposure and respiratory health
20           effects" which include a spectrum of respiratory-related morbidity endpoints and
21           respiratory-related mortality;
22
23       2.  "there is likely to be a causal relationship between short-term 63 exposures and total
24           mortality";
25
26       3.  "there is likely to be a causal relationship between long-term exposure to 63 and
27           respiratory health effects" which include respiratory-related morbidity endpoints,
28           including new-onset  asthma, and respiratory-related mortality;
29
30       4.  "the evidence is suggestive of a causal relationship between long-term 63 exposures and
31           total mortality";
32
33       5.  "the overall body of evidence across disciplines is suggestive of a causal relationship for
34           both relevant short- and long-term exposures to 63 and cardiovascular effects";
35
36       6.  "the evidence from studies of short- and long-term exposure to Os is suggestive of a
37           causal relationship between 63 exposure and central nervous system effects";
      2The strength of the association refers to the magnitude of the association (i.e., the size of the effect estimate in
      epidemiologic studies) and its statistical precision.
                                                  2-3

-------
 1
 2       7.  "the evidence is suggestive of a causal relationship between long-term exposures to Os
 3           and reproductive and developmental effects"; and
 4
 5       8.  "the evidence is inadequate to determine if a causal relationship exists between ambient
 6           Os exposures and cancer."
 7
 8           The following sections provide an overview of the scientific evidence for the health
 9    endpoints linked to short-term (section 2.2) and long-term (section 2.3) O3 exposures, with a
10    particular focus on studies that evaluate health endpoints judged in the third draft ISA to be
11    caused by, or likely to be caused by, exposures to Os.

12    2.2     HEALTH EFFECTS FOLLOWING SHORT-TERM EXPOSURES TO O3

13           Given the weight-of-evidence conclusions in the ISA, staffs consideration of health
14    effects linked to short-term 63 exposures focuses on respiratory effects (section 2.2.1) and all-
15    cause mortality (section 2.2.2). Other effects linked to short-term exposures are also considered,
16    including cardiovascular effects (section 2.2.3) and central nervous system (CNS) and
17    developmental effects (section 2.2.4). In discussing the evidence in this draft PA, although we
18    consider the total body of available evidence of effects following short-term exposures to Os, we
19    focus the majority of our discussion on the studies that are most likely to inform policy decisions
20    regarding the adequacy of the current standard and potential alternative standards, as discussed in
21    chapter 4.
22    2.2.1   Respiratory Effects
23           The ISA concludes that "the clearest evidence for health effects associated with exposure
24    to Os is provided by studies of respiratory effects" (US EPA, 2012a, chapter 1, p. 1-5).
25    Collectively, there is a vast amount of evidence spanning several  decades that supports a causal
26    association between exposures to Os and a continuum of respiratory effects (US EPA, 2012a,
27    section 2.5).  The majority of this evidence is derived from studies investigating short-term
28    exposures (i.e., hours to weeks). In this section, we revisit the overarching question of this
29    chapter, as it relates to respiratory  effects following short-term Os exposures.
30           In the last review, the 2006 Os Air Quality Criteria Document (AQCD) concluded that
31    there was clear, consistent evidence of a causal relationship between short-term exposure to Os
32    and respiratory health effects (US EPA, 2006).  This causal association was  substantiated by the
33    coherence of effects observed across controlled human exposure, epidemiologic, and
34    toxicological studies indicating associations of short-term Os exposures with a range of

                                                 2-4

-------
 1    respiratory health endpoints from respiratory tract inflammation to respiratory emergency
 2    department (ED) visits and hospital admissions. Across disciplines, short-term Os exposures
 3    induced or were associated with statistically significant declines in lung function. An equally
 4    strong body of evidence from controlled human exposure and toxicological studies demonstrated
 5    Os-induced inflammatory responses, increased epithelial permeability, and airway
 6    hyperresponsiveness. Toxicological studies provided additional evidence for (Vinduced
 7    impairment of host defenses. Combined, these findings from experimental studies suggested a
 8    mechanism whereby Os reacts with lipids and antioxidants in airway epithelial lining fluid and
 9    epithelial cell membranes, activating a cascade of events that can lead to oxidative stress,
10    inflammation, and epithelial cell damage in the airways. These experimental and mechanistic
11    results provided support for epidemiologic evidence, in which short-term O?, exposure was
12    consistently associated with increases  in respiratory symptoms and asthma medication use in
13    asthmatic children, respiratory-related hospital admissions, and asthma-related emergency
14    department visits. The combined evidence across disciplines supported a causal relationship
15    between short-term O3 exposure and respiratory effects (US EPA, 2006).
16           Consistent with the strong body of evidence presented in the 2006 Os AQCD, recent
17    studies continue to support the conclusion that short-term Os exposures cause a variety of
18    respiratory effects (US EPA, 2012, sections 2.5, 6.2). As discussed in more detail below, recent
19    controlled human exposure studies report decrements in pulmonary function, increased
20    respiratory symptoms, and increased airway inflammation following exposures of young, healthy
21    adults to 63 concentrations as low as 60-70 ppb.  In addition, a number of recent epidemiologic
22    studies report associations with respiratory hospital admissions and emergency department visits,
23    as well as respiratory-related mortality, across the U.S., Europe, and Canada.  This recent
24    evidence is supported by a large body  of epidemiologic panel studies reporting associations with
25    respiratory symptoms in children with asthma; by some recent studies reporting CVassociated
26    increases in indicators of airway inflammation and oxidative stress in children with asthma; by
27    controlled human exposure and animal toxicological studies reporting (Vinduced airway
28    hyperresponsiveness, decreased pulmonary function, allergic responses, lung injury, impaired
29    host defense, and airway inflammation; and by recent evidence indicating that antioxidant
30    capacity may  modify the risk of respiratory morbidity associated with O3 exposures. Together,
31    the IS A judges that the total body of evidence integrated across controlled human exposure,
32    epidemiologic, and toxicological studies, and across the spectrum of respiratory health endpoints,
33    continues to demonstrate that there is a causal relationship between short-term Os exposure and
34    respiratory health effects.
                                                2-5

-------
 1          The extensive body of evidence supporting a causal relationship between short-term 63
 2    exposures and adverse respiratory health effects is discussed in detail in the ISA (US EPA, 2012,
 3    Chapter 6) and is summarized below for lung function (section 2.2.1.1); pulmonary inflammation
 4    and injury (section 2.2.1.2); airway hyperresponsiveness (section 2.2.1.3); respiratory symptoms
 5    and medication use (section 2.2.1.4); lung host defense (section 2.2.1.5); hospital admissions and
 6    emergency department visits (section 2.2.1.6); and respiratory mortality (section 2.2.1.7).
 7          2.2.1.1 Lung function
 8          In the last review, a large number of controlled human exposure studies had reported Os-
 9    induced lung function decrements in healthy adults engaged in intermittent, moderate exertion
10    following 6.6 hour exposures to Os concentrations at or above 80 ppb, while a relatively smaller
11    number of studies had reported effects following exposures to lower concentrations.  In addition,
12    epidemiologic panel studies had reported (Vassociated lung function decrements in a variety of
13    different populations likely to experience increased exposures. In the current review,  additional
14    controlled human exposure studies are available that have evaluated exposures to 63
15    concentrations  of 60 or 70 ppb.
16          In considering the overall body of evidence, the ISA notes that the link between Os
17    exposures and changes in lung function has been evaluated in a large number of controlled
18    human exposure studies and epidemiologic studies, as well as in some animal toxicological
19    studies. The majority of controlled human exposure studies of lung function have investigated
20    effects, such as the forced expiratory volume in one second (FEVi), in young healthy
21    nonsmoking adults (18-35 years of age) (US EPA, 2012, section 6.2.1.1) and during exposures to
22    fixed concentrations of O3 under carefully regulated environmental conditions and subject
23    activity levels.  Epidemiologic studies have evaluated associations between ambient Os and lung
24    function in children and adults, including in outdoor working and  exercising populations.
25    Controlled human exposure and epidemiologic studies are discussed in detail in section 6.2.1 of
26    the ISA (US EPA, 2012) and are summarized briefly in this section.
27    Controlled human exposure studies
28          In assessing controlled human exposure studies of lung function, the ISA considers both
29    group mean changes in lung function and inter-individual variability in such changes (US EPA,
30    2012, section 6.2.1.1). Consideration of group mean changes is important for discerning whether
31    observed effects are due to Os exposures themselves, as opposed to chance alone. Consideration
32    of inter-individual variability in responses is important when assessing the fraction of the
33    population that might experience clinically relevant effects following Os exposure (US EPA,
34    2012, section 6.2.1.1).

                                                2-6

-------
 1           In healthy adults, increasing the duration of 63 exposures and increasing ventilation rates
 2    decreases the Os exposure concentrations required to impair lung function.  Ozone exposure
 3    concentrations well above those typically found in ambient air are required to impair lung
 4    function in healthy resting individuals, while exposure to 63 concentrations at or below those in
 5    the ambient air have been reported to impair lung function in healthy individuals exposed for
 6    longer durations while undergoing intermittent, moderate exertion (US EPA, 2012a, section
 7    6.2.1.1). Figure 6-1 in the ISA summarizes the available evidence from multiple studies
 8    evaluating group mean changes in FEVi following prolonged Os exposures (i.e.,  6.6 hours) at
 9    moderate levels of physical activity (US EPA, 2012, section 6.2.1.1). With regard to the group
10    mean changes reported in these studies, the ISA specifically notes the following (US EPA,
11    2012a, section 6.2.1.1, Figure 6-1):
12       1.  Prolonged exposure to 40 ppb  Os results in a small decrease in group mean FEVi that is
13           not statistically different from  responses following exposure to filtered air (Adams, 2002;
14           Adams, 2006).
15       2.  Prolonged exposure to an average Os concentration of 60 ppb results in group mean FEVi
16           decrements ranging from 1.8% to 3.6% (Adams 2002; Adams, 2006;3 Schelegle et al.,
17           2009; Kim et al., 2011). Based on data from multiple studies, the weighted average
18           group mean decrement was 2.7%. In some analyses, these group mean decrements in
19           lung function were statistically significant (Brown et al., 2008; Kim et al., 2011) while in
20           other analyses, they were not (Adams, 2006; Schelegle et al., 2009).4
21       3.  Prolonged exposure to an average 63 concentration of 70 ppb results in a statistically
22           significant group mean decrement in FEVi of about 6% (Schelegle et al.,  2009).
23       4.  Prolonged square-wave exposure to average Os concentrations of 80 ppb, 100 ppb, or 120
24           ppb 63 results in statistically significant group mean decrements in FEVi ranging from 6
25           to 8%, 8 to 14%, and 13 to 16%, respectively (Folinsbee et al., 1988; Horstman et al.,
26           1990; McDonnell et al., 1991;  Adams,  2002; Adams, 2003; Adams, 2006).
      3 Adams (2006a); (2002) both provide data for an additional group of 30 healthy subjects that were exposed via
      facemask to 60 ppb (square-wave) O3 for 6.6 hours with moderate exercise (VE = 23 L/min per m2 BSA). These
      subjects are described on page 133 of Adams (2006) and pages 747 and 761 of Adams (2002). The FEVi decrement
      may be somewhat increased due to a target VE of 23 L/min per m2 BSA relative to other studies with which it is
      listed having the target VE of 20 L/min per m2 BSA. The facemask exposure is not expected to affect the FEVi
      responses relative to a chamber exposure.
      4Adams (2006) did not find effects on FEVi at 60 ppb to be statistically significant.  In an analysis of the Adams
      (2006) data, Brown et al. (2008) addressed the more fundamental question of whether there were statistically
      significant differences in responses before and after the 6.6 hour exposure period and found the average effect on
      FEVi at 60 ppb to be small, but highly statistically significant using several common statistical tests, even after
      removal of potential outliers.
                                                  2-7

-------
 1    As illustrated in Figure 6-1 of the ISA, there is a smooth dose-response curve without evidence
 2    of a threshold for exposures between 40 and 120 ppb O3 (US EPA, 2012a, Figure 6-1). When
 3    these data are taken together, the ISA concludes that "mean FEVi is clearly decreased by 6.6-h
 4    exposures to 60 ppb 63 and higher concentrations in subjects performing moderate exercise" (US
 5    EPA, 2012a, p. 6-9).
 6          Given the considerable inter-individual variability in the FEVi response in controlled
 7    human exposure studies of O?, (US EPA, 2012a, section 6.2.1.1), the ISA notes that the
 8    interpretation of biologically  small group mean decrements requires careful consideration.
 9    Specifically, the ISA notes that, even when the group mean FEVi decrement is small, some
10    individuals within the group could experience clinically meaningful decrements in lung function.
11    When considering what constitutes a clinically meaningful decrement in lung function, the ISA
12    notes that a 10% FEVi decrement is generally accepted as a clinically relevant abnormal
13    response (Dryden et al., 2010) (US EPA, 2012, section 6.2.1.1).  In addition, CASAC has
14    previously stated that "[a] 10% decrement in FEVi can lead to respiratory symptoms, especially
15    in individuals with pre-existing pulmonary or cardiac disease.  For example, people with  chronic
16    obstructive pulmonary disease have decreased ventilatory reserve (i.e., decreased baseline FEVi)
17    such that  a >10% decrement could lead to moderate to severe respiratory symptoms" (Samet,
18    2011).
19          The ISA notes that following prolonged exposures to an average concentration of 60 ppb
20    63, the proportion of study subjects with 10% or greater FEVi decrements was 20% in Adams
21    (2002), 3% in Adams (2006), 16% in Schelegle et al. (2009), and 5% in Kim et al. (2011). When
22    the results from these studies were combined, the ISA notes that 10% of subjects exposed to an
23    average 63 concentration of 60 ppb had FEVi decrements at or above 10%.5 In addition, the ISA
24    notes that responses within an individual tend to be reproducible over a period of several  months,
25    indicating that inter-individual differences reflect differences in the intrinsic responsiveness.
26    Given this, the  ISA concludes that "a considerable fraction" of healthy individuals experience
27    clinically meaningful decrements in lung function when exposed for 6.6 hours to 60 ppb Os
28    during intermittent, moderate exertion (US EPA, 2012a, section 6.2.1.1, p. 6-18).
29          Controlled human exposure studies of (Vinduced lung function decrements suggest that
30    an initial phase of recovery proceeds relatively rapidly following the cessation of O3  exposure in
31    healthy individuals, with acute decrements resolving within about 2 to 4 hours (Folinsbee and
32    Hazucha, 1989). Residual lung function effects are almost completely resolved within 24 hours.
      5The ISA also notes that by considering uncorrected responses, 10% is an underestimate of the proportion of healthy
      individuals that are likely to experience clinically meaningful changes in lung function following exposure for 6.6
      hours to 60 ppb O3 during intermittent moderate exertion (US EPA, 2012, section 6.2.1.1).
                                                2-8

-------
 1    In addition, with repeated 63 exposures over several days, (Vinduced lung function decrements
 2    are attenuated (US EPA, 2012a, section 6.2.1.1). Although this attenuation in Os-induced lung
 3    function decrements is correlated with an attenuation of respiratory symptoms, ongoing cellular
 4    damage persists as indicated by markers of inflammation and injury (section 2.2.1.2 below and
 5    US EPA, 2012a, section 6.2.3).
 6    Epidemiologic studies
 7          Epidemiologic studies have consistently reported that short-term ambient O3
 8    concentrations are associated with lung function decrements in diverse populations (US EPA,
 9    2012a, section 6.2.1.2), including in groups expected to experience elevated 63 exposure
10    concentrations and/or with higher exertion levels, such as children attending summer camps and
11    adults exercising or working outdoors, and in groups with pre-existing respiratory diseases such
12    as asthmatic children (US EPA, 2012a, section 6.2.1.2). The ISA notes that, among
13    epidemiologic studies, those that evaluate individuals engaged in outdoor recreation, exercise, or
14    work are more comparable to controlled exposure studies because of improved estimates of 63
15    exposures, measurement of lung function before and after discrete periods of outdoor activity,
16    and examination of Os effects during exertion when the dose of Os reaching the lungs may be
17    higher because of higher ventilation and inhalation of larger volumes of air (US EPA, 2012a,
18    section 6.2.1.2).
19          Studies of children attending summer camps have provided important insights into the
20    impact of ambient Os exposure on respiratory effects in young, healthy children. These studies
21    have been noted for their on-site measurement of ambient 63 and daily assessment of lung
22    function by trained staff (US EPA, 2012a, section 6.2.1.2). In groups mostly comprising healthy
23    children (ages 7-17 years), lung function decrements, as measured by within subject changes in
24    FEVi and peak expiratory flow (PEF), have been consistently reported to be associated with
25    ambient Os concentrations averaged over the 1-8 hours preceding lung function measurement
26    (US EPA, 2012a, Figure 6-3 and Table 6-3).  For FEVi, group mean decrements ranged from
27    approximately 0.3% to 2.2% per 40 ppb increase in 1-hour 63 concentrations (US EPA, 2012a,
28    Table 6-3). In most cases, associations between Os and lung function were statistically
29    significant (US EPA, 2012a, Figure 6-3, Table 6-3).  Maximum 1-hour Os concentrations in
30    these studies ranged from 95 ppb to 245 ppb (US EPA, 2012a, Table 6-2). In the one study that
31    specifically evaluated different ranges of ambient Os concentrations (Spektor et al., 1988a),
32    associations with lung function decrements remained statistically significant when the analysis
33    was restricted to 1-hour Os concentrations below 60 ppb (US EPA, 2012a, Table 6-6).
34          Similar to the camp studies discussed above, studies of individuals exercising outdoors
35    have evaluated subjects over days with a wide range in ambient Os concentrations and have used
                                               2-9

-------
 1    onsite assessment of 63 exposures during discrete periods of outdoor exercise.  Collectively,
 2    these studies report that ambient Os concentrations are associated with small (< 1% to 4% per
 3    standardized increment in O^6) group mean decrements in lung function in adults and children
 4    during exercise of variable duration and intensity (US EPA, 2012a, Figure 6-4 and Table 6-4).
 5    Most of these associations were statistically significant and, in studies that presented analyses
 6    restricted to 63 concentrations below 61 ppb (Brunekreef et al., 1994) or below 80 ppb (Spektor
 7    et al.,  1988b), associations with lung function decrements remained  (though associations in the
 8    restricted analysis of Brunekreef were no longer statistically significant) (US EPA, 2012a, Table
 9    6-6). In addition, when Korrick et al. (1998) restricted their analyses to 2-12 hour 63
10    concentrations greater than 40 ppb (maximum concentration was 74  ppb), Os was associated
11    with larger lung function decrements than in their analysis encompassing the full distribution of
12    O3 concentrations (US EPA, 2012a, Table 6-6).7
13          In addition to the camp studies and the outdoor exercise studies, studies of outdoor
14    workers have also consistently reported that ambient Os concentrations are associated with
15    decrements in lung function (US EPA, 2012a, section 6.2.1.2, Figure 6-5,  Table 6-5). Although
16    most of these studies assessed 63 exposures using central site measurements, they were
17    noteworthy for the long periods of time spent outdoors (6-14 hours) and because associations
18    with lung function decrements were reported for time periods with relatively low ambient Os
19    concentrations (Chan and Wu, 2005; Brauer et al., 1996; Hoppe et al.,  1995) (US EPA, 2012a,
20    section 6.2.1.2, Figure 6-5, Table 6-1).8 In the study by Brauer et al. (1996), associations
21    between ambient 63 and lung function decrements remained statistically significant in analyses
22    restricted to maximum hourly concentrations below 40 ppb. Similar to other outdoor exposure
23    studies, group mean Os-associated lung function decrements in outdoor workers were relatively
24    small (>1% to 3.6% per standardized increment in 63 concentration) (US EPA, 2012a, section
25    6.2.1.2).
26          The majority of the epidemiologic studies discussed above were not specifically focused
27    on populations with respiratory diseases such as asthma. Epidemiologic studies of children and
28    adults with respiratory disease provide additional support for the link between exposure to
29    ambient Os and lung function decrements. Although these studies typically rely on central site
30    monitoring and self-administered lung function tests, they provide important information on
      6 Effect estimates are standardized to a 40 ppb increase in O3 concentrations averaged over 1-hour or less and to a 30
      ppb increase in O3 concentrations averaged over 12 hours (US EPA, 2012a, Table 6-3).
      7Ozone averaging times ranged from 10 minutes to 2.4 hours in Brunekreef et al. (1994), were 30-minutes in
      Spektor et al. (1988b), and ranged from 2 to 12 hours in Korrick et al. (1988) (US EPA, 2012a, Table 6-4).
      8The maximum reported 8-hour concentration in the study by Chan and Wu (2005) was 65 ppb. In the other studies,
      maximum 1-hour (Brauer et al., 1996) and 30-minute (Hoppe et al., 1995) concentrations were 84 ppb and 77 ppb,
      respectively (US EPA, 2012a, Table 6-2). Maximum 8-hour concentrations in these studies would have been lower.
                                                 2-10

-------
 1    factors that may confer increased susceptibility to the respiratory effects of 63 exposures (US
 2    EPA, 2012a, section 6.2.1.2).
 3          Collectively, the large body of evidence for O3-associated lung function decrements in
 4    asthmatic children, which includes two U.S. multi-city studies (Mortimer et al., 2000; Mortimer
 5    et al., 2002; O'Connor et al., 2008), demonstrates that increases in ambient 63 exposure are
 6    associated with decrements in lung function in asthmatic children and with increases in the
 7    percent of asthmatic children who experience clinically significant lung function decrements (US
 8    EPA, 2012a, section 6.2.1.2, Figures 6-6 and 6-7, Tables 6-7 and 6-8). Such associations have
 9    been reported to persist even when analyses are restricted to days with relatively low ambient Os
10    concentrations (i.e., to 8-hour concentrations below 80 ppb in Mortimer et al., 20029).  With
11    regard to inter-individual variability, Mortimer et al. (2002) reported that for a 30 ppb increase in
12    8-hour Os  concentrations, there was a 30% increase in the incidence of PEF decrements greater
13    than 10%.  In addition, Hoppe et al. (2003) reported that 47% of asthmatic children experienced
14    lung function decrements greater than 10% on days with 30-minute maximum Os concentrations
15    greater than 50 ppb, relative to days with 30-minute maximum concentrations less than 40 ppb.
16          Epidemiologic studies have reported that Os-associated lung function decrements may be
17    clinically significant in asthmatic children, based on concurrent increases in respiratory
18    symptoms (section 2.2.1.4, below). Similar lung function decrements have been reported in
19    studies evaluating children without respiratory disease but without concurrent increases in
20    respiratory symptoms. Because of the higher overall lung function, the decrements may not be
21    large enough to be clinically significant in healthy children (section 2.2.1.4 below; US EPA,
22    2012a, section 6.2.1.2).
23           A  number of epidemiologic studies of Os-associated decrements in lung function have
24    evaluated the impacts of potential confounders, including temperature, pollen, and co-occurring
25    pollutants.  Available studies have consistently reported that associations between Os and
26    decrements in lung function are not driven by the effects of either temperature or pollen (US
27    EPA, 2012a, section 6.2.1.2). With regard to co-pollutants, several studies have reported low co-
28    pollutant concentrations or the lack of associations between co-pollutants and lung function
29    decrements, suggesting that co-pollutants were not responsible for reported associations with 63.
30    Other studies have evaluated statistical models that include co-pollutants such as PM2.s, PMi0,
31    sulfate, NO2 and/or SO2.
      9The 80 ppb cutoff in the restricted analysis is by Mortimer et al. (2002) reflects O3 concentrations averaged across
      multiple monitors.
                                                2-11

-------
 1          Among studies that have evaluated co-pollutant models, associations between 63 and
 2    lung function decrements generally remained robust, though results have been somewhat
 3    variable for co-pollutant models that include sulfate (US EPA, 2012a, section 6.2.1.2, Figure 6-9,
 4    Table 6-14). In one study (Neas et al., 1999), the association with 63 was attenuated in a co-
 5    pollutant model that included sulfate, while another study (Thurston et al., 1997) reported a
 6    larger (Vassociated decrement in lung function when sulfate was included in the model. In the
 7    study by Thurston (1997), when a single influential day was removed (a day when both sulfate
 8    and O3 were at their peaks), the sulfate effect was attenuated while the O3 effect remained robust.
 9    Several studies did not provide quantitative results from co-pollutant models but reported that O3
10    effects on lung function remained statistically significant in models that included PMio, sulfate,
11    NO2, nitrate, or ammonium (US EPA, 2012a, section 6.2.1.2).
12          2.2.1.2 Pulmonary inflammation and injury
13          As discussed in detail in the ISA (US EPA, 2012a,  section 6.2.3), O3 exposure can
14    increase respiratory tract inflammation.  In the last review, controlled human exposure studies
15    reported O3-induced airway inflammation following exposures at or above 80 ppb. In the current
16    review, additional epidemiologic studies, as well as a controlled human exposure study
17    conducted following exposures to 60 ppb O3, are also available.
18          Inflammation is a host response to injury and the induction of inflammation is evidence
19    that injury has occurred.  The link between O3 exposures and inflammation and injury has been
20    evaluated in controlled human exposure studies, epidemiologic studies, and animal toxicological
21    studies. Controlled human exposure studies have generally been conducted in healthy
22    individuals or asthmatics and have evaluated one or more indicators of inflammation, including
23    neutrophil (PMN) influx, increased permeability of the respiratory epithelium, and/or prevalence
24    of proinflammatory cytokines (US EPA, 2012a, section 6.2.3.1). Epidemiologic studies have
25    generally evaluated associations between ambient O3 and markers of inflammation and/or
26    oxidative stress, which has been reported to play a key role in initiating and sustaining
27    inflammation (US EPA, 2012a, section 6.2.3.2). These two types of studies are discussed in
28    detail in the ISA (US EPA, 2012a, section 6.2.3) and are summarized briefly below.
29    Controlled human exposure studies
30          Short-term exposures to O3 concentrations at or above typical ambient concentrations
31    (i.e., 60 ppb to  400 ppb) result in inflammation and tissue damage in the respiratory tract, as
32    indicated by a variety of endpoints including PMN influx, increased epithelial permeability, and
33    elevated levels of proinflammatory markers (US EPA, 2012a, section 6.2.3.1). As with lung
34    function, inflammatory responses to O3 are generally reproducible within an individual and some

                                               2-12

-------
 1    individuals (e.g., asthmatics) appear to be intrinsically more at-risk than others.  Unlike (V
 2    induced decrements in lung function, which are attenuated following repeated exposures over
 3    several days (US EPA, 2012a, section 6.2.1.1), some markers of Os-induced inflammation and
 4    tissue damage remain elevated during repeated exposures, indicating ongoing damage to the
 5    respiratory system (US EPA, 2012a, section 6.2.3.1).
 6           Many of the controlled human exposure studies evaluating Os-induced inflammation
 7    have used Os concentrations above those typically found in the ambient air in the United States
 8    (e.g., 200 to 400 ppb). However, a few studies have reported increases in markers of respiratory
 9    inflammation following exposures to relatively low Os concentrations.  In a recent study, Kim et
10    al. (2011) reported a statistically significant increase in sputum neutrophil levels following a 6.6-
11    hour exposure to 60 ppb O3 in young healthy adults engaged in intermittent, moderate exertion.
12    Other studies (Devlin et al., 1991; Alexis et al., 2010) have reported increases in inflammatory
13    markers following 6.6-hour exposures to 80 ppb 63 in healthy adults (US EPA, 2012a, 6.2.3.1),
14    though one of these studies (Alexis et al., 2010) did not use a filtered air control.
15    Epidemiologic studies
16           In the 2006  O3 AQCD (US EPA, 2006), epidemiologic evidence for O3-associated
17    changes in pulmonary inflammation was limited.  Since 2006, as a result of the development of
18    less invasive methods, the number of studies assessing ambient (Vassociated changes in airway
19    inflammation and oxidative stress has increased.  Most of these recent studies have evaluated
20    biomarkers of inflammation or oxidative stress10 in exhaled breath, nasal lavage fluid, or induced
21    sputum. As noted in the ISA, these recent studies form a larger body of evidence to help
22    establish  coherence with findings from controlled human exposure and animal toxicological
23    studies that have measured similar or related endpoints and provide further biological plausibility
24    for associations of ambient 63 exposure  with other effects such as respiratory symptoms.
25    However, important uncertainties limit the interpretation of recent biomarker studies, including
26    uncertainty regarding the clinical relevance of observed changes in biomarker levels.  In
27    addition,  the limited available information on the time course of changes in biomarker levels, on
28    the subject factors that contribute to inter-individual variability, and on standardized sample
29    collection approaches contributes to differences among study results (US EPA, 2012a, section
30    6.2.3.2).
31           While epidemiologic studies have reported mixed results when evaluating associations
32    between ambient Os concentrations and  different markers of oxidative stress or inflammation
      10As discussed in the ISA (US EPA, 20112a, section 6.2.3.2), oxidative stress has been used as an indicator of
      pulmonary inflammation in epidemiologic studies.
                                                2-13

-------
 1    (US EPA, 2012a, section 6.2.3.2, Figure 6-10, Tables 6-15, 6-16, and 6-17), these studies do
 2    provide some insight into the mechanism of Os-induced inflammation. With regard to
 3    mechanism, in a study in Mexico City, (^-associated increases in pulmonary inflammation were
 4    attenuated with higher antioxidant intake (Sienra-Monge et al., 2004).  This result suggests that
 5    inhaled Os may be an important source of reactive oxygen species in airways and may increase
 6    airway inflammation via oxidative stress mediated mechanisms. In addition, though results of
 7    epidemiologic studies have been mixed, a study conducted in Mexico reported associations
 8    between Os and markers of airway inflammation with 8-hour Os concentrations at or below
 9    about 61 ppb (Romieu et al., 2008) (US EPA, 2012a, Table 6-17).
10          The clinical significance of (Vinduced increases in oxidative stress and inflammation is
11    supported by the small number of studies that have correlated O3-associated inflammation with
12    other endpoints, including respiratory symptoms.  In subjects with asthma, ambient Os exposure
13    was associated with increases in markers of oxidative stress and inflammation that were
14    accompanied by a concomitant increase in cough  (Barraza-Villarreal et al.,  2008) and by a
15    decrease in a quality of life score (Khatri et al., 2009). These findings, though limited, support
16    the clinical relevance of (Vassociated airway inflammation in asthmatics.
17          2.2.1.3 Airways hyperresponsiveness
18          Airway hyperresponsiveness refers to a condition in which the conducting airways
19    undergo enhanced bronchoconstriction in response to a variety of stimuli (US EPA, 2012a,
20    section 6.2.2). Increased airway responsiveness is an important consequence of exposure to
21    ambient 63 because the  airways are then predisposed to narrowing upon inhalation of a variety
22    of ambient stimuli including specific allergens, SO2, and cold air. Asthmatics often exhibit
23    increased airway responsiveness at baseline relative to healthy controls, and they can experience
24    further increases in  responsiveness following exposures to 63 (US EPA, 2012a, section 6.2.2.1).
25          Airway responsiveness is often quantified by measuring changes in  pulmonary function
26    following the inhalation of an aerosolized allergen or a nonspecific bronchoconstricting agent
27    (e.g., methacholine) or following exposure to a bronchoconstricting stimulus such as cold air. In
28    the last review, controlled human exposure studies reported that exposures to 63 concentrations
29    at or above 80 ppb increased airway responsiveness in human subjects, as indicated by a
30    reduction in the concentration of specific (e.g., ragweed) and non-specific (e.g., methacholine)
31    agents required to produce a given reduction in lung function (e.g., as measured by FEVi or
32    specific airway resistance) (US EPA, 2012a, section 6.2.2.1).  Horstman et  al. (1990) reported
33    increased airway responsiveness in healthy volunteers following 6.6 hour exposures to Os
34    concentrations at or above 80 ppb (subjects engaged in intermittent, moderate exertion).
                                               2-14

-------
 1          A number of animal toxicology studies, including some recent studies conducted since
 2    the last review, provide support for the Os-induced airway hyperresponsiveness reported in
 3    humans (US EPA, 2011, section 6.2.2.2).  Although most of these studies evaluated Os
 4    concentrations above those typically found in ambient air in cities in the United States (i.e., most
 5    studies evaluated Os concentrations of 100 ppb or greater), one study reported that a very low
 6    exposure concentration (50 ppb for 4 hours) induced airway hyperresponsiveness in some rat
 7    strains, suggesting a genetic component (Depuydt et al., 1999). Additional rodent studies have
 8    reported Os-induced airway hyperresponsiveness following exposures to Os concentrations from
 9    100 to 500 ppb (Johnston et al., 2005; Chhabra et al., 2010; Larsen et al., 2010).  In
10    characterizing the relevance of these exposure concentrations, the ISA noted that a study using
11    radiolabeled Os suggests that even very high Os exposure concentrations in rodents could be
12    equivalent to much lower exposure concentrations in humans. Specifically, a 2000 ppb (2 ppm)
13    Os exposure concentration in resting rats was reported to be roughly equivalent to a 400 ppb
14    exposure concentration in exercising humans (Hatch et al., 1994). Given this relationship, the
15    ISA noted  that animal data obtained  in resting conditions could underestimate the risk of effects
16    for humans, though the ISA also cautioned that there are important limitations in the approach
17    used in this study, leading to uncertainty in such interspecies comparisons (US EPA, 2012a,
18    section 5.5.1).
19          Changes in airway responsiveness  after O3 exposure appear to resolve more slowly than
20    changes  in FEVi or respiratory symptoms  (e.g.,  18 to 24 hours, though longer in some
21    individuals, as reported in Folinsbee and Hazucha (1989, 2000)) and tend to be somewhat less
22    likely to attenuate with consecutive exposures (Gong et al., 1997; Folinsbee et al., 1994; Kulle et
23    al., 1982; Dimeo et al., 1981) (US EPA, 2012a, section 6.2.2.1).  In animal studies a 3-day
24    continuous exposure resulted in attenuation of (Vinduced airway hyperresponsiveness (Johnston
25    et al., 2005) while repeated exposures for 2 hours per day over 10 days did not (Chhabra et al.,
26    2010), suggesting that attenuation could be lost when repeated exposures are interspersed with
27    periods of  rest (US EPA, 2012a, section 6.2.2.2).  Studies reporting increased airway
28    responsiveness after Os exposure contribute to a plausible link between ambient Os exposure and
29    increased respiratory symptoms in asthmatics, and increased hospital admissions and emergency
30    department visits for asthma (see below and US EPA, 2012a, section 6.2.2).
31          2.2.1.4 Respiratory symptoms and medication use
32          Because respiratory symptoms are  associated with limitations in activity  and are the
33    primary reason for using asthma medication and seeking medical care, studies evaluating the link
34    between Os exposures and such symptoms allow a characterization of the clinical and public
35    health significance of ambient Os exposure (US EPA, 2012a, section 6.2.4).  The link between

                                               2-15

-------
 1    subjective respiratory symptoms and 63 exposures has been evaluated in both controlled human
 2    exposure and epidemiologic studies.  In the last review, controlled human exposure studies
 3    reported statistically significant increases in respiratory symptoms following exposures to O?,
 4    concentrations at or above 80 ppb, with non-significant increases reported following exposures
 5    to lower concentrations.  In addition, epidemiologic studies reported associations between
 6    ambient 63 and respiratory symptoms in a variety of locations and populations, including
 7    asthmatic children living in U.S. cities. In the current review, additional controlled human
 8    exposure studies have evaluated respiratory symptoms following exposures to Os concentrations
 9    below 80 ppb and recent epidemiologic studies have evaluated associations with respiratory
10    symptoms.  These studies are discussed in detail in the ISA (US EPA, 2012a, sections 6.2.1 and
11    6.2.4) and summarized briefly below.
12    Controlled human exposure studies
13          In controlled human exposures studies, statistically significant increases in respiratory
14    symptoms have been consistently reported in healthy volunteers  engaged in intermittent,
15    moderate exertion following 6.6 hour exposures to average O3 concentrations at or above 80 ppb
16    (Adams, 2003; Adams, 2006; Schelegle et al., 2009). In addition, Schelegle et al. (2009)
17    reported a statistically significant increase in respiratory symptoms following 6.6 hour exposures
18    to an average Os concentration of 70 ppb. In contrast,  increased  respiratory symptoms following
19    exposures to Os concentrations below 70 ppb have not been consistently reported. Adams
20    (2006) reported an increase in respiratory symptoms in healthy volunteers during a 6.6 hour
21    exposure protocol with an average Os exposure concentration of 60 ppb. With one of the
22    exposure protocols tested in this study (i.e.,  the triangular exposure protocol), the increase in
23    symptoms was significantly different from initial respiratory symptoms but not from filtered air
24    controls.  Neither Schelegle et al.  (2009) nor Kim et al. (2011) reported a statistically significant
25    increase in respiratory symptoms  following exposures  to 60 ppb  63, though Schelegle reported a
26    non-significant trend toward increased symptoms.
27    Epidemiologic studies
28          In epidemiologic studies, respiratory symptom  data are typically self-reported by subjects
29    or their parents.  Such symptom diaries are a convenient and useful tool for collecting individual-
30    level data from a large number of subjects.  However, several limitations of this approach are
31    well-recognized, including recall  error, differences among subjects in the interpretation of
32    symptoms, and biased reporting between participants with and without asthma. Most of these
33    limitations are sources of random error that can bias effect estimates to the null or increase the
34    uncertainty around effect estimates.  Most epidemiologic studies of O3 and respiratory symptoms
                                                2-16

-------
 1    have been conducted in children and/or adults with asthma, with fewer studies, and less
 2    consistent results, in non-asthmatic populations.
 3           Single-city and multi-city epidemiologic studies have generally reported positive
 4    associations between ambient Os concentrations and respiratory symptoms and medication use in
 5    children with asthma (US EPA, 2012a, Figures 6-11, 6-12; Tables 6-19, 6-20). The ISA
 6    concludes that the epidemiologic evidence clearly demonstrates that short-term ambient Os
 7    exposure is associated with increases in respiratory symptoms and asthma medication use in
 8    asthmatic children, with a smaller body of literature also supporting such associations in
 9    asthmatic adults (US EPA, 2012a, section 6.2.4.2). In studies that evaluated the potential for
10    confounding by meteorological factors, pollen counts, or co-pollutants, associations with 63
11    remained robust (US EPA, 2012a, Table 6-25), though disentangling the independent effects of
12    Os exposure in these studies is complicated due to the high correlations observed between
13    pollutants and the different averaging times and lags of exposure examined for co-pollutants (US
14    EPA, 2012a, section 6.2.4.4). The consistency of associations among individuals with asthma,
15    with and without adjustment for co-pollutants, combined with evidence from controlled human
16    exposure studies for the direct effects of 63 exposure (see above), provide substantial evidence
17    for the independent effects of ambient Os exposure on increases in respiratory symptoms (US
18    EPA, 2012a, section 6.2.4.5). Multi-city and single-city epidemiologic studies are discussed in
19    detail in the ISA (US EPA, 2012a, section 6.2.4) and summarized briefly below.
20           A multi-city  study by Mortimer et al. (2002) reported a positive and statistically
21    significant association between Os and morning asthma symptoms in children across eight U.S.
22    communities. This association remained positive and statistically significant when the analysis
23    was restricted to 8-hour 63 concentrations below 80 ppb.11 In co-pollutant models that included
24    SC>2, NC>2, or PMio, the association between Os and respiratory symptoms remained positive,
25    though not statistically significant. The ISA notes that the interpretation of these co-pollutant
26    models is complicated because of the different averaging times and lags of exposure examined
27    for O3 and co-pollutants (US EPA, 2012a, section 6.2.4.4).
28           Other multi-city studies conducted in the United States and North America have reported
29    mixed results in asthmatic children. Specifically, Schildcrout et al. (2006) reported positive
30    associations with respiratory symptoms across nine North American cities, though these
31    associations were not statistically significant. Schildcrout did not report an association with
32    medication use. In addition, O'Connor et al. (2008) reported both positive and negative
33    associations with wheeze and nighttime asthma, respectively, in seven U.S. communities. These
      1 JThe 80 ppb cutoff in the restricted analysis is by Mortimer et al. (2002) reflected O3 concentrations that were
      averaged across multiple monitors.
                                                2-17

-------
 1    associations were not statistically significant (US EPA, 2012a, section 6.2.4; Figures 6-11, 6-12;
 2    Tables 6-19, 6-20).
 3          Single-city studies conducted in the United States generally reported associations with
 4    symptoms and/or medication use that were positive, but not statistically significant, in asthmatic
 5    children (US EPA, 2012a, Figures 6-11, 6-12; Tables 6-19, 6-20).  For associations in locations
 6    with relatively low ambient Os concentrations, positive associations with respiratory symptoms
 7    were statistically significant in a cool-season study by Rabinovitch et al. (2004) (maximum 1-
 8    hour concentration was 70 ppb) and not statistically  significant in a study by Delfmo et al.  (2003)
 9    (maximum  1-hour concentration was 52 ppb).  In addition to these studies conducted in the
10    United States, a study conducted in Amsterdam, with a maximum 8-hour 63 concentration of 57
11    ppb during the study period, reported a positive and  statistically significant association between
12    Os concentrations and upper respiratory symptoms and a positive, but not statistically  significant,
13    association  with medication use (Gielen, et al., 1997) (US EPA, 2012a, Figures 6-11, 6-12;
14    Tables 6-19, 6-20).
15          A relatively small number of epidemiologic studies have evaluated associations with
16    respiratory symptoms and/or activity levels in adults with respiratory disease (US EPA, 2012a,
17    section 6.2.4.2).  In Los Angeles, Eiswerth et al. (2005) reported that increased 63 concentrations
18    were associated with lower probabilities of participating in both indoor and outdoor activities,
19    though only the association with indoor activities was statistically significant.  Khatri et al.
20    (2009) reported that an increase in ambient 63 concentrations was  associated with lower scores
21    on a quality of life assessment that characterizes symptoms, mood, and activity limitations. In  a
22    panel study of asthmatic adults and children, conducted in an area with relatively low
23    concentrations of 63 and aeroallergens (i.e., in East Moline, Illinois,  with a maximum  8-hour 63
24    concentration of 78 ppb), Os was associated with increased morning  and evening respiratory
25    symptoms (Ross et al., 2002). These associations remained statistically significant with
26    adjustment for weather and aeroallergens.
27          2.2.1.5 Lung host defense
28          The mammalian respiratory tract has a number of closely integrated defense mechanisms
29    that, when functioning normally, provide protection from the potential health effects attributed to
30    exposure to a wide variety of inhaled particles and microbes. The previous Os AQCD (US EPA,
31    2006) concluded that animal toxicological studies provided evidence that acute 63 exposures as
32    low as 80 to 500 ppb can increase susceptibility to infectious diseases due to modulation of lung
33    host defenses. A few recent studies have added to this body of evidence, which includes
34    controlled human exposure and/or animal toxicological evidence for O3 effects on mucociliary
35    clearance, alveolar macrophage functioning, and effects on infection and adaptive immunity (US
                                                2-18

-------
 1    EPA, 2012a, section 6.2.5). Of the immune-related effects with clear functional implications,
 2    only impaired alveolar macrophage functioning and increased susceptibility to bacterial
 3    infections have been reported following exposures to Os concentrations near the level of the
 4    current 63 standard.
 5           With regard to macrophage function, a controlled human exposure study reported
 6    decrements in the ability of alveolar macrophages to phagocytize yeast following 6.6 hour
 7    exposures of healthy adult volunteers undergoing intermittent, moderate exertion to Os
 8    concentrations from 80 to  100 ppb  (Devlin et al., 1991). In addition, 2-hour exposures of rabbits
 9    to 100 ppb Os was reported to inhibit phagocytosis while a 3-hour exposure to 250 ppb decreased
10    lysosomal enzyme activities (Driscoll et al., 1987; Hurst et al., 1970). Alveolar macrophages
11    from rats exposed to 100 ppb  O3 for 1 or 3 weeks exhibited reduced hydrogen peroxide
12    production (Cohen et al., 2002).
13           With regard to susceptibility to bacterial infections, the ISA notes (US EPA, 2012a,
14    section 6.2.5.4) that in animal models of bacterial infection, exposure to 80 ppb 63 has been
15    reported to increase streptococcus-induced mortality, regardless of whether O3 exposure
16    precedes or follows infection (Miller et al., 1978; Coffin and Gardner, 1972; Coffin et al., 1967).
17    Although increases in mortality were due to the infectious agent, thereby reflecting an (V
18    induced functional impairment of host defenses, results have been inconsistent across species.
19    For example, although both mice and rats exhibit impaired bactericidal macrophage activity after
20    63 exposure, mortality due to infection is only observed in mice. In addition, although mice and
21    humans share many host defense mechanisms, there is little compelling evidence from
22    epidemiologic studies to suggest an association between Os exposures and decreased resistance
23    to bacterial infection (US EPA, section 6.2.7).
24           2.2.1.6 Hospital admissions and emergency department visits
25           The 2006 O3 AQCD (US EPA, 2006) identified a number of epidemiologic studies
26    reporting positive associations between ambient O3-exposures and increased respiratory hospital
27    admissions and emergency department visits.  Overall, the AQCD concluded that these studies
28    provided evidence of a causal relationship between short-term ambient ozone exposures and
29    increased respiratory morbidity during the warm season (US EPA, 2012a, section 6.2.7.1). A
30    number of recent studies conducted in the United States, Canada, Europe, and Asia provide
31    additional support for (Vassociated increases in both hospital admissions and emergency
32    department visits for respiratory outcomes (US EPA 2012a, section 6.2.7, Figures 6-18 and 6-
33    19). Consistent with studies available in the last review, recent studies have reported stronger
34    associations during the warm  season than the cold season (US EPA, 2012a, Figure 6-18, Table 6-
35    28) and that O3 effect estimates remained relatively robust upon the inclusion of PM and gaseous
                                               2-19

-------
 1    co-pollutants in two-pollutant models (US EPA, 2012a, Figure 6-19; Table 6-29 and section
 2    6.2.7.5).
 3           This section discusses multi-city and single-city studies that have evaluated the
 4    relationship between Os and respiratory-related hospital admissions and emergency department
 5    visits (US EPA, 2012a, section 6.2.7).  In the majority of these studies, all respiratory-related
 6    admissions or emergency department visits have been evaluated, though several recent studies
 7    have evaluated cause-specific endpoints, including admissions or emergency department visits
 8    for COPD,  asthma, and pneumonia. This section is divided by endpoint into discussions of
 9    hospital admissions for all respiratory causes, cause-specific hospital admissions, emergency
10    department visits for all respiratory causes, and cause-specific emergency department visits.
11    Hospital admissions for all respiratory causes
12           The APHENA study (APHENA is for Air Pollution and Health: A European and North
13    American Approach) analyzed air pollution and health outcome data from existing Canadian,
14    European, and U.S.  multi-city studies and examined the influence of varying model specification
15    to control for season and weather (Katsouyanni et al., 2009).  The U.S.-based portion of the
16    APHENA study utilized the National Morbidity, Mortality, and Air Pollution Study (NMMAPS)
17    cohort which, for the Katsouyanni et al. (2009) analysis, comprised respiratory hospital
18    admissions among individuals 65 years of age and older from 14 US cities with 63 data from
19    1985-1994  (7 cities  had summer only O3 data). For the year round analysis, Katsouyanni et al.
20    (2009)  reported consistently positive, and statistically significant in models with 8 degrees of
21    freedom per year (US EPA, 2012a, section 6.2.7.2), associations between 1-hour 63
22    concentrations and respiratory hospital admissions across the datasets from the U.S., Canada, and
23    Europe (US EPA 2012a, Figure 6-14).12 In co-pollutant models adjusting for PMio, O3 effect
24    estimates remained positive, though effect estimates were somewhat attenuated in the U.S. and
25    European datasets, possibly due to the PM sampling schedule (US EPA 2012a, Figure 6-14).
26    Effect estimates for the warm season were larger than for the year-round analysis in the
27    Canadian dataset,  but generally similar in magnitude to the year-round analysis in the U.S. and
28    European datasets.
29           Several additional multicity studies examined  respiratory disease hospital admissions in
30    Canada and Europe.  Cakmak et al. (2006) reported a statistically significant increase in
31    respiratory  hospital  admissions in 10 Canadian cities (4.4% increase per 20 ppb increase in 24-
      12The study by Katsouyanni et al. (2009) evaluated different statistical models. Although the investigators did not
      identify the model they deemed to be the most appropriate for comparing the results across study locations, they did
      specify that "overall effect estimates (i.e., estimates pooled over several cities) tended to stabilize at high degrees of
      freedom" (Katsouyanni et al., 2009). In discussing of the results of this study, the ISA focused on models with 8
      degrees of freedom per year (US EPA, 2012a, section 6.2.7.2).
                                                 2-20

-------
 1    hour average 63, 95% CI: 2.2, 6.5%).  In analyses of potential effect modifiers of the (V
 2    respiratory hospital admission relationship, individuals with an education level less than the 9th
 3    grade were found to be at greater risk. Dales et al. (2006) reported a 5.4% (95% CI: 2.9, 8.0%)
 4    increase in neonatal respiratory hospital  admissions for a 20 ppb increase in 24-h avg 63
 5    concentrations in 11 Canadian cities from 1986 to 2000.  In contrast, Biggeri et al. (2005) did not
 6    detect an association between short-term 63 exposure and respiratory hospital admissions in four
 7    Italian cities from 1990 to 1999.
 8          In addition to the large multi-city studies discussed above, several smaller-scale studies
 9    have also reported associations with total respiratory hospital admissions.  Specifically, Lin et al.
10    (2008) reported a positive association between 63 and pediatric (i.e., <18 years) respiratory
11    admissions, though results were not presented quantitatively, in an analysis of 11 geographic
12    regions in New York  state from 1991 to  2001.  In co-pollutant models with PMio, the authors
13    reported that region-specific 63 associations with respiratory hospital admissions remained
14    relatively robust. A recent study (Neidell,  2009; Neidell  and Kinney, 2010) conducted in
15    Southern California reported positive and statistically significant O?, effect estimates for children,
16    working age adults, and older adults. This study  reported that controlling for avoidance behavior
17    on days with Os alerts increases Os effect estimates for respiratory hospital admissions in
18    children and older adults, providing preliminary evidence that epidemiologic studies may
19    underestimate associations between O3 exposure  and health effects by not accounting for
20    behavioral modification when public health alerts are issued.
21    Cause-specific hospital admissions
22          With regard to cause-specific respiratory outcomes, the limited evidence available in the
23    last review indicated that the strongest findings were for ambient Os associated asthma and
24    chronic obstructive pulmonary disease (COPD) respiratory hospital admissions (US EPA 2012a,
25    6.2.7.2). Since the last review, a few additional studies have investigated cause-specific
26    respiratory admissions (i.e., COPD, asthma, pneumonia)  in relation to O?, exposure (Medina-
27    Ramon et al, 2006; Yang et al., 2005; Zanobetti and Schwartz, 2006; Silverman and Ito, 2010).
28          With regard to COPD, Medina-Ramon  et al. (2006) examined the association between
29    short-term ambient Os concentrations and Medicare hospital admissions among individuals >
30    65 years of age for COPD in 35 cities in the U.S. for the years 1986-1999. The authors reported
31    a 1.6% increase (95% CI: 0.48, 2.9%) in COPD admissions for lag 0-1 in the warm season  for a
32    30 ppb increase in 8-h max Os concentrations.  The authors found no evidence for such
33    associations in cool season or in year round analyses. In  a co-pollutant model with PMio, the
34    association between O3 and COPD hospital admissions remained robust.  In a single-city study
35    conducted in Vancouver from  1994-1998, a location with low ambient Os concentrations (US
                                                2-21

-------
 1   EPA, 2012a, Table 6-26), Yang et al. (2005) reported a statistically non-significant 8.8% (95%
 2   CI: -12.5, 32.6%) increase in COPD admissions per 20 ppb increase in 24-hour average Os
 3   concentrations. In two-pollutant models with every-day data for NC>2, SC>2, CO, and PMio, Os
 4   risk estimates remained robust, though not statistically significant (US EPA, 2012a, Figure 6-19;
 5   Table 6-29). In addition, Wong et al. (2009) reported increased Os-associated COPD admissions
 6   during periods of increased influenza activity in Hong Kong.
 7          The ISA assessed a study that evaluated asthma-related hospital admissions in New York
 8   City (US EPA, 2012a, section 6.2.7.2) (Silverman and Ito, 2010).  This  study examined the
 9   association of 8-hour maximum Os  concentrations with severe acute asthma admissions (i.e.,
10   those admitted to the Intensive Care Unit [ICU]) during the warm season in the years 1999
11   through 2006 (Silverman and Ito, 2010)). The investigators reported positive associations
12   between Os and ICU asthma admissions for the 6- to 18-year age group (26.8% [95% CI: 1.4,
13   58.2%] for a 30 ppb increase in maximum 8-hour average Os concentrations, but little evidence
14   of associations for the other age groups examined (<6 years, 19-49, 50+, and all ages). However,
15   positive associations were  observed for each age-stratified group and all ages for non-ICU
16   asthma admissions, but again the strongest association was reported for the 6- to  18-years age
17   group (28.2% [95% CI: 15.3, 41.5%]; lag 0-1).  In two-pollutant models, O3 effect estimates for
18   both non-ICU and ICU hospital admissions remained robust to adjustment for PM2.5.  In an
19   additional analysis, using a smooth  function, the authors examined whether the shape of the
20   concentration-response curve for Os and asthma hospital admissions (i.e.,  both general and ICU
21   for all ages) is linear.  When comparing the curve to a linear fit line, the authors found that the
22   linear fit was a reasonable  approximation of the concentration-response relationship between O3
23   and asthma hospital admissions, but the limited data density at relatively low Os concentrations
24   contributes to uncertainty in the shape of the concentration-response relationship at the low end
25   of the distribution of Os concentrations (US EPA, 2012a, Figure 6-15).
26          In contrast to COPD and asthma, the evidence for pneumonia-related admissions was less
27   consistent. Medina-Ramon et al. (2006) examined the association between short-term ambient
28   Os concentrations and Medicare hospital admissions among individuals > 65 years of age for
29   pneumonia. The authors reported an increase in pneumonia hospital admissions in the warm
30   season (2.5% [95% CI: 1.6, 3.5%] for a 30 ppb increase in 8-h max Os concentrations, with no
31   evidence of an association in the cool season or year round.  In two-pollutant models restricted to
32   days for which PMio data was available, the association between Os exposure and pneumonia
33   hospital admissions remained robust. In contrast, Zanobetti  and Schwartz (2006) reported a
34   6.0% (95% CI: -11.1, -1.4%) decrease in pneumonia admissions for a 20 ppb increase in 24-h
35   average Os concentrations in Boston for the average of lags 0 and 1.

                                               2-22

-------
 1    Emergency department visits for all respiratory causes
 2          A large single-city study conducted in Atlanta by Tolbert et al. (2007), and subsequently
 3    reanalyzed by Darrow et al. (2011) using different air quality data and evaluating associations
 4    with different metrics, provides evidence for associations between short-term exposures to
 5    ambient 63 concentrations and respiratory emergency department visits. Tolbert et al. (2007)
 6    reported a 3.9% (95% CI: 2.7, 5.2%) increase in respiratory emergency department visits for a 30
 7    ppb increase in 8-h max O?, concentrations during the warm season. In copollutant models with
 8    CO, NC>2, and PMio, limited to days in which data for all pollutants were available, associations
 9    between Os and respiratory emergency department visits remained positive, but were attenuated.
10    Darrow et al. (2011) reported the strongest associations with respiratory emergency department
11    visits for 8-hour daily maximum, 1-hour daily maximum, and day-time O3 exposure metrics (all
12    associations positive and statistically significant), while positive, but statistically non-significant,
13    associations were reported with 24-hour average  and commuting period exposure metrics.  In
14    addition, a negative association was observed when using the night-time exposure metric (US
15    EPA, 2012a, Figure 6-16). The results of Darrow et al. (2011) suggest that averaging over
16    nighttime hours may lead to smaller 63 effect estimates for respiratory emergency department
17    visits due to dilution of relevant Os concentrations (i.e., the higher concentrations that occur
18    during the daytime); and potential negative confounding by other pollutants (e.g., CO, NO2)
19    during the nighttime hours (US EPA, 2012a, section 6.2.7.3).
20    Cause-specific emergency department visits
21          In evaluating cause-specific emergency department visits in an all-year analysis, a
22    Canadian multi-city study (Stieb et al., 2009) reported that 24-hour O3 concentrations were
23    positively associated with emergency department visits for asthma (4.7% [95% CI: -1.4, 11.1%]
24    at lag 1 and 3.5% [95% CI: 0.33, 6.8%] at lag 2). Though the authors did not present seasonal
25    analyses, they stated that no associations were observed with emergency department visits in the
26    winter season, suggesting that the positive associations reported in the all-year analysis were due
27    to the warm season (Stieb et al., 2009). In addition to asthma, the authors reported that 63 was
28    positively associated with COPD emergency department visits in all-year analyses, but that
29    associations with COPD visits were statistically significant only for the warm season (i.e., April-
30    September) 6.8% [95% CI: 0.11, 13.9%].
31          Several single-city studies have also provided evidence for positive associations between
32    asthma emergency department visits and ambient Os concentrations.  Ito et al. (2007) reported
33    positive  and statistically significant associations with asthma emergency department visits in
34    New York City during the warm season (percent increased risk ranged from 8.6 to 16.9% across
35    models that controlled for the potential confounding effects of weather using different
                                                2-23

-------
 1    approaches), and an inverse association in the cool season (ranging from -23.4 to -25.1%), for a
 2    30 ppb increase in 8-hour maximum O3 concentrations.  In two-pollutant models with PM2.s,
 3    NO2, SO2, and CO, the authors found that O3 risk estimates were not substantially changed
 4    during the warm season (US EPA, 2012a, Figure 6-19; Table 6-29).
 5          Strickland et al. (2010) examined the association between O3 exposure and pediatric
 6    asthma emergency department visits (ages 5-17 years) in Atlanta using air quality data over the
 7    same years as Darrow et al. (201 Ib) and Tolbert et al. (2007), but using population-weighting to
 8    combine daily pollutant concentrations across monitors.  Strickland et al. (2010) reported a 6.4%
 9    (95% CI: 3.2, 9.6%) increase in emergency department visits for a 30 ppb increase in 8-h max O3
10    concentrations in an all-year analysis.  In seasonal analyses, stronger associations were observed
11    during the warm season (i.e., May-October) (8.4% [95%  CI: 4.4, 12.7%]; lag 0-2) than the cold
12    season (4.5% [95% CI: -0.82, 10.0%]; lag 0-2).  In co-pollutant analyses that included CO, NO2,
13    PM2.5  elemental carbon, or PM2.5 sulfate, Strickland et al. (2010) reported that O3 risk estimates
14    were not substantially changed. The authors also examined the concentration-response
15    relationship between O3 exposure and pediatric asthma emergency  department visits and
16    reported that positive associations with O3 persist at 8-hour ambient O3 concentrations (3-day
17    average of 8-hour daily maximum concentrations) at least as low as 30 ppb.
18          In a single-city study conducted in Seattle, WA, Mar and Koenig (2009) examined the
19    association between O3 exposure and asthma emergency  department visits for children (< 18)
20    and adults (>  18).  For children, positive and statistically  significant associations were reported
21    across multiple lags, ranging from a 19.1-36.8% increase in asthma emergency department visits
22    for a 30 ppb increase in  8-hour maximum O3 concentrations, with the strongest associations
23    observed at lag 0 (33.1% [95% CI: 3.0, 68.5]) and lag 3 (36.8% [95% CI: 6.1, 77.2]). Ozone was
24    also found to be positively associated with asthma emergency department visits for adults at all
25    lags, ranging from 9.3-26.0%, except at lag 0. The slightly different lag times for children and
26    adults suggest that children may be more immediately responsive to O3 exposures than adults
27    (Mar and Koenig, 2009).
28          In addition to the U.S. single-city studies discussed above, a single-city study conducted
29    in Alberta, Canada (Villeneuve et al., 2007) provides support for the findings from Stieb et al.
30    (2009), but also attempts to identify those lifestages at greatest risk for O3-associated asthma
31    emergency department visits. Villeneuve et al. reported an increase in asthma emergency
32    department visits in an all-year analysis across all ages (12.0% [95% CI: 6.8, 17.2] for a 30 ppb
33    increase in max 8-h average O3 concentrations at lag 0-2) with associations being stronger during
34    the warmer months (19.0% [95% CI: 11.9, 28.1]). When stratified  by age, the strongest
35    associations were observed in the warm season for individuals 5-14 (28.1% [95% CI: 11.9, 45.1];

                                                2-24

-------
 1    lag 0-2) and 15-44 (19.0% [95% CI: 8.5, 31.8]; lag 0-2).  These associations were not found to
 2    be confounded by the inclusion of aeroallergens in age-specific models.
 3           2.2.1.7 Respiratory mortality
 4           The 2006 Os AQCD found inconsistent evidence for an association between short-term
 5    Os exposure and respiratory mortality (US EPA, 2006).  In contrast, recent multicity studies have
 6    consistently reported positive associations, particularly during the summer months (US EPA,
 7    2012a, Figure 6-37).  Specifically, Zanobetti and Schwartz (2008) reported a positive and
 8    statistically significant association with respiratory mortality across 48 U.S. cities in the summer
 9    [2.51% (95% CI 1.1.4 to 3.89) increase in respiratory mortality per 30 ppb increase in 8-hour
10    daily maximum Os].  Consistent with this study, the APHENA study reported positive
11    associations with respiratory mortality across cities in the U.S., Canada, and Europe, with larger
12    effect estimates in the warm season than in all-year analyses and with some effect estimates
13    statistically significant (US EPA, 2012a, Figure 6-36).13 In analyses  of potential co-pollutant
14    confounding, Katsouyanni et al. (2009) reported 63 respiratory mortality risk estimates were
15    robust to the inclusion of PMi0 in co-pollutant models in all year analyses  of the Canadian and
16    US datasets.  In summer-only analyses, Os respiratory mortality risk  estimates remained robust
17    to the inclusion of PMio in co-pollutant models using the US dataset, but were attenuated in
18    analyses of the European dataset. In addition to these U.S. studies, associations with respiratory
19    mortality have been reported to be positive, and in some cases statistically significant, in multi-
20    city studies conducted in Europe and Asia (Samoli et al., 2009; Stafoggia et al., 2010; Wong et
21    al., 2010).
22           2.2.1.8 Summary of respiratory effects
23           As discussed in detail in the ISA (US EPA, 2012a, section 2.5.2 and chapter 6), recent
24    studies build upon the strong body of evidence presented in the 1996 and 2006 Os AQCDs,
25    supporting the conclusion that there is a causal relationship between  short-term 63 exposure and
26    respiratory health effects. Recent controlled human exposure studies demonstrate statistically
27    significant group mean decreases in pulmonary function following exposures as low as 60 and 70
28    ppb 63 in young, healthy adults. These studies are supported by the  strong, cumulative evidence
29    from epidemiologic studies. Equally strong evidence demonstrated associations of ambient Os
30    with respiratory hospital admissions and emergency department visits across the U.S., Europe,
31    and Canada.  Several multicity studies and a multi-continent study reported associations between
      13As noted above (section 2.2.1.6), the study by Katsouyanni et al. (2009) evaluated different statistical models.
      Although the investigators did not identify the model they deemed to be the most appropriate for comparing the
      results across study locations, they did specify that "overall effect estimates (i.e., estimates pooled over several
      cities) tended to stabilize at high degrees of freedom" (Katsouyanni et al., 2009). In discussing of the results of this
      study, the ISA focused on models with 8 degrees of freedom per year (US EPA, 2012a, section 6.2.7.2).
                                                 2-25

-------
 1    short-term increases in ambient 63 concentrations and increases in respiratory mortality. This
 2    evidence is supported by a large body of individual-level epidemiologic panel studies that
 3    demonstrate associations of ambient Os with respiratory symptoms in children with asthma.
 4    Further support is provided by recent studies that found (Vassociated increases in indicators of
 5    airway inflammation and oxidative stress in children with asthma. Across respiratory endpoints,
 6    evidence indicates antioxidant capacity may modify the risk of respiratory morbidity associated
 7    with O3 exposure. The potentially elevated risk of populations with diminished antioxidant
 8    capacity and the reduced risk of populations with enhanced antioxidant capacity identified in
 9    epidemiologic studies is strongly supported by similar findings from controlled  human exposure
10    studies and by evidence that characterizes Os-induced decreases in intracellular  antioxidant
11    levels as a mode of action for downstream effects. By demonstrating Os-induced airway
12    hyperresponsiveness, decreased pulmonary function, allergic responses, lung injury, impaired
13    host defense, and airway inflammation, toxicological studies have characterized Os modes of
14    action and provided biological plausibility for epidemiologic associations of ambient 63
15    concentrations with lung function and respiratory symptoms, hospital admissions, emergency
16    department visits, and mortality. Together, the ISA  concludes that the evidence integrated across
17    controlled human exposure, epidemiologic, and toxicological studies and across the spectrum of
18    respiratory health endpoints continues to demonstrate that there is a causal relationship between
19    short-term O?, exposure and respiratory health effects (US EPA, 2012a, section 2.5.2).
20    2.2.2  Total Mortality
21          The 2006 Os AQCD reviewed a large number of time-series studies of associations
22    between short-term Os exposures and total mortality including single- and multicity studies, and
23    meta-analyses. In the large U.S. multicity studies that examined all-year data, effect estimates
24    corresponding to single-day lags ranged from a 0.5-1% increase in all-cause (nonaccidental) total
25    mortality per a 20 ppb (24-hour), 30 ppb (8-hour maximum), or 40 ppb (1-hour  maximum)
26    increase in ambient O3 (US EPA, 2012a, section 6.6.1). Available studies reported some
27    evidence for heterogeneity in Os mortality risk estimates across cities and across studies.  Studies
28    that conducted seasonal analyses reported larger 63  mortality risk estimates during the warm
29    season. Overall, the 2006 Os AQCD identified robust associations between various measures of
30    daily ambient Os concentrations and all-cause mortality, which could not be readily explained by
31    confounding due to time, weather, or copollutants. With regard to cause-specific mortality,
32    consistent positive associations were reported between short-term Os exposure and
33    cardiovascular mortality, with less consistent evidence for associations with respiratory
34    mortality.  The majority of the evidence for associations between O3 and cause-specific mortality
35    were from single-city studies, which had small daily mortality counts and subsequently limited

                                                2-26

-------
 1    statistical power to detect associations.  The 2006 63 AQCD concluded that "the overall body of
 2    evidence is highly suggestive that Os directly or indirectly contributes to non-accidental and
 3    cardiopulmonary-related mortality" (US EPA, 2012a, section 6.6.1).
 4          Recent studies have further confirmed the association between short-term Os
 5    concentrations and mortality, including a number of studies reporting associations with non-
 6    accidental as well as cause-specific mortality. Multi-continent and multicity studies have
 7    consistently reported positive and statistically significant associations between short-term Os
 8    concentrations and all-cause  mortality, with evidence for larger mortality risk estimates during
 9    the warm or summer months (Figure 2-1 below, reprinted from the ISA) (US EPA, 2012a, Figure
10    6-26; Table 6-42).  Similarly, evaluations of cause-specific mortality have reported consistently
11    positive associations with O3, particularly in analyses restricted to the warm season (US EPA,
12    2012a, Figure 6-36; Table 6-53).14
      14Respiratory mortality is discussed in more detail in section 2.2.1.6, above.
                                                 2-27

-------
Figure 2-1.  Summary of mortality risk estimates for short-term 63 and all-cause (nonaccidental) mortality
                                                                                                               15
 Study

 Grypariset al. (2004;57276)
 Bell etal. (2007; 93256)
 Schwartz (2005; 57333)
 Bel I and Dominici (2008; 193828)
 Bell etal. (2004; 94417)a
 Levyetal.(2005;74347)a
 Katsouyanni et al. (2009; 199899)
 Bell etal. (2005; 74345)a
 Ito etal. (2005; 74346)a
 Wong etal. (2010; 732535)
 Katsouyanni et al. (2009; 199899)
 Cakmak et al. (2011, 699135)
 Katsouyanni et al. (2009; 199899)
 Katsouyanni et al. (2009; 199899)b

 Samoli etal. (2009; 195855)
 Bell etal. (2004; 94417)a
 Schwartz (2005; 57333)
 Zanobetti and Schwartz (2008; 195755
 Zanobetti and Schwartz (2008; 101596
 Franklin and Schwartz (2008; 156448)
 Grypariset al. (2004;57276)
 Medina-Ramon and Schwartz (2008)
 Katsouyanni et al. (2009; 199899)
 Bell etal. (2005; 74345)a
 Katsouyanni et al. (2009; 199899)
 Katsouyanni et al. (2009; 199899)b
 Levyetal.(2005;74347)a
 Ito etal. (2005; 74346)a
 Katsouyanni et al. (2009; 199899)
 Stafoggia et al. (2010; 625034)
     Location

 APHEA2 (23 cities)
98 U.S. communities
    14 U.S. cities
98 U.S. communities
95 U.S. communities
 U.S. and  Non-U.S.
  APHENA-Europe
 U.S. and  Non-U.S.
 U.S. and  Non-U.S.
   PAPA (4 cities
    APHENA-U.S.
   7 Chilean cities
  APHENA-Canada
  APHENA-Canada

 21 European cities
95 U.S. communities
    14 U.S. cities
    48 U.S. cities
    48 U.S. cities
18 U.S. communities
 APHEA2 (21 cities)
    48 U.S. cities
  APHENA-Europe
 U.S. and  Non-U.S.
  APHENA-Canada
  APHENA-Canada
 U.S. and  Non-U.S.
 U.S. and  Non-U.S.
    APHENA-U.S.
   10 Italian cities
  Lag

  0-1
  0-1
   0
  0-6
  0-6

DL(0-2)
                                                                    All-Year
DL
DL
DL
DL
  0-1
0-2)
0-6
0-2
0-2
  0-1
  0-6
   0
   0
  0-3
   0
  0-1
  0-2
DL(0-2)

DL(0-2
DLO-2
DL(0-2
DLO-5
                                                                       Summer
                                                                          -1
                                                                                                          11
15Figure 2-1 is reprinted from the ISA (US EPA, 2012a, Figure 6-26).
                                                                                                        % Increase
                                                                    2-28

-------
 1           In assessing the evidence for (Vrelated mortality, the 2006 AQCD also noted that
 2    multiple uncertainties remained regarding the relationship between Os and mortality, including
 3    the extent of residual confounding by co-pollutants; characterization of the factors that modify
 4    the (Vmortality association; the appropriate lag structure for identifying (^-mortality effects;
 5    and the shape of the Os-mortality concentration-response function and whether a threshold
 6    exists.  Many of the studies published since the last review have attempted to address one or
 7    more of these uncertainties. The following sections discuss the extent to which recent studies
 8    have evaluated these uncertainties in the relationship between Os and mortality.
 9    Confounding by co-pollutants
10           Recent epidemiologic studies have examined potential confounders of the Os-mortality
11    relationship, with a focus on PM and its constituents. However, because of the temporal
12    correlation among these PM components and 63, and their possible interactions, the
13    interpretation of results from co-pollutant models that attempt to disentangle the health effects
14    associated with each pollutant is challenging. Further complicating the interpretation of
15    copollutant results, in some cases, is the every-3rd or -6th day PM sampling schedule employed in
16    most locations, which limits the number of days where both PM and Os data are available.
17           Katsouyanni et al. (2009) investigated the influence of PMio on Os-mortality estimates
18    among the three APHENA datasets (i.e., US, Canada, Europe). The  sensitivity of 63 effect
19    estimates varied across the datasets and age groups. In the U.S. dataset, O3 risk estimates for all-
20    cause mortality remained positive, but were reduced, in both the year-round and summer-only
21    analyses (Figure 6-29; Table 6-45). Risk estimates for cause-specific mortality were more
22    variable in co-pollutant models in the U.S. dataset,  with risk estimates for respiratory mortality
23    larger in co-pollutant models than in single pollutant models and risk estimates for
24    cardiovascular mortality both larger and smaller than in single pollutant models, depending on
25    the age groups being evaluated (Figure 6-29; Table 6-45).  In the Canadian dataset, Os risk
26    estimates for all-cause mortality remained positive, though they were modestly reduced, when
27    adjusted for PMi0.  Variable results were reported for cause-specific  mortality, with 63 risk
28    estimates reduced for cardiovascular mortality and  increased for respiratory mortality in co-
29    pollutant models with PMio (US EPA, 2012a, Figure 6-29; Table 6-45).  In the European  dataset,
30    O3 risk estimates for total mortality and cause-specific mortality generally remained robust in co-
31    pollutant models that included PMio in the year-round analyses.  When analyses were restricted
32    to the summer months, moderate reductions were observed in 63 risk estimates.
33           Bell et al. (2007) used data on 98 U.S. urban communities from the NMMAPS study to
34    evaluate the potential for confounding effects of PMio and PM2.s on the O3-mortality
35    relationship.  An examination of the correlation between PM (PMio and PM2.s) and Os across
                                                2-29

-------
 1    various strata of daily PMio and PM2.5 concentrations found that neither PM size fraction was
 2    highly correlated with daily, 8-hour maximum, or 1-hour maximum Os concentrations. National
 3    and community-specific effect estimates of the association between short-term Os exposure and
 4    mortality were robust to inclusion of PMio or PM2.5 in time-series models, though Os risk
 5    estimates were not statistically significant in co-pollutant models, likely due to the limited
 6    number of days with both 63 and PM data available (i.e., Bell et al. (2007) reported that only
 7    9.2% of days had both O3 and PM2.5 data). Smith et al. (2009) re-analyzed the publicly available
 8    NMMAPS database used in Bell et al. (2007).  In analyses conducted to examine the potential
 9    confounding effects of PMio, the authors reported that, in most cases, 63 mortality risk estimates
10    were reduced by between 22% and 33% in co-pollutant models.
11          Franklin and Schwartz (2008) examined the sensitivity of Oj mortality risk estimates to
12    the inclusion of PM2.5 or PM chemical components associated with secondary aerosols.  The
13    association between 63 and non-accidental mortality was examined in single-pollutant models
14    and after adjustment for PM2.5, sulfate, organic carbon, or nitrate concentrations. In the single-
15    pollutant model, the authors found a 0.89% (95% CI: 0.45, 1.33%) increase in nonaccidental
16    mortality with a 10 ppb increase in same-day 24-h summertime 63 concentrations across the 18
17    U.S. communities.  Adjustment for PM2.5 mass, which was available for 84% of the days,
18    decreased the CVmortality  risk estimate only slightly (e.g., from 0.88% to 0.79% for PM2.5
19    mass). Similar results were reported for nitrate. In contrast, the inclusion of sulfate in the model
20    reduced the Os risk estimate by 31%. However, this could be attributed to  only 18% of days
21    having both 63 and sulfate  data.
22    Effect modification
23          Several multicity studies have reported that (Vrelated mortality risk estimates vary
24    regionally across the United States.  For example, Bell and Dominici (2008) reported that
25    (Vmortality risk estimates were larger in the Northeast (1.44% [95% CI: 0.78, 2.10%]) and
26    Industrial Midwest (0.73% [95% CI: 0.11, 1.35%]) than in other regions of the country, where
27    null associations were found (US EPA, 2012a, Table 6-49; Figure 6-31). Similar regional
28    variability was reported by Smith et al. (2009) (US EPA, 2012a, Figure 6-32). In addition,
29    Franklin and Schwartz (2008) indicated that between-city heterogeneity in  the effect of 63
30    exposure on mortality may contribute to greater uncertainty in assessing this association than co-
31    pollutant confounding. In light of this regional heterogeneity, multicity studies have evaluated a
32    number of factors that may modify the (Vmortality relationship and explain the observed
33    regional heterogeneity. These potential effect modifiers can be categorized as either individual-
34    level or community-level characteristics.
                                                2-30

-------
 1          Of the individual-level characteristics that have been evaluated, the most consistent
 2    evidence for effect modification has been for older adults. Medina-Ramon and Schwartz (2008)
 3    evaluated the warm season in 48 U.S. cities to identify populations potentially at greatest risk for
 4    (Vrelated mortality.  Across cities, the authors reported a 1.96% (95% CI: 1.14-2.82%) increase
 5    in mortality (lag 0-2) for a 30 ppb increase in 8-hour maximum Os concentrations. Older adults
 6    (i.e., > 65 years of age) were at increased risk for Os-related mortality compared to younger
 7    individuals [i.e., older adults had an additional 1.10% (95% CI: 0.44, 1.77) increase in O3-related
 8    mortality] (US EPA, 2012a, Table 6-47). Other multeity studies conducted outside the U.S.
 9    have reported similar results for individuals 85 years or older (Stafoggia et al., 2010) and for
10    individuals older than 75 years (Cakmak et al., 2011).
11          Other individual-level characteristics have also been reported to modify the O3-mortality
12    relationship, though these factors have typically either not been examined across multiple studies
13    or different studies have reported inconsistent results. For example, Medina-Ramon et al. (2008)
14    reported larger Os-associated mortality risks in women > 60 years of age compared to men, in
15    black individuals compared to non-black individuals, and in individuals with atrial fibrillation
16    compared to individuals without atrial fibrillation (US EPA, 2012a, Table 6-47).  Cakmak et al.
17    (2011) reported the largest Os-related mortality risks in males and in individuals at the low end
18    of the socioeconomic spectrum. In addition, studies have not consistently demonstrated that
19    individuals with diabetes are potentially at increased risk for O3 associated mortality (Schwartz,
20    2005; Medina-Ramon et al., 2008, Stafoggia et al., 2010).
21          Several studies have also examined community-level variables in an attempt to explain
22    the observed city-to-city variation in estimated Os-mortality risk estimates. Bell and Dominici
23    (2008) analyzed 98 U.S. urban communities from NMMAPS for the period 1987-2000. In the
24    all-year regression model that included no community-level variables, a 20 ppb increase in 24-h
25    average Os concentrations during the previous week was associated with a 1.04% (95% CI: 0.56,
26    1.55) increase in mortality.  The authors reported that larger O3-mortality effect estimates were
27    associated with higher percent unemployment, fraction of the population Black/African-
28    American, and percent of the population that take public transportation to work. In addition, the
29    authors reported larger Os mortality effect estimates in locations with lower temperatures and
30    with lower prevalence of central air conditioning (US EPA, 2012a, Figure 6-30).  In the
31    APHENA study, Katsouyanni  et al. (2009) reported generally inconsistent results across the
32    U.S., Canadian, and European datasets  for potential effect modifiers, though larger Os mortality
33    risks were reported for cities with higher unemployment and lower temperatures (a surrogate for
34    lower prevalence of air conditioning) in the U.S. dataset (US EPA, 2012a, Table 6-48).
                                                2-31

-------
 1          Ren et al. (2008) examined the possible interaction between 63 and temperature during
 2    the warm months in the 60 largest eastern U.S. communities from the NMMAPS dataset. In the
 3    northeast region, a 20 ppb increase in 24-h avg Os concentrations at lag 0-2 was associated with
 4    an increase in mortality of 4.49% (95% posterior interval [PI]: 2.39, 6.36%), 6.21% (95% PI:
 5    4.47, 7.66%) and 12.8% (95% PI: 9.77, 15.7%) for low, moderate, and high temperature levels,
 6    respectively. The corresponding percent increases in mortality in the southeast region were
 7    2.27% (95% PI: -2.23, 6.46%) for low temperature, 3.02% (95% PI: 0.44, 5.70%) for moderate
 8    temperature, and 2.60% (95% PI: -0.66, 6.01%) for high temperature. This observed difference
 9    may in part reflect the higher air conditioning prevalence in communities with higher long-term
10    average temperatures.  Therefore, the findings from Ren et al. (2008)  indicating generally lower
11    Os risk estimates in the southeast region, where the  average temperature is higher, than in the
12    northeast region is consistent with the regional results reported by Bell and Dominici (2008).
13    Lag structure
14          Several studies have evaluated different lag  structures in order to assess the possibility
15    that O3 mortality risks reported for short lags could  result from small shifts in the day of death
16    for individuals who are already frail, such that those frail individuals die slightly earlier than
17    would otherwise have been the case (mortality displacement). Zanobetti and Schwartz (2008a)
18    examined this issue in 48 U.S. cities during the warm  season (i.e., June-August) for the years
19    1989-2000. The authors reported a 0.96% (95% CI: 0.60, 1.30%) increase in all-cause mortality
20    across all 48 cities for a 30 ppb increase in 8-hour maximum 63 concentrations at lag 0, whereas
21    a combined estimate using an unconstrained distributed lag model (lag 0-20) was 1.54% (95%
22    CI: 0.15,  2.91%).  Similarly, when examining the cause-specific mortality results, larger risk
23    estimates were observed for the distributed lag model  compared to estimates based on a 0 day lag
24    (US EPA, 2012a, Table 6-50). Samoli et al. (2009) conducted a similar analysis in 21 European
25    cities. The authors reported that the 21 day distributed lag model resulted in (Vrelated mortality
26    estimates that were approximately 10-fold larger than estimates from the 0 lag model, suggesting
27    that using single-day exposures may underestimate  (Vrelated respiratory mortality.  In contrast,
28    the 21-day distributed lag models yielded non-significant negative estimates of (Vrelated total
29    mortality and cardiovascular mortality (US EPA, 2012a, Table 6-51).
30          Overall, the evidence for mortality displacement remains mixed, with inconsistent results
31    across studies and for total mortality and cardiovascular mortality versus respiratory mortality.
32    Interpretation of these studies is also complicated by the finding that (Vrelated mortality effect
33    estimates are reduced late in the summer (i.e., 0.84% in August) compared to earlier in the
34    summer (1.96% in July), potentially supporting the  existence of an adaptive response for CV
35    related mortality (Zanobetti and Schwartz, 2008b).  Such an adaptive  response may complicate

                                                2-32

-------
 1    the interpretation of the distributed lag coefficients with long lag periods because the decreased
 2    coefficients may reflect diminished effects of the late summer, rather than mortality
 3    displacement.  Although interpretation of these studies is complicated, both Samoli et al. (2009)
 4    and Zanobetti and Schwartz (2008a), as well as other studies that evaluated multiple lags (e.g.,
 5    Katsouyanni et al., 2009), suggest that the positive associations between Os and mortality are
 6    observed mainly in the first few days after exposure.
 7    Concentration-response relationship
 8           Recent studies have evaluated different statistical  approaches to examine the shape of the
 9    (Vmortality concentration-response relationship and to evaluate whether a threshold exists for
10    CVrelated mortality. In an analysis of the NMMAPS data, Bell et al. (2006) evaluated the
11    potential for a threshold in the (Vmortality relationship.  The authors reported positive and
12    statistically significant associations with mortality in a variety of restricted analyses, including
13    analyses restricted to days with 24-hour average Os concentrations below 60, 55, 50, 45, 40,  35,
14    and 30 ppb. In these restricted analyses 63 effect estimates were of similar magnitude, were
15    statistically significant, and had similar statistical precision.  In analyses restricted to days with
16    24-hour average Os concentrations below 25 ppb, the Os effect estimate was similar in
17    magnitude  to the effect estimates resulting from analyses with the higher cutoffs, but had
18    somewhat lower statistical precision, with the estimate approaching statistical significance (i.e.,
19    based on observation of Figure 2 in Bell et al.,  2006).  In  analyses restricted to days with lower
20    24-hour average 63 concentrations (i.e., below 20 and 15 ppb), effect estimates were similar in
21    magnitude  to analyses with higher cutoffs, but with notably less statistical precision, and were
22    not statistically significant (i.e., confidence intervals included no (^-associated mortality based
23    on observation  of Figure 2 in Bell et al., 2006). Ozone was no longer positively associated with
24    mortality when the analysis was restricted to days with 24-hour Os concentrations below 10 ppb.
25    Given the relatively small number of days included in these restricted analyses, especially for cut
26    points of 20 ppb and below,16 statistical uncertainty is increased.
27           Bell et al. (2006) also evaluated the shape of the concentration-response relationship
28    between Os and mortality. Although the results of this analysis suggested the lack of threshold
29    in the (Vmortality relationship, the ISA noted that it is difficult to interpret such a curve
30    because: (1) there is uncertainty around the shape of the concentration-response curve at 24-h
31    average Os concentrations generally below 20  ppb  and (2) the concentration-response curve does
32    not take into consideration the heterogeneity in (Vmortality risk estimates across cities (US
33    EPA, 2012a, section 6.6.2.4).
      16For example, Bell et al. (2006) reported that for analyses restricted to 24-hour O3 concentrations at or below 20
      ppb, 73% of days were excluded on average across the 98 communities.
                                                 2-33

-------
 1           Several additional studies have used the NMMAPS dataset to evaluate the (Vmortality
 2    concentration-response relationship.  For example, using the same data as Bell et al. (2006),
 3    Smith et al. (2009) conducted a subset analysis, but instead of restricting the analysis to days
 4    with 63 concentrations below a cutoff the authors only included days above a defined cutoff.
 5    The results of this analysis were consistent with those reported by Bell et al.  (2006).
 6    Specifically, the authors reported consistent positive associations for all cutoff concentrations up
 7    to concentrations where the total number of days available were so limited that the variability
 8    around the central estimate was increased (US EPA, 2012a, section 6.6.2.4).  In addition, using
 9    NMMAPS data for 1987-1994 for Chicago, Pittsburgh, and El Paso, Xia and long (2006)
10    reported evidence for a threshold around a 24-hour average Os concentration of 25 ppb, though
11    the threshold values estimated in the analysis were sometimes in the range of where data density
12    was low (US EPA, 2012a, section 6.6.2.4).  Stylianou and Nicolich (2009) examined the
13    existence of thresholds following an approach similar to Xia and Tong (2006) using data from
14    NMMAPS for nine major U.S. cities (i.e., Baltimore, Chicago, Dallas/Fort Worth, Los Angeles,
15    Miami, New York, Philadelphia, Pittsburgh, and  Seattle) for the years 1987-2000. The authors
16    reported that the estimated Os-mortality risks varied across the nine cities, with the models
17    exhibiting apparent thresholds in the 10-45 ppb range for 63. However, given the city-to-city
18    variation in risk estimates, combining the city-specific estimates into an overall estimate
19    complicated the interpretation of the results.  Additional studies in Europe, Canada, and Asia did
20    not report the existence of a threshold (Katsouyanni  et al., 2009), with inconsistent and/or
21    inconclusive results across cities, or a non-linear  relationship in the Os-mortality concentration-
22    response curve (Wong et al., 2010).
23           In light of the above evidence, the ISA concluded that Os mortality studies do not
24    provide evidence for the existence of a threshold  within the range of 24-hour Os concentrations
25    most commonly observed in the U.S. during the Os season (US EPA, 2012a, section 2.5.4.4).
26    Although recent evidence suggests that the shape of the Os-mortality relationship remains linear
27    across the this range of 63 concentrations, evidence indicates less certainty in the shape of the
28    concentration-response curve at the lower end of the distribution of Os concentrations, and city-
29    to-city heterogeneity complicates the interpretation of a combined concentration-response curves
30    (US EPA, 2012a, section 2.5.4.4).  Overall, recent evidence continues to support the conclusion
31    from the 2006 Os AQCD, which stated that "if a population threshold level exists in Os health
32    effects, it is likely near the lower limit of ambient 63 concentrations in the United States" (US
33    EPA, 2012a, p. 6-266).
                                                2-34

-------
 1    2.2.3  Other Effects
 2          In contrast to the extensive bodies of evidence linking short-term 63 exposures to
 3    respiratory effects and to total mortality (discussed above), more limited evidence links short-
 4    term Os exposures to other health endpoints. Specifically, in recognition of the limitations and
 5    uncertainties in the health evidence, the ISA concluded that the evidence is suggestive of a causal
 6    relationship with cardiovascular effects and CNS effects. In addition, the evidence was judged
 7    inadequate to determine if a causal relationship exists between ambient Os exposures and cancer
 8    (US EPA, 2012a, section 2.5.2). The health evidence supporting these conclusions is discussed
 9    in detail in the ISA (US EPA, 2012a) and is summarized briefly below for cardiovascular effects
10    (2.2.3.1) and CNS effects (2.2.3.2).
11          2.2.3.1 Cardiovascular effects
12          In past Os AQCDs the effects of short-term exposure to Os on the cardiovascular system
13    could not be thoroughly evaluated due to limitations in the evidence available.  However, some
14    recent experimental and epidemiologic studies have investigated Os-related cardiovascular
15    events (US EPA, 2012a, summarized in section 2.5.2).
16          Animal toxicological studies provide evidence that short-term Os exposure can lead to
17    cardiovascular morbidity. These studies have reported (Vinduced atherosclerosis and
18    ischemia/reperfusion injury, in some cases in conjunction with a systemic oxidative, pro-
19    inflammatory environment;  disruption of NO-induced vascular reactivity; decrease in cardiac
20    function; and increase in heart rate variability  (HRV).  The observed increase in HRV is
21    supported by a recent controlled human exposure study that reported increased high frequency
22    HRV, but not altered blood pressure, following Os exposure.  The mechanism by which Os
23    inhalation may cause systemic toxicity remains unclear, though the cardiovascular effects of 63
24    found in animals are consistent with the development of an extra-pulmonary oxidative,
25    proinflammatory environment that may result from pulmonary inflammation.
26          There is only limited and inconsistent evidence for Os-related cardiovascular morbidity in
27    epidemiologic studies examining short-term exposures to 63. This is highlighted by the multiple
28    studies that examined the association between short-term increases in ambient Os concentrations
29    and cardiovascular-related hospital admissions and emergency department visits and other
30    various cardiovascular effects and found no evidence of a consistent relationship with O3
31    exposure. Positive associations between short-term increases in Os concentration and
32    cardiovascular mortality have been consistently reported in multiple epidemiologic studies.
33    However, the lack of coherence between the results from studies that examined associations
34    between short-term increases in Os concentration and cardiovascular morbidity and subsequently

                                                2-35

-------
 1    cardiovascular mortality, complicate the interpretation of the evidence for (Vinduced
 2    cardiovascular mortality.

 3          Overall, animal toxicological studies provide some evidence for Os-induced
 4    cardiovascular effects, but the effects observed were not consistently supported by controlled
 5    human exposure studies or epidemiologic studies. Although the toxicological evidence provides
 6    initial support to the relatively strong body of evidence indicating Os-induced cardiovascular
 7    mortality, there is a lack of coherence with controlled human exposure and epidemiologic studies
 8    of cardiovascular morbidity which together do not support Os-induced cardiovascular effects.
 9    Thus, the ISA concludes that the overall body of evidence across disciplines is suggestive of a
10    causal relationship for both relevant short- and long-term exposures to 63 and cardiovascular
11    effects (US EPA, 2012a, section 2.5.2).

12          2.2.3.2 CNS effects
13          Recent evidence suggests that Os may impart health effects through biological
14    mechanisms not previously considered.  For example, recent toxicological studies add to earlier
15    evidence that short-term exposures to O3 can produce a range of effects on the central nervous
16    system and behavior.  Additionally, an epidemiologic study demonstrated that long-term
17    exposure to 63 affects memory in humans as well (US EPA, 2012a, section 2.5.2).

18    2.3    HEALTH EFFECTS FOLLOWING LONG-TERM EXPOSURES TO O3

19          Given the weight-of-evidence conclusions in the ISA, staffs consideration of health
20    effects linked to long-term 63 exposures focuses on respiratory effects (section 2.3.1).  Other
21    effects linked to long-term Os exposures, including cardiovascular, reproductive, and CNS
22    effects are also considered (section 2.3.2).  In discussing the evidence in this draft PA, although
23    we consider the total body of available evidence of effects of long-term exposures to 63, we
24    focus the majority of our discussion on the studies that are most likely to inform policy decisions
25    regarding the adequacy of the current standard and potential alternative standards with respect to
26    long-term exposures to O3, as discussed  in chapter 4.
27          Long-term exposure has been defined in the ISA (EPA, 2012a) to include durations of
28    approximately 30 days (1 month) or longer. Epidemiologic  studies generally present Os-related
29    effect estimates for mortality and morbidity health outcomes based on an incremental change in
30    an exposure period. For example, studies traditionally present the relative risk per an
31    incremental change equal to the interquartile range in Os concentrations  or some other arbitrary
32    value (e.g., 10 ppb). Additionally, various short-term exposure metrics are used in 63
33    epidemiologic studies, with the three most common being the maximum 1-hour average within a

                                               2-36

-------
 1    24-hour period (1-hour maximum), the maximum 8-hour average within a 24-hour period (8-
 2    hour maximum), and 24-hour average (24-hour average). For the purpose of presenting results
 3    from short-term studies that use different exposure metrics, the ISA (EPA, 2012a) consistently
 4    applies the same 63 increments to facilitate comparisons between the results of various studies
 5    that may present results for different incremental changes.  Differences due to the use of varying
 6    exposure metrics (e.g., 1-hour maximum, 24-hour average) become less apparent when averaged
 7    across longer exposure periods, because levels are typically lower and less variable. As such,
 8    throughout the ISA and hence in this chapter, an increment of 10 ppb was consistently applied
 9    across long-term exposure studies, regardless of exposure metric, to facilitate comparisons
10    between the results from these studies.
11    2.3.1   Respiratory Effects
12           The ISA concludes that "the clearest evidence for health effects associated with exposure
13    to Os is provided by studies of respiratory effects" (US EPA, 2012a, section 1, p. 1-5).
14    Collectively, there is a vast amount of evidence spanning several decades that supports a causal
15    association between exposure to O3 and a continuum of respiratory effects (US EPA, 2012a,
16    section 2.5).  While the majority of this evidence is derived from studies investigating short-term
17    exposures, evidence from animal toxicological studies and recent epidemiologic evidence
18    demonstrate that long-term exposures (i.e., months to years) may also be detrimental to the
19    respiratory system. In this section, we revisit the overarching question of this chapter, as it
20    relates to respiratory effects following long-term 63 exposures.
21           In the 2006 Os  AQCD, evidence was examined for relationships between long-term  Os
22    exposure (several months to yearly) and effects on respiratory health outcomes including
23    declines in lung function, increases in inflammation, and development of asthma in children and
24    adults.  Animal toxicology data provided a clearer picture indicating that long-term 63 exposure
25    may have lasting effects.  Chronic17 exposure studies in animals have reported biochemical  and
26    morphological changes suggestive of irreversible long-term O?, impacts on the lung. In contrast
27    to supportive evidence from chronic animal studies, the epidemiologic studies on longer-term
28    (annual) lung function declines, inflammation, and new asthma development remained
29    inconclusive.
30           Several epidemiologic studies collectively indicated that Os exposure averaged over
31    several summer months was associated with smaller increases in lung function growth in
32    children.  For longer averaging periods (annual), the analysis in the Children's Health Study
      17 Unless otherwise specified, the term "chronic" generally refers to an annual exposure duration for epidemiology
      studies and a duration of greater than 10% of the lifespan of the animal in toxicological studies.
                                                2-37

-------
 1    (CHS) reported by Gauderman et al. (2004) provided little evidence that such long-term
 2    exposure to ambient Os was associated with significant deficits in the growth rate of lung
 3    function in children. Limited epidemiologic research examined the relationship between long-
 4    term 63 exposures and inflammation.  Cross-sectional studies detected no associations between
 5    long-term Os exposures and asthma prevalence, asthma-related symptoms or allergy to common
 6    aeroallergens in children. However, longitudinal studies provided evidence that long-term 63
 7    exposure influences the risk of asthma development in children and adults.
 8          The currently available body of evidence supporting a likely causal relationship between
 9    long-term Os exposures and adverse respiratory health effects is discussed in detail in the ISA
10    (EPA 2012a, section 7.2). New evidence reports interactions between genetic variants and long-
11    term Oj exposure in effects on new-onset asthma in U.S. cohorts in multi-community studies
12    where protection by specific oxidant gene variants was restricted to children living in low Os
13    communities.  A new line of evidence reports a positive concentration-response relationship
14    between first asthma hospitalization and long-term Os exposure.  Related studies report coherent
15    relationships between asthma severity and control, and respiratory symptoms among asthmatics
16    and long-term 63 exposure. These studies are summarized briefly below for new-onset asthma
17    (section 2.3.1.1), asthma hospital admissions and other morbidity effects (section 2.3.1.2),
18    pulmonary structure and function (section 2.3.1.3), and respiratory mortality (section 2.3.1.4).
19          2.3.1.1 New-onset asthma
20          Asthma is a heterogeneous disease with a high degree of temporal variability. The
21    progression and symptoms can vary within an individual's lifetime, and the course of asthma
22    may vary markedly in young children, older children, adolescents and adults. In the previous
23    review, longitudinal cohort studies that examined associations between long-term Os exposures
24    and the onset of asthma in adults and children indicated a direct effect of long-term 63 exposures
25    on asthma risk in adults and effect modification by Os in children. Since that review, important
26    new evidence has become available about the association between long-term exposures to Os and
27    new-onset asthma that has increased our understanding of the gene-environment interaction and
28    the mechanisms and biological pathways most relevant to assessing Os-related effects.
29          Associations between  annual mean Os  exposure and new cases of asthma were reported
30    in a cohort of non-smoking adults in California (McDonnell et al., 1999a, 15-year follow-up;
31    Greer et al., 1993, 10-year follow-up). Both the  10- and 15-year follow-up studies reported a
32    positive association between new-onset asthma and Os, but only in males. No other pollutants
33    were associated with the development of asthma in either males or females, and adjusting for
34    copollutants did not diminish  the association between O3 and asthma incidence in males. The
35    consistency of the results of the two studies with different follow-up times, as well as the
                                                2-38

-------
 1    independent and robust association between annual mean 63 concentrations and asthma
 2    incidence provide supportive evidence that long-term Os exposure may be associated with the
 3    development of asthma in adult males.
 4          In children, the relationship between long-term Os exposure and new-onset asthma has
 5    been extensively studied in the CHS; a long-term study that was initiated in the early 1990's and
 6    has evaluated effects in several cohorts of children.  The CHS was initially designed to examine
 7    whether long-term exposure to ambient pollution was related to chronic respiratory outcomes in
 8    children in 12 communities in southern California. In the CHS, new-onset asthma was classified
 9    as having no prior history of asthma at study entry with subsequent report of physician-
10    diagnosed asthma at follow-up, with the date of onset assigned to be the midpoint of the interval
11    between the interview date when asthma diagnosis was first reported and the previous interview
12    date.
13          The results of one study (McConnell et al., 2002) available in the previous review
14    indicated that within high Os communities, asthma risk was 3.3 times greater for children who
15    played three or more sports as compared with children who played no sports.  There was no
16    evidence of an association in low-Os communities. Communities were stratified by 4-year
17    average 1-hour maximum Os levels, with six high-Os communities (mean 75.4 ppb) and six low-
18    Os communities (mean 50.1  ppb). Analyses aimed at distinguishing the effects of Os from
19    effects of other pollutants indicated that in communities with high Os and low levels of other
20    pollutants there was a 4.2-fold increased risk of asthma in children playing three or more sports,
21    compared to children who played no sports. These results provide additional support that the
22    effects of physical activity on asthma are modified by long-term Os exposure. Overall, the
23    results from McConnell et al. (2002) suggest that playing sports may indicate greater outdoor
24    activity when Os levels are higher and an increased ventilation rate, which may lead to increased
25    63 exposure.
26          For this review, as discussed in section 7.2.2.1 of the ISA (US EPA 2012a), recent studies
27    from the CHS provide evidence for gene-environment interactions in effects on new-onset
28    asthma by indicating that the lower risks associated with specific genetic variants are found in
29    children who live in lower Os communities. Risk for new-onset asthma is related in part to
30    genetic susceptibility, as well as behavioral factors and environmental exposure.  The onset of a
31    chronic disease, such as asthma, is partially the result of a sequence of biochemical reactions
32    involving exposures to various environmental agents metabolized by enzymes related to a
33    number of different genes. Oxidative stress has been proposed to underlie these mechanistic
34    hypotheses. Understanding the relation between genetic polymorphisms and environmental
                                               2-39

-------
 1    exposure can help identify high-risk subgroups in the population and provide better insight into
 2    pathway mechanisms for these complex diseases.
 3          The CHS analyses have found that asthma risk is related to interactions between O3 and
 4    variants in genes for enzymes such as heme-oxygenase (HO-1), arginases (ARG1 and 2), and
 5    glutathione S transferase PI (GSTP1). Biological plausibility for these findings is provided by
 6    evidence that these enzymes have antioxidant and/or anti-inflammatory activity and participate
 7    in well-recognized modes of action in asthma pathogenesis. Further, several lines of evidence
 8    demonstrate that secondary oxidation products of 63 initiate the key modes of action that
 9    mediate downstream health effects (ISA, Section 5.3.2, US EPA,  2012a).
10          One study (Islam et al., 2008) found that functional polymorphisms of the heme
11    oxygenase-1 gene (HMOX-1) influenced the risk of new-onset asthma, depending on ethnicity
12    and long-term community 63 concentrations. Analyses were restricted to children of Hispanic18
13    or non-Hispanic white ethnicity and were conducted with long-term pollutant levels averaged
14    from 1994 to 2003.  For HMOX-1, the interaction indicated a greater protective effect of one
15    allele among non-Hispanic white children who lived in the low-Os community (shown in Figure
16    7-1, p. 7-8, ISA, EPA, 2012a).  Among children residing in low-Os communities (community
17    mean 63 level 38.4 ppb), the hazard ratio (HR) of new-onset asthma associated with this allele
18    was significantly reduced compared to non-Hispanic white children who lived in low-Os
19    communities without it. Biological plausibility for these results is provided by evidence that the
20    expression of the protective allele is more readily induced than the other. However, this allele
21    was found to have a less protective effect in non-Hispanic white children who resided in high-Os
22    communities (community mean Os level 55.2 ppb) compared to non-Hispanic white children in
23    low Os communities without the allele, indicating that in environments of low ambient Os,
24    enzymes with greater antioxidative activity may have the capacity to counter any temporary
25    imbalance in an oxidant-antioxidant relationship. In the presence of high background 63, the
26    protective effect may be attenuated because with higher exposure to oxidants, the antioxidant
27    genes may be  at their maximal level of expression, and variation in promoters no longer affects
28    levels of expression. No significant interactions were observed between PMio or other pollutants
29    and the HMOX-1 gene, and average Os levels  showed low correlation with the other monitored
30    pollutants.
31          Expanding on the results of McConnell et al. (2002), Islam et al. (2009) provided
32    evidence that variants in GSTM1 and GSTP1 may influence associations between outdoor
33    exercise and new-onset asthma. A primary conclusion that the authors (Islam et al., 2009)
       ! HMOX-1 variants were not associated with asthma risk in Hispanic children.
                                               2-40

-------
 1    reported was that variants of the GSTP1 genotype and the GSTM1 null genotype increased risk
 2    of new-onset asthma during adolescence.  The highest risk was found for participation in three or
 3    more team sports (compared to no sports) among children with variants of the GSTP1 genotype
 4    living in high-Os communities (community mean Os level  55.2 ppb).  No three-way interaction
 5    was found for GSTM1.  These results demonstrate the potential importance of a combination of
 6    genetic variability, 63 exposure, and outdoor activity on asthma risk.
 7           The CHS also provided evidence of interactions between Os exposure and variants in
 8    genes for arginase (Salam et al., 2009).  Higher arginase activity can limit production of NO and
 9    subsequent nitrosative stress. While epidemiologic evidence of associations of arginase variants
10    with asthma is limited, asthmatic subjects have been found to have higher arginase activity than
11    non-asthmatic subjects.  The modifying effect of O3 and atopy on the association between
12    variants in genes for arginase and asthma were evaluated.  Different haplotypes were associated
13    with different odds of childhood-onset asthma, modified by community 63 levels.  High 63
14    communities were defined as having an annual mean Os level greater than 50 ppb, and low Os
15    communities were defined as having an annual mean Os level less than 50 ppb.  The implications
16    of these findings are somewhat limited because the functional relevance of the variants is not
17    clear.
18           Two cross-sectional analyses provide further evidence relating Os exposure and the risk
19    of asthma. In a nationwide study of more than 32,000 Taiwanese school children, Hwang et al.
20    (2005)  assessed  the effects of air pollutants on the risk of asthma. The study population was
21    recruited from elementary and middle schools within 1 km of air monitoring stations. The risk
22    of asthma was related to Os in the one-pollutant model. The addition of other pollutants (NOx,
23    CO, SC>2, and PMio), in two-pollutant and three-pollutant models, increased the 63 risk
24    estimates.  In another cross-sectional analysis, Akinbami et al. (2010) examined the association
25    between chronic exposure to outdoor pollutants (12-month average levels by county) and asthma
26    outcomes in a national sample of children ages 3-17 years living in U.S. metropolitan areas
27    (National Health Interview  Survey). A 5-ppb increase in estimated 8-hour max Os concentration
28    (annual average) yielded a positive association for both currently having asthma and for having
29    at least 1 asthma attack in the previous year, while the adjusted odds ratios for other pollutants
30    were not statistically significant.  Models in which pollutant value ranges were divided into
31    quartiles produced comparable results, as did multi-pollutant models (862 and PM). The median
32    value for 12-month  average Os levels was 39.5 ppb and the IQR was 35.9-43.7 ppb, with a
33    positive concentration-response relationship apparent from the lowest quartile to the highest.
                                               2-41

-------
 1          2.3.1.2 Asthma hospital admissions and other morbidity effects
 2          The studies on (Vrelated hospital discharges and emergency department (ED) visits for
 3    asthma and respiratory disease that were available in the 2006 O3 AQCD mainly looked at the
 4    daily time metric.  Collectively, the short-term Os  studies presented in section 6.2.7.5 of the ISA
 5    EPA 2012a) and discussed above in section 2.2.1.6 indicate that there is evidence for increases in
 6    both hospital admissions and ED visits related to both all respiratory outcomes and asthma with
 7    stronger associations in the warm months. New studies also evaluated long-term Os exposure
 8    metrics,  providing a new line of evidence that suggests a positive exposure-response relationship
 9    between first asthma hospital admission and long-term Os exposure.
10          An ecologic study (Moore et al., 2008) evaluated time trends in associations between
11    declining warm-season O?, concentrations and hospitalization for asthma in children in
12    California's South Coast Air Basin. Quarterly 1-hour average daily maximum 63 concentrations
13    were used (median 87.8 ppb).  Ozone was the only pollutant associated with increased hospital
14    admissions over the study period. A linear relationship was observed for asthma hospital
15    discharges.
16          In a cross-sectional study, Meng et al.  (2010) examined  associations between air
17    pollution and asthma morbidity in the  San Joaquin Valley in California by using the 2001
18    California Health Interview  Survey data from subjects ages 1 to 65+ who reported physician-
19    diagnosed asthma. Subjects were assigned annual average concentrations for O3 based on
20    residential ZIP code and the closest air monitoring station within 8 km, but did not have data on
21    duration of residence.  Co-pollutant models for 63 and PM did not differ substantially from
22    single-pollutant estimates, indicating that pollutant multi-collinearity was not a problem.  The
23    authors reported increased asthma-related ED visits or hospitalizations for Os for all ages.
24    Positive and statistically significant associations for symptoms in adults (ages 18 +) were
25    observed; positive associations were obtained for symptoms for all ages.
26          Evidence associating long-term O3 exposure to first asthma hospital admission in a
27    concentration-response relationship is provided in a retrospective cohort study (Lin et al.,
28    2008b).  This  study investigated the association between chronic exposure to 63 and childhood
29    asthma admissions by following a birth cohort of more than 1.2 million babies born in New York
30    State (1995-1999) to first asthma admission or until 31 December 2000. Three annual indicators
31    (all  8-hour maximum from 10:00 a.m. to 6:00 p.m.) were used to define chronic Os exposure: (1)
32    mean concentration during the follow-up period (41.06 ppb);  (2) mean concentration during the
33    63 season (50.62 ppb); and (3) proportion of follow-up days with 63 levels >70 ppb. The effects
34    of co-pollutants were controlled, and interaction terms were used to assess potential effect
35    modifications. A positive association  between chronic exposure to Os and childhood asthma
                                                2-42

-------
 1    hospital admissions was observed, indicating that children exposed to high O3 levels over time
 2    are more likely to develop asthma severe enough to be admitted to the hospital. The various
 3    factors were examined and differences were found for younger children (1-2 years), poor
 4    neighborhoods, Medicaid/self-paid births, geographic region and others. As shown in the ISA,
 5    Figure 7-3 (EPA 2012a, p. 7-18), positive concentration-response relationships were observed.
 6    Asthma admissions were significantly associated with increased O3 levels for all chronic
 7    exposure indicators. Thus, this study provides evidence associating long-term O3 exposure to
 8    first asthma hospital admission in a concentration-response relationship.
 9    Asthma severity and control, bronchitic symptoms and school absences
10           In a cross-sectional study (Rage et al., 2005) involving an adult cohort in five French
11    cities, asthma severity over the previous 12 months was assessed using both clinical events and
12    treatment, as well as reported symptoms. Two measures of exposure were also assessed, using
13    the monitoring data closest to the participant's residence, and a validated spatial model assigning
14    air pollutants to the geocoded residential addresses of all participants. Higher asthma severity
15    scores were significantly related to both the 8-hour average O3 levels during April-September,
16    and the number of days with 8-hour O3  averages above 55 ppb.  Both exposure assessment
17    methods and severity score methods resulted in very similar findings. Effect estimates of O3
18    were similar in three-pollutant models, which included NC>2. Although no PM data were
19    available, since there are usually substantial correlations between PM and NC>2, the authors
20    expressed the view that the findings are not likely explained by ambient PM.
21           A follow-up study (Jacquemin et al., In Press) examines the relationship between asthma,
22    O3, NO2 and PMi0. New aspects considered include examination of three domains of asthma
23    control (symptoms, exacerbations, and lung function), levels of asthma control (controlled,
24    partially controlled, and uncontrolled asthma), and a multi-pollutant analysis including PMi0.
25    The results of a separate analysis suggest that the effects of O3 and PMio on asthma control are
26    independent.  Both annual and summer (April-September) O3 levels were used as variables.
27    Both O3-sum and PMio were positively  associated with partly controlled and uncontrolled
28    asthma, with a clear gradient from controlled, partly controlled and uncontrolled asthma. The
29    analysis of the associations between air pollution for all asthma subjects and each one of the
30    three asthma control domains showed positive associations between O3-sum and lung function,
31    symptoms, and exacerbations. Since the estimates for both pollutants were more stable and
32    significant when using the integrated measure of asthma control, this  indicates that the  results are
33    not driven by  one domain.  These results support an effect of long-term exposure to O3 on
34    asthma control in adults with pre-existing asthma.
                                                2-43

-------
 1          The CHS also examined interactions between TNF-a 308 genotype and long-term 63
 2    exposure in the occurrence of bronchitic symptoms among children with asthma (Lee et al.,
 3    2009b).  Increased airway levels of the cytokine TNF-a have been related to inflammation, and
 4    the GG genotype has been linked to lower expression of TNF-a.  Asthmatic  children with the
 5    GG genotype had a lower prevalence of bronchitic symptoms compared with children carrying at
 6    least one A-allele (e.g., GA or AA genotype).  Low- versus high-Os communities was defined as
 7    an average of less than or greater than 50 ppb O3. Asthmatic children with TNF-308 GG
 8    genotype had a significantly reduced risk of bronchitic symptoms in low-O3 communities,
 9    whereas the risk was not reduced in children living in high-Os communities. The difference in
10    genotypic effects between low- and high-Os communities was statistically significant among
11    asthmatics, but not significant among non-asthmatic children. Figure 7-2 (ISA, EPA 2012a, p.7-
12    12) presents adjusted 63 community-specific regression coefficients plotted  against ambient 63
13    concentrations. Investigators further reported no substantial differences in the effect of the GG
14    genotype on bronchitic symptoms by long-term exposure to PMio, PM2.5, NC>2, acid vapor, or
15    second-hand smoke.
16          Another CHS analyses reported interrelationships between variants in CAT and 1
17    myeloperoxidase (MPO) genes, ambient pollutants, and respiratory-related school absences for
18    Hispanic and non-Hispanic white cohort children (Wenten et al., 2009). A related study
19    evaluated in the 2006 O3 AQCD, Gilliland et al. (2001), found increased O3  exposure to be
20    related to greater school absenteeism  due to respiratory illness but did not consider genetic
21    variants. Wenten et al. (2009) hypothesized that variation in the level or function of antioxidant
22    enzymes would modulate respiratory  illness risk, especially under high levels of oxidative stress
23    expected from high ambient Os exposure. The joint effect of variants in these two genes (genetic
24    epistasis) on respiratory illness was examined because the enzyme products operate on the same
25    substrate within the same biological pathway.  Risk of respiratory-related school absences was
26    elevated for children with one set of variants in the two genes and reduced for children with a
27    different set of variants.  In analyses that stratified communities into high and low 63 exposure
28    groups by median levels (46.9 ppb Os), the protective effect of variants in the two genes was
29    largely limited to children living in communities with high ambient Os levels.  The association of
30    respiratory-illness absences with functional variants in CAT and MPO that differ by air pollution
31    levels illustrates the need to consider genetic epistasis in assessing gene-environment
32    interactions.  Collective evidence from CHS provides an important demonstration of gene-
33    environment interactions, which helps to dissect disease mechanisms in humans by using
34    information on susceptibility genes to focus on the biological pathways that  are most relevant to
35    that disease.
                                               2-44

-------
 1          2.3.1.3 Pulmonary structure and function
 2    Evidence from epidemiology studies
 3          In the 2006 Os AQCD, few studies had investigated the effect of chronic Os exposure on
 4    pulmonary function. The strongest evidence was for medium-term effects of extended 63
 5    exposures over several summer months on lung function (FEVi) in children, i.e., reduced lung
 6    function growth being associated with higher ambient 63 levels.  Longer-term studies (annual),
 7    investigating the association of chronic O3 exposure on lung function (FEVi) such as the 8-year
 8    follow-up analysis of the first cohort (Gauderman et al., 2004) provided little evidence that long-
 9    term exposure to ambient 63 at current levels is associated with significant deficits in the growth
10    rate of lung function in children. Analyses indicated that there was no evidence that either 8-
11    hour average Os (10 a.m. to 6 p.m.) or 24-hour average Os was associated with any measure of
12    lung function growth over a 4-year (age 10 to  14 years; Gauderman et al., 2000) or 8-year (age
13    10 to 18 years; Gauderman et al., 2004) period. However, most of the other pollutants examined
14    (including PM2.5, NC>2, acid vapor, and elemental carbon) were found to be significantly
15    associated with reduced growth in lung function.  In addition, there was only about a 2- to 2.5-
16    fold difference in Os concentrations from the least to most polluted communities (mean annual
17    average of 8-hour average 63 ranged from 30 to 65 ppb), versus the ranges observed  for the other
18    pollutants (which had 4- to 8-fold differences in concentrations).
19          A later CHS study (Islam et al., 2007) examined relationships between air pollution, lung
20    function, and new-onset asthma and reported no substantial differences in the effect of Os on
21    lung function. In a more recent CHS study, Breton et al. (2011) hypothesized that genetic
22    variation in genes on the glutathione metabolic pathway may influence the association between
23    ambient air pollutant exposures and lung function growth in children.  They investigated whether
24    genetic variation in glutathione genes was associated with lung function growth in healthy
25    children using data collected on more than 2,100 children over an 8-year time-period. Breton et
26    al. (2011) found that variation in these genes was associated with altered risk of children for lung
27    function growth deficits associated with NC>2, PMio, PM2.5, elemental carbon, organic carbon,
28    and Os.  When compared to the other pollutants, Os was associated with larger decreases in lung
29    function in children with a different genetic variation than the other pollutants, and only the
30    association with maximal midexpiratory flow (MMEF) was statistically significant.
31          Short-term 63 exposure studies presented  in ISA (EPA, 2012a, Section 6.2.1.2) provide a
32    cumulative body of epidemiologic evidence that strongly supports associations between ambient
33    63 exposure and decrements in lung function among children.  A recent study of long-term
34    exposure to O3; not described above, observed a relationship with pulmonary function declines in
35    school-aged children where Os and other pollutant levels were higher (90 ppb at high end of the
                                                2-45

-------
 1    range) than those in the CHS. Two studies of adult cohorts provide mixed results where long-
 2    term exposures were at the high end of the range with levels of 49.5 ppb in one study and 27 ppb
 3    IQR in the other.  Toxicological studies examining monkeys have provided data for airway
 4    resistance in an asthma model but this is difficult to compare to FEVi results.  Thus there is little
 5    new evidence to build upon the very limited studies of pulmonary function (FEVi) from the 2006
 6    O3 AQCD.
 7    Evidence from toxicological studies and non-human primate models
 8          Long-term studies in animals allow for greater insight into the potential effects of
 9    prolonged exposure to 63 that may not be easily measured in humans, such as structural changes
10    in the respiratory tract. As reviewed in the 1996 and 2006 63 AQCDs and Chapter 5  of the ISA
11    (US EPA 2012a), there are both qualitative and quantitative uncertainties in the extrapolation of
12    data generated by rodent toxicology studies to the understanding of health effects in humans.
13    Despite these uncertainties, epidemiologic studies  observing functional changes in humans can
14    attain biological plausibility in conjunction with long-term toxicological studies,  particularly (V
15    inhalation studies performed in non-human primates whose respiratory system most closely
16    resembles that of the human. An important series of studies have used nonhuman primates to
17    examine the effect of 63 alone or in combination with an inhaled allergen, house dust mite
18    antigen (FIDMA), on morphology and lung function. These animals exhibit the hallmarks of
19    allergic asthma defined for humans. Hyde et al. (2006) compared asthma models of rodents
20    (mice) and the nonhuman primate model to responses in humans and concluded that the unique
21    responses to inhaled allergen shown in the rhesus monkeys make it the most appropriate animal
22    model of human asthma.  These studies and others have demonstrated changes in pulmonary
23    function and airway morphology in adult and infant nonhuman primates repeatedly exposed to
24    environmentally relevant concentrations of Os. Many of the observations found in adult
25    monkeys have also been noted in infant rhesus monkeys,  although a direct comparison of the
26    degree of effects between adult and infant monkeys has not been reported. The findings of these
27    nonhuman primate studies have also been observed in rodent studies discussed in section 7.2.3.2
28    and included in Table 7-lof the ISA (US EPA, 2012a, p. 7-28).
29          Collectively, evidence from animal studies strongly suggests that chronic 63 exposure is
30    capable of damaging the distal airways and proximal alveoli, resulting in lung tissue remodeling
31    and leading to apparent irreversible changes. Potentially, persistent inflammation and interstitial
32    remodeling play an important role in the progression and development of chronic lung disease.
33    Further discussion of the modes of action that lead to Os-induced morphological  changes can be
34    found in Section 5.3.7 of the ISA (US EPA, 2012a).  The findings reported in chronic animal
35    studies offer insight into potential biological mechanisms for the suggested association between

                                               2-46

-------
 1    seasonal 63 exposure and reduced lung function development in children as observed in
 2    epidemiologic studies (see Section 7.2.3). Discussion of mechanisms involved in lifestage
 3    susceptibility and developmental effects 10 can be found in Section 5.4.2.4. 11
 4           2.3.1.4 Respiratory mortality
 5           A limited number of epidemiologic studies have assessed the relationship between long-
 6    term exposure to 63 and mortality.  The 2006 63 AQCD concluded that an insufficient amount
 7    of evidence existed "to suggest a causal relationship between chronic O3 exposure and increased
 8    risk for mortality in humans" (U.S. EPA, 2006b).  Though total and cardio-pulmonary mortality
 9    were considered in these studies, respiratory mortality was not specifically considered.  In the
10    most recent follow-up analysis of the ACS cohort (Jerrett et al., 2009), cardiopulmonary deaths
11    were separately subdivided into respiratory and cardiovascular deaths, rather than combined as in
12    the Pope et al. (2002) work. Increased 63 exposure was associated with the risk of death from
13    respiratory causes, and this effect was robust to the inclusion of PM2.5. The association between
14    increased 63 concentrations and increased risk of death from respiratory causes was insensitive
15    to the use of different models and to adjustment for several ecologic variables considered
16    individually.  Additionally, a recent multi-city time series study (Zanobetti and Schwartz, 2011),
17    which followed (from 1985 to 2006) four cohorts of Medicare enrollees with chronic conditions
18    that might predispose to Os-related effects, observed an association between long-term  (warm
19    season) exposure to Os and elevated risk of mortality in the cohort that had previously
20    experienced an emergency hospital admission due to COPD. A key limitation of this study was
21    the inability to control for PM2.5, because data were not available in these cities until 1999.
22           2.3.1.5 Summary of respiratory effects
23           The ISA (US EPA, 2012a, section 7.2.8) concludes that taken together, the recent
24    epidemiologic studies of respiratory  health effects (including respiratory symptoms, new-onset
25    asthma and respiratory mortality) combined with toxicological studies in rodents and nonhuman
26    primates, provide biologically plausible evidence that there is likely to be a causal relationship
27    between long-term exposure to Os and respiratory effects.  The strongest epidemiologic evidence
28    for a relationship between long-term 63 exposure and respiratory effects is provided by studies
29    that demonstrate interactions between exercise or different genetic variants and long-term
30    measures of Os  exposure on new-onset asthma in children; and increased respiratory symptom
31    effects in asthmatics. Additional studies of respiratory health effects and a study of respiratory
32    mortality provide a collective body of evidence supporting these relationships.  Studies
33    considering other pollutants provide data suggesting that the effects related to 63 are independent
34    from potential effects of the other pollutants.  Some studies provide evidence for a positive
35    concentration-response relationship. Short-term studies provide supportive evidence with
                                                2-47

-------
 1    increases in respiratory symptoms and asthma medication use, hospital admissions and ED visits
 2    for all respiratory outcomes and asthma, and decrements in lung function in children.  The recent
 3    epidemiologic and toxicological data base provides a compelling case to support the hypothesis
 4    that a relationship exists between long-term exposure to ambient 63 and measures of respiratory
 5    health effects.
 6    2.3.2  Other Effects
 7          In contrast to the more extensive bodies of evidence linking long-term O3 exposures to
 8    respiratory effects (discussed above), more limited evidence links long-term Os exposures to
 9    other health endpoints. Specifically, in recognition of the limitations and uncertainties in the
10    health evidence, the ISA concluded that the evidence is suggestive of a causal relationship with
11    cardiovascular effects, reproductive and developmental effects and CNS effects (US EPA,
12    2012a, sections 7.3.3, 7.4.11 and 7.5.2, respectively). In addition, the evidence was judged
13    inadequate to determine if a causal relationship exists between ambient Os exposures and cancer
14    (US EPA, 2012a, section 7.6.4).  The health evidence supporting these conclusions is discussed
15    in detail in the ISA (US EPA, 2012a) and is summarized briefly below for reproductive and
16    developmental effects (2.3.3.1) and CNS effects (2.3.3.2).
17          2.3.2.1 Cardiovascular effects
18          Previous AQCDs did not address the cardiovascular effects of long-term Os exposure due
19    to limited data availability. The evidence remains limited; however the emerging data is
20    suggestive of a role for 63 in chronic cardiovascular diseases.  Few epidemiologic studies have
21    investigated cardiovascular morbidity after long-term Os exposures.  The majority evaluated only
22    cardiovascular disease biomarkers such as lipid peroxidation, overall antioxidant capacity,
23    inflammation, coagulation, and blood pressure, with mixed results. However, three new animal
24    studies suggest that long-term Os exposure may result in cardiovascular effects.  These studies
25    demonstrate (Vinduced atherosclerosis and injury. In addition, evidence is presented for a
26    potential mechanism for the development of vascular pathology that involves increased oxidative
27    stress and proinflammatory mediators and upregulation of genes responsible for proteolysis,
28    thrombosis and vasoconstriction.
29          A very limited number of epidemiologic studies have assessed the relationship between
30    long-term O3 exposure and cardiovascular mortality. The 2006 O3 AQCD concluded that an
31    insufficient amount of evidence existed "to  suggest a causal relationship between chronic O3
32    exposure and increased risk for mortality in humans" (U.S.  EPA, 2006b). Though total and
33    cardio-pulmonary mortality were considered in these studies, cardiovascular mortality was not
34    specifically considered.  As noted in section 2.3.1.3 above,  in the most recent follow-up analysis

                                                2-48

-------
 1    of the ACS cohort (Jerrett et al., 2009), cardiopulmonary deaths were subdivided into respiratory
 2    and cardiovascular deaths.  While increased exposure to Os was associated with the risk of death
 3    from cardiopulmonary, cardiovascular, and ischemic heart disease in a single pollutant model,
 4    inclusion of PM2.5 as a copollutant attenuated the association with exposure to 63 for all of the
 5    cardiovascular endpoints to null. In addition, a recent study (Zanobetti and Schwartz 2011)
 6    discussed above, observed  an association between long-term exposure to 63 and elevated risk of
 7    mortality among Medicare enrollees that had previously experienced an emergency hospital
 8    admission due to congestive heart failure (CHF) or myocardial infarction (MI).  Toxicological
 9    evidence is also limited, but three strong toxicological studies have been published since the
10    2006 AQCD. These studies provide evidence for Os-enhanced atherosclerosis and injury along
11    with the potential mechanisms for vascular pathology. Taking into consideration the findings of
12    toxicological studies and the emerging evidence from epidemiologic studies, the ISA (US EPA,
13    2012a, section 7.3.3) concludes that the  generally limited body of evidence is suggestive of a
14    causal relationship between long-term exposure to 63 and cardiovascular effects.
15            2.3.2.2 Reproductive and developmental effects
16          The 2006 Os AQCD concluded that the limited number of studies that investigated Os
17    demonstrated no associations between 63 and birth outcomes, with the possible exception of
18    birth defects. The current review included an expanded body of evidence on the associations
19    between Os and reproductive and developmental  effects. Recent epidemiologic and toxicological
20    studies provide evidence for an effect of prenatal exposure to 63 on pulmonary structure and
21    function, including lung function changes in the newborn, incident asthma, ultrastructural
22    changes in bronchiole development, alterations in placental and pup cytokines, and increased pup
23    airway hyper-reactivity. Also, there is limited toxicological evidence for an effect of prenatal and
24    early life exposure on CNS effects, including laterality,  brain morphology, neurobehavioral
25    abnormalities, and sleep aberration. Recent epidemiologic studies have begun to explore the
26    effects of O3 on sperm quality, and provide limited evidence for decrements in sperm
27    concentration, while there is limited toxicological evidence for testicular degeneration associated
28    with 63. The weight of evidence does not indicate that prenatal or early life 63 concentrations
29    are associated with infant mortality.
30          Some of the key challenges to interpretation of these study results include the difficulty in
31    assessing exposure as most studies use existing monitoring networks to estimate individual
32    exposure to ambient air pollution; the inability to control for potential confounders such as other
33    risk factors that affect birth outcomes (e.g., smoking); evaluating  the exposure window  (e.g.,
34    trimester) of importance; integrating the results from both short- and long-term exposure periods;
                                                2-49

-------
 1    integrating the results across a variety of reproductive and developmental outcomes; and limited
 2    evidence on the physiological mechanism of these effects.
 3          2.3.2.3 CNS effects
 4          The 2006 63 AQCD included toxicological evidence that acute exposures to 63 are
 5    associated with  alterations in neurotransmitters, motor activity, short and long term memory, and
 6    sleep patterns. Additionally, histological signs of neurodegeneration have been observed.
 7    However, evidence regarding chronic exposure and neurobehavioral effects was not available.
 8    Recent research in the area of Os-induced neurotoxicity has included several long-term exposure
 9    studies. Notably, the first epidemiologic study to examine the relationship between 63 exposure
10    and neurobehavioral effects observed an association between annual Os levels and an aging-
11    related cognitive performance decline in tests measuring attention/short-term memory. This
12    observation is supported by studies in rodents which demonstrate progressive oxidative stress
13    and damage in the brain and associated decrements in behavioral tests, including those
14    measuring memory, after subchronic exposure to 0.25 ppm 63. Additionally, neurobehavioral
15    changes are evident in animals whose only exposure to O3 occurred in utero.

16    2.4    PUBLIC HEALTH IMPLICATIONS

17          This section discusses the public health implications of 63 exposures with respect to the
18    adversity of responses (section 2.4.1), the populations at potentially increased risk from
19    exposures (section 2.4.2), the potential effects of averting behavior on reducing Os exposures and
20    thereby the incidence of health effects (section 2.4.3), and an estimate of the size of the at-risk
21    population (section 2.4.4).  Providing appropriate public health protection requires identification
22    of populations potentially at greater risk from O?, exposures, and that a distinction is made
23    between those effects that are considered adverse health  effects and those that are not adverse.
24    What constitutes an adverse health effect depends not only on the type and magnitude of the
25    effect but also on the population group being affected. While some changes in healthy
26    individuals would not be considered adverse,  similar changes in at-risk groups would be seen as
27    adverse. In order to estimate potential overall for public health impacts, it is important to
28    consider not only the adversity of the health effects, but also the populations at greater risk and
29    potential behaviors that may reduce exposure.
30    2.4.1  Adversity of Responses
31          In this section we pose the following question:
32    •     To what extent does the currently available scientific evidence expand our
33          understanding  of the adversity of O3-related health effects?

                                                2-50

-------
 1          In making judgments as to when various (Vrelated effects become regarded as adverse
 2    to the health of individuals, in previous NAAQS reviews staff has relied upon the guidelines
 3    published by the American Thoracic Society (ATS) and the advice of CASAC.  While
 4    recognizing that perceptions of "medical significance" and "normal activity" may differ among
 5    physicians, lung physiologists and experimental subjects, the ATS (1985) defined adverse
 6    respiratory health effects as "medically significant physiologic changes generally evidenced by
 7    one or more of the following: (1) interference with the normal activity of the affected person or
 8    persons, (2) episodic respiratory illness, (3) incapacitating illness, (4) permanent respiratory
 9    injury, and/or (5) progressive respiratory dysfunction."
10          During the 2008 review,  it was concluded that for ethical reasons, clear causal evidence
11    from controlled human exposure studies covers only effects in the first category. However, there
12    were results from epidemiologic studies, upon which to base judgments about adversity, for
13    effects in all of the categories. Statistically significant and robust associations were reported in
14    epidemiology studies falling into the second and third categories. These more serious effects
15    included respiratory illness that may require medication (e.g., asthma), but not necessarily
16    hospitalization, as well as respiratory hospital admissions and ED visits for respiratory causes.
17    Less conclusive, but still positive associations have been reported for school absences and
18    cardiovascular hospital admissions.  Human health effects for which associations had been
19    suggested through evidence from epidemiologic and animal toxicology studies,  but had not been
20    conclusively demonstrated still fell primarily into the last two categories.
21          In this review, the new evidence strengthens the relationship between Os exposure and
22    health effects in all of the categories defined by the ATS in 1985. The ISA judgment that there is
23    a causal relationship between short-term 63 exposure and a full range of respiratory morbidity
24    effects, including hospital admissions and ED visits, provides support for concluding that short-
25    term 63 exposure is associated with incapacitating effects.  Overall, the evidence supporting an
26    association between short-term O3 exposures and respiratory mortality is much stronger. And
27    the demonstration of associations between long-term measures of Os exposure and new-onset
28    asthma provides evidence of permanent respiratory injury or progressive respiratory decline.
29          In 2000, the American Thoracic  Society (ATS) published an official statement on "What
30    Constitutes an Adverse Health Effect of Air Pollution?" (ATS, 2000), which updated its earlier
31    guidance (ATS, 1985). The revised guidance was intended to address new investigative
32    approaches used to identify the effects of air pollution, and to reflect the concern for the impacts
33    of air pollution on specific groups that had been expressed through the environmental justice
34    movement.
                                                2-51

-------
 1          The new guidance builds upon and expands the 1985 definition of adversity in several
 2    ways. There is an increased focus on quality of life measures as indicators of adversity.  There is
 3    also a more specific consideration of population risk. Exposure to air pollution that increases the
 4    risk of an adverse effect to the entire population is adverse, even though it may not increase the
 5    risk of any individual to an unacceptable level. For example, a population of asthmatics could
 6    have a distribution of lung function such that no individual has a level associated with significant
 7    impairment. Exposure to air pollution could shift the distribution to lower levels that still do not
 8    bring any individual to a level that is associated with clinically relevant effects.  However, this
 9    would be considered to be adverse because individuals within the population would have
10    diminished reserve function, and therefore would be at increased risk to further environmental
11    insult, if affected by another agent.
12          Reflecting new investigative approaches, the ATS statement describes the potential
13    usefulness of research into the genetic basis for disease, including responses to environmental
14    agents that will provide insights into the mechanistic basis for susceptibility, and provide
15    markers of risk status. The committee also observed that elevations of biomarkers, such as cell
16    number and types, cytokines and reactive oxygen species, may signal risk for ongoing injury and
17    clinical effects or may simply indicate transient responses that can provide insights into
18    mechanisms of injury, thus illustrating the lack of clear boundaries that separate adverse from
19    nonadverse effects. These newer guidelines, while providing a basis for evaluating new types of
20    scientific evidence with respects to its adversity, do not change the fundamental conclusions that
21    exposure to 63 at ambient concentrations can cause adverse health effects in the general
22    population, but especially in at-risk groups.
23          While 63 has been associated with effects that are clearly  adverse, including premature
24    mortality, application of these guidelines, in particular to the least serious category of effects
25    related to ambient 63 exposures, which are also the most numerous and therefore are  also
26    important from a policy perspective, involves judgments about which medical experts on the
27    CASAC panel and public commenters have in the past expressed  diverse views.  To help frame
28    such judgments, we have defined gradations of individual functional responses (e.g., decrements
29    in FEVi and airway responsiveness) and symptomatic responses (e.g., cough, chest pain,
30    wheeze), together with judgments as to the potential impact on individuals experiencing varying
31    degrees of severity of these responses, that have been used in previous NAAQS reviews. These
32    gradations and impacts are summarized in Gradation of Individual Responses to Short-Term Os
33    Exposure Tables 2-1 and 2-2, below.
34          For active healthy people, moderate levels of functional responses (e.g., FEVi
35    decrements of >10% but < 20%, lasting up to 24 hours) and/or moderate symptomatic responses

                                                2-52

-------
 1    (e.g., frequent spontaneous cough, marked discomfort on exercise or deep breath, lasting up to
 2    24 hours) would likely interfere with normal activity for relatively few sensitive individuals;
 3    whereas large functional responses (e.g., FEVi decrements > 20%, lasting longer than 24 hours)
 4    and/or severe symptomatic responses (e.g., persistent uncontrollable cough, severe discomfort on
 5    exercise or deep breath, lasting longer than 24 hours) would likely interfere with normal
 6    activities for many sensitive individuals and therefore would be considered adverse under ATS
 7    guidelines.
 8          For the purpose of estimating potentially adverse lung function decrements in active
 9    healthy people, the CAS AC indicated that a focus on the mid to upper end of the range of
10    moderate levels of functional responses is most appropriate (e.g., FEVi decrements > 15% but <
11    20%) (Henderson, 2006).  However, for people with lung disease, even moderate functional
12    (e.g., FEVi decrements > 10% but < 20%, lasting up to 24 hours) or symptomatic responses
13    (e.g., frequent spontaneous cough, marked discomfort on exercise or with deep breath, wheeze
14    accompanied by shortness of breath, lasting up to 24 hours) would likely interfere with normal
15    activity for many individuals, and would likely result in additional and more frequent use of
16    medication.  For people with lung disease, large functional responses (e.g., FEVi decrements >
17    20%, lasting longer than 24 hours) and/or severe symptomatic responses (e.g., persistent
18    uncontrollable cough, severe discomfort on exercise or deep breath, persistent wheeze
19    accompanied by shortness of breath, lasting longer than 24 hours) would likely interfere with
20    normal activity for most individuals and would increase the likelihood that these individuals
21    would seek medical treatment. For the purpose of estimating potentially adverse lung function
22    decrements in people with lung disease, the CASAC indicated that a focus on the lower end of
23    the range of moderate levels of functional responses is most appropriate (e.g., FEVi decrements
24    > 10%) (Henderson, 2006).
25          Responses measured in controlled human exposure studies indicate that humans exposed
26    to ambient O3 concentrations include: decreased inspiratory capacity; mild bronchoconstriction;
27    rapid, shallow breathing pattern during exercise; and symptoms of cough and pain on deep
28    inspiration (EPA, 2012a, section 6.2.1.1). Reflex inhibition of inspiration results in a decrease in
29    forced vital capacity and, in combination with mild bronchoconstriction, contributes to a
30    decrease in FEVi. Some healthy young adults exposed to OB concentrations > 60 ppb develop
31    statistically significant reversible, transient decrements in lung function, symptoms of breathing
32    discomfort, and inflammation if minute ventilation or duration of exposure is increased
33    sufficiently.  Among healthy subjects there is considerable interindividual variability in the
34    magnitude of the FEVi responses. For example, at 60 ppb  (EPA, 2012a, p.  6-17), the proportion
35    of healthy subjects with >10% FEVi decrements was 20% (n = 30) by Adams (2002), 3% (n =

                                               2-53

-------
 1    30) by Adams (2006a), 16% (n = 31) by Schelegle et al. (2009), and 5% (n = 59) by Kim et al.
 2    (2011). Based on these studies, the weighted average proportion of individuals with >10% FEVi
 3    decrements is 10% following 6.6 hour exposure with moderate, quasi-continuous exercise to 60
 4    ppb.19 Subjects with asthma appeared to be more sensitive to acute effects of 63 in terms of
 5    FEVi and inflammatory responses than healthy non-asthmatic subjects, such that controlled
 6    human exposure studies of healthy adults may underestimate effects in people with asthma.
 7          As discussed above, relatively small, reversible declines in lung function parameters may
 8    be of questionable significance in healthy people. However, a 5 to 15% change in FEVi is
 9    considered to have clinical importance to asthma morbidity (ATS 1991; Lebowitz et al. 1987;
10    Lippmann, 1988).  This is in line with the view expressed by the CAS AC that a focus on the
11    lower end of the range of moderate levels of functional responses is most appropriate (e.g., FEVi
12    decrements >10%) to estimate the risk of potentially adverse lung function responses in people
13    with lung disease.

14
      19 Due to limited data within the published papers, these proportions were not corrected for responses to FA
      exposure where lung function typically improves in healthy adults. For example, uncorrected versus O3-induced
      (i.e., adjusted for response during FA exposure) proportions of individuals having >10% FEVi decrements in the
      Adams (2006a) study were,  respectively, 3% versus 7% at 60 ppb and 17% versus 23% at 80 ppb. Thus,
      uncorrected proportions underestimate the actual fraction of healthy individuals affected.
                                                 2-54

-------
Table 2-1. Gradation of Individual Responses to ShorfTerm Ozone Exposure in Healthy
People
Functional
Response
None
Small
Moderate
Large
FEM
Nonspecific
airway
responsiveness
Chest pain
Within       Decrements of  Decrements of
normal       3 to=10%      > 10 but <20%
range (±3%)

Within       Increases of    Increases of
normal range  <100%        =300%
                               Decrements of
                               =20%
                               Increases of
                               >300%
Duration of
response
Symptom
Response
Cough
None
Normal
Infrequent
cough
<4 hours
Mild
Cough with
deepbreath
>4 hours bu£24
hours
Moderate
>24 hours
Severe
Frequent Persistent
spontaneous cough uncontrollable
None
                               cough

Discomfort just Marked discomfortSevere discomfort
noticeable on   on exercise or     on exercise or
exercise or    deepbreath       deep breath
deep breath
Duration of
response
Impact of
Responses
None
Normal
<4 hours
Mild
>4 hours but24
hours
Moderate
>24 tours
Severe
Interference with
normal
activity
None
None
A few sensitive    Many sensitive
individuals choose individuals
to limit activity    chooseto limit
                 activity
                                       2-55

-------
Table 2-2. Gradation of Individual Responses to Sfiertn Ozone Exposure in People
with Respiratory Disease
Functional
Response
FEV


Nonspecific
airway
responsiveness
Airway resistance
(SRaw)

Duration of
response
Symptom
Resnonse
Wheeze


Cough


Chest pain



Duration of
response
Impact of
ResDonses
Interference with
normalactivity


Medical Treatment



None

Within
normal
range (±3%)
Within
normal range

Within
normal range
(±20%)
None

Normal

None


Infrequent
cough

None



None

Normal

None



No change



Small

Decrements of
3 to=10%

Increases of
<100%

SRaw
increased
<100%
<4 hours

Mild

With otherwise
normal
breathing
Cough with
deep
breath
Discomfort just
noticeable on
exercise or
deep breath
<4 hours

Mild

None



Normal
medication as
needed

Moderate

Decrements of
>10but<20%

Increases of
=300%

SRaw increased up
to 200% or up to 15
cm H2O/S
>4 hours bu£24
hours
Moderate

With shortness of
breath

Frequent
spontaneous cough

Marked discomfort
on exercise or deep
breath

>4 hours but24
hours
Moderate

A few sensitive
individuals choose
to limit activity

Increased frequency
of medication use or
additional
medication
Large

Decrements^
=20%

Increases of
>300%

SRawincreased
>200% or more than
15cmH2O/s
>24 hours

Severe

Persistent with
shortness^ breath

Persistent
uncontrollable
cough
Severediscomfort
on exercise or
deep breath

>24 tours

Severe

Many sensitive
individuals
choose to limit
activity
Physician or
emergency room
visit

                                      2-56

-------
 1           In judging the extent to which these impacts represent effects that should be regarded as
 2    adverse to the health status of individuals, an additional factor that has been considered in
 3    previous NAAQS reviews is whether such effects are experienced repeatedly during the course
 4    of a year or only on a single  occasion. While some experts would judge single occurrences of
 5    moderate responses to be a "nuisance," especially for healthy individuals, a more general
 6    consensus view of the adversity of such moderate responses emerges as the frequency of
 7    occurrence increases. Thus it has been judged that repeated occurrences  of moderate responses,
 8    even in otherwise healthy individuals, may be considered to be adverse since they could well set
 9    the stage for more serious illness (61 FR 65723). The CAS AC panel in the 1997  review
10    expressed a consensus view that these "criteria for the determination of an adverse physiological
11    response were reasonable" (Wolff, 1995).  In the 2008 review, estimates  of repeated occurrences
12    continued to be an important policy factor in judging the adversity of moderate lung function
13    decrements in healthy and asthmatic people.
14    2.4.2   At-Risk Populations
15           In this section we pose the following question:
16    •      To what extent does the currently available scientific evidence expand our
17           understanding of at-risk populations?
18
19           In order to identify populations and lifestages at greater risk for Os-related health effects,
20    studies have evaluated factors that may contribute to the susceptibility and/or vulnerability of an
21    individual to air pollutants. The definitions of susceptibility and vulnerability have been found to
22    vary across studies,  but in most instances "susceptibility" refers to biological or intrinsic factors
23    (e.g., lifestage, sex,  preexisting disease/conditions) while "vulnerability" refers to non-biological
24    or extrinsic factors (e.g., socioeconomic status [SES]) (U.S. EPA, 2010c, 2009d). In the ISA and
25    this PA, "at-risk" is  the all-encompassing term used for groups with specific factors that increase
26    the risk of Os-related health effects in a population.
27           There are multiple avenues by which individuals, and ultimately populations, could
28    experience increased risk for (Vinduced health effects. A population or lifestage20 may exhibit
29    greater effects with  the same concentration or dose as the general population, or they may be at
30    greater risk due to increased exposure to an air pollutant (e.g., time  spent outdoors). A group
31    with intrinsically increased risk would have some factor(s) that increases risk through a
      20 Lifestages, which in this case includes children and older adults, are factors that most people go through over the
      course of a lifetime, unlike the other factors associated with at-risk populations.
                                                 2-57

-------
 1    biological mechanism and, in general, would have a steeper concentration-risk relationship,
 2    compared to those not in the group. Factors that are often considered intrinsic include asthma
 3    and genetic background. A group of people could also have extrinsically increased risk, which
 4    would be through an external, non-biological factor, including, for example, socioeconomic
 5    status (SES) and diet.  Some groups are at risk of increased internal dose at a given exposure
 6    concentration, for example, because of breathing patterns. This category would include people
 7    who work or exercise outdoors. Finally, there are those who might be placed at increased risk
 8    for experiencing greater exposures by being exposed at higher concentrations. This would
 9    include, for example, groups of people with greater exposure to ambient 63 due to less
10    availability or use of home air conditioners (i.e., more open windows on high ozone days).
11           Some factors described above are multifaceted and may influence the risk of O3-related
12    health effects through a combination of factors. For example, children tend to spend more time
13    outdoors at higher levels of activity than adults, which leads to increased exposure and dose, but
14    they also have biological (i.e., intrinsic) differences when compared to adults.
15           The following sections  discuss factors for which the ISA concludes that there is evidence
16    to support potentially increased risk of Os-related health effects, and the overall  weight of
17    evidence for the factor. This discussion includes factors that potentially increase the risk of
18    Os-related health effects, regardless of whether the increased risk is due to intrinsic factors,
19    extrinsic factors, or a combination, due to the often connected pathways between factors.
20           2.4.2.1 Approach to classifying factors that increase risk
21           To identify factors that potentially lead to some populations being at greater risk to Os-
22    related health effects, the evidence across relevant scientific disciplines (i.e., exposure sciences,
23    dosimetry, controlled human exposure, toxicology, and epidemiology) was evaluated in Chapter
24    8 of the ISA. In this systematic approach, the collective evidence is used to examine coherence
25    of effects across disciplines and determine biological plausibility.  The ISA first focuses on
26    studies (i.e., epidemiologic or controlled human exposure) that conduct stratified analyses to
27    identify factors  that  may result in some populations being at greater risk of an air pollutant
28    related health effect. These types of studies allow for an evaluation of populations exposed to
29    similar air pollutant  (e.g., Os) concentrations within the same study design. Experimental studies
30    also provide important lines of evidence in the evaluation of factors that may lead to increased
31    risk of an air pollutant related-health effect.  Toxicological studies conducted  using animal
32    models of disease and controlled human exposure studies that examine individuals with
33    underlying disease or genetic polymorphisms may provide additional evidence of at-risk
34    populations in the absence of stratified epidemiologic analyses. Experimental and toxicological
35    studies can provide evidence of biological plausibility as well as provide support for coherence
                                                2-58

-------
 1    with the health effects observed in epidemiologic studies. The collective results across the
 2    scientific disciplines comprise the overall weight of evidence that is used to determine whether a
 3    specific factor results in a population being at increased risk of an air pollutant related health
 4    effect.
 5           Building on the causal framework discussed the Preamble to the ISA, and discussed in
 6    section 2.1 above, conclusions are made regarding the strength of evidence for each factor that
 7    may contribute to increased risk of an O3-related health effect based on the evaluation and
 8    synthesis of evidence across scientific disciplines. The  conclusions drawn considered the
 9    "Aspects to Aid in Judging Causality" discussed in Table 1 of the Preamble to the ISA.  The
10    categories considered for evaluating the evidence for potential increased risk of an O3-related
11    health effect are "adequate evidence," "suggestive evidence,"  "inadequate evidence," and
12    "evidence of no effect." They are described in more detail in Table 8-1, Classification of
13    Evidence for Potential At-Risk Factors, in the ISA (US EPA, 2012a).
14           2.4.2.2 Factors that increase risk
15           Specific groups within the general population, referred to as at-risk populations, are at
16    increased risk of experiencing adverse health effects related to O3 exposures. As discussed
17    above, these groups can exhibit a greater risk of O3-related health effects than the general
18    population because of a number of different types of factors, which may be intrinsic or extrinsic
19    in nature.  Given the heterogeneity of individual responses to O3 exposures, the severity of
20    effects  experienced by at-risk populations may be much greater than that experienced by the
21    general population. Providing appropriate public health protection requires identification of
22    populations potentially at greater risk from O3 exposures.
23           A summarized below, the currently available evidence expands our understanding of
24    populations identified to be at greater risk from the last review (i.e., people who are active
25    outdoors, people with lung disease, children and older adults and people with increased
26    responsiveness to O3) (US EPA 2006, section 3.6.2) and supports the identification of additional
27    factors that may lead to increased risk.  For the purpose of this PA, only the factors for which
28    there is adequate evidence for increased risk (this section) or suggestive evidence for potential
29    increased risk (section 2.4.2.3) of O3-related health effects will be considered.
30    Asthma
31           Previous O3 AQCDs identified individuals with  asthma as a population at increased risk
32    of O3-related health effects.  Within the U.S., approximately 8.2% of adults have reported
33    currently having asthma (Schiller et al., 2012) and 9.5% of children have reported currently
                                                 2-59

-------
1   having asthma (Bloom et al., 2011). Table 2-3 below provides more detailed information on
2   prevalence of asthma by age in the U.S.
3
4   Table 2-3. Prevalence of asthma by age in the U.S.








Age (years)
0-4
5-11
12-17
18-44
45-64
65-74
75+
N (in thousands)
1,285
3,020
2,672
8,902
6,704
1,849
1,279
Percent
6.0
10.5
10.9
8.1
8.4
8.7
7.4
Asthma prevalence is reported for "still has asthma"
Source: Statistics for adults: Schiller et al. (2012); Statistics for children: Bloom et al. (201 1)
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21



Multiple new epidemiologic studies included in the ISA have evaluated the potential for
increased risk of Os-related health effects in people with asthma,
including: lung function;
symptoms; medication use; airway hyperresponsiveness (AHR); and airway inflammation (also
measured as exhaled nitric oxide fraction, or FeNO). A study of lifeguards in Texas reported
decreased lung function with short-term Os exposure among both
individuals with and without
asthma, however, the decrease was greater among those with asthma (Thaller et al., 2008). A
Mexican study of children ages
6-14 detected an association between short-term O3 exposure and
wheeze, cough, and bronchodilator use among asthmatics but not
non-asthmatics, although this
may have been the result of a small non-asthmatic population (Escamilla-Nufiez et al., 2008). A
study of modification by AHR (an obligate condition among asthmatics) reported greater short-
term Os-associated decreases in
lung function in elderly individuals with AHR, especially among
those who were obese (Alexeeff et al., 2007). With respect to airway inflammation, in one
study, a positive association was reported for airway inflammation among asthmatic children
following short-term Os exposure, but the observed association was similar in magnitude to that
of non-asthmatics (Barraza-Villarreal et al., 2008). Similarly, another study of children in
California reported an association between Os concentration and FeNO that persisted both among
                                             2-60

-------
 1    children with and without asthma as well as those with and without respiratory allergy (Berhane
 2    et al., 2011). Finally, Khatri et al. (2009) found no association between short-term Os exposure
 3    and altered lung function for either asthmatic or non-asthmatic adults, but did note a decrease in
 4    lung function among individuals with allergies.
 5          New evidence for difference in effects among asthmatics has been observed in studies
 6    that examined the association between Os exposure and altered lung function by asthma
 7    medication use.  A study of children with asthma living in Detroit reported a greater association
 8    between short-term Os and  lung function for corticosteroid users compared with
 9    noncorticosteroid users (Lewis et al., 2005). Conversely, another study found decreased lung
10    function among  noncorticosteroid users compared to users, although in this study, a large
11    proportion of non-users were considered to be persistent asthmatics (Hernandez-Cadena et al.,
12    2009). Lung function was not related to short-term Os exposure among corticosteroid users and
13    non-users in a study taking  place during the winter months in Canada (Liu et al., 2009a).
14    Additionally, a study of airway inflammation reported a counterintuitive inverse association with
15    Os of similar magnitude for all groups of corticosteroid users and non-users (Qian et al., 2009).
16          Controlled human exposure studies that have examined the effects of Os on individuals
17    with asthma and healthy controls are limited. Based on studies reviewed in the 1996 and 2006
18    Os AQCDs, subjects with asthma appeared to be more  sensitive to acute effects of Os in terms of
19    FEVi and inflammatory responses than healthy non-asthmatic  subjects.  For instance, Horstman
20    et al. (1995) observed that mild-to-moderate asthmatics, on average, experienced double the
21    Os-induced  FEVi decrement of healthy subjects (19% versus 10%, respectively, p = 0.04).
22    Moreover, a statistically significant positive correlation between FEVi responses to Os  exposure
23    and baseline lung function was observed in individuals with  asthma, i.e., responses increased
24    with severity of  disease. Minimal evidence exists suggesting that  individuals with asthma have
25    smaller Os-induced FEVi decrements than healthy subjects (3% versus 8%, respectively)
26    (Mudway et al.,  2001). However, the asthmatics in that study also tended to be older than the
27    healthy subjects, which could partially explain their lesser response since FEVi responses to Os
28    exposure diminish with age. Individuals with  asthma also had significantly more neutrophils in
29    the BALF (18 hours postexposure) than similarly exposed healthy individuals (Peden et al.,
30    1997; Scannell et al., 1996; Basha et al., 1994). Furthermore, a study examining the effects of
31    Os on individuals with atopic asthma and healthy controls reported that greater numbers of
32    neutrophils, higher levels of cytokines and hyaluronan, and greater expression of macrophage
33    cell-surface markers were observed in induced sputum of atopic asthmatics compared with
34    healthy controls (Hernandez et al., 2010). Differences  in O3-induced epithelial cytokine
35    expression were noted in bronchial biopsy samples from asthmatics and healthy controls (Bosson

                                                2-61

-------
 1    et al., 2003). Cell-surface marker and cytokine expression results, and the presence of
 2    hyaluronan, are consistent with Os having greater effects on innate and adaptive immunity in
 3    these asthmatic individuals. In addition, studies have demonstrated that Os exposure leads to
 4    increased bronchial reactivity to inhaled allergens in mild allergic asthmatics (Kehrl et al., 1999;
 5    Torres et al., 1996) and to the influx of eosinophils in individuals with pre-existing allergic
 6    disease (Vagaggini et al., 2002; Peden et al., 1995).  Taken together, these results point to several
 7    mechanistic pathways which could account for the enhanced sensitivity to O3 in subjects with
 8    asthma (see Section 5.4.2.2 in the ISA).
 9           Toxicological studies (see section 8.2.2 in the ISA) provide biological plausibility for
10    greater effects of 63 among those with asthma or AHR. In animal toxicological studies,  an
11    asthmatic phenotype is modeled by allergic sensitization of the respiratory tract. Many of the
12    studies that provide evidence that Os exposure is an inducer of AHR and remodeling utilize these
13    types of animal models.  For example, a series of experiments in infant rhesus monkeys have
14    shown these effects, but only in monkeys sensitized to house dust mite allergen. Similarly,
15    adverse changes in pulmonary function were demonstrated in mice exposed to O^; enhanced
16    inflammatory responses were in rats exposed to 63, but only in animals sensitized to allergen. In
17    general, it is the combined  effects of Os and allergic sensitization which result in measurable
18    effects on pulmonary function.  In a pulmonary fibrosis model, exposure 63 for 5 days increased
19    pulmonary inflammation and fibrosis, along with the frequency of bronchopneumonia in rats.
20    Thus,  short-term exposure to Os may enhance damage in a previously injured lung.
21           In the 2006 Os AQCD, the potential for individuals with asthma to have greater risk of
22    Os-related health effects was supported by a number of controlled human exposure studies,
23    evidence from toxicological studies, and a limited number of epidemiologic studies. In section
24    8.2.2, the ISA reports that in the recent epidemiologic literature some, but not all, studies report
25    greater risk of health effects among individuals with asthma.  Studies examining effect measure
26    modification of the relationship between short-term O3 exposure and altered lung function by
27    corticosteroid use provided limited evidence of Os-related health effects.  However, recent
28    studies of behavioral responses have found that studies do not take into account individual
29    behavioral adaptations to forecasted air pollution levels (such as avoidance and reduced time
30    outdoors), which may underestimate the observed associations in studies that examined the effect
31    of 63 exposure on respiratory health (Neidell and Kinney, 2010).  This could explain some
32    inconsistency observed among recent epidemiologic studies.  The evidence from controlled
33    human exposure studies provides support for increased detriments in FEVi and greater
34    inflammatory responses to  Oj in individuals with asthma than in healthy individuals without a
35    history of asthma. The collective evidence for increased risk of Os-related health effects among

                                                2-62

-------
 1    individuals with asthma from controlled human exposure studies is supported by recent
 2    toxicological studies which provide biological plausibility for heightened risk of asthmatics to
 3    respiratory effects due to O?, exposure.  Overall, the ISA finds there is adequate evidence for
 4    asthmatics to be an at-risk population.
 5    Children
 6           The 2010 Census reported that 27% of the U.S. population, or more than 83 million
 7    people, was under 20 years of age, with 13.1%, or more than 40 million people, under the age of
 8    10 (Howden and Meyer, 2011). Children are considered to be at greater risk from Os exposure
 9    because their respiratory systems undergo lung growth until about 18-20 years of age and are
10    therefore thought to be intrinsically more  at risk for (Vinduced damage (U.S. EPA, 2006b).  It is
11    generally recognized that children spend more time outdoors than adults, and therefore would be
12    expected to have higher exposure to Os than adults. The ventilation rates also vary between
13    children and adults, particularly during moderate/heavy activity. Children aged 11 years and
14    older and adults have higher absolute ventilation rates than children aged 1-11 years.  However,
15    children have higher ventilation rates relative to their lung volumes, which tends to increase dose
16    normalized to lung surface area.  Exercise intensity has a substantial effect on ventilation rate,
17    with high intensity activities resulting in nearly double the ventilation rate during moderate
18    activity among children and those adults less than 31 years of age. For more information on time
19    spent outdoors  and ventilation rate differences by age group, see Section 4.X in the ISA.
20          The 1996 Os AQCD reported clinical evidence that children, adolescents, and young
21    adults (<18 years of age) appear, on average, to have nearly equivalent spirometric responses to
22    O3 exposure, but have greater responses than middle-aged and older adults (U.S. EPA, 1996a).
23    Symptomatic responses (e.g., cough, shortness of breath,  pain on deep inspiration) to Os
24    exposure, however, appear to increase with age until early adulthood and then gradually decrease
25    with increasing age (U.S. EPA, 1996a). Complete lung growth  and development is not achieved
26    until 18-20 years of age in women and the early 20s for men; pulmonary function is at its
27    maximum  during this time as well.
28          Recent  epidemiologic studies have examined different age groups and their risk to
29    (Vrelated  respiratory hospital admissions and emergency department (ED) visits. Evidence for
30    greater risk in children was reported in several studies.  A study in Cyprus of short-term O3
31    concentrations  and respiratory hospital admissions (HA) detected possible effect measure
32    modification by age with a larger association among individuals < 15 years of age compared
33    with those  > 15 years of age; the effect was apparent only with a 2-day lag (Middleton et al.,
34    2008).  Similarly, a Canadian study of asthma-ED visits reported the strongest Os-related
35    associations among 5- to 14-year olds compared to the other age groups (ages examined 0-75+)
36    (Villeneuve et al., 2007). Greater Os-associated risk in asthma-related ED visits were also
                                                2-63

-------
 1    reported among children (<15 years) as compared to adults (15 to 64 years) in a study from
 2    Finland (Halonen et al., 2009). A study of New York City hospital admissions demonstrated an
 3    increase in the association between Os exposure and asthma-related hospital admissions for 6- to
 4    18-year olds compared to those < 6 years old and those > 18 years old (Silverman and Ito, 2010).
 5    When examining long-term Os exposure and asthma HA among children, associations were
 6    determined to be larger among children 1 to 2 years old compared to children 2 to 6 years old
 7    (Lin et al., 2008b).  A few studies reported positive associations among both children and adults
 8    and no modification of the effect by age.
 9          The evidence reported in  epidemiologic studies is supported by recent toxicological
10    studies which observed Os-induced health effects in immature animals.  Early life exposures of
11    multiple species of laboratory animals, including infant monkeys, resulted in changes in
12    conducting airways at the cellular, functional, ultra-structural, and morphological levels. The
13    studies conducted on infant monkeys are most relevant for assessing effects in children.  Carey et
14    al. (2007) conducted a study of 63 exposure in infant rhesus macaques, whose respiratory tract
15    closely resemble that of humans. Monkeys were exposed either acutely or in episodes designed
16    to mimic human exposure.  All monkeys acutely exposed to Os had moderate to marked
17    necrotizing rhinitis, with focal regions of epithelial exfoliation, numerous infiltrating neutrophils,
18    and some eosinophils. The distribution, character, and severity of lesions in episodically exposed
19    infant monkeys were similar to that of acutely exposed animals. Neither exposure protocol for
20    the infant monkeys produced mucous cell metaplasia proximal to the lesions, an adaptation
21    observed in adult monkeys exposed in another study (Harkema et al., 1987a). Functional and
22    cellular changes in conducting airways were common manifestations of exposure to Os among
23    both the adult and infant monkeys (Plopper et al., 2007).  In addition, the lung structure of the
24    conducting airways in the infant monkeys was significantly stunted by Os and this aberrant
25    development was persistent 6 months postexposure (Fanucchi et al., 2006).
26          Age may also affect the inflammatory response to O3 exposure.  Toxicological studies
27    reported that the difference in effects among younger lifestage test animals may be due to
28    age-related changes in antioxidants levels and sensitivity to oxidative stress. Further discussion
29    of these studies may be found in  section 8.3.1.1 of the ISA (US EPA, 2012a, p. 8-20).
30          The previous and recent human clinical and toxicological studies reported evidence of
31    increased risk from 63 exposure for younger ages, which provides coherence and biological
32    plausibility for the findings from epidemiologic studies. Although there was some inconsistency,
33    generally, the epidemiologic studies reported positive associations among both children and
34    adults or just among children.  The interpretation of these studies is limited by the lack of
35    consistency in comparison age groups and outcomes examined.  However, overall, the
                                               2-64

-------
 1    epidemiologic, controlled human exposure, and toxicological studies provide adequate evidence
 2    that children are potentially at increased risk of Os-related health effects.
 3    Older adults
 4          The ISA notes that older adults are at greater risk of health effects associated with Os
 5    exposure through a variety of intrinsic pathways (US EPA, 2012a, section 8.3.1.2). In addition,
 6    older adults may differ in their exposure and internal dose. Older adults were outdoors for a
 7    slightly longer proportion of the day than adults aged 18-64 years. Older adults also have
 8    somewhat lower ventilation rates than adults aged 31 - less than 61 years. For more information
 9    on time spent outdoors and ventilation rate differences by age group, see Section 4.4 in the ISA
10    (US EPA, 2012a).  The gradual decline in physiological processes that occur with aging may
11    lead to increased risk of O3-related health effects  (U.S. EPA, 2006a). Respiratory symptom
12    responses to Os exposure appears to increase with age until early adulthood and then gradually
13    decrease with increasing age  (U.S.  EPA, 1996a);  lung function responses to O3  exposure also
14    decline from early adulthood (US EPA, 1996a). The reductions of these responses with age may
15    put older adults at increased risk for continued Os exposure. In addition, older adults, in general,
16    have a higher prevalence of preexisting diseases compared to younger age groups  and this may
17    also lead to increased risk of O3-related health effects (see Table 8-3 in the ISA, US EPA, 2012a,
18    p. 8-10). With the number of older Americans increasing in upcoming years (estimated to
19    increase from 12.4% of the U.S. population to 19.7% between 2000 to 2030, which is
20    approximately 35 million and 71.5  million individuals, respectively) this group represents a large
21    population potentially at risk  of Os-related health effects (SSDAN CensusScope, 2010a; U.S.
22    Census Bureau, 2010).
23          The majority of recent studies reported greater effects of short-term Os exposure and
24    mortality among older adults, which is consistent with the findings of the 2006 Os AQCD. A
25    study (Medina-Ramon and Schwartz, 2008) conducted in 48 cities across the U.S. reported larger
26    effects among adults >65 years  old compared to those < 65 years; further investigation of this
27    study population revealed a trend of Os-related mortality risk that gets larger with  increasing age
28    starting at age (Zanobetti and Schwartz, 2008). Another study conducted in 7 urban centers in
29    Chile reported similar results, with greater effects in adults >65 years old (Cakmak et al., 2007).
30    More recently, a study conducted in the same area reported similar associations  between Os
31    exposure and mortality in adults aged < 64 years old and 65 to 74 years old, but the risk was
32    increased among older age groups (Cakmak et al., 2011).  A study performed in China reported
33    greater effects in populations >45 years old (compared to 5 to 44 year olds), with statistically
34    significant effects  present only among those >65 years old (Kan et al., 2008).  An Italian study
35    reported higher risk of all-cause mortality associated with increased Os concentrations among

                                                2-65

-------
 1    individuals >85 year old as compared to those 35 to 84 years old (Stafoggia et al., 2010). The
 2    Air Pollution and Health: A European and North American Approach (APHENA) project
 3    examined the association between Os exposure and mortality for those <75 and > 75 years of
 4    age.  In Canada, the associations for all-cause and cardiovascular mortality were greater among
 5    those >75 years old. In the U.S., the  association for all-cause mortality was slightly greater for
 6    those <75 years of age compared to those >75 years old in summer-only analyses. No consistent
 7    pattern was observed for CVD mortality. In Europe, slightly larger associations for all-cause
 8    mortality were observed in those <75 years old in all-year and summer-only analyses.  Larger
 9    associations were reported among those <75years for CVD mortality in all-year analyses, but the
10    reverse was true for summer-only analyses (Katsouyanni et al., 2009).
11          With respect to epidemiologic studies of O3 exposure and hospital admissions,  a positive
12    association was reported between short-term Os exposure and respiratory hospital admissions for
13    adults >65 years old but not for those adults aged 15 to 64 years (Halonen et al., 2009). In the
14    same study, no association was observed between Os concentration and respiratory mortality
15    among those >65 years old or those 15 to 64 years old. No modification by age (40 to  64 year
16    olds versus >64 year olds) was observed in  a study from Brazil examining Os levels and COPD
17    ED visits.
18          Although some outcomes reported mixed findings regarding an increase in risk for older
19    adults, recent epidemiologic studies report consistent positive associations between short-term
20    63 exposure and mortality in older adults. The evidence from mortality studies is consistent with
21    the results reported in the 2006 Os AQCD and is supported by toxicological studies providing
22    biological  plausibility for increased risk of effects in older adults. Also, older adults may be
23    experiencing increased exposure compared to younger adults.  Overall, the ISA concludes
24    adequate evidence is available indicating that older adults are at increased risk of Os-related
25    health effects.
26    Diet
27          Diet was not examined as  a factor potentially affecting risk in previous Os AQCDs, but
28    recent studies have examined modification of the association between 63 and health effects by
29    dietary factors. Because Os mediates some of its toxic effects through oxidative stress, the
30    antioxidant status of an individual is  an important factor that may contribute to increased risk of
31    (Vrelated health effects.  Supplementation with vitamins C and E has been investigated in a
32    number of studies as a means of inhibiting Os-mediated damage.
33          Two epidemiologic studies have examined effect measure modification by diet and found
34    evidence that certain dietary components are related to the effect Os has on respiratory outcomes.

                                                2-66

-------
 1    In one recent study the effects of fruit/vegetable intake and Mediterranean diet were examined.
 2    Increases in these food patterns, which have been noted for their high vitamins C and E and
 3    omega-3 fatty acid content, were positively related to lung function in asthmatic children living
 4    in Mexico City, and modified by 63 exposure (Romieu et al., 2009). Another study examined
 5    supplementation of the diets of asthmatic children in Mexico with vitamins C and E (Sienra-
 6    Monge et al., 2004).  Associations were detected between short-term 63 exposure and nasal
 7    airway inflammation among children in the placebo group but not in those receiving the
 8    supplementation.
 9          The epidemiologic evidence is supported by controlled human exposure studies,
10    discussed in section 8.4.1 of the ISA, that have shown that the first line of defense against
11    oxidative stress is antioxidants-rich extracellular lining fluid (ELF) which scavenge free radicals
12    and limit lipid peroxidation. Exposure to Os depletes antioxidant levels in nasal ELF probably
13    due to scrubbing of 63; however, the concentration and the activity of antioxidant enzymes either
14    in ELF or plasma do not appear to be related to Os responsiveness. Controlled studies of dietary
15    antioxidant supplementation have demonstrated some protective effects of a-tocopherol (a form
16    of vitamin E) and ascorbate (vitamin C) on spirometric measures of lung function after 63
17    exposure but not on the intensity of subjective symptoms and inflammatory responses.  Dietary
18    antioxidants have also afforded partial protection to asthmatics by attenuating postexposure
19    bronchial hyperresponsiveness. Toxicological studies discussed in section 8.4.1 of the ISA
20    provide evidence of biological plausibility  to the epidemiologic and  controlled human exposure
21    studies.
22          There is adequate evidence that individuals with diets lower in vitamins C and E are at
23    risk for (Vrelated health effects.  The evidence from epidemiologic studies is supported by
24    controlled human exposure and toxicological studies.
25    Outdoor workers
26           Studies included in the 2006 O3 AQCD reported that individuals who participate in
27    outdoor activities or work outside to be a population at increased risk based on consistently
28    reported associations between 63  exposure and respiratory health outcomes in these groups (U.S.
29    EPA, 2006b). Outdoor workers are exposed to ambient Os concentrations for a greater period of
30    time than individuals who spend their days indoors. As discussed in Section 4.Y of the ISA (US
31    EPA, 2012) outdoor workers sampled during the work shift had a higher ratio of personal
32    exposure to fixed-site monitor concentrations than health clinic workers who spent most of their
33    time indoors. Additionally, an increase in  dose to the lower  airways is possible during outdoor
34    exercise due to both increases in the amount of air breathed (i.e., minute ventilation) and a shift
                                                2-67

-------
 1    from nasal to oronasal breathing.  The association between FEVi responses to 63 exposure and
 2    minute ventilation is discussed more fully in Section 6.2.3.1 of the 2006 Os AQCD.
 3          Previous studies have shown that increased exposure to O3 due to outdoor work leads to
 4    increased risk of Os-related health effects, specifically decrements in lung function (U.S. EPA,
 5    2006b).  The strong evidence from the 2006 Os AQCD which demonstrated increased exposure,
 6    dose, and ultimately risk of Os-related health effects in this population supports the conclusion
 7    that there is adequate evidence to indicate that increased exposure to Os through outdoor work
 8    increases the risk of (Vrelated health effects.
 9          2.2.4.3 Factors that potentially increase risk
10          There were four factors for which the ISA concludes that there is suggestive evidence for
11    potential increased risk of O3-related health effects. These include genetic factors,  sex, SES and
12    obesity.  Each factor is discussed briefly below.
13          For variants in multiple genes there is suggestive evidence for potential increased risk to
14    some populations from 63 exposure. Controlled human exposure and epidemiologic studies
15    have reported some evidence of O3-related increases in respiratory symptoms or decreases in
16    lung function with variants including GSTM1, GSTP1, HMOX1 and NQO1, although the results
17    are not consistent across studies and gene variants. Future studies of these and other genes in
18    human populations will be important for determining the role of each genotype and its effect on
19    risk as well as finding coherence across the disciplines.
20          With respect to effect measure modification by sex, most studies examining the
21    associations 63 and mortality report females to be at greater risk than males, but minimal
22    evidence is available regarding a difference between the sexes for other outcomes. Inconsistent
23    findings were reported for respiratory and cardiovascular hospital admissions and ED visits,
24    although there is some indication that females are at increased risk of (Vrelated respiratory
25    hospital  admissions and ED visits. While Os-related effects may occur in both men and women,
26    there is suggestive evidence exists indicating that females are at potentially increased risk of Cv
27    related health effects as there are consistent findings among epidemiologic studies of mortality.
28          Overall, most studies have reported that individuals with low SES and those living in
29    neighborhoods with low SES are more at risk for Os-related health effects, resulting in increased
30    risk of respiratory hospital  admissions and ED visits. Inconsistent results have been observed in
31    the few studies examining effect modification of associations between 63 exposure and mortality
32    and reproductive outcomes. Also, a controlled human exposure  study does not support evidence
33    of increased risk of respiratory morbidity among individuals with lower SES.  Overall, evidence
34    is suggestive of SES as a factor affecting risk of O3-related health outcomes based on collective

                                                2-68

-------
 1    evidence from epidemiologic studies of respiratory hospital admissions, but there is
 2    inconsistency among studies of mortality and reproductive outcomes. Further studies are needed
 3    to confirm this relationship, especially in populations within the U.S.
 4           Multiple epidemiologic, human clinical and toxicological studies have reported
 5    suggestive evidence for increased CVrelated respiratory health effects among obese individuals.
 6    Future research of the effect modification by body mass index on the relationship between Os
 7    and nonrespiratory-related health outcomes, and studies examining the role of physical
 8    conditioning will advance understanding of obesity as a factor potentially increasing risk.
 9           2.4.2.4 Summary  of factors that increase or potentially increase risk
10           In this section, epidemiologic, controlled human exposure and toxicological studies have
11    been discussed which indicate that various factors may lead to increased risk of O3-related health
12    effects.  The populations identified in chapter 8 of the ISA that have "adequate" evidence for
13    (Vrelated health effects are individuals with asthma, younger and older age groups, individuals
14    with certain dietary deficiencies, and outdoor workers, based on consistency in findings across
15    studies and evidence of coherence in results from different scientific disciplines. Asthma as a
16    factor affecting risk was supported by controlled human exposure and toxicological studies, as
17    well as some evidence from epidemiologic studies. Generally, studies of age groups reported
18    positive associations for respiratory hospital admissions and ED visits among children.
19    Biological plausibility for this increased risk is supported by toxicological and clinical research.
20    Children have higher exposure and dose due to increased time spent outdoors and ventilation
21    rate, their lungs are still developing, and they are more likely than adults to have asthma. Most
22    studies comparing age groups reported greater effects of short-term O3 exposure on mortality
23    among older adults, although studies of other health outcomes had inconsistent findings
24    regarding whether older adults were at increased risk. Older adults may also withstand greater
25    O3 exposure and not seek relief as quickly as younger adults.  Multiple epidemiologic, controlled
26    human exposure and toxicological studies reported that diets deficient in vitamins E and C are
27    associated with risk of (Vrelated health effects.  Previous studies have shown that increased
28    exposure to Os due to outdoor work leads to increased risk of CVrelated health effects and it is
29    clear that outdoor workers  have higher exposures, and possibly greater internal doses, of 63,
30    which may lead to increased risk of (Vrelated health effects.
31           In some cases, it is  difficult to determine a factor that results in increased risk of effects.
32    For example, previous assessments have included controlled human exposure studies in which
33    some healthy individuals demonstrate greater CVrelated health effects compared to other healthy
34    individuals.  Intersubject variability has been observed for lung function decrements,
35    symptomatic responses, pulmonary inflammation, AHR, and altered epithelial permeability in
                                                2-69

-------
 1    healthy adults exposed to 63 and these results tend to be reproducible within a given individual
 2    over a period of several months indicating differences in the intrinsic responsiveness.  In many
 3    cases the reasons for the variability is not clear. This may be because one or some of the factors
 4    described above have not been evaluated in studies, or it may be that additional, unidentified
 5    factors influence individual responses to Os.
 6          As discussed in chapter 8 of the ISA, the challenges and limitations in evaluating the
 7    factors that can increase risk for experiencing O3-related health effects may contribute to a lack
 8    of information about the factors that may increase risk from Os exposures.  This lack of
 9    information may contribute to conclusions that evidence for some factors, such as genetic
10    factors, sex,  SES, and obesity  provided "suggestive" evidence of increased risk, or that for a
11    number of factors the evidence was inadequate to draw conclusions about potential increase in
12    risk of effects.  Overall, the factors most strongly supported as contributing to increased risk of
13    populations for experience Os-related effects were related to asthma, lifestage (children and older
14    adults), dietary factors, and working outdoors.
15    2.4.3  Averting Behavior
16          The activity pattern of individuals is an important determinant of their exposure.
17    Variation in  Os concentrations among various microenvironments means that the amount of time
18    spent in each location, as well as the level of activity, will influence an individual's exposure to
19    ambient O^.  Activity patterns vary both among and within individuals, resulting in
20    corresponding variations in exposure across a population and over time. Individuals can reduce
21    their exposure to Os by altering their behaviors, such as by staying indoors, being active outdoors
22    when air quality is better, and  by reducing their activity levels or reducing the time being active
23    outdoors on high-Os days.  This is a topic that was not discussed in the 2006 AQCD.  The
24    evidence discussed below is new in this review.
25          The EPA has developed the Air Quality Index (AQI) to provide the public information
26    about ambient levels of common air pollutants and associated health effects, if any
27    (www.airnow.gov).  The AQI describes the potential for health effects from Os (and other
28    individual pollutants) in six color-coded categories of air-quality, ranging from Good (green),
29    Moderate (yellow), Unhealthy for  Sensitive Groups (orange), Unhealthy (red), and Very
30    Unhealthy (purple), and Hazardous (maroon).  Levels in the unhealthy ranges (i.e., Unhealthy for
31    Sensitive Groups and above) come with recommendations about reducing exposure. Forecasted
32    and actual AQI values for 63 are reported to the public during the 63 season.
33          The AQI advisories explicitly state that children, older adults, people with lung disease,
34    and people who are active outdoors, may be at greater risk from exposure to O3. People are
35    advised to reduce exposure depending on the predicted Os levels  and the likelihood of risk.  This
                                                2-70

-------
 1    advice includes being active outdoors when air quality is better, and reducing activity levels or
 2    reducing the time being active outdoors on high-Os days.  Staying indoors to reduce exposure is
 3    not recommended until air quality reaches the Very Unhealthy or Hazardous categories.
 4          Evidence of individual averting behaviors in response to AQI advisories has been found
 5    in several studies, including activity pattern and epidemiologic studies, especially for the at-risk
 6    populations, such as children, older adults, and people with asthma, who are targeted by the
 7    advisories. Such effects are less pronounced in the general population, possibly due to the
 8    opportunity cost of behavior modification. Epidemiologic evidence from a study (Neidell and
 9    Kinney,  2010) conducted in the 1990's in Los Angeles, CA reports increased asthma hospital
10    admissions among children and older adults when 63 alert days (1-hour max 63 concentration
11    >200 ppb) were excluded from the analysis of daily hospital admissions and O3 concentrations
12    (presumably thereby eliminating averting behavior based on high Os forecasts).  The lower rate
13    of admissions observed when alert days were included in the analysis suggests that estimates of
14    health effects based on concentration-response functions that do not account for averting
15    behavior may be biased towards the null.
16    2.4.4  Size of At-Risk Populations in the United States
17          One consideration in the assessment of potential public health impacts is the size of
18    various population groups for which there is adequate evidence of increased risk for health
19    effects associated with O3-related air pollution exposure.  The factors for which the ISA (EPA,
20    2012a, section 8.5, p. 8-36) judged the evidence to be "adequate" with respect to contributing to
21    increased risk of (Vrelated effects among various populations and lifestages included: asthma;
22    age group (children and older adults); dietary factors; and, working outdoors.
23          With regard to asthma, Table 2.3 above summarizes information on the prevalence of
24    current asthma by age in the U.S. adult population in 2010 (Schiller et al. 2012; children - Bloom
25    et al., 2011). Individuals with current asthma constitute a fairly large proportion of the
26    population, including more than 25 million people.  Asthma prevalence tends to be higher in
27    children than adults.
28          With regard to lifestages, based on U.S. census data from 2010 (Howden and Meyer,
29    2011), about 74 million people, or 24% of the U.S. population, are under 18 years of age and
30    more than 40 million people, or about 13% of the U.S. population, are 65 years of age or older.
31    Hence, large proportions of the U.S. population are included in age groups that are considered
32    likely to be at increased risk for health effects from ambient 63 exposure.
33          With regard to dietary factors, no statistics are available to estimate the size of an at-risk
34    population based on nutritional status.  However, in order to get an approximate estimate of the

                                                2-71

-------
 1    potential size of this population, inference can be made that people living below the poverty level
 2    would be more likely to have reduced nutritional status. Based on data from the 2010 Census,
 3    about 46 million people (about 15% of the total population) live below the poverty level
 4    (DeNavas-Walt et al., 2011). Of these, about 26 million people were in the age range of 18 to 64
 5    years (about 14 % of the people in that age range), and therefore would not be included in the
 6    groups (i.e., children and older adults) considered to be at increased risk from 63 exposure
 7    simply because of their age.
 8          With regard to outdoor workers, in 2010 approximately 11.7% of the total number of
 9    people (143 million people) employed, or about 16.8 million people, worked outdoors one or
10    more day per week (O*NetOnline:
11    http://www.onetonline.Org/fmd/descriptor/result/4.C.2.a. l.c?a=l). Of these approximately 7.4%
12    of the workforce, or about 7.8 million people, worked outdoors three or more days per week.
13          The health statistics data illustrate what is known as the "pyramid" of effects.  At the top
14    of the pyramid, there are approximately 2.5 million deaths from all causes per year in the U.S.
15    population, with about 250 thousand respiratory-related deaths (CDC-WONDER, 2008). For
16    respiratory health diseases, there are nearly 3.3 million hospital discharges per year (HCUP,
17    2007), 8.7 million respiratory ED visits (HCUP, 2007), 112 million ambulatory care visits
18    (Woodwell and Cherry, 2004), and an estimated 700 million restricted activity days per year due
19    to respiratory conditions (Adams et al., 1999). Combining small risk estimates with relatively
20    large baseline levels of health outcomes can result in quite large public health impacts. Thus,
21    even a small percentage reduction in Os health impacts on cardiopulmonary diseases  would
22    reflect a large number of avoided cases.

23    2.5    INTEGRATED DISCUSSION OF THE EVIDENCE

24          In this section, we revisit the overarching question for this chapter:
25    •   To what extent has scientific information become available that alters or substantiates
26       our understanding of the health effects that occur following short-term or long-term
27       exposures to Os, and our understanding of the Os concentrations at which such effects
28       occur?
29          As in the last review, the clearest evidence for health effects associated with exposure to
30    63 is provided by studies of respiratory effects. Collectively, there is a vast amount of evidence
31    spanning several decades that supports a causal association between exposure to Os and a
32    continuum of respiratory effects (US EPA,  2012a, section 2.5). The majority of this evidence is
33    derived from studies investigating short-term exposures (i.e., hours to weeks) to 63, although
                                               2-72

-------
 1    animal toxicological studies and recent epidemiologic evidence demonstrate that long-term
 2    exposure (i.e., months to years) may also be detrimental to the respiratory system.
 3          The 2006 O3 AQCD concluded that there was clear, consistent evidence of a causal
 4    relationship between short-term exposure to Os and respiratory health effects (U.S. EPA, 2006).
 5    This causal relationship was substantiated by the coherence of effects observed across controlled
 6    human exposure, epidemiologic, and toxicological studies indicating associations of short-term
 7    Os exposures with a range of respiratory health endpoints. Across disciplines, short-term Os
 8    exposures induced or were associated with statistically significant declines in lung function.  An
 9    equally strong body of evidence from controlled human exposure and toxicological studies
10    demonstrated Os-induced inflammatory responses, increased epithelial permeability, and airway
11    hyperresponsiveness. Toxicological studies provided additional evidence for O3-induced
12    impairment of host defenses. Combined, these findings from experimental studies provided
13    support for epidemiologic evidence, in which short-term Os exposure was consistently associated
14    with increases in respiratory symptoms and asthma medication use in asthmatic children,
15    respiratory-related hospital admissions, and asthma-related emergency department visits.
16    Although Os was consistently associated with non-accidental and cardiopulmonary mortality, the
17    contribution  of respiratory causes to these findings was uncertain. The combined evidence
18    across disciplines supported a causal relationship between short-term Os exposure and
19    respiratory effects (US EPA, 2006).
20          Studies conducted since the last review have generally supported the conclusions of the
21    2006 AQCD regarding the strength of the evidence for respiratory effects following short-term
22    Os exposures, with more recent studies reporting such effects following exposures to lower Os
23    concentrations than previously reported. Specifically, as discussed above (section 2.2.1), recent
24    controlled human exposure studies have reported lung function decrements, respiratory
25    symptoms, and airway inflammation following exposures to Os concentrations from 60 to 80 ppb
26    and recent epidemiologic studies have reported O3-associated airway inflammation in locations
27    with ambient Os concentrations below 75 ppb.  These recent studies reinforce the large body of
28    existing  experimental and epidemiologic evidence for respiratory effects following exposures to
29    somewhat higher Os concentrations.
30          The frequency of emergency department visits and hospital admissions due to respiratory
31    symptoms, asthma exacerbations and other respiratory  diseases is associated with short- and
32    long-term exposure to ambient Os (US EPA,  2012a, section 2.5.3). Summertime daily hospital
33    admissions for respiratory causes in various locations of eastern North America were
34    consistently associated with ambient levels of Os in studies reviewed in the 1996 Os AQCD.
35    The 2006 Os AQCD concluded that aggregate population time-series studies demonstrate a

                                               2-73

-------
 1    positive and robust association between ambient 63 concentrations and respiratory-related
 2    hospitalizations and asthma emergency department visits during the warm season. Recent
 3    epidemiologic time-series studies that include additional multicity studies and a multicontinent
 4    study further support that short-term exposures to ambient 63 concentrations are consistently
 5    associated with increases in respiratory hospital admissions and emergency department visits
 6    specifically during the warm/summer months in multiple geographic locations and across a range
 7    of O3 concentrations (Section 6.2.7). Recent evidence from several multicity studies and a
 8    multicontinent study also demonstrate  consistent positive associations between short-term
 9    exposure to ambient 63 concentrations and increases in respiratory mortality (Section 6.6.2.5).
10    Evidence from recent mortality studies is consistent and coherent with the evidence from
11    epidemiologic, controlled human exposure, and animal toxicological studies for the respiratory
12    effects of 63 exposures.  Additionally,  the evidence for respiratory morbidity after short- and
13    long-term exposure  provides biological plausibility for mortality due to respiratory disease.
14           In addition, recent epidemiologic studies provide greater insight into factors that may
15    increase susceptibility to (^-associated respiratory morbidity (e.g., presence of asthma and
16    respiratory infection; presence of airway hyperresponsiveness or elevated body mass index,
17    particularly in older adults; and groups with diminished antioxidant capacity). Across endpoints,
18    recent studies indicate that groups with diminished antioxidant capacity or comorbidities such as
19    atopy, AHR, or elevated body mass index may have increased susceptibility to respiratory
20    morbidity associated with Os exposure. The potential susceptibility of these populations
21    identified in recent epidemiologic studies are strongly supported by findings from experimental
22    studies that demonstrated O3-induced decreases in intracellular antioxidant levels, increases in
23    airway responses with co-exposures to allergens, and increases in airway responses in animal
24    models of obesity.  By demonstrating (Vinduced airway hyperresponsiveness, decreased
25    pulmonary function, allergic responses, lung injury, impaired host defense, and airway
26    inflammation, recent toxicological studies have characterized Os modes of action and have
27    provided biological  plausibility for epidemiologic associations of ambient 63 exposure with lung
28    function and respiratory symptoms, hospital admissions, emergency department visits, and
29    mortality. Together, the evidence integrated across controlled human exposure, epidemiologic,
30    and toxicological studies and across the spectrum of respiratory health endpoints continues to
31    demonstrate that there is a causal relationship between short-term Os exposure and respiratory
32    health effects (US EPA, 2012a, section 6.2.1).
33           The strongest evidence for a relationship between long-term 63 exposure and respiratory
34    morbidity is contributed by recent studies from a single cohort demonstrating associations
35    between long-term measures of Os exposure and new-onset asthma in children and increased

                                                2-74

-------
 1    respiratory symptom effects in asthmatics. While the evidence is limited, this U.S. multi-
 2    community prospective cohort demonstrates that asthma risk is affected by interactions among
 3    genetic variability, environmental O?, exposure, and behavior. Other recent studies provide
 4    coherent evidence for long-term 63 exposure and respiratory morbidity effects such as first
 5    asthma hospitalization and respiratory symptoms in asthmatics.  Generally, the epidemiologic
 6    and toxicological evidence provides a compelling case that supports the hypothesis that a
 7    relationship exists between long-term exposure to ambient O3 and measures of respiratory
 8    morbidity. The evidence for short-term exposure to Os and effects on respiratory endpoints
 9    provides coherence and biological plausibility for the effects of long-term exposure to 63.
10    Building upon that evidence, the more recent epidemiologic evidence, combined with
11    toxicological studies in rodents and non-human primates, provides biologically plausible
12    evidence that there is likely to be a causal relationship between long-term exposure to 63 and
13    respiratory health effects.

14    2.6    REFERENCES

15    Adams, W. C. (2006a) Comparison of chamber 6.6 hour exposures to 0.04-0.08 ppm  ozone via
16         square-wave and triangular profiles on pulmonary responses. Inhalation Toxicol. 18: 127-
17         136. http://dx.doi.org/10.1080/08958370500306107
18    Adams, WC. (2006b). Human pulmonary responses with 30-minute time intervals of exercise
19         and rest when exposed for 8 hours to 0.12 ppm ozone via square-wave and acute triangular
20         profiles. Inhal Toxicol 18: 413-422. http://dx.doi.org/10.1080/08958370600563599
21    Adams, WC. (2003a). Comparison of chamber and face mask 6.6-hour exposure to 0.08 ppm
22         ozone via square-wave  and triangular profiles on pulmonary responses. Inhal Toxicol 15:
23         265-281.
24    Adams, WC. (2003b). Relation of pulmonary responses induced by 66-h exposures to 0.08 ppm
25         ozone and 2-h exposures to 0.30 ppm ozone via chamber and face-mask inhalation. Inhal
26         Toxicol 15: 745-759.
27    Adams, WC. (2002). Comparison of chamber and face-mask 6.6-hour exposures to ozone on
28         pulmonary function and symptoms responses. Inhal Toxicol 14: 745-764.
29         http://dx.doi.org/10.1080/08958370290084610
30    Adams, WC. (1998). Dose-response effect of varied equivalent minute ventilation rates on
31         pulmonary function responses during exposure to ozone. Washington, DC: American
32         Petroleum Institute.
                                               2-75

-------
 1    Akinbami, LJ; Lynch, CD; Parker, JD; Woodruff, TJ. (2010). The association between childhood
 2         asthma prevalence and monitored air pollutants in metropolitan areas, United States, 2001-
 3         2004. Environ Res 110: 294-301. http://dx.doi.Org/10.1016/j.envres.2010.01.001

 4    Alexeeff, SE; Litonjua, AA; Suh, H; Sparrow, D; Vokonas, PS; Schwartz, J. (2007). Ozone
 5         exposure and lung function: Effect modified by obesity and airways hyperresponsiveness in
 6         the VA Normative Aging Study. Chest 132: 1890-1897. http://dx.doi.org/10.1378/chest.07-
 7         1126

 8    Alexis, NE; Lay, JC; Hazucha, M; Harris, B; Hernandez, ML; Bromberg, PA; Kehrl, H; Diaz-
 9         Sanchez, D; Kim, C; Devlin, RB; Peden, DB. (2010). Low-level ozone exposure induces
10         airways inflammation and modifies cell surface phenotypes in healthy humans. Inhal
11         Toxicol 22: 593-600.

12    ATS (American Thoracic Society). (2000). What constitutes an adverse health effect of air
13         pollution? Am. J. Respir. Crit. Care Med.  161: 665-673.

14    ATS (American Thoracic Society). (1991). Lung function testing: selection of reference values
15         and interpretive strategies. Am.  Rev. Respir. Dis. 144: 1202-1218.

16    ATS (American Thoracic Society). (1985) Guidelines as to what constitutes an adverse
17         respiratory health effect, with special reference to epidemiological studies of air pollution.
18         Am. Rev. Respir. Dis. 131: 666-668.

19    Barraza-Villarreal, A; Sunyer, J; Hernandez-Cadena, L; Escamilla-Nunez, MC; Sienra-Monge,
20         JJ; Ramirez-Aguilar, M; Cortez-Lugo, M; Holguin, F; Diaz-Sanchez, D; Olin, AC;
21         Romieu, I. (2008). Air pollution, airway inflammation, and lung function in a cohort study
22         of Mexico City schoolchildren. Environ Health Perspect 116: 832-838.
23         http://dx.doi.org/10.1289/ehp.10926

24    Basha, MA; Gross, KB; Gwizdala,  CJ; Haidar, AH; Popovich, J, Jr. (1994). Bronchoalveolar
25         lavage neutrophilia in asthmatic and healthy volunteers after controlled exposure to ozone
26         and filtered purified air. Chest 106: 1757-1765.

27    Bell, ML; Dominici, F. (2008). Effect modification by community characteristics on the short-
28         term effects of ozone exposure and mortality in 98 US communities. Am J Epidemiol 167:
29         986-997. http://dx.doi.org/10.1093/aje/kwm396

30    Bell, ML; Kim, JY; Dominici, F. (2007). Potential confounding of particulate matter on the
31         short-term association between ozone and mortality in multisite time-series studies.
32         Environ Health Perspect 115: 1591-1595.  http://dx.doi.org/10.1289/ehp.10108
                                               2-76

-------
 1    Bell, ML; Peng, RD; Dominici, F. (2006). The exposure-response curve for ozone and risk of
 2        mortality and the adequacy of current ozone regulations. Environ Health Perspect 114:  532-
 3        536.

 4    Bell, M. L.; McDermott, A.; Zeger, S. L.; Samet, J. M.; Dominici, F. (2004) Ozone and short-
 5        term mortality in 95 US urban communities, 1987-2000. JAMA J. Am. Med. Assoc. 292:
 6        2372-2378.

 7    Berhane, K; Zhang, Y; Linn, WS; Rappaport, EB; Bastain, TM; Salam, MT; Islam, T; Lurmann,
 8        F; Gilliland, FD. (2011). The effect of ambient air pollution on exhaled nitric oxide in the
 9        Children's Health Study. Eur Respir J 37: 1029-1036.
10        http://dx.doi.org/10.1183/09031936.00081410

11    Biggeri, A; Baccini, M; Bellini, P; Terracini, B. (2005). Meta-analysis of the Italian studies of
12        short-term effects of air pollution (MISA), 1990-1999. Int J Occup Environ Health 11:  107-
13        122.

14    Bloom B, Cohen RA, Freeman G.  Summary health statistics for U.S. children: National Health
15        Interview Survey, 2010. National Center for Health Statistics. Vital Health Stat 10(250).
16        2011. http://www.cdc.gov/nchs/data/series/sr_10/srl0_250.pdf
17    Bosson, J; Stenfors, N; Bucht, A; Helleday, R; Pourazar, J; Holgate, ST; Kelly, FJ; Sandstrom,
18        T; Wilson, S; Frew, AJ; Blomberg, A. (2003). Ozone-induced bronchial epithelial cytokine
19        expression differs between healthy and asthmatic subjects. Clin Exp Allergy  33: 777-782.

20    Brauer, M; Blair, J; Vedal, S. (1996). Effect of ambient ozone exposure on lung function in farm
21        workers. Am J Respir Crit Care Med 154:  981-987.

22    Brauer, M; Brook, JR. (1997). Ozone personal exposures and health effects for selected groups
23        residing in the Fraser Valley. Atmos Environ 31: 2113-2121.

24    Breton, CV; Salam, MT; Vora, H; Gauderman, WJ; Gilliland, FD. (2011). Genetic variation  in
25        the glutathione synthesis pathway, air pollution, and children's lung function growth. Am J
26        Respir Crit Care Med 183: 243-248. http://dx.doi.org/10.1164/rccm.201006-0849OC

27    Brown, JS; Bateson,  TF; McDonnell, WF. (2008). Effects of exposure to 0.06 ppm ozone on
28        FEV1 in humans: A secondary analysis of existing data. Environ Health Perspect 116:
29        1023-1026. http://dx.doi.org/10.1289/ehp.11396

30    Brunekreef, B; Hoek, G; Breugelmans, O; Leentvaar, M. (1994). Respiratory effects of low-level
31        photochemical  air pollution in amateur cyclists. Am J Respir Crit Care Med 150: 962-966.

32    Cakmak, S; Dales, RE; Angelica Rubio, M; Blanco Vidal, C. (2011). The risk of dying on days
33        of higher air pollution among the socially disadvantaged elderly. Environ Res 111: 388-
34        393. http://dx.doi.0rg/10.1016/i.envres.2011.01.003

                                               2-77

-------
 1    Cakmak, S; Dales, RE; Judek, S. (2006a). Do gender, education, and income modify the effect of
 2         air pollution gases on cardiac disease? J Occup Environ Med 48: 89-94.
 3         http://dx.doi.org/10.1097/01.iom.0000184878.11956.4b

 4    Cakmak, S; Dales, RE; Judek, S. (2006b). Respiratory health effects of air pollution gases:
 5         Modification by education and income. Arch Environ Occup Health 61: 5-10.
 6         http://dx.doi.Org/10.3200/AEOH.61.l.5-10

 7    Carey, SA; Minard, KR; Trease, LL; Wagner, JG; Garcia, GJ; Ballinger, CA; Kimbell, JS;
 8         Plopper, CG; Corley, RA; Postlethwait, EM; Harkema, JR; Einstein, DR. (2007). Three-
 9         dimensional mapping of ozone-induced injury in the nasal airways of monkeys using
10         magnetic resonance imaging and morphometric techniques. Toxicol Pathol 35: 27-40.
11         http://dx.doi.org/10.1080/01926230601072343

12    Chan, CC; Wu, TH. (2005). Effects of ambient ozone exposure on mail carriers' peak expiratory
13         flow rates. Environ Health Perspect 113: 735-738. http://dx.doi.org/10.1289/ehp.7636

14    Chhabra, SK; Yasir, A;  Chaudhry, K; Shah, B. (2010). Effect of ozone on response to ovalbumin
15         & its modulation by vitamins C & E in sensitized guinea pigs. Indian J Med Res 132: 87-
16         93.

17    Coffin, DL; Blommer, EJ; Gardner, DE; Holzman, R. (1967). Effect of air pollution on alteration
18         of  susceptibility to pulmonary infection. Cincinnati, OH: U.S. Department of Health,
19         Education, and Welfare.

20    Coffin, DL; Gardner, DE.  (1972). Interaction of biological agents and chemical air pollutants.
21         Ann Occup Hyg 15:  219-234.

22    Cohen, MD; Sisco, M; Baker, K; Li, Y; Lawrence, D; Van Loveren, H; Zelikoff, JT;
23         Schlesinger, RB. (2002). Effects of inhaled  ozone  on pulmonary immune cells critical to
24         antibacterial responses in situ. Inhal Toxicol 14: 599-619.
25         http://dx.doi.org/10.1080/08958370290084520

26    Dales, RE; Cakmak, S; Doiron, MS. (2006). Gaseous air pollutants and hospitalization for
27         respiratory disease in the neonatal period. Environ Health Perspect 114: 1751-1754.
28         http://dx.doi.org/10.1289/ehp.9044

29    Darrow, LA; Klein, M;  Sarnat, JA; Mulholland, JA; Strickland, MJ; Sarnat, SE; Russell, AG;
30         Tolbert, PE. (201 la). The use  of alternative pollutant metrics in time-series studies of
31         ambient air pollution and respiratory emergency department visits. J Expo Sci Environ
32         Epidemiol 21: 10-19. http://dx.doi.org/10.1038/ies.2009.49
3 3    DeNavas-Walt, Carmen, Bemadette D. Proctor, and Jessica C. Smith, U.S. Census Bureau, Current Population Reports, P60-239,
34         Income, Poverty, and Health Insurance Coverage in the United States: 2010, U.S. Government Printing Office,
35         Washington, DC,2011. http://www.census.gov/prod/2011pubs/p60-239.pdf

                                                2-78

-------
 1   Depuydt, P; Joos, GF; Pauwels, RA. (1999). Ambient ozone concentrations induce airway
 2        hyperresponsiveness in some rat strains. Eur Respir J 14: 125-131.
 3   Devlin, RB; Mcdonnell, WF; Mann, R; Becker, S; House, DE; Schreinemachers, D; Koren, HS.
 4        (1991). Exposure of humans to ambient levels of ozone for 6.6 hours causes cellular and
 5        biochemical changes in the lung. Am J Respir Cell Mol Biol 4: 72-81.

 6   Dimeo, MJ; Glenn, MG; Holtzman, MJ; Shelter, JR; Nadel, JA; Boushey, HA. (1981). Threshold
 7        concentration of ozone causing an increase in bronchial reactivity in humans and
 8        adaptation with repeated exposures. Am Rev Respir Dis 124: 245-248.

 9   Driscoll, KE; Vollmuth, TA; Schlesinger, RB. (1987). Acute and subchronic ozone inhalation in
10        the rabbit: Response of alveolar macrophages. J Toxicol Environ Health 21: 27-43.
11        http://dx.doi.org/10.1080/15287398709531000

12   Dryden, DM; Spooner, CH; Stickland, MK; Vandermeer, B; Tjosvold, L; Bialy, L; Wong, K;
13        Rowe, BH. (2010). Exercise-induced bronchoconstriction and asthma. (AHRQ Publication
14        No. 10-E001). Rockville, MD: Agency for Healthcare Research and Quality.

15   Eiswerth, ME; Shaw, WD; Yen, ST. (2005). Impacts of ozone on the activities of asthmatics:
16        Revisiting the data. J Environ Manage 77: 56-63.
17        http://dx.doi.0rg/10.1016/j.jenvman.2005.02.010

18   Escamilla-Nufiez, MC; Barraza-Villarreal, A; Hernandez-Cadena, L; Moreno-Macias, H;
19        Ramirez-Aguilar, M; Sienra-Monge, JJ; Cortez-Lugo, M; Texcalac, JL; del Rio-Navarro,
20        B; Romieu, I. (2008). Traffic-related air pollution and respiratory symptoms among
21        asthmatic  children, resident in Mexico City: The EVA cohort study. Respir Res 9: 74.
22        http://dx.doi.org/10.1186/1465-9921-9-74

23   Fanucchi, MV;  Plopper, CG; Evans, MJ; Hyde, DM; Van Winkle, LS; Gershwin, LJ; Schelegle,
24        ES. (2006). Cyclic exposure to ozone alters  distal  airway development in infant rhesus
25        monkeys.  Am J Physiol Lung Cell Mol Physiol 291: L644-L650.
26        http://dx.doi.org/10.1152/aiplung.00027.2006

27   Folinsbee, LJ; Hazucha, MJ. (2000). Time course of response to ozone exposure in healthy adult
28        females. Inhal Toxicol 12:  151-167.

29   Folinsbee, LJ; Hazucha, MJ. (1989). Persistence of ozone-induced changes in lung function and
30        airway responsiveness. In Atmospheric ozone  research and its policy implications.
31        Amsterdam, The Netherlands: Elsevier.

32   Folinsbee, LJ; Horstman, DH; Kehrl, HR; Harder, S; Abdul-Salaam, S; Ives, PJ. (1994).
33        Respiratory responses to repeated prolonged exposure to 0.12 ppm ozone. Am J Respir Crit
34        Care Med 149:98-105.
                                              2-79

-------
 1   Franklin, M; Schwartz, J. (2008). The impact of secondary particles on the association between
 2        ambient ozone and mortality. Environ Health Perspect 116: 453-458.
 3        http://dx.doi.org/10.1289/ehp.10777

 4   Gauderman, WJ; Avol, E; Gilliland, F; Vora, H; Thomas, D; Berhane, K; McConnell, R;
 5        Kuenzli, N; Lurmann, F; Rappaport, E; Margolis, H; Bates, D; Peters, J. (2004). The effect
 6        of air pollution on lung development from 10 to 18 years of age. N Engl J Med 351: 1057-
 7        1067.

 8   Gauderman, WJ; McConnell, R; Gilliland, F; London, S; Thomas, D; Avol, E; Vora, H; Berhane,
 9        K; Rappaport, EB; Lurmann, F; Margolis, HG; Peters, J. (2000). Association between air
10        pollution and lung function growth in  southern California children. Am J Respir Crit Care
11        Med 162: 1383-1390.

12   Gielen, MH; Van Der Zee, SC; Van Wijnen, JH; Van Steen, CJ; Brunekreef, B. (1997). Acute
13        effects of summer air pollution on respiratory health of asthmatic children. Am J Respir
14        Crit Care Med 155: 2105-2108.

15   Gong, H, Jr; McManus, MS; Linn, WS. (1997). Attenuated response to repeated daily ozone
16        exposures in asthmatic subjects.  Arch  Environ Occup Health 52: 34-41.

17   Greer, JR; Abbey, DE; Burchette, RJ.  (1993). Asthma related to occupational and ambient air
18        pollutants in nonsmokers. J Occup Environ Med 35: 909-915.

19   Harkema, JR; Plopper, CG; Hyde, DM; St George, JA;  Dungworth, DL.  (1987a). Effects of an
20        ambient level of ozone on primate nasal epithelial mucosubstances: quantitative
21        histochemistry. Am J Pathol 127: 90-96.

22   Hatch, GE; Slade, R; Harris, LP; Mcdonnell, WF; Devlin, RB; Koren, HS; Costa, DL; Mckee, J.
23        (1994). Ozone dose and effect in humans  and rats: A comparison using oxygen-18 labeling
24        and bronchoalveolar lavage. Am J Respir Crit Care Med 150: 676-683.

25   Henderson, R. (2008) Letter from CASAC Chairman Rogene Henderson to EPA Administrator
26        Stephen Johnson. April 7, 2008,  EPA-CASAC-08-001.
27   Henderson, R. (2007) Letter from CASAC Chairman Rogene Henderson to EPA Administrator
28        Stephen Johnson. March 26, 2007, EPA-CASAC-07-002.

29   Henderson, R. (2006) Letter from CASAC Chairman Rogene Henderson to EPA Administrator
30        Stephen Johnson. October 24, 2006, EPA-CASAC-07-001.
31   Hernandez, ML; Lay, JC; Harris, B; Esther, CR; Brickey, WJ; Bromberg, PA; Diaz-Sanchez, D;
32        Devlin, RB; Kleeberger, SR; Alexis, NE;  Peden, DB. (2010). Atopic asthmatic subjects but
33        not atopic subjects without asthma have enhanced inflammatory response to ozone. J
34        Allergy Clin Immunol 126: 537-544. http://dx.doi.Org/10.1016/j.jaci.2010.06.043

                                             2-80

-------
 1   Hernandez-Cadena, L; Holguin, F; Barraza-Villarreal, A; Del Rio-Navarro, BE; Sienra-Monge,
 2        JJ; Romieu, I. (2009). Increased levels of outdoor air pollutants are associated with reduced
 3        bronchodilation in children with asthma. Chest 136: 1529-1536.
 4        http://dx.doi.org/10.1378/chest.08-1463

 5   Halonen, JI; Lanki, T; Tiittanen, P; Niemi, JV; Loh, M; Pekkanen, J. (2009). Ozone and cause-
 6        specific cardiorespiratory morbidity and mortality. J Epidemiol Community Health 64:
 7        814-820. http://dx.doi.org/10.1136/iech.2009.087106

 8   Holz, O; Torres, RA; Timm, P; Mucke, M; Richter, K; Koschyk, S; Magnussen, H. (1999).
 9        Ozone-induced airway inflammatory changes differ between individuals and are
10        reproducible. Am J Respir Crit Care Med 159: 776-784.

11   Hoppe, P; Peters, A; Rabe, G; Praml, G; Lindner, J; Jakobi, G; Fruhmann, G; Nowak, D. (2003).
12        Environmental ozone effects in different population subgroups. Int J Hyg Environ Health
13        206: 505-516. http://dx.doi.org/10.1078/1438-4639-00250

14   Hoppe, P; Praml, G; Rabe, G; Lindner, J; Fruhmann, G; Kessel, R. (1995). Environmental ozone
15        field study on pulmonary and subjective responses of assumed risk groups. Environ Res 71:
16        109-121.

17   Horstman, DH; Ball, BA; Brown, J; Gerrity, T; Folinsbee, LJ. (1995). Comparison of pulmonary
18        responses of asthmatic and nonasthmatic subjects performing light exercise while exposed
19        to a low level of ozone. Toxicol Ind Health 11: 369-385.

20   Horstman, DH; Folinsbee, LJ; Ives, PJ;  Abdul-Salaam, S; McDonnell, WF. (1990). Ozone
21        concentration and pulmonary response relationships for 6.6-hour exposures with five hours
22        of moderate exercise to 0.08, 0.10, and 0.12 ppm. Am J Respir Crit Care Med 142: 1158-
23        1163.

24   Howden, Lindsay M. and Meyer, Julie A., U.S. Census Bureau, 2010 Census Briefs, C2010BR-
25        03, Age and Sex Composition: 2010, U.S. Department of Commerce, Economics and
26        Statistics Administration, U.S. Census Bureau, Washington, DC 20233.
27        http://www.census.gov/prod/cen2010/briefs/c2010br-03.pdf

28   Hurst,  DJ; Gardner, DE; Coffin, DL. (1970). Effect of ozone on acid hydrolases of the
29        pulmonary alveolar macrophage. J Reticuloendothel Soc 8: 288-300.

30   Hwang, BF; Lee, YL; Lin, YC; Jaakkola, JJK; Guo, YL. (2005). Traffic related air pollution as a
31        determinant of asthma among Taiwanese school  children. Thorax 60: 467-473.

32   Hyde,  DM; Miller, LA; Schelegle, ES; Fanucchi, MV; Van Winkle, LS; Tyler, NK; Avdalovic,
33        MV; Evans, MJ; Kajekar, R; Buckpitt, AR; Pinkerton, KE; Joad, JP; Gershwin, LJ; Wu, R;
                                              2-81

-------
 1        Plopper, CG. (2006). Asthma: a comparison of animal models using stereological methods.
 2        EurRespirRev 15: 122-135. http://dx.doi.org/10.1183/09059180.00010103

 3   Islam, T; Berhane, K; McConnell, R; Gauderman, WJ; Avol, E; Peters, JM; Gilliland, FD.
 4        (2009). Glutathione-S-transferase (GST) PI, GSTM1, exercise, ozone and asthma
 5        incidence in school children. Thorax 64: 197-202.
 6        http://dx.doi.org/10.1136/thx.2008.099366

 7   Islam, T; McConnell, R; Gauderman, WJ; Avol, E; Peters, JM; Gilliland, FD. (2008). Ozone,
 8        oxidant defense genes and risk of asthma during adolescence. Am J Respir Crit Care Med
 9        177: 388-395. http://dx.doi.org/10.1164/rccm.200706-863OC

10   Islam, T; Gauderman, WJ; Berhane, K; McConnell, R; Avol, E; Peters, JM; Gilliland, FD.
11        (2007). The relationship between air pollution, lung function and asthma in adolescents.
12        Thorax 62: 957-963.

13   Ito, K; Thurston, GD; Silverman,  RA. (2007b). Characterization of PM2.5, gaseous pollutants,
14        and meteorological interactions in the context of time-series health effects models. J Expo
15        Sci Environ Epidemiol  17: S45-S60.

16   Jacquemin, B; Kauffmann, F; Pin, I; Le Moual, N; Bousquet, J; Gormand, F; Just, J; Nadif, R;
17        Pison, C; Vervloet, D; Kiinzli, N; Siroux, V. (In Press) Air pollution and asthma control in
18        the epidemiological study on the genetics and  environment of asthma. J Epidemiol
19        Community Health, http://dx.doi.org/10.1136/jech.2010.130229

20   Jerrett, M; Burnett, RT; Pope, CA, III; Ito, K; Thurston, G; Krewski, D; Shi, Y; Calle, E; Thun,
21        M. (2009). Long-term ozone exposure and mortality. N Engl J Med 360: 1085-1095.
22        http://dx.doi.org/10.1056/NEJMoa0803894

23   Johnston, RA; Schwartzman, IN; Flynt, L; Shore, SA. (2005). Role of interleukin-6 in murine
24        airway responses to ozone. Am J Physiol Lung Cell Mol Physiol 288: L390-L397

25   Jorres, R; Nowak, D; Magnussen, H; Speckin, P; Koschyk, S. (1996). The effect of ozone
26        exposure on allergen responsiveness in subjects with asthma or rhinitis. Am J Respir Crit
27        Care Med 153: 56-64.

28   Kan, H; London, SJ; Chen, G; Zhang, Y; Song, G; Zhao, N; Jiang, L; Chen, B. (2008). Season,
29        sex, age, and education as modifiers of the effects of outdoor air pollution on daily
30        mortality in Shanghai, China: The Public Health and Air Pollution in Asia (PAPA) Study.
31        Environ Health Perspect 116: 1183-1188. http://dx.doi.org/10.1289/ehp.10851

32   Katsouyanni, K;  Samet, JM; Anderson, HR; Atkinson, R; Le Tertre, A; Medina,  S; Samoli, E;
33        Touloumi, G; Burnett, RT; Krewski, D; Ramsay, T; Dominici, F; Peng, RD; Schwartz, J;
34        Zanobetti, A. (2009). Air pollution and health: A European and North American approach

                                              2-82

-------
 1         (APHENA). (Research Report 142). Boston, MA: Health Effects Institute.
 2         http://pubs.healtheffects.org/view.php?id=327

 3   Kehrl, HR; Peden, DB; Ball, BA; Folinsbee, LJ; Horstman, DH. (1999). Increased specific
 4         airway reactivity of persons with mild allergic asthma after 7.6 hours of exposure to 0.16
 5         ppm ozone. J Allergy Clin Immunol 104: 1198-1204.

 6   Khatri, SB; Holguin, FC; Ryan, PB; Mannino, D; Erzurum, SC; league, WG. (2009).
 7         Association of ambient ozone exposure with airway inflammation and allergy in adults
 8         with asthma. J Asthma 46: 777-785. http://dx.doi.org/10.1080/02770900902779284

 9   Kim, CS; Alexis, NE; Rappold, AG; Kehrl, H; Hazucha, MJ; Lay, JC; Schmitt, MT; Case, M;
10         Devlin, RB; Peden, DB; Diaz-Sanchez, D. (2011). Lung function and inflammatory
11         responses in healthy young adults exposed to 0.06 ppm ozone for 6.6 hours. Am J Respir
12         Crit Care Med 183: 1215-1221. http://dx.doi.org/10.1164/rccm.201011-1813OC

13   Ko, FWS; Tarn, W; Wong, TW; Lai, CKW. (2007). Effects of air pollution on asthma
14         hospitalization rates in different age groups in Hong Kong. Clin Exp Allergy 37: 1312-
15         1319. http://dx.doi.0rg/10.llll/i.1365-2222.2007.02791.x

16   Korrick, SA; Neas, LM; Dockery, DW; Gold, DR; Allen, GA; Hill, LB; Kimball, KD; Rosner,
17         BA; Speizer, FE. (1998). Effects of ozone and other pollutants on the pulmonary function
18         of adult hikers. Environ Health Perspect 106:93-99. http://dx.doi.org/10.1289/ehp.9810693

19   Kulle, TJ; Sauder, LR; Kerr, HD; Farrell, BP; Bermel, MS; Smith, DM. (1982). Duration of
20         pulmonary function adaptation to ozone in humans. Am Ind Hyg Assoc J 43: 832-837.

21   Larsen, ST; Matsubara, S; Mcconville, G; Poulsen, SS; Gelfand, EW. (2010). Ozone increases
22         airway hyperreactivity and mucus hyperproduction in mice previously exposed to allergen.
23         J Toxicol Environ Health A 73: 738-747.  http://dx.doi.org/10.1080/15287391003614034

24   Lee, YL; McConnell, R; Berhane, K; Gilliland, FD. (2009b). Ambient ozone modifies the effect
25         of tumor necrosis factor G-308A on bronchitic symptoms among children with asthma.
26         Allergy 64: 1342-1348. http://dx.doi.Org/10.llll/i.1398-9995.2009.02014.x

27   Lebowitz, M. D.; Camilli, A. E.; Bronnimann, D.; Quackenboss, J. (1987). The significance and
28         meaningfulness of intraindividual changes in objective test results as responses to air
29         contaminants. Presented at:  80th annual meeting of the Air Pollution Control Association;
30         June; New York, NY.  Pittsburgh, PA: Air Pollution Control 30 Association; paper no. 87-
31         32.1.

32   Lewis, TC; Robins, TG; Dvonch, JT; Keeler, GJ; Yip, FY; Mentz, GB; Lin, X; Parker, EA;
33         Israel, BA; Gonzalez, L; Hill, Y. (2005). Air pollution-associated changes in lung function
34         among asthmatic children in Detroit. Environ Health Perspect 113: 1068-1075.

                                              2-83

-------
 1   Lin, S; Bell, EM; Liu, W; Walker, RJ; Kim, NK; Hwang, SA. (2008a). Ambient ozone
 2         concentration and hospital admissions due to childhood respiratory diseases in New York
 3         State, 1991-2001. Environ Res 108: 42-47. http://dx.doi.Org/10.1016/i.envres.2008.06.007

 4   Lin, S; Liu, X; Le, LH; Hwang, SA. (2008b). Chronic exposure to ambient ozone and asthma
 5         hospital admissions among children. Environ Health Perspect 116: 1725-1730.
 6         http://dx.doi.org/10.1289/ehp. 11184

 7   Lippmann, M. (1988) Health significance of pulmonary function responses to airborne irritants.
 8         JAPCA 38: 881-887.
 9   Liu, L; Poon, R; Chen, L; Frescura, AM; Montuschi, P; Ciabattoni,  G; Wheeler, A; Dales, R.
10         (2009a). Acute effects of air pollution on pulmonary function, airway inflammation,  and
11         oxidative stress in asthmatic children. Environ Health Perspect 117: 668-674.
12         http://dx.doi.org/10.1289/ehpll

13   Mar, TF; Koenig, JQ. (2009). Relationship between visits to emergency departments for asthma
14         and ozone exposure in greater Seattle, Washington.  Ann Allergy Asthma Immunol 103:
15         474-479.

16   McConnell, R; Berhane, K; Gilliland, F; London, SJ; Islam, T; Gauderman, WJ; Avol, E;
17         Margolis, HG; Peters, JM. (2002). Asthma in exercising children exposed to ozone: A
18         cohort study. Lancet 359: 386-391. http://dx.doi.org/10.1016/S0140-6736(02)07597-9

19   McDonnell, WF; Abbey, DE; Nishino, N; Lebowitz, MD. (1999a). Long-term ambient ozone
20         concentration and the incidence of asthma in nonsmoking adults: the Ahsmog study.
21         Environ Res 80: 110-121.

22   Medina-Ramon, M; Schwartz, J. (2008). Who is more vulnerable to die from ozone air
23         pollution? Epidemiology 19: 672-679.

24   Medina-Ramon, M; Zanobetti, A; Schwartz, J. (2006). The effect of ozone and PM10 on hospital
25         admissions for pneumonia and chronic obstructive pulmonary disease: A national multicity
26         study. Am J Epidemiol 163: 579-588. http://dx.doi.org/10.1093/aje/kwj078

27   Meng, YY; Rull, RP; Wilhelm, M; Lombard!, C; Balmes, J; Ritz, B. (2010). Outdoor air
28         pollution and uncontrolled asthma in the San Joaquin Valley, California. J Epidemiol
29         Community Health 64: 142-147. http://dx.doi.org/10.1136/iech.2008.083576

30   Middleton, N; Yiallouros, P; Kleanthous, S; Kolokotroni, O; Schwartz, J; Dockery, DW;
31         Demokritou, P; Koutrakis, P. (2008). A 10-year time-series  analysis of respiratory and
32         cardiovascular morbidity in Nicosia, Cyprus: The effect of short-term changes in air
33         pollution and dust storms. Environ Health 7: 39.
                                              2-84

-------
 1    Miller, FJ; Tiling, JW; Gardner, DE. (1978). Effect of urban ozone levels on laboratory-induced
 2         respiratory infections. Toxicol Lett 2: 163-169.

 3    Moore, K; Neugebauer, R; Lurmann, F; Hall, J; Brajer, V; Alcorn, S; Tager, I. (2008). Ambient
 4         ozone concentrations cause increased hospitalizations for asthma in children: An 18-year
 5         study in Southern California. Environ Health Perspect 116: 1063-1070.
 6         http://dx.doi.org/10.1289/ehp.10497

 7    Mortimer, KM; Neas, LM; Dockery, DW; Redline, S; Tager, IB. (2002). The effect of air
 8         pollution on inner-city children with asthma. Eur Respir J 19: 699-705.
 9         http://dx.doi.org/10.1183/09031936.02.00247102

10    Mortimer, KM; Tager, IB; Dockery, DW; Neas, LM; Redline, S. (2000). The effect of ozone on
11         inner-city children with asthma: Identification of susceptible subgroups. Am J Respir Crit
12         Care Med 162: 1838-1845.

13    Mudway, IS; Stenfors, N; Blomberg, A; Helleday, R; Dunster, C; Marklund, SL; Frew, AJ;
14         Sandstrom,  T; Kelly, FJ. (2001). Differences in basal airway antioxidant concentrations are
15         not predictive of individual responsiveness to ozone: A comparison of healthy and mild
16         asthmatic subjects. Free Radic Biol Med 31: 962-974.

17    Neas, LM; Dockery, DW; Koutrakis, P; Speizer, FE. (1999).  Fine particles and peak flow in
18         children: Acidity versus mass. Epidemiology 10: 550-553.

19    Neidell, M. (2009). Information, avoidance behavior, and health: The effect of ozone on asthma
20         hospitalizations. Journal of Human Resources 44: 450-478.

21    Neidell, M; Kinney, PL. (2010). Estimates of the association between ozone and asthma
22         hospitalizations that account for behavioral responses to air quality information. Environ
23         SciPol 13: 97-103. http://dx.doi.Org/10.1016/i.envsci.2009.12.006

24    O'Connor, GT; Neas, L; Vaughn, B; Kattan, M; Mitchell, H;  Crain, EF; III, ER; Gruchalla, R;
25         Morgan, W; Stout, J; Adams, GK; Lippmann, M. (2008). Acute respiratory health effects
26         of air pollution on children with asthma in US inner cities. J Allergy Clin Immunol  121:
27         1133-1139.  http://dx.doi.0rg/10.1016/i.iaci.2008.02.020

28    O*Net OnLine (2012). Work Context — Outdoors, Exposed to Weather.
29         http://www.onetonline.Org/fmd/descriptor/result/4.C.2.a. 1 .c?a=l

30    Paulu, C;  Smith, AE. (2008). Tracking associations between ambient ozone and asthma-related
31         emergency department visits using case-crossover analysis. J Public Health Manag Pract
32         14:581-591.
                                               2-85

-------
 1   Peden, DB; Boehlecke, B; Horstman, D; Devlin, R. (1997). Prolonged acute exposure to 0.16
 2         ppm ozone induces eosinophilic airway inflammation in asthmatic subjects with allergies. J
 3         Allergy Clin Immunol 100: 802-808.

 4   Peden, DB; Setzer, RW, Jr; Devlin, RB. (1995). Ozone exposure has both a priming effect on
 5         allergen-induced responses and an intrinsic inflammatory action in the nasal airways of
 6         perennially allergic asthmatics. Am J Respir Crit Care Med 151: 1336-1345.

 7   Plopper, CG; Smiley-Jewell, SM; Miller, LA; Fanucchi, MV; Evans, MJ; Buckpitt, AR;
 8         Avdalovic, M; Gershwin, LJ; load, JP; Kajekar, R; Larson, S; Pinkerton, KE; Van Winkle,
 9         LS; Schelegle, ES; Pieczarka, EM; Wu, R; Hyde, DM. (2007). Asthma/allergic airways
10         disease: Does postnatal exposure to environmental toxicants promote airway pathobiology?
11         Toxicol Pathol 35: 97-110. http://dx.doi.org/10.1080/01926230601132030

12   Pope, CA, III; Burnett, RT; Thun, MJ; Calle, EE; Krewski, D; Ito, K; Thurston, GD. (2002).
13         Lung cancer, cardiopulmonary mortality, and long-term exposure to fine particulate air
14         pollution. JAMA 287: 1132-1141.

15   Qian, Z; Lin, FIM; Chinchilli, VM; Lehman, EB; Duan, Y; Craig, TJ; Wilson, WE; Liao, D;
16         Lazarus, SC; Bascom, R.  (2009). Interaction of ambient air pollution with asthma
17         medication on exhaled nitric oxide among asthmatics. Arch Environ Occup Health 64: 168-
18         176. http://dx.doi.org/10.1080/19338240903240616

19   Rabinovitch, N; Zhang, LN; Murphy, JR; Vedal, S; Dutton, SJ; Gelfand, EW. (2004). Effects of
20         wintertime ambient air pollutants on asthma exacerbations in urban minority children with
21         moderate to severe disease. J Allergy Clin Immunol  114: 1131-1137.
22         http://dx.doi.0rg/10.1016/i.iaci.2004.08.026

23   Rage, E; Siroux, V; Kunzli, N;  Pin, I; Kauffmann, F. (2009b). Air pollution and asthma severity
24         in adults. Occup Environ Med 66:  182-188. http://dx.doi.org/10.1136/oem.2007.038349

25   Ren,  C; Williams, GM; Mengersen, K; Morawska, L; Tong, S. (2008). Does temperature modify
26         short-term effects of ozone on total mortality in 60 large eastern US communities? An
27         assessment using the NMMAPS data. Environ Int 34: 451-458.

28   Romieu, I; Barraza-Villarreal, A; Escamilla-Nufiez, C; Texcalac-Sangrador, JL; Hernandez-
29         Cadena, L; Diaz-Sanchez, D; De Batlle, J; Del Rio-Navarro, BE. (2009). Dietary intake,
30         lung function and airway inflammation in Mexico City school children exposed to air
31         pollutants. Respir Res 10: 122.

32   Romieu, I; Barraza-Villarreal, A; Escamilla-Nunez, C; Almstrand, AC; Diaz-Sanchez, D; Sly,
33         PD; Olin, AC. (2008). Exhaled breath malondialdehyde as a marker of effect of exposure to
34         air pollution in children with asthma. J Allergy Clin Immunol 121: 903-909.
35         http://dx.doi.0rg/10.1016/i.iaci.2007.12.004

                                               2-86

-------
 1   Ross, MA; Persky, VW; Scheff, PA; Chung, J; Curtis, L; Ramakrishnan, V; Wadden, RA;
 2        Hryhorczuk, DO. (2002). Effect of ozone and aeroallergens on the respiratory health of
 3        asthmatics. Arch Environ Occup Health 57: 568-578.
 4        http://dx.doi.org/10.1080/00039890209602090
 5   Salam, MT; Islam, T; Gauderman, WJ; Gilliland, FD. (2009). Roles of arginase variants, atopy,
 6        and ozone in childhood asthma. J Allergy Clin Immunol 123:  596-602.
 7        http://dx.doi.0rg/10.1016/j.jaci.2008.12.020
 8   Samet, J.M. Clean Air Scientific Advisory Committee (CASAC) Response to Charge Questions
 9        on the Reconsideration of the 2008 Ozone National Ambient Air Quality Standards. EPA-
10        CASAC-11-004.  March 30,2011. Available online at:
11        http://vosemite.epa.gOv/sab/sabproduct.nsf/0/F08BEB48Cl 139E2A8525785E006909AC/$
12        File/EPA-CASAC-ll-004-unsigned+.pdf
13   Samoli, E; Zanobetti, A; Schwartz, J; Atkinson, R; Le Tertre, A; Schindler, C; Perez, L; Cadum,
14        E; Pekkanen, J; Paldy, A; Touloumi, G; Katsouyanni, K. (2009). The temporal pattern of
15        mortality responses to ambient ozone in the APFIEA project. J Epidemiol Community
16        Health 63: 960-966. http://dx.doi.org/10.1136/iech.2008.084012
17   Scannell, C; Chen, L; Aris, RM; Tager, I; Christian, D; Ferrando, R; Welch, B; Kelly, T;
18        Balmes, JR. (1996). Greater ozone-induced inflammatory responses in subjects with
19        asthma. Am J Respir Crit Care Med 154: 24-29.
20   Schelegle, ES; Morales, CA; Walby, WF; Marion, S; Allen, RP. (2009). 6.6-hour inhalation of
21        ozone concentrations from 60  to 87 parts per billion in healthy humans. Am J Respir Crit
22        Care Med 180: 265-272. http://dx.doi.org/10.1164/rccm.200809-1484OC
23   Schildcrout, JS; Sheppard, L; Lumley, T; Slaughter, JC; Koenig, JQ; Shapiro, GG. (2006).
24        Ambient air pollution and asthma exacerbations in children: An  eight-city analysis. Am J
25        Epidemiol 164: 505-517.  http://dx.doi.org/10.1093/aie/kwi225
26   Schiller JS, Lucas JW, Ward BW, Peregoy JA. Summary health statistics for U.S. adults:
27        National Health Interview Survey, 2010. National Center for Health Statistics. Vital Health
28        Stat 10(252). 2012. http://www.cdc.gov/nchs/data/series/sr  10/srlO 252.pdf
29   Schwartz, J. (2005a). How sensitive is the association between ozone  and daily deaths to control
30        for temperature? Am J Respir  Crit Care Med 171: 627-631.
31   Schwartz, J. (2005b). Who is sensitive to extremes of temperature?  A  case-only analysis.
32        Epidemiology 16: 67-72.  http://dx.doi.org/10.1097/01.ede.0000147114.25957.71
33   Sienra-Monge, JJ; Ramirez-Aguilar, M; Moreno-Macias, H; Reyes-Ruiz, NI; Del Rio-Navarro,
34        BE; Ruiz-Navarro, MX; Hatch, G; Crissman, K; Slade, R; Devlin, RB; Romieu, I. (2004).
35        Antioxidant supplementation and nasal inflammatory responses  among young asthmatics
36        exposed to high levels of ozone. Clin Exp Immunol 138: 317-322.
37        http://dx.doi.0rg/10.llll/i.1365-2249.2004.02606.x

                                               2-87

-------
 1    Silverman, RA; Ito, K. (2010). Age-related association of fine particles and ozone with severe
 2         acute asthma in New York City. J Allergy Clin Immunol 125: 367-373.
 3         http://dx.doi.0rg/10.1016/i.iaci.2009.10.061
 4    Smith, RL; Xu, B; Switzer, P. (2009b). Reassessing the relationship between ozone and short-
 5         term mortality in U.S. urban communities. Inhal Toxicol 21: 37-61.
 6         http://dx.doi.org/10.1080/08958370903161612

 7    Spektor, DM; Lippmann, M; Lioy, PJ; Thurston, GD; Citak, K; James, DJ; Bock, N; Speizer,
 8         FE; Hayes, C. (1988a). Effects of ambient ozone on respiratory function in active, normal
 9         children. Am Rev Respir Dis 137: 313-320.

10    Spektor, DM; Lippmann, M; Thurston, GD; Lioy, PJ; Stecko, J; O'Connor, G; Garshick, E;
11         Speizer, FE; Hayes, C. (1988b). Effects of ambient ozone on respiratory function in healthy
12         adults exercising outdoors. Am Rev Respir Dis 138: 821-828.

13    Stafoggia, M; Forastiere, F; Faustini, A; Biggeri, A; Bisanti, L; Cadum, E; Cernigliaro, A;
14         Mallone, S; Pandolfi, P; Serinelli, M; Tessari, R; Vigotti, MA; Perucci, CA. (2010).
15         Susceptibility factors to ozone-related mortality: A population-based case-crossover
16         analysis. Am J Respir Crit Care Med  182: 376-384. http://dx.doi.org/10.1164/rccm.200908-
17         1269OC

18    Stieb, DM; Szyszkowicz, M; Rowe,  BH; Leech, JA. (2009). Air pollution and emergency
19         department visits for cardiac and respiratory conditions: A multi-city time-series analysis.
20         Environ Health Global Access  Sci Source 8: 25. http://dx.doi.org/10.1186/1476-069X-8-25

21    Strickland, MJ; Darrow, LA; Klein, M; Flanders, WD; Sarnat, JA; Waller, LA; Sarnat, SE;
22         Mulholland, JA; Tolbert, PE. (2010).  Short-term associations between ambient air
23         pollutants and pediatric asthma emergency department visits. Am J Respir Crit Care Med
24         182: 307-316. http://dx.doi.org/10.1164/rccm.200908-1201OC

25    Thurston, GD; Lippmann, M; Scott,  MB; Fine, JM. (1997). Summertime haze air pollution and
26         children with asthma. Am J Respir Crit Care Med 155:  654-660.

27    Tolbert, PE; Klein, M; Peel, JL; Sarnat, SE; Sarnat, JA. (2007). Multipollutant modeling issues
28         in a study of ambient air quality and emergency department visits in Atlanta. J Expo Sci
29         EnvironEpidemiol 17: S29-S35. http://dx.doi.org/10.1038/si.ies.7500625

30    US EPA (2012a). Integrated Science Assessment of Ozone and Related Photochemical Oxidants
31         (Third External Review Draft). U.S. Environmental Protection Agency, Washington, DC.
32         EPA/600/R-10/076C. June 2012. Available at:
33         http://www.epa.gOv/ttn/naaqs/standards/ozone/s o3 2008 isa.html

-------
 1   US EPA (2012b). Health Risk and Exposure Assessment for Ozone, First External Review
 2        Draft.  Office of Air Quality Planning and Standards, Research Triangle Park, NC, 27711.
 3        EPA 452/P-12-001. July 2012. Available at:
 4        http://www.epa.gov/ttn/naaqs/standards/ozone/s  o3 2008 rea.html
 5
 6   US EPA (2012c). Welfare Risk and Exposure Assessment for Ozone, First External Review
 7        Draft.  Office of Air Quality Planning and Standards, Research Triangle Park, NC, 27711.
 8        EPA452/P-12-004. July 2012. Available at:
 9        http://www.epa.gov/ttn/naaqs/standards/ozone/s  o3 2008 rea.html
10
11   US EPA (2006). Air Quality Criteria for Ozone and Related Photochemical Oxidants (2006
12        Final). U.S. Environmental Protection Agency, Washington, DC. EPA/600/R-05/004aF-
13        cF. March 2006.  Available at:
14        http ://www. epa. gov/ttn/naaqs/standards/ozone/s_o3_cr_cd.html
15
16   US EPA (1996a). Air quality criteria for ozone and related photochemical oxidants [EPA
17        Report]. (EPA/600/P-93/004AF).  U.S. Environmental Protection Agency, Research
18        Triangle Park, NC.
19
20   Vagaggini, B; Taccola, M;  Clanchetti, S; Carnevali, S; Bartoli, ML; Bacci, E; Dente, FL; Di
21        Franco, A; Giannini, D; Paggiaro, PL. (2002). Ozone exposure increases eosinophilic
22        airway response induced by previous allergen challenge. Am J Respir Crit Care Med 166:
23        1073-1077.
24
25   Villeneuve, PJ;  Chen, L; Rowe, BH; Coates, F. (2007). Outdoor air pollution and emergency
26        department visits for asthma among  children and adults: A case-crossover study in northern
27        Alberta, Canada. Environ Health Global Access Sci Source 6: 40.
28        http://dx.doi.org/10.1186/1476-069X-6-40
29
30   Wenten, M;  Gauderman, WJ; Berhane, K; Lin, PC; Peters, J; Gilliland, FD. (2009). Functional
31        variants in the catalase and myeloperoxidase genes, ambient air pollution, and respiratory -
32        related school absences: An example of epistasis in gene-environment interactions. Am J
33        Epidemiol 170: 1494-1501. http://dx.doi.org/10.1093/aje/kwp310
34   Wolff, G.T. (1995) Letter to EPA Administrator Carol Browner: "CASAC Closure on the
35        Primary Standard Portion of the Staff Paper for Ozone" EPA-SAB-CASAC-LTR-96-002,
36        November 30, 1995.
37   Wong, CM;  Vichit-Vadakan, N; Vajanapoom, N; Ostro,  B; Thach, TQ; Chau, PY; Chan, EK;
38        Chung, RY; Ou, CQ;  Yang, L; Peiris, JS; Thomas,  GN; Lam, TH; Wong, TW; Hedley, AJ;
39        Kan, H; Chen, B;  Zhao, N; London,  SJ; Song, G; Chen, G; Zhang, Y; Jiang, L; Qian, Z;
40        He, Q; Lin, HM; Kong, L; Zhou, D;  Liang, S; Zhu, Z; Liao, D; Liu, W; Bentley, CM; Dan,

                                              2-89

-------
 1         J; Wang, B; Yang, N; Xu, S; Gong, J; Wei, H; Sun, H; Qin, Z. (2010). Part 5. Public health
 2         and air pollution in Asia (PAPA): A combined analysis of four studies of air pollution and
 3         mortality. In Public Health and Air Pollution in Asia (PAPA): Coordinated Studies of
 4         Short-Term Exposure to Air Pollution and Daily Mortality in Four Cities (pp. 377-418).
 5         Boston, MA: Health Effects Institute. http://pubs.healtheffects.org/view.php?id=348
 6    Wong, CM; Yang, L; Thach, TQ; Chau, PY; Chan, KP; Thomas, GN; Lam, TH; Wong, TW;
 7         Hedley, AJ; Peiris, JS. (2009). Modification by influenza on health effects of air pollution
 8         in Hong Kong. Environ Health Perspect  117: 248-253. http://dx.doi.org/10.1289/ehp.11605
 9    Woodwell, D. A.; Cherry, D. K. (2004) National Ambulatory Medical Care Survey: 2002
10         summary. Hyattsville, MD: National Center for Health Statistics; DHHS publication no.
11         (PHS) 2004-1250. (Advance data from vital and health statistics; no. 346). Available:
12         http://www.cdc.gov/nchs/data/ad/ad346.pdff3 August, 2005]
13    Xia, Y; long, H. (2006). Cumulative effects of air pollution on public health. Stat Med 25: 3548-
14         3559. http://dx.doi.org/10.1002/sim.2446
15    Yang, Q; Chen, Y; Krewski, D; Burnett, RT; Shi, Y; Mcgrail, KM. (2005). Effect of short-term
16         exposure to low levels of gaseous pollutants on chronic obstructive pulmonary disease
17         hospitalizations. Environ Res 99: 99-105. http://dx.doi.Org/10.1016/i.envres.2004.09.014
18    Zanobetti, A; Schwartz, J. (2011). Ozone and survival in four cohorts with potentially
19         predisposing diseases. Am J Respir Crit Care Med 184: 836-841.
20         http://dx.doi.org/10.1164/rccm.201102-0227OC
21    Zanobetti, A; Schwartz, J. (2008a). Is there adaptation in the ozone mortality relationship: A
22         multi-city case-crossover analysis. Environ Health 7: 22.  http://dx.doi.org/10.1186/1476-
23         069X-7-22
24    Zanobetti, A; Schwartz, J. (2008b). Mortality displacement in the association of ozone with
25         mortality: An analysis of 48 cities in the  United States. Am J Respir Crit Care Med 177:
26         184-189. http://dx.doi.org/10.1164/rccm.200706-823OC
27    Zanobetti, A; Schwartz, J. (2006). Air pollution and emergency admissions in Boston, MA.  J
28         Epidemiol Community Health 60: 890-895. http://dx.doi.org/10.1136/iech.2005.039834
29
30
31
                                               2-90

-------
 1             3.     ASSESSMENT OF O3-RELATED EXPOSURES AND RISKS

 2           To put judgments about (Vrelated health effects into a broader public health context, the
 3    first draft REA has developed and applied models to estimate  human exposures to Os and Os-
 4    associated health risks across the United States, with a specific focus on several urban study
 5    areas (US EPA, 2012b). Using such models, the first draft REA has estimated the 63 exposures
 6    and health risks that are associated with recent air quality and  with air quality adjusted to
 7    simulate just meeting the current 63 standard (US EPA, 2012b). The second draft and final REA
 8    will also include estimates of O3 exposures and health risks associated with air quality adjusted
 9    to simulate just meeting potential alternative Oj standards.1 The first draft REA has identified
10    the following goals for the exposure and risk assessments: (1) to provide estimates of the number
11    of people in the general population and in sensitive populations with Os exposures above short-
12    term benchmark levels; (2) to provide estimates of the number of people in the general
13    population and in sensitive populations with impaired lung function resulting from short-term
14    exposures to Ch; (3) to provide estimates of the potential magnitude of premature mortality
15    and/or selected morbidity health effects in the population, including sensitive populations,
16    associated with recent short-term ambient concentrations of O3 and with just meeting the current
17    Os NAAQS in selected urban study areas; (4) to develop a better understanding of the influence
18    of various inputs and assumptions on the risk estimates to more clearly differentiate alternative
19    standards that might be considered including potential impacts on various sensitive populations;
20    (5) to gain insights into the distribution of risks and patterns of risk reduction and uncertainties in
21    those risk estimates, and (6) to understand the national mortality burden associated with recent
22    ambient Os, and  how well the risk estimates for the set of urban areas modeled reflect the
23    national distribution of mortality risk (US EPA, 2012b, section 1.2).
24           Air quality inputs to the exposure and risk assessments include: (1) recent air quality data
25    for 63 from selected urban study areas; (2) simulated air quality for the selected urban areas that
26    reflects changes  in the distribution of Os air quality estimated  to occur when an area just meets
27    the current 63 NAAQS, and (3) 63 air quality surfaces for recent years covering the entire
28    continental U.S.  for use in the national scale assessment. The exposure and risk analyses are
29    based on the five most recent years of air quality data available at this time, 2006-2010, in order
30    to reflect the considerable variability in meteorological conditions and the variation in 63
      1 Air quality simulations are meant to provide perspective on the O3-associated exposures and health risks under
      different air quality scenarios. These simulations do not reflect any consideration of specific control programs or
      strategies designed to achieve the reductions in emissions required to meet the specified standards. Further, these
      simulations do not represent predictions of when, whether, or how areas might meet the specified standards.
                                                     3-1

-------
 1    precursor emissions that have occurred in recent years.2  The analyses in the draft REA focus
 2    primarily on the months of May to September, when Os concentrations are likely to be highest in
 3    urban areas (US EPA, 2012b, section 3.2.1).
 4           Simulation of just meeting the current Os standard is accomplished in this first draft REA
 5    using a quadratic rollback method, similar to that used in the previous risk and exposure analysis
 6    for the 2008 O3 NAAQS review (US EPA, 2007). In evaluating just meeting the current
 7    standard, the first draft REA focused on air quality changes that were likely to  occur as the U.S.
 8    puts in place programs to meet the standard.  As such, the REA used U.S. background
 9    concentrations as a floor for the quadratic rollback. The first draft REA also explored alternative
10    simulation approaches based on modeled sensitivities of ozone to U.S. emissions. These
11    alternative approaches, which will be evaluated more fully in the second draft REA,3 will
12    remove the need for imposing a specific floor to prevent adjustments beyond those likely to
13    occur due to U.S. emissions reductions.
14           In selecting health endpoints on which to base estimates of (Vrelated health impacts, the
15    first draft REA considers the weight-of-evidence conclusions from the ISA. Specifically, the
16    first draft REA notes the ISA conclusions that there is a causal relationship between short-term
17    63 exposures and respiratory effects and a likely causal relationship between short-term 63
18    exposures and all-cause mortality. In light of these conclusions, the first draft REA estimates
19    respiratory effects and all-cause mortality following short-term O?, exposures (section 2.1,
20    above).4  The remainder of this chapter discusses the assessment of 63 exposures (section 3.1)
21    and Os-associated health risks (section 3.2) for recent Os air quality  concentrations and for Os air
22    quality adjusted to simulate just meeting the current 8-hour Os NAAQS.

23    3.1    EXPOSURE ASSESSMENT

24           The exposure assessment, which provides estimates of the number of people exposed to
25    different concentrations of ambient 63 while at specified exertion levels, serves two purposes.
       The national-scale risk analyses are based on air quality data from 2006-2008, given available air quality modeling
      data for that time period.
      3In the second draft, the REA will evaluate approaches for simulating attainment of current and alternative standards
      that are based on modeling the response of O3 concentrations to reductions in anthropogenic NOx and VOC
      emissions, using the Higher-order Decoupled Direct Method (HDDM) capabilities in the Community Multi-scale
      Air Quality (CMAQ) model.  This modeling incorporates all known emissions, including emissions from
      nonanthropogenic sources and anthropogenic emissions from sources in and outside of the U.S. As a result, the
      need to specify values for U.S. background is not necessary, as it is incorporated in the modeling directly. The
      evaluation of this new approach is presented in Chapter 4 of the draft REA (US EPA, 2012b) and in Simon et al.
      (2012).
      4The first draft REA does not evaluate risks associated with long-term exposures, but notes that such risks could be
      assessed in the second draft REA (US EPA, 2012b, section 7.7).
                                                      3-2

-------
 1    First, the entire range of modeled personal exposures to ambient 63 is an essential input to the
 2    portion of the health risk assessment based on exposure-response functions from controlled
 3    human exposure studies, discussed in the next section (i.e., section 3.2).5 Second, estimates of
 4    personal exposures (both the number and percent of total populations and sensitive
 5    subpopulations) to ambient Os concentrations at and above specific health benchmark levels
 6    provide perspective on the potential public health impacts of (Vrelated health effects, including
 7    effects that cannot currently be evaluated in a quantitative risk assessment.
 8          As part of the 2008 review of the Os NAAQS, the EPA conducted exposure analyses in
 9    12 metropolitan areas representing different regions of the United States where the then-current
10    8-hour 63 standard was not met. For each of these metropolitan areas, 63 exposures were
11    estimated for the general population, for school age children (ages 5-18), and for school age
12    children with asthma.  The emphasis on children reflected the finding  of the 1997 Os NAAQS
13    review that children are an important at-risk group.  The  12 modeled areas included 89 million
14    people, including  18 million  school age children, 2.6 million of whom had asthma. The selection
15    of urban areas to include in the exposure analysis took into consideration the location of Os
16    epidemiological studies, the availability of ambient 63 monitoring data,  and the desire to
17    represent a range of geographic areas, population demographics, and Os climatology. These
18    selection criteria are discussed further in chapter 5 of the 2007 Staff Paper (EPA, 2007).
19          The exposure analysis conducted for the current review in the first draft REA builds upon
20    the methodology and lessons learned from the exposure analyses conducted in previous reviews,
21    as well as information provided in the third draft ISA (US EPA, 2012a). EPA will be conducting
22    exposure modeling for 16 urban areas located across the United States (US EPA, 2012b).  In the
23    first draft REA, results are presented for four of these areas, Atlanta, Denver, Los Angeles, and
24    Philadelphia. The criteria and considerations that went into selection of these urban areas for the
25    Os risk assessment included the following:
         •  The overall set of urban locations should represent a range of geographic areas, urban
            population demographics, and climatology.
         •  The locations should be focused on areas that do not meet or are close to not meeting the
            current 8-hr 63 NAAQS and should include the largest areas with major 63
            nonattainment problems.
         •  There must be sufficient 63 air quality data for the recent 2006-2010 period.
      5As noted in the first draft REA, the quantitative assessment of respiratory health risks based on controlled human
      exposure studies is under development and will be made available along with this first draft Policy Assessment.
      Therefore, the characterization of such respiratory health risks will be considered in the second draft Policy
      Assessment.
                                                    3-3

-------
         •  The areas should include the 12 cities modeled in the epidemiologic-based risk
            assessment.
 1
 2          In considering exposure estimates, the draft REA noted that there are several sources of
 3    variability and uncertainty inherent in assessment inputs and that there is uncertainty in the
 4    resulting O3 exposure estimates. The exposure modeling approach accounts for variability in
 5    ambient O3 concentrations, demographic characteristics, physiological attributes, activity
 6    patterns, and factors affecting microenvironmental (e.g., indoor) concentrations.  Among the
 7    most important uncertainties affecting the exposure estimates are those related to the modeling of
 8    human activity patterns over an O3 season, the modeling of variations in ambient concentrations
 9    across urban areas, the modeling of air exchange rates that affect the amount of O3 that
10    penetrates indoors, and the characterization of energy expenditure and breathing rates for
11    children engaged in various activities. These uncertainties will be characterized more fully in the
12    second draft REA (US EPA, 2012b, section 5.5.2).
13          The remainder of this section discusses the approach taken in the draft REA to assess O3
14    exposures (3.1.1) and the key observations from the REA analyses of O3 exposures (section
15    3.1.2).
16    3.1.1  Approach to Assessing Os Exposures
17          Population exposures to O3 are driven primarily by exposures to ambient concentrations,
18    which vary by time of day, location, and peoples' activities.  In the absence of large scale
19    exposure  studies that encompass the overall population as well as at-risk subpopulations,
20    exposure  modeling is the preferred approach to estimating population exposures to O3.  In the
21    first draft REA, population exposures to ambient ozone concentrations were evaluated using the
22    current version of the Air Pollutants Exposure (APEX) model.  The APEX model simulates the
23    movement of individuals through time and space and estimates their exposures to a given
24    pollutant in indoor, outdoor, and in-vehicle microenvironments (US EPA, 201 Ic, section 3.2).
25    APEX takes into  account the most significant factors that contribute to total human O3 exposure,
26    including the temporal and spatial distribution of people and O3 concentrations throughout an
27    urban area, the variation of O3 concentrations within each microenvironment, and the effects of
28    exertion on breathing rate in exposed individuals.
29          The first draft REA developed exposure estimates for four urban areas for recent O3
30    concentrations during the O3 season, based on 2006-2010 air quality data, and for ozone
31    concentrations adjusted to simulate just meeting the current 8-hour ozone NAAQS, based on
32    adjusting  2006-2010 air quality data to simulate attainment for the 2006-2008 and 2008-2010

                                                    3-4

-------
 1    periods, reflecting the 3-year average period over which the current standard is evaluated.
 2    Exposure estimates based on adjusted air quality allow consideration of the extent to which Os-
 3    related exposures and associated health impacts might be reduced by meeting the current Os
 4    NAAQS. Multiple years were modeled in order to reflect the substantial year-to-year variability
 5    that occurs in ambient Os concentrations and related meteorological conditions, and because the
 6    standard is specified in terms of a three-year period. The year-to-year variability observed in 63
 7    levels is due to a combination of different weather patterns and the variation in emissions of O3
 8    precursors.  For each year and in each urban location, the Os season was modeled in order to
 9    characterize the period of the year where the highest 63 concentrations tend to occur. Exposures
10    were estimated for the general population, school-age children (ages 5 to 18), and asthmatic
11    school-age children. This choice of population groups includes a strong emphasis on children,
12    which reflects the  findings of the last O3 NAAQS review (EPA, 2007) and the third draft ISA in
13    the current review (EPA, 2012a, Chapter  8) that children are an important at-risk group.
14    Children breathe more air per pound of body weight, are more likely than adults to have asthma,
15    and their lungs continue to develop until they are fully grown.
16           As noted above, one of the main purposes of 63 exposure estimates is to provide
17    information on population exposures to Os concentrations exceeding health benchmarks, which
18    are based on the available evidence for respiratory effects following exposures to specific 63
19    concentrations, as  assessed in the third draft ISA (US EPA, 2012a, section 6.2).  Estimates of
20    exposures at or above discrete benchmark concentrations provide some perspective on the public
21    health impacts of (Vrelated health effects that have been demonstrated in controlled human
22    exposure and toxicological studies but that cannot be evaluated in quantitative risk assessments,
23    such as lung inflammation, increased airway responsiveness, and changes in lung host defenses.
24    They also help in understanding the extent to which such impacts have the potential to be
25    reduced by meeting the current and potential alternative standards. Identification of Os
26    benchmark concentrations for analysis are based on consideration of the range of health effects
27    reported in controlled human exposure and epidemiologic studies, including effects that are not
28    assessed in the quantitative health risk assessment discussed in section 3.2 below.
29           Though this analysis is conducted using  discrete benchmark concentrations, health-
30    relevant exposures are more appropriately viewed as a continuum with greater confidence and
31    less uncertainty  about the existence of health effects at higher 63 exposure concentrations and
32    less confidence and greater uncertainty as one considers increasingly lower exposure
33    concentrations.  In considering these results, it is important to balance concerns  about the
34    potential for health effects and their severity with the increasing uncertainty associated with our
35    understanding of the likelihood of such effects following exposures to lower Os concentrations.

                                                    3-5

-------
 1          The draft REA evaluates specific benchmark concentrations from 60 to 80 ppb (i.e., 60,
 2    70, 80 ppb) (US EPA, 2012b, section 5.5), based on controlled human exposure studies that
 3    report respiratory effects following Os exposures in healthy exercising adults. As discussed
 4    above (section 2.2.1) and in more detail in the ISA (US EPA, 2012a, section 6.2), there is a
 5    substantial body of evidence from controlled human exposure studies reporting that prolonged
 6    (e.g., 6.6 hour) exposures to 63 concentrations from 60 to 80 ppb during moderate levels of
 7    exertion result in a range of respiratory effects in healthy adult subjects. These effects include
 8    decrements in lung function, increased airway inflammation, increased respiratory symptoms,
 9    and increased airway responsiveness (section 2.2.1, above). As 63 exposure concentrations
10    decrease from 80 to 60 ppb, the breadth of reported Os-induced effects decrease, the magnitudes
11    of those effects decrease, and the consistency and statistical precision of those effects decrease,
12    with 60 ppb the lowest exposure concentration for which such respiratory effects have been
13    reported.
14    3.1.2  Key Observations: Exposures to Os Concentrations Above Health Benchmarks
15          The first draft REA presents a series of figures and tables characterizing the percents and
16    numbers of school-age children who experience at least one 8-hour average Os exposure above
17    the different benchmark concentrations (60, 70, 80 ppb), while at the same time undergoing
18    exercise of moderate or greater intensity (US EPA, 2012b, Figures 5-1 through 5-15), in four
19    cities (Atlanta, Denver, Los Angeles, Philadelphia). In characterizing the results of these
20    analyses, the REA noted that the APEX model is not proficient at modeling activity patterns that
21    lead to repeated exposures to elevated ozone concentrations. As a result, while the REA reported
22    the numbers and percentages of children with at least one exposure greater than the alternative
23    exposure benchmarks, the REA did not report the numbers or percentages of children estimated
24    to experience more than one exposure. Children with repeated exposures may be at greater risk
25    of significant health effects, an issue that could be explored more fully in the second draft REA.
26    Key observations from the REA exposure analyses are discussed below.
27          Across the years included in the analysis (2006-2010), the pattern of exposures differed.
28    For example, in the worst Os year (2006) the percent of children exposed to 8-hour Os
29    concentrations at or above 60 ppb  while at moderate or greater exertion ranged from 30 to 37%
30    across the 4 study areas; the percent exposed to 8-hour concentrations at or above 70 ppb ranged
31    from 10 to 21%; and the percent exposed to 8-hour concentrations at or above 80 ppb ranged
32    from 1 to 10%.  In the year with the lowest Os concentrations (2009), the percent of children
33    exposed to 8-hour Os concentrations at or above 60 ppb while at moderate or greater exertion
34    ranged from 9 to 32%; the percent exposed to 8-hour concentrations at or above 70 ppb ranged
35    from 1 to 15%; and the percent exposed to 8-hour concentrations at or above 80 ppb ranged from

                                                   3-6

-------
 1    0 to 5%. With unadjusted air quality, the average (i.e., average across years 2006 to 2010)
 2    percentages of school age children estimated to experience one or more exposures per year to 8-
 3    hour Os concentrations above 60, 70, or 80 ppb, while at moderate or greater exertion, were as
 4    follows (US EPA, 2012b, Figures 5-1 to 5-15, Tables 4-5 to 4-6, and section 9.1):

 5       •  For Denver, approximately 20% above 60 ppb (corresponding  to 109,000 children), 4%
 6          above 70 ppb (corresponding to approximately 22,000 children), and 0.4% above 80 ppb
 7          (corresponding to approximately 2,200 children)
 8
 9       •  For Atlanta, approximately 22% above 60 ppb (corresponding  to 189,000 children), 9%
10          above 70 ppb (corresponding to approximately 75,000 children), and 3% above 80 ppb
11          (corresponding to approximately 24,000 children)
12
13       •  For Philadelphia,  approximately 26% above 60 ppb (corresponding to 297,000 children),
14          10% above 70 ppb (corresponding to 117,000 children), and 2% above 80 ppb
15          (corresponding to approximately 28,000 children)
16
17       •  For Los Angeles,  approximately 32% above 60 ppb (corresponding to 1,150,000
18          children), 15% above 70 ppb (corresponding to 559,000 children), and 6%  above 80 ppb
19          (corresponding to approximately 218,000 children)
20

21          When air quality was adjusted to simulate just meeting the current Os NAAQS, these
22    estimated benchmark exceedances decreased.  Specifically, for adjusted air quality, the average
23    percentages of school age children estimated to experience one or more exposures per year to 8-
24    hour Os concentrations above 60, 70, or 80 ppb, while at moderate or greater exertion, were as
25    follows6 (US EPA, 2012b, Figures 5-1 to 5-15, Tables 4-5 to 4-6):

26       •  For Denver, approximately 9-11% above 60 ppb (corresponding to approximately
27          51,000-61,000 children), 0.6-0.7% above 70 ppb (corresponding to approximately 3,300-
28          4,200 children), and 0.0% above 80 ppb
29
30       •  For Atlanta, approximately 8-10% above 60 ppb (corresponding to approximately
31          72,000-83,000 children), 1% above 70 ppb (corresponding to approximately 9,600-
32          11,200 children),  and 0.1% above 80 ppb (corresponding to approximately 700-900
33          children)
34
35       •  For Philadelphia,  approximately 9-13% above 60 ppb (corresponding to approximately
36          105,000-145,000  children), 1-2% above 70 ppb (corresponding to approximately 10,000-
37          26,000 children),  and 0.0-0.1% above 80 ppb  (corresponding to approximately 500-1,600
38          children)
      6The ranges presented represent the range of estimates across each of the 3-year periods for which air quality was
      adjusted to simulate just meeting the current O3 NAAQS (i.e., 2006-2008, 2008-2010).

                                                   3-7

-------
 1
 2       •  For Los Angeles, approximately 4% above 60 ppb (corresponding to approximately
 3          126,000-148,000 children), 0.2-0.3% above 70 ppb (corresponding to approximately
 4          7,500-11,300 children), and 0.0% above 80 ppb
 6          When considering exposures above benchmark concentrations in asthmatic children at
 7    moderate or greater exertion, the results were similar in term of percentages (US EPA, 2012b,
 8    Tables 5-11 and 5-12). When air quality was adjusted to simulate just meeting the current 63
 9    NAAQS, the numbers of asthmatic children estimated to be exposed one or more times per year
10    to 8-hour Os concentrations at or above 60 ppb, while at moderate exertion, ranged from
11    approximately 5,000 to 17,000 across these four cities; the numbers of asthmatic children
12    estimated to be exposed to Os concentrations at or above 70 ppb ranged from approximately 300
13    to 3,000 across cities; and the numbers of asthmatic children estimated to be exposed to Os
14    concentrations at or above 80 ppb ranged from 0 to approximately 200 across cities (US EPA,
15    2012b, Tables 5-11 and 5-12).

16    3.2    RISK ASSESSMENT

17          For some health endpoints, there is sufficient scientific evidence and information
18    available to support the development of quantitative estimates of Os-related health risks.  In the
19    last review of the 63 NAAQS, the quantitative health risk assessment estimated the risks of lung
20    function decrements in all children and in asthmatic school age children, of respiratory
21    symptoms in asthmatic children, of respiratory-related hospital admissions, and of non-accidental
22    and cardiorespiratory-related mortality. Both controlled human exposure and epidemiologic
23    studies were used for the quantitative assessment of Os-related human health risks.
24          In the current review, the health risk assessments in the first draft REA estimated Os-
25    attributable all-cause mortality; hospital admissions and emergency department visits for
26    respiratory causes; and respiratory symptoms in asthmatics. The first draft REA will also
27    estimate Os-attributable lung function decrements, though these analyses will be made available
28    in conjunction with the release of this first draft Policy Assessment and, therefore, will be fully
29    considered in the second draft Policy Assessment (US EPA, 2012b, chapter 6). Ozone-
30    attributable health effects have been estimated  for recent ambient Os concentrations, based on
31    2006-2010 air quality data, as well as for ambient concentrations associated with just attaining
32    the current 8-hour ozone NAAQS, based on adjusting 2006-2010 air quality data.7
      7As with the exposure assessment, the second draft REA will also estimate O3 health risks associated with just
      meeting different potential alternative O3 standards.
                                                    3-8

-------
 1           Section 3.2.1 below discusses the risk assessment based on concentration-response
 2    relationships from epidemiologic studies, which are used as the health basis for assessing all-
 3    cause mortality; respiratory-related hospital admissions and emergency room visits; and
 4    respiratory symptoms. Section 3.2.2 provides a placeholder for the discussion of the risk
 5    assessment based on exposure-response relationships for lung function derived from controlled
 6    human exposure studies.
 7    3.2.1   Risk Assessment Based on Epidemiologic Studies
 8           As discussed in the third draft ozone ISA (EPA, 2012a, chapter 6), a number of Os
 9    epidemiologic studies have evaluated associations between short-term 63 concentrations and
10    mortality or morbidity in different locations across the United States. The first draft REA
11    evaluated risks of mortality and morbidity from short-term exposures to Os based on application
12    of concentration-response functions derived from such epidemiologic studies. The analyses
13    included both a set of urban area case studies and a national scale assessment.
14           The urban case study analyses evaluated mortality and morbidity risks, including
15    emergency department visits, hospitalizations, and respiratory symptoms associated with recent
16    63 concentrations (2006-2010) and with 63 concentrations adjusted to simulate just meeting the
17    current Os standard.  Mortality and hospital admissions (HA) were evaluated in 12 urban areas,
18    while emergency department visits and respiratory symptoms were evaluated in subsets of these
19    12 areas.  The 12 urban areas were: Atlanta, GA; Baltimore, MD; Boston, MA; Cleveland, OH;
20    Denver, CO; Detroit, MI; Houston, TX; Los Angeles, CA; New York, NY; Philadelphia, PA;
21    Sacramento, CA; and St. Louis, MO. The urban case study analyses focused on risk estimates
22    for the middle year of each three-year attainment simulation period (2006-2008 and 2008-2010)
23    in order to provide estimates of risk for a year with generally higher Os levels (2007) and a year
24    with generally lower O3 levels (2009) (US EPA,2012b, section 9.3).
25           The national  scale assessment evaluated only mortality associated with recent Os
26    concentrations  across the entire U.S for 2006-2008.  The national scale  assessment is a
27    complement to the urban scale analysis, providing both a broader assessment of Os-related health
28    risks across the U.S., as well as an evaluation of how well the 12 urban study areas represented
29    the full distribution of ozone-related health risks in the U.S.
30           Both the urban area and national  scale assessments provided the absolute incidence and
31    percent of incidence  attributable to Os.  In modeling risks, the REA employed continuous non-
32    threshold concentration-response functions, reflecting the conclusions reached in the ISA (US
33    EPA, 2012a, sections 6.2.7 and 6.6.2.4).  However, the draft REA also recognized that
34    confidence is increased in the concentration-response function as it is applied to O?,

                                                   3-9

-------
 1    concentrations closer to the central mass of 63 measurements used in the underlying
 2    epidemiological study.  Specifically, the REA noted that estimates of risk associated with Os
 3    concentrations below the lowest measured level (LML) for the underlying epidemiological study
 4    would be associated with reduced confidence since these estimates involve applying the
 5    concentration-response function outside of the range of data used in its derivation.  In light of
 6    this, the REA has characterized mortality risks in excess of lowest measured 63 concentrations8
 7    as well as total risks associated with O3 concentrations down to zero (US EPA, 2012b, sections
 8    7.3.3 and 8.1.1 A)9 In considering these different approaches, the REA concluded that the two
 9    sets of estimates provide a reasonable bound on estimated total risks, reflecting uncertainties
10    about the concentration-response functions below the lowest ozone concentrations evaluated in
11    the studies.
12           The remainder of this section discusses the key observations from the analyses in the
13    REA with regard to CVattributable all-cause mortality (section 3.2.1.1) and CVattributable
14    morbidity effects, including hospital admissions,  emergency department visits, and respiratory
15    symptoms (section 3.2.1.2). Section 3.2.1.3 identifies key sources of variability and uncertainty
16    in these risk assessments, as discussed in the REA.
17           3.2.1.1 All-cause mortality
18           As noted above, the REA estimated (Vattributable mortality in a national assessment for
19    2006-2008 air quality and in 12 specific urban areas for 2007 and 2009 air quality and, in the
20    case of the 12 urban areas, for 2007 and 2009 air quality adjusted to simulate just meeting the
21    current standard.  Mortality estimates were based on concentration-response relationships from
22    two studies (Bell et al., 2004; Zanobetti and Schwartz, 2008). For the 12 urban area risk
23    analysis, estimates of mortality attributable to short-term Os exposures under recent conditions
24    varied widely across urban study areas, reflecting differences in ambient 63 concentrations and
25    populations, as well as  differences in city-specific effect estimates.  The estimates based on
26    Zanobetti and Schwartz (2008) showed the largest (^-associated mortality risks in Boston,
27    Detroit, Los Angeles, and New York, while the estimates based on Bell et al (2004) showed the
28    largest risks in Atlanta, Boston, Houston, Los Angeles, and New York (US EPA, 2012b, section
29    9.3).
30           Across the 12 study cities, and using estimates based on both Bell et al. (2004) and
31    Zanobetti and Schwartz (2008), (Vattributable mortality risk estimates for unadjusted air quality
      8 Due to data limitations, the REA did not identify the actual LMLs from the epidemiologic studies used in the risk
      assessment. Rather, as a surrogate for the study-based LMLs, the REA used the lowest O3 concentrations from the
      composite monitor O3 distributions used to model health risks (US EPA, 2012b, section 7.1.1).
      9For morbidity endpoints, risks were estimated down to the LML but not for total O3 concentrations down to zero
      (US EPA, 2012b, chapter 7).
                                                    3-10

-------
 1    ranged from approximately 20 to 930 deaths, accounting for approximately 0.5 to 4.9% of total
 2    baseline all-cause mortality (for 2007 air quality).  When risk estimates were focused on Os
 3    concentrations above the LML, risk estimates were somewhat smaller (i.e., approximately 10 to
 4    730 deaths across the different cities).  For 2009 unadjusted air quality, the CVattributable
 5    mortality risk estimates ranged from 20 to approximately 980 deaths across the 12 cities,
 6    accounting for approximately 0.6 to 4.3% of total baseline all-cause mortality.  When risk
 7    estimates were focused on O3 concentrations above the LML, these risk estimates ranged from
 8    approximately 10 to 780  deaths across cities, accounting for approximately 0.4 to 3.0% of total
 9    baseline all-cause mortality. Mortality estimates based on 63 effect estimates from Bell et al.
10    (2004) were generally larger than estimates based on effect estimates from Zanobetti and
11    Schwartz (2008), likely due to the larger effect estimates reported by Bell et al. and to the longer
12    O3 season modeled in Bell et al. (US EPA, 2012b, Table 7-4).
13          After simulating just meeting the current standard, estimates of (Vattributable mortality
14    decreased across the 12 study cities, as noted in Table 4-1  of the first draft REA (US EPA,
15    2012b, section 7.5).  Specifically, in considering estimates based on effect estimates from both
16    epidemiologic studies, we note that reductions in (Vattributable mortality upon simulation of
17    just meeting the current Os NAAQS were estimated to range from approximately 10 to 50% for
18    2007 air quality and approximately 0.1 to 35% for 2009 air quality. Although this suggests the
19    potential for important risk reductions as precursor emissions are reduced to meet the current
20    NAAQS, particularly when these percentages are considered within the context of national
21    estimates of (Vattributable mortality,  estimates also suggest that substantial CVattributable risks
22    will remain after meeting the current NAAQS.  With regard to the (Vattributable mortality risks
23    estimated to remain after air quality was adjusted to simulate just meeting the current Os
24    NAAQS, we specifically note the following:
25       •  Using effect estimates from Zanobetti and Schwartz (2008) for the 2007 simulation year,
26          the REA estimated that the range of remaining CVattributable deaths in the 12 U.S. study
27          cities was approximately 20-850, based on estimates with no O3 concentration cutoff, and
28          10-630, based on  estimates  down to the LML. The cities with the largest remaining CV
29          attributable mortality risks were New York City [849 (no cutoff), 626 (LML)], Detroit
30          [212 (no cutoff), 122 (LML)], and Boston [209 (no cutoff),  110 (LML)].
31
32       •  Using effect estimates from Zanobetti and Schwartz (2008) for the 2009 simulation year,
33          the REA estimated that the range of remaining CVattributable deaths in the 12 U.S. study
34          cities was approximately 20-780, based on estimates with no O3 concentration cutoff, and
35          10-520, based on  estimates  down to the LML. The cities with the largest remaining CV
36          attributable mortality risks were New York City [777 (no cutoff), 521 (LML)], Detroit
37          [178 (no cutoff), 127 (LML)], Boston [180 (no cutoff), 93 (LML)], and Los Angeles [175
38          (no cutoff), 83 (LML)].

                                                   3-11

-------
 1
 2       •  Using effect estimates from Bell et al. (2004) for the 2007 simulation year, the REA
 3          estimated that the range of remaining CVattributable deaths in the 12 U.S. study cities
 4          was approximately 30-830, based on estimates with no Os concentration cutoff, and 30-
 5          590, based on estimates down to the LML. The cities with the largest remaining CV
 6          attributable mortality risks were New York City [827 (no cutoff), 585 (LML)], Los
 7          Angeles [786 (no cutoff), 567 (LML)], Boston [404 (no cutoff), 282 (LML)], Atlanta
 8          [415 (no cutoff), 260 (LML)], Houston [270 (no cutoff), 217 (LML)], and St. Louis [193
 9          (no cutoff), 157 (LML)].
10
11       •  Using effect estimates from Bell et al. (2004) for the 2009 simulation year, the REA
12          estimated that the range of remaining CVattributable deaths in the 12 U.S. study cities
13          was approximately 30-820, based on estimates with no Os concentration cutoff, and 20-
14          630, based on estimates down to the LML. The cities with the largest remaining CV
15          attributable mortality risks were Los Angeles [821 (no cutoff), 628 (LML)], New York
16          City [764  (no cutoff), 576 (LML)], Atlanta [364 (no cutoff), 315 (LML)], Boston [369
17          (no cutoff), 250 (LML)], and Houston [272 (no cutoff), 211 (LML)].
18
19          In the national analysis of CVattributable mortality for the years 2006 to 2008, the REA
20    estimated 18,000 (Vattributable deaths based on O3 effect estimates from Bell et al. (2004) and
21    15,000 CVattributable deaths based on Os effect estimates from Zanobetti and Schwartz (2008),
22    accounting for approximately 1.9 to 2.5% of total mortality (US EPA, 2012b, chapter 8, Tables
23    1.2 and 1.3). Of these O3-attributable deaths, the first draft REA estimated that 85-90% occur in
24    locations where the seasonal average 8-hr daily maximum or 8-hr daily mean (10am-6pm) O3
25    concentration is greater than 40 ppb, corresponding to 4th high 8-hour daily maximum 63
26    concentrations ranging from approximately 50 ppb to 100 ppb. In addition, the national analysis
27    showed that the 12 urban study areas considered in the city-specific risk assessment represent the
28    overall distribution of risk across the nation well, with a potential for better characterization of
29    the high end of the risk distribution.
30          3.2.1.2 Respiratory morbidity
31          As noted above, the REA also estimated CVattributable respiratory hospital admissions
32    for the 12 urban study areas, and emergency department visits and asthma exacerbations for
33    subsets of the 12 cities, based on availability of data. These estimates were based on
34    concentration-response relationships from several available epidemiologic studies, using
35    different statistical approaches (US EPA, 2012b, section 7.5, Table 7-4), and they characterized
36    risks for O3 concentrations down to the LML.  With regard to these estimates, we specifically
37    note the following:
38

                                                   3-12

-------
 1    •   In Atlanta for 2007 unadjusted air quality, using different studies and statistical models, the
 2       REA estimated approximately 3,000 to 6,000 respiratory emergency department visits
 3       attributable to O3.  Upon simulating just meeting the current O3 NAAQS, the REA estimated
 4       that (Vattributable emergency department visits decreased by approximately 20%, but that
 5       approximately 2,000 to 5,000 (Vattributable emergency department visits remained.  For
 6       2009, the REA estimated that approximately 3,000 to 7,000 (Vattributable emergency
 7       department visits remained (US EPA, 2012b, Table 7-21).
 8
 9    •   In New York City  for 2007 unadjusted air quality, using different statistical models, the REA
10       estimated approximately 7,000 to 11,000 asthma emergency department visits attributable to
11       (V  Upon simulating just meeting the current 63 NAAQS, the REA estimated that 
-------
 1          3.2.1.3 Key sources of variability and uncertainty
 2          As discussed in the first draft REA (US EPA, 2012a, section 7.4), an important
 3    component of a population risk assessment is the characterization of both variability and
 4    uncertainty in risk estimates.  Variability refers to the heterogeneity of a variable of interest
 5    within a population or across different populations while uncertainty refers to the lack of
 6    knowledge regarding the actual values of inputs to an analysis. Key sources of variability in the
 7    REA risk assessments, as identified in section 7.4.1 of the first draft REA (US EPA, 2012b),
 8    include heterogeneity in (Vrelated health effects across different urban areas; intra-urban
 9    variability in ambient Os concentrations; variability in the patterns of ambient Os reductions
10    across urban areas (i.e., for rollback to current standard); co-pollutant concentrations;
11    demographics and socioeconomic status-related factors; and baseline incidence of disease (US
12    EPA, 2012b, section 7.4.1).  Key sources of uncertainly include the use of ambient Os
13    concentrations measured by existing monitors as surrogates for population exposures;
14    characterizing U.S. background Os concentrations; characterizing intra-urban population
15    exposures; shape and statistical fit of concentration-response functions; surrogate LML
16    concentrations used in risk assessment; potential for co-pollutant confounding; appropriate lag
17    structure; using studies from a particular geographic area to estimate risks for locations outside
18    study area (US EPA, 2012b, Table 7-6). The potential implications of these sources of
19    variability and uncertainty are discussed in section 7.4.1 of the first draft REA (US EPA,  2012b).
20    3.2.2  Risk Assessment Based on Controlled Human Exposure Studies
21          As noted above, the risk assessment of (Vinduced lung function decrements will  be
22    released in parallel with this first draft Policy Assessment. The results of this assessment will be
23    fully considered in the second draft Policy Assessment.

24    3.3    REFERENCES

25    Bell, ML; McDermott, A; Zeger, SL; Samet, JM; Dominici,  F. (2004). Ozone  and short-term
26          mortality in 95 US urban communities, 1987-2000. JAMA 292: 2372-2378.
27    US EPA (2007) Review of the national ambient air quality standards for ozone: assessment of
28          scientific and technical information. OAQPS staff paper. (Updated Final) July 2007.
29          Research Triangle Park, NC: Office of Air Quality Planning and Standards; EPA report
30          no. EPA-452/R-07-007. Available online at:
31          http ://epa. gov/ttn/naaqs/standards/ozone/s_o3_cr_sp.html.
32    U.S. Environmental Protection Agency. (2011). Ozone National Ambient Air Quality Standards:
33          Scope and Methods Plan for Health Risk and Exposure Assessment, U.S. Environmental
34          Protection Agency, Research Triangle  Park, NC. EPA-452/P-11-001.

                                                   3-14

-------
 1   US EPA (2012a). Integrated Science Assessment of Ozone and Related Photochemical Oxidants
 2          (Third External Review Draft).  U.S. Environmental Protection Agency, Washington,
 3          DC. EPA/600/R-10/076C.  June 2012.  Available at:
 4          http://www.epa.gOv/ttn/naaqs/standards/ozone/s o3 2008 isa.html

 5   US EPA (2012b). Health Risk and Exposure Assessment for Ozone, First External Review
 6          Draft. Office of Air Quality Planning and Standards, Research Triangle Park, NC,
 7          27711. EPA 452/P-12-001. July 2012.  Available at:
 8          http://www.epa.gOv/ttn/naaqs/standards/ozone/s o3 2008 rea.html

 9   Zanobetti, A; Schwartz, J. (2008b). Mortality displacement in the association of ozone with
10          mortality: An analysis of 48 cities in the United States. Am J Respir Crit Care Med 177:
11          184-189.
                                                   5-15

-------
 1                4.  PRELIMINARY STAFF CONCLUSIONS REGARDING THE
 2                                    PRIMARY O3 NAAQS

 3          The purpose of this chapter is to present staffs preliminary considerations and
 4    conclusions regarding the adequacy of the current primary Os standard and additional analyses
 5    that would be appropriate to inform consideration of potential alternative standards in the second
 6    draft PA, including additional exposure and risk analyses.  Our preliminary conclusions in this
 7    first draft chapter are based on the assessment and integrative synthesis of the scientific evidence
 8    presented in the third draft 63 ISA and the exposure and risk analyses presented in the first draft
 9    Health REA (discussed above in chapters 2 and 3, respectively). In the final PA, this chapter
10    will present staffs final considerations and conclusions for the Administrator to consider
11    regarding the adequacy of the current primary 63 standard and, if appropriate, staffs
12    considerations and conclusions regarding the range of potential alternative standards that could
13    be supported by the scientific evidence and exposure/risk information available in this review.
14          In this first draft PA, we consider the following overarching questions:
15       •  To  what extent does the available scientific evidence and exposure/risk information
16          support or call into question the adequacy of the public health protection afforded
17          by the current Os primary standard?
18
19       •  What additional analyses would be appropriate to help inform consideration of
20          potential alternative standards in the second draft of the PA?
21          In considering our approach to developing preliminary conclusions regarding the primary
22    Os standard, we first note that the CAA charges the Administrator with setting primary NAAQS
23    that are "requisite" (i.e., neither more nor less stringent than necessary) to protect public health
24    with an adequate margin of safety. In light of this requirement, we note that a decision on the
25    adequacy of the public health protection provided by the current Os standard, and by potential
26    alternative standards, will be a public health policy judgment in which the Administrator weighs
27    the available evidence, exposure/risk information, and the uncertainties and limitations inherent
28    in that evidence and information. Therefore,  in developing preliminary conclusions in this first
29    draft PA, we are mindful that the Administrator's ultimate judgments on the  current and potential
30    alternative standards will most appropriately reflect an interpretation of the available scientific
31    evidence and exposure/risk information that neither overstates nor understates the strengths and
32    limitations of that evidence and information.
33          Section 4.1 discusses the approach to reviewing the 63 NAAQS in the last review and
34    provides a  general overview of the preliminary approach to reviewing the current Os primary
35    NAAQS. Section 4.2 presents staffs preliminary considerations and conclusions regarding the
                                                   4-1

-------
 1    adequacy of the current 8-hour 63 primary standard.  Section 4.3 presents staffs views regarding
 2    additional analyses that would be appropriate to inform consideration of potential alternative
 3    standards in the second draft PA, including additional exposure and risk analyses for the second
 4    draft Health REA.  In the second draft PA, section 4.4 will present a summary of staffs
 5    conclusions on the current standard and, if appropriate,  on alternative standards for the
 6    Administrator's consideration. Section 4.5 will be added to outline areas for future research and
 7    data collection to address key uncertainties identified in this review.

 8    4.1    APPROACH TO REVIEWING THE PRIMARY O3 NAAQS
 9          This section discusses the approach taken to reviewing the primary Os NAAQS in the
10    review completed in 2008 (section 4.1.1) and the approach being taken in the current review
11    (section 4.1.2).
12    4.1.1  Approach Taken in 2008 Review
13          In the last review of the Os NAAQS, the Administrator revised the level of the 8-hour Os
14    primary standard to 75 ppb.1 In  making this decision, the Administrator placed primary
15    emphasis on the body of available scientific evidence, while viewing the results of exposure and
16    risk assessments as providing information in support of the decision. Specifically, the
17    Administrator judged that a standard set at 75 ppb would be appreciably below 80 ppb, the level
18    in controlled human exposure studies at which adverse  effects had been demonstrated  at the
19    time, and would provide a significant increase in protection compared to the then-current
20    standard.  Based on results of the exposure assessment,  he also noted that exposures to Os
21    concentrations at and above a benchmark level of 80 ppb would be essentially eliminated with a
22    standard level of 75 ppb, and that exposures at and above a 70 ppb benchmark level would be
23    substantially reduced or eliminated for the vast majority of people in at-risk groups. In addition,
24    the Administrator concluded that the body of evidence did not support setting a lower  standard
25    level, specifically judging that the available evidence for effects following exposures to Os
26    concentrations of 60 ppb was "too limited to support a primary focus at this level" (FR 75 2938).
27    In light of this conclusion regarding the  evidence, the Administrator did not emphasize
28    quantitative estimates of exposures to Os concentrations at or above a 60 ppb benchmark level.
29          In making his final decision about the level of the primary Os standard, the Administrator
30    noted that the level of 75 ppb was above the range recommended by the CASAC (i.e.,  70 to 60
31    ppb).  The Administrator concluded that "CASAC's recommendation appeared to be a mixture
      lrThe level of the O3 standard is specified as 0.075 ppm rather than 75 ppb. However, in this draft PA we refer to
      ppb, which is most often used in the scientific literature and in the ISA, in order to avoid the confusion that could
      result from switching units when discussing the evidence versus the standard level.
                                                    4-2

-------
 1    of scientific and policy considerations" (FR 75 2938).  The Administrator reached a different
 2    policy judgment than the CASAC Panel based on placing less weight than CASAC on the
 3    available controlled human exposure studies reporting effects following exposures to 60 ppb Os
 4    (Adams, 2002; 2006) and less weight on the results from exposure and risk assessments,
 5    particularly on estimates of exposures to Os concentrations at or above 60 ppb.
 6    4.1.2  Approach in the Current Review
 7          In this review our approach to considering the adequacy of the current primary Os
 8    standard and to identifying a range of potential alternative primary standards for consideration,
 9    draws from the approaches used in previous reviews.  As discussed above, past approaches have
10    been based most fundamentally on using information from Os health studies and exposure/risk
11    analyses in order to inform the selection of 63 standards that, in the Administrator's judgment,
12    protect the public health with an adequate margin of safety.  As in past reviews, staffs approach
13    in this review relies most heavily on consideration of the health evidence, including the
14    controlled human exposure, epidemiologic,  and animal toxicological evidence as assessed in the
15    ISA and on consideration of the Os exposure/risk information, as assessed in the Health REA.
16          In considering our approach in this review, we note that using the available scientific
17    evidence and technical information to inform conclusions on the adequacy of the current primary
18    O3 standard and on potential alternative standards appropriate for consideration is complicated
19    by the recognition that no population threshold, below which it can be concluded with
20    confidence that (Vrelated effects do not occur, can be discerned within the range of 63
21    concentrations commonly observed during the O3 season in the U.S. (US EPA, 2012a, section
22    2.5.4.4). As a result, any approach that uses the available scientific evidence and exposure/risk
23    information to inform decisions on the primary 63 NAAQS would most appropriately require
24    judgments to be made about how to consider the evidence and information, including
25    consideration of how to weigh associated uncertainties.
26          In light of the above, staffs consideration of the adequacy of the current primary Os
27    standard in this first draft PA and of potential  alternative standards in the second draft PA
28    recognizes that the available health effects evidence reflects a continuum from relatively high Os
29    exposure concentrations, at which scientists generally agree that adverse health effects are likely
30    to occur, through lower exposure concentrations, at which the likelihood and magnitude of a
31    response become increasingly uncertain. Therefore, in considering the evidence and information
32    we consider the likelihood that particular 63 exposure concentrations will result in one or more
33    health effects, as well as the likely magnitude  and severity of such effects. In this first draft PA,
34    such considerations include the nature,  magnitude, and statistical precision of Os-related effects,
35    as well as the 63 exposure concentrations reported in controlled human exposure studies and the
                                                    4-3

-------
1    Os air quality concentrations in locations where epidemiologic studies have been conducted. The
2    second draft PA will further refine these considerations, particularly consideration of the
3    approach to using epidemiologic evidence to inform decisions on the primary Os NAAQS, in
4    reaching preliminary conclusions regarding the current and potential alternative primary
5    standards.
6           The approach described above for considering the evidence and information on Os-
7    related health effects in this review is consistent with setting standards that are neither more nor
8    less stringent than necessary, recognizing that a zero-risk standard is not required by the CAA.
9    This approach is outlined below in Figure 4-1.
                                                    4-4

-------
1     Figure 4-1.  Overview of Approach to Reviewing Primary Os NAAQS
                                     Adequacy of CurrentS-HourPrimary 03 Standard
                Evidence-Based Considerations
        >ISA weight-of-evidence conclusions
        >Evidence for03 effects following exposures to
        concentrations below current standard level (controlled
        human exposure, animal toxicology)
        >Evidence for03 effects in locations that would likely
        have met current 03 standard (epidemiologic)
        >Uncertainties in the evidence at those exposure/air
        quality concentrations
                         Exposure-/Risk-Based Considerations
                       > What is the nature, magnitude, and importance
                       of estimated exposures and risks associated with
                       current 03 standard?
                       > Uncertainties in the exposure and risk
                       estimates
                                                   Does information call
                                                   into question adequacy
                                                     of current 8-hour
                                                   Primary 03 standard?
                                                         Considerretaining
                                                          current 8-hour 03
                                                              standard
                                        Consider Potential Alternative Standards
                 Indicator
          > Support for retaining 03?
     Averaging Time
> Support for 8-hour only?
> Support for longer-term
averaging time?
              Form
> Support for retaining annual 4th
highest?
                                                           Level
                > Evidence-based considerations
                         >03exposure concentrations in controlled human exposure and animal toxicology studies
                         >03air quality concentrations  in U.S. epidemiologic study locations
                         > Uncertainties in the evidence at those exposure/air quality concentrations
                >Exposure and risk considerations
                         > What is the nature, magnitude, and importance of estimated exposures and risks associated
                         with potential alternative 03 standards?
                         > Uncertainties in the exposure and risk estimates
                                  Identify range of potential alternative standards for consideration
                                                                  4-5

-------
 1    4.2    PRELIMINARY STAFF CONCLUSIONS REGARDING THE ADEQUACY OF
 2          THE CURRENT PRIMARY O3 STANDARD
 3          In this section, we consider the extent to which the available Os health evidence and
 4    exposure/risk information, as discussed in detail in the third draft ISA and first draft REA
 5    respectively, and as summarized above (chapters 2 and 3), supports or calls into question the
 6    adequacy of the protection afforded by the current 8-hour Os standard against effects that have
 7    been linked to short-term or long-term 63 exposures. Section 4.2.1 presents evidence-based
 8    considerations, section 4.2.2 presents exposure- and risk-based considerations, section 4.2.3
 9    presents the advice and recommendations from CASAC in the last review of the Os NAAQS,
10    and section 4.2.4 presents an integrated synthesis of staffs preliminary conclusions regarding the
11    adequacy of the current Os primary NAAQS.
12    4.2.1  Evidence-Based Considerations
13          In translating information from controlled human exposure, epidemiologic, and animal
14    toxicological studies into the basis for reaching preliminary  staff conclusions on the adequacy of
15    the current O3 standard, we apply the policy framework outlined in Figure 4-1 above. In so
16    doing, we  consider the strength of the evidence for Os-related health effects, the Os
17    concentrations at which such health effects  have been reported, and the potential public health
18    implications of exposures in at-risk populations to Os concentrations that would be allowed by
19    the current Os standard.
20          In first considering the strength of the evidence, we note the weight-of-evidence
21    conclusions in the ISA. Specifically, as discussed above (section 2.1), we note the ISA
22    conclusions that there is a causal relationship between short-term Os exposures and respiratory
23    effects, including respiratory-related morbidity and mortality; there is likely to be  a causal
24    relationship between short-term 63 exposures and all-cause  mortality; and there is likely to be a
25    causal relationship between  long-term Os exposures and respiratory effects (including morbidity
26    and mortality).  These conclusions were based on extensive  bodies of scientific evidence
27    reporting coherent and consistent results using different approaches and study designs to evaluate
28    Os-related health effects. For other health endpoints, we note that the more limited bodies of
29    scientific evidence were judged in the ISA to be either suggestive of a causal relationship or
30    inadequate to infer a causal relationship (US EPA, 2012a, Table 1-1). Specifically, the evidence
31    for cardiovascular and central nervous system effects following short-term exposures; the
32    evidence for cardiovascular, reproductive and developmental, and central nervous system effects
33    following long-term exposures; and the evidence for mortality following long-term exposures
34    was judged to be suggestive of a causal relationship.  The evidence for cancer was judged
35    inadequate to determine whether a causal relationship exists. In light of these conclusions in the

                                                    4-6

-------
 1    ISA, which reflect the strength of the available evidence for different health endpoints, our
 2    evidence-based consideration of the current Os primary standard places emphasis on studies that
 3    have evaluated respiratory effects, including an array of respiratory-related morbidity endpoints
 4    and respiratory-related mortality following short- and long-term exposures, and on studies that
 5    have evaluated total mortality following short-term exposures.
 6           In considering such Os-related health effects within the context of the adequacy of the
 7    current Os NAAQS, we specifically consider the following question:
 8       •   To what extent does the available scientific evidence support the occurrence of
 9           respiratory-related morbidity and mortality effects or total mortality following
10           exposures to Os concentrations that would be allowed by the current Os standard?
11           In evaluating this question, we consider the 63 exposure concentrations at which
12    respiratory effects have been reported in experimental studies (i.e., controlled human exposure
13    and animal toxicological studies) and the ambient 63 concentrations in locations where (V
14    related morbidity or mortality has been reported in epidemiologic studies. The following
15    sections consider the evidence for Os-induced lung function decrements, respiratory symptoms,
16    airway inflammation, and other respiratory effects (section 4.2.1.1); (Vrelated respiratory
17    hospital admissions, emergency department visits, and mortality (section 4.2.1.2); and Os-related
18    total mortality (section 4.2.1.3).  Section 4.2.1.4 discusses other health endpoints and section
19    4.2.1.5 presents an integrated synthesis of staffs evidence-based considerations.
20           4.2.1.1 Lung function, respiratory symptoms, airway inflammation, and other
21                 respiratory effects
22           As discussed above (section 2.2.1), and as assessed and discussed in detail in the ISA (US
23    EPA, 2012a, Chapter 6), controlled human exposure, animal toxicological, and epidemiologic
24    studies provide strong evidence that short-term exposures to Os result in a variety of respiratory
25    effects, including lung function decrements, respiratory symptoms, airway inflammation,
26    increased airway responsiveness,  and decreased lung host defense. When considering studies of
27    these Os-induced respiratory effects within the context of the adequacy of the current Os
28    NAAQS, we consider (1) the extent to which controlled human exposure studies have reported
29    such effects following exposures to Os concentrations below the level of the current standard and
30    (2) the extent to which 63 epidemiologic studies have reported associations with such effects in
31    locations likely to have  met the current O3 standard.  In considering such studies, we also note
32    the extent to which effects reported following exposures to Os concentrations below the current
33    standard level are plausibly  linked to the larger body of evidence for effects following exposures
34    to higher Os concentrations.
                                                    4-7

-------
 1          In first considering controlled human exposure studies of Os-induced lung function
 2    decrements within the context of the current Os standard, we note that such studies have
 3    consistently reported decrements in group mean lung function in healthy exercising adults
 4    following 6.6-hour exposures to Os concentrations ranging from 60 to 120 ppb (US EPA, 2012a,
 5    Figure 6-1). When data were combined across a number of studies, group mean decrements
 6    were approximately 3% following exposures to 60 ppb Os, 6% to 8% following exposures to 80
 7    ppb O3, 8% to 14% following exposures to  100 ppb O3, and 13% to 16% following exposures to
 8    120 ppb Os. In addition, a study that evaluated exposures to 70 ppb Os (Schelegle et al., 2009)
 9    reported a 6% decrement in group mean FEVi (section 2.2.1.1, above). In considering these
10    studies as a whole, we note that reported group mean decrements have consistently been
11    statistically significant following exposures at or above 70 ppb Os. Following exposures to 60
12    ppb Os, decrements have been reported to be statistically significant in a study by Kim et al.
13    (2011) and in an analysis of existing data by Brown et al. (2008), though not in the original  study
14    that generated the data used by Brown et al. (Adams, 2006) or in a study by Schelegle et al.
15    (2009) (US EPA, 2011b, section 6.2.1.1). Statistically significant lung function decrements have
16    not been reported following exposures to Os concentrations below 60 ppb (i.e., 40 ppb) (US
17    EPA, 2012a, Figure 6-1).
18          In further considering these controlled human exposure studies of (Vinduced lung
19    function decrements, we note that while it is important to consider the statistical  precision of
20    group mean decrements when evaluating whether  reported effects are due to Os exposures rather
21    than chance alone, when considering the potential  public health implications of study results it is
22    also important to consider the potential for some individuals to experience larger decrements
23    than average (US EPA, 2012a,  section 6.2.1.1). Such inter-individual variability in lung function
24    responses is relevant when evaluating the proportion of the population that might experience
25    clinically  meaningful effects during an Os exposure.
26          As noted in the ISA (US EPA, section 6.2.1.1), for individuals with relatively normal
27    lung function, a within-day change in FEVi of 5% or larger is clinically meaningful and a 10%
28    reduction in FEVi has been generally accepted as  an abnormal response (Dryden et al., 2010;
29    American Thoracic Society, 2000). In addition, in the last review of the Os  standard CAS AC
30    recommended that EPA focus on FEVi decrements at or above 10% when estimating potentially
31    adverse lung function decrements in people with lung disease. The CASAC noted that while
32    such moderate levels of lung function impairment  would likely not interfere with normal activity
33    for most healthy individuals, even moderate functional impairment would likely interfere with
34    normal activity for many individuals with lung disease (Henderson, 2006).
                                                   4-8

-------
 1          In light of the above, in considering the potential for some individuals to experience
 2    potentially adverse CVinduced lung function decrements (as defined by CASAC in the last
 3    review) following exposures to Os concentrations below the level of the current standard, we
 4    consider the extent to which exposures to 63 concentrations below 75 ppb result in FEVi
 5    decrements of 10% or more. With regard to this, we note that following 6.6-hour exposures to
 6    an average 63 concentration of 60 ppb, the proportion of healthy adult study subjects at
 7    intermittent, moderate exertion, with such potentially adverse FEVi decrements ranged from 3%
 8    to 20% across several  studies (US EPA, 2012a, section 6.2.1.1).  When the results from such
 9    studies were combined, the ISA noted that 10% of healthy subjects experienced such FEVi
10    decrements following  exposures to an average Os concentration of 60 ppb (US EPA, 2012a,
11    section 6.2.1.1).  In addition, in a study that evaluated the exposure-response relationship in the
12    percentage of subjects experiencing FEVi decrements greater than or equal to 10%, the
13    percentage increased with increasing Os exposure concentrations such that no subjects
14    experienced such decrements following exposures to filtered air, 16% of subjects  experienced
15    such decrements following exposures to 60 ppb O3; 19% experienced such decrements following
16    exposures to 70 ppb Ch; and 29% experienced such decrements following exposures to 80 ppb
17    63 (US EPA, 2012a, Figure 6-2). Taken together these results indicate that, although group
18    mean lung function decrements are relatively small following exposures to Os concentrations
19    below the level of the  current standard, on average more than 10% of healthy individuals
20    experience CVinduced decrements that are potentially adverse for those with lung diseases such
21    as asthma.
22          With regard to the O3-induced FEVi decrements reported in controlled human exposure
23    studies, we also note that these studies were conducted in healthy adults, while individuals in at-
24    risk groups could experience larger (Vinduced decrements. Specifically, the ISA concludes that
25    there is adequate evidence to identify several groups as particularly at-risk for CVrelated effects,
26    including asthmatics and children (US EPA, 2012a, section 2.4.5.1). Several studies have
27    reported larger (Vinduced FEVi decrements in asthmatics than non-asthmatics (US EPA, 2012a,
28    section 8.2.2), with one study reporting that decrements increased with increasing disease
29    severity (Horstman et  al., 1995). In addition, children experience higher 63 exposures and doses,
30    compared to adults, due to increased time spent outdoors and increased ventilation rates  (US
31    EPA, 2012a, section 2.4.5.1).  In light of this evidence, we note that studies conducted in healthy
32    adults may underestimate (Vinduced lung function decrements in asthmatics and children (US
33    EPA, 2012a, Preamble).
34          In further considering the potential clinical implications of (Vinduced effects reported in
35    controlled human exposure studies, we note that a number of such studies have  also evaluated

                                                   4-9

-------
 1    respiratory symptoms.  Statistically significant increases in respiratory symptoms (which
 2    includes pain on deep inspiration, shortness of breath, and cough) have been consistently
 3    reported in healthy volunteers engaged in intermittent, moderate exertion following 6.6 hour
 4    exposures to average 63 concentrations at or above 80 ppb (US EPA, 2012a, section 6.2.1.1). In
 5    addition, in a recent study Schelegle et al. (2009) reported a statistically significant increase in
 6    respiratory symptoms following 6.6 hour exposures to an average 63 concentration of 70 ppb.
 7    With regard to lower exposure concentrations, we note that two studies, Adams (2006) and
 8    Schelegle et al. (2009), reported a tendency for increased respiratory symptoms in healthy
 9    volunteers during 6.6 hour exposure protocols with average 63 exposure concentrations of 60
10    ppb. For one of the exposure protocols tested in the study by Adams (i.e., the triangular
11    exposure protocol), the increase in symptoms was reported to be statistically different from
12    initial respiratory symptoms, though neither study reported the increase in symptoms to be
13    statistically significant relative to filtered air controls.
14           Consistent with the results of the controlled human exposure studies in healthy adults, we
15    note the large number of Os epidemiologic studies that have reported positive and statistically
16    significant associations with lung function decrements and respiratory symptoms in a variety of
17    populations including children attending summer camps, adults exercising outdoors, outdoor
18    workers, and asthmatic children living in urban areas (US EPA, 2012a, section 6.2.1).
19    Consistent with controlled human exposure studies, O3-related group mean decrements in lung
20    function were often small in magnitude, though some individuals experienced decrements in
21    excess of 10%. When studies of (Vassociated lung function decrements or respiratory
22    symptoms have evaluated the potential for confounding by co-pollutants, meteorological factors,
23    or pollen counts, we note that associations with Os generally remained robust (sections 2.2.1.1
24    and 2.2.1.4, above). In considering these studies within the context of the adequacy of the
25    current Os standard, we take particular note of studies that have been conducted in locations
26    likely to have met the current Os standard, though we also note that the results of such studies are
27    consistent with the larger body of epidemiologic studies reporting (Vinduced lung function
28    decrements and respiratory symptoms in locations that would not have likely met the current Os
29    standard (US EPA, 2012a, section 6.2).2
30           As discussed above (section 2.2.1), associations with lung function decrements and
31    respiratory symptoms typically remained, and in several  studies remained statistically significant,
      Because attainment status is based on ambient O3 concentrations measured at individual monitors, with the highest
      relevant monitored concentration in an area determining whether that area meets or violates the NAAQS, it is most
      appropriate to use O3 concentrations from individual monitoring sites when drawing conclusions about whether a
      study area would likely have met or violated the O3 NAAQS. Therefore, in considering ambient O3 concentrations
      in epidemiologic study locations, we focus on locations with data available from individual monitoring sites.
                                                     4-10

-------
 1    in analyses restricted to ambient 63 concentrations near or below the level of the current 63
 2    standard (US EPA, 2012a, section 6.2.1.2 and Table 6-6). For example, Os-associated lung
 3    function decrements in children remained statistically significant in an analysis restricted to 1-
 4    hour average 63 concentrations below 60 ppb (Spektor et al., 1988) and remained, though were
 5    not statistically significant, in a separate analysis restricted to concentrations below 61 ppb
 6    (Brunekreef et al., 1994).  In studies of outdoor workers, (Vassociated lung function decrements
 7    have been reported in analyses restricted to maximum hourly concentrations below 40 ppb
 8    (Brauer et al., 1996); in a location where the highest 8-hour concentration was 65 ppb (Chan and
 9    Wu, 2005); and in a location with a maximum 30-minute concentration of 77 ppb3 (Hoppe et al.,
10    1995).  In addition, positive associations with respiratory symptoms were statistically significant
11    in asthmatic children in Denver, a location with a maximum 1-hour O?, concentration of 70 ppb
12    during the study period (Rabinovitch et al., 2004). Considered together, we note that these
13    epidemiologic results, combined with the results of the controlled human exposure studies
14    discussed above, provide strong support for the conclusion that short-term exposures to ambient
15    O3 concentrations can result in direct and independent effects on respiratory function in healthy
16    children and adults, and in asthmatics, even when ambient Os  concentrations remain below the
17    level of the current standard.
18           In further considering Os-induced decrements in lung function and increases in
19    respiratory symptoms, we note the American Thoracic Society (ATS) conclusions that (1)
20    reversible loss of lung function in combination with respiratory symptoms should be  considered
21    adverse and that (2) a downward shift in the population distribution of lung function should be
22    considered adverse, even in the absence of overt illness, because individuals within that
23    population would have diminished reserve function placing them at increased risk from other
24    agents such as viral infections (ATS, 2000, p. 672).  Consistent with these conclusions, the ISA
25    states the following (US EPA, 201 Ib, section 6.2.1.2, pp. 6-54 to 6-55):
26           "In considering the clinical significance of more subtle health outcomes such  as lung
27           function changes, it is important to note that  a small shift in the population mean likely
28           will have a disproportionate effect in the extreme ends of the distribution of lung function
29           where these small  magnitudes of decrease lead to clinically-significant airway resistance
30           or obstruction and where individuals likely have concurrent symptoms.  Several
31           epidemiologic studies have demonstrated the clinical significance of (^-associated lung
32           function decrements,  primarily in individuals with asthma, by finding concomitant
33           increases in respiratory symptoms."
      3While it is not possible to conclude definitively that the maximum 8-hour concentration would have been at or
      below 75 ppb in the study by Hoppe et al. (1995), it would have been somewhat lower than 77 ppb, the maximum
      30-minute concentration.
                                                   4-11

-------
 1          In light of these conclusions, and in considering the controlled human exposure and
 2    epidemiologic studies discussed above, we note that exposures to Os concentrations below the
 3    level of the current standard can result in respiratory effects that meet ATS criteria for adversity,
 4    given that (1) controlled human exposure and epidemiologic studies have reported lung function
 5    decrements and respiratory symptoms in healthy adults following exposures to Os concentrations
 6    below 75 ppb, 8-hour average, with one controlled human exposure study reporting both lung
 7    function decrements and respiratory symptoms following exposures to 70 ppb. and (2) controlled
 8    human exposure and epidemiologic studies have reported a downward shift in the population
 9    distribution of lung function (e.g., as indicated by decrements in group mean FEVi) following
10    exposures to  Os concentrations below 75 ppb, including 8-hour average concentrations as low as
11    60 ppb in some controlled human exposure studies.
12          In addition to Os-induced lung function decrements and respiratory symptoms, controlled
13    human exposure, epidemiologic,  and animal toxicological studies have reported airway
14    inflammation following Os exposures. Although there is little quantitative information available
15    to relate changes in markers of airway inflammation to specific clinical effects, inflammation is
16    the host response to injury and the induction of inflammation is evidence that injury has occurred
17    (US EPA, 201 Ib, section 6.2.3).  Animal toxicology studies suggest that continued airway
18    inflammation could have a variety of important respiratory consequences,  including (1) the
19    development of a chronic inflammatory state; (2) altered structure and function of pulmonary
20    tissues; (3) alteration of the body's host defense response to inhaled microorganisms; and (4)
21    alteration of the lung's response to other agents such as allergens or toxins (US EPA, 2012a,
22    section 6.2.3). Some controlled human exposure studies have reported more pronounced O3-
23    induced inflammatory responses  in asthmatics than non-asthmatics and some studies have
24    reported that tissue damage persists for days or weeks following repeated 63 exposures (US
25    EPA, 201 Ib, sections 6.2.1.3, 6.2.3.1). Ozone-induced airway inflammation is generally
26    reproducible within individuals over time (Holz et al., 1999), and some individuals are
27    intrinsically more susceptible to increased inflammatory responses following 63 exposures (Holz
28    et al., 2005).
29          We note that several controlled human exposure studies have reported Os-induced airway
30    inflammation in humans following exposures to Os concentrations of 80 and 100 ppb (Alexis et
31    al., 2010; Peden et al., 1997; Devlin et al., 1991), and that the only controlled human exposure
32    study to have evaluated airway inflammation following exposures below the level of the current
33    63 standard (Kim et al., 2011) reported a statistically significant increase in neutrophilic
34    inflammation following 6.6 hour exposures to 60 ppb Oj in healthy adults  engaged in
35    intermittent,moderate exertion (section 2.2.1.2, above). Relative to clean air, exposures to 60

                                                   4-12

-------
 1    ppb 63 resulted in a 16% increase in inflammation, on average across subjects. While
 2    information is not available to quantitatively link markers of airway inflammation with specific
 3    clinical effects, given evidence that asthmatics can experience larger Os-induced airway
 4    inflammation than non-asthmatics and evidence supporting the potential for clearly adverse
 5    respiratory effects with continued airway inflammation, these findings in healthy  humans further
 6    support the potential for important respiratory effects following exposures to 63 concentrations
 7    below the level of the current standard.
 8           In addition to lung function decrements, respiratory symptoms, and airway inflammation,
 9    we note that several additional Os-induced respiratory effects have been reported  in controlled
10    human exposure and animal toxicological studies, including airway hyperresponsiveness,
11    impaired lung host defense, and structural changes in the respiratory system. However, few
12    studies have evaluated these endpoints following exposures to Os concentrations below 75 ppb.
13    With regard to studies evaluating these respiratory effects, we specifically note the following:
14       •   Airway hyperresponsiveness has been reported following short-term exposures of
15           humans to Os concentrations at or above 80 ppb  and in animals following  exposures to
16           63 concentrations as low as 50 ppb (US EPA, 201 Ib, section 6.2.2). Airway
17           hyperresponsiveness following 63 exposures is more pronounced in asthmatics than non-
18           asthmatics and can exacerbate asthma by placing allergic asthmatics at greater risk for
19           prolonged bouts of breathing difficulties due to airway constriction.
20
21       •   Impaired lung host defenses have been reported in healthy humans following short-term
22           exposures to Os concentrations as low as 80 ppb, and a number of human  and animal
23           studies have reported immune cell alterations following exposures to 63 concentrations at
24           or above 80 ppb (US EPA, 201 Ib, section 6.2.5). Such impairments could increase the
25           susceptibility of exposed individuals to respiratory infections.
26
27       •   Animal toxicological studies in rodents and non-human primates have reported that long-
28           term (i.e, chronic or repeated exposures to Os with typical concentrations ranging from
29           120 to 500 ppb) during gestation or development resulted in irreversible morphological
30           changes in the lung, which in turn can  influence  pulmonary function (US EPA,  201 Ib,
31           section 7.2.3).
32
33           While studies of most of these additional endpoints have not evaluated exposures to Os
34    concentrations at or below the level of the current 63 standard, they do provide insight into the
35    range of O3-related respiratory effects and into the potential for at-risk individuals (e.g.,
36    asthmatics) to experience more severe effects.
37           Controlled human exposure studies of Os-induced lung function decrements,  respiratory
38    symptoms, and airway inflammation  provide strong evidence that exposures to 63 concentrations
39    below those allowed by the current O3 standard (i.e., 60-70 ppb) can impair respiratory function.
                                                   4-13

-------
 1    As noted above, members of at-risk groups, including asthmatics and children, would likely
 2    experience larger Os-induced respiratory effects than indicated by controlled human exposure
 3    studies, most of which have been conducted in healthy adults.  Based on available evidence from
 4    controlled human exposure studies, the weighted average proportion of healthy individuals with
 5    >10% FEVi decrements is 10% following exposure to 60 ppb Os. For the purpose of estimating
 6    potentially adverse lung function decrements in people with lung disease, the CASAC indicated
 7    that a focus on the lower end of the range of moderate levels of functional responses is most
 8    appropriate (i.e., FEVi decrements >10%) (Henderson, 2006).  Thus the proportion of healthy
 9    individuals experiencing >10% FEVi decrements in these controlled human exposure studies
10    may be viewed as important from a public health perspective, and the results are important to
11    consider in assessing the overall public health impacts associated with exposures to 03.  In
12    addition, controlled human exposure and animal toxicological studies of airway
13    hyperresponsiveness, impaired host defense, and structural changes in the respiratory system
14    provide perspective on the broad range of (Vrelated respiratory effects.  In some individuals, the
15    types of O3-induced respiratory responses reported in  controlled human exposure studies could
16    become severe enough that they result in increased use of medication,  emergency room visits,
17    and/or hospital admissions. Thus, the strong evidence for lung function decrements, respiratory
18    symptoms, airway inflammation, and other respiratory effects following exposures to Os
19    concentrations commonly encountered in U.S. urban locations supports the biological
20    plausibility of the conclusions that exposures to ambient 63 concentrations can result in
21    respiratory-related hospital admissions and emergency department visits, as well as the most
22    severe (^-associated effect, premature mortality.  Studies of these Os-related effects are
23    discussed below, within the context of considering the adequacy of the current 63 standard.
24          4.2.1.2 Respiratory-related hospital admissions, emergency department visits, and
25                 mortality
26          Ozone-related respiratory hospital admissions, emergency department visits, and
27    mortality have been reported in a number of epidemiologic  studies conducted in the United
28    States and around the world.  In translating information from these epidemiologic studies into the
29    basis for reaching preliminary conclusions on the adequacy of the current 63 standard, we
30    consider available information on the extent to which  the current standard provides protection
31    against health effects that have been reported to be associated with both short- and long-term 63
32    concentrations. Our general approach to considering epidemiologic studies with regard to the
33    adequacy of the current standard is to characterize our degree of confidence in associations
34    reported for 63 concentrations that would likely be allowed by the current standard. In doing so,
35    we note the ambient Os concentrations in study locations and, where available, we note
36    information on the concentration-response relationship between O?, and morbidity or mortality.
                                                   4-14

-------
 1           In placing relative weight on specific epidemiologic studies, we take several factors into
 2    account.  These factors include the geographic coverage provided by the study, the extent to
 3    which analyses control for potential confounders, and the statistical precision of results.  In
 4    considering specific epidemiologic studies, we place considerable weight on information from
 5    U.S. multi-city studies.  These studies have a number of advantages compared to single-city
 6    studies, including representing ambient 63 concentrations and potential health impacts across a
 7    range of diverse locations, providing spatial coverage for different regions reflecting differences
 8    in Os exposure patterns and co-pollutants; lack of publication bias; consideration of larger study
 9    populations that afford the possibility of generalizing to the broader national population; and
10    providing increased statistical power.
11           While we emphasize multi-city studies, we also take into account relevant information
12    from single-city studies, where appropriate.  Specifically, although multi-city studies have more
13    power to detect associations and provide broader geographic coverage, the extent to which
14    effects reported in multi-city studies are associated with the specific Os air quality in any
15    particular location is less clear than in single-city studies. Regional heterogeneity in O^-
16    associated health risks has led some researchers (e.g., Smith et al., 2009) to question the
17    appropriateness of using multi-city approaches to generate national Os effect estimates.
18    Therefore, while single-city studies are more limited than multi-city studies in terms of power
19    and geographic coverage, the link between reported health effects and the air quality in a given
20    study area is more straightforward to establish.  In light of this, we consider relevant information
21    from single-city studies, where appropriate.
22           In light of the above considerations, and given the weight-of-evidence conclusions in the
23    ISA as discussed above, in considering multi-city and single-city epidemiologic studies within
24    the  context of the adequacy of the current Os standard, we consider studies that have evaluated
25    associations between short-term  63 concentrations and respiratory effects or total mortality, as
26    well as studies that have evaluated associations between long-term O3 concentrations and
27    respiratory effects.
28           Short-term exposure studies
29           In considering the available epidemiologic evidence for associations with short-term Os
30    concentrations within the context of considering the adequacy of the current Os standard, we first
31    take into account the 63 air quality concentrations in the locations where U.S. studies have
32    reported associations with respiratory-related hospital admissions, respiratory-related emergency
33    department visits, respiratory-related mortality, and total mortality.  In order to facilitate
                                                     4-15

-------
 1    consideration of this issue, we have used EPA's Air Quality System (AQS)4 to identify 63 design
 2    values5 for the locations where U.S. studies have been conducted.
 3           With regard to U.S. multi-city studies that evaluated O3-associated respiratory hospital
 4    admissions, emergency department visits, or mortality (i.e., Medina-Ramon et al., 2006; Lin et
 5    al., 2008; Katsouyanni et al., 2009; Zanobetti and Schwartz, 2008) (US EPA, 2012a, Chapter 6;
 6    sections 2.2.1 and 2.2.2, above), we note that 50th percentiles of the distributions of Oj design
 7    values across cities ranged from 83 to 94 ppb (Table 4-1).  These median concentrations indicate
 8    that most of the cities included in available U.S. multi-city 63 studies would likely not have met
 9    the current Os standard over the study periods.  In considering design values for individual cities
10    evaluated in these studies we note the following: (1) For Medina-Ramon et al. (2006) 21 of the
11    27 cities evaluated had average (i.e., averaged over the time periods of the studies) O3 design
12    values above 75 ppb; (2) for Lin et al. (2008) 22 of the 26 cities evaluated had average Os design
13    values above 75 ppb;6 (3) for Katsouysanni et al. (2009) 9 of the  14 cities evaluated for hospital
14    admissions and 79 of the 90 cities evaluated for respiratory mortality had average Os design
15    values above 75 ppb; and (4) for Zanobetti and Schwartz (2008) 42 of the 48 cities evaluated had
16    average 63 design values above 75 ppb (Wells  et al., 2012). Given this, we note that the large
17    majority of the cities included in U.S. multi-city studies of Os-related respiratory hospital
18    admissions, emergency department visits, or mortality would likely not have met the current 63
19    standard during the study periods. Therefore, we note that while U.S.  multi-city epidemiologic
20    studies provide strong support for the occurrence of clearly adverse Os-related respiratory
21    effects, consideration of the 63 design values for the cities included in these studies does not
22    provide insight into the appropriateness of the degree of public health protection provided by the
23    current Os standard.
24           Of the U.S. single-city studies evaluating Os-related respiratory effects, only the study by
25    Mar and Koenig (2009) was conducted in a location with an 63 design value lower than 75 ppb.
26    As discussed in section 2.2.1.6 above and as indicated in Table 4-1, this study reported  positive
27    and statistically significant associations with asthma emergency department visits in both
28    children  and adults in Seattle. The average (i.e., over the years of the study) 63 design value for
29    Seattle during the study period was 71 ppb (Table 4-1), and the highest design value for any
30    three-year period during the study was 75 ppb (Wells et al., 2012), indicating that the study area
      4Accessible at http://www.epa.gov/ttn/airs/airsaqs/
      5 A design value is a statistic that describes the air quality status of a given area relative to the level of the NAAQS
      (http://www.epa.gov/airtrends/values.html). In the case of O3, the design value for an area is based on the 3-year
      average of the annual 4th highest 8-hour daily maximum concentration in an O3 season, measured at the monitor
      recording the highest such concentration.
      6The study by Lin et al. (2008) identified regions rather than specific cities. In identifying design values for the
      study area, we considered the cities that were encompassed by the regions evaluated in the study (Wells et al., 2012).
                                                     4-16

-------
 1    would have met the current 63 standard during the entire study period.  Average design values
 2    for other U.S. single-city studies ranged from 104 ppb to 107 ppb (Table 4-1).
 3           Although most of the U.S. O3 epidemiologic studies of respiratory-related hospital
 4    admissions, emergency department visits, and mortality have not been conducted in locations
 5    likely to have met the current 63 standard, a study conducted in New York City (Silverman and
 6    Ito, 2010) has evaluated the  concentration-response relationship between Os and asthma hospital
 7    admissions (US EPA, 2012a, Figure 6-15), potentially providing insight into (^-associated
 8    health risks at air quality concentrations that would have been allowed by the current standard.7
 9    In considering such concentration-response relationships, we take note of the statistical precision
10    of estimates of that relationship at different 63 concentrations across the air quality distribution.
11    We specifically note the O3 concentrations below which confidence bounds indicate appreciably
12    less confidence in the nature of the association. In taking this approach to evaluating
13    concentration-response relationships we acknowledge that the decreasing precision in effect
14    estimates reported  as concentrations approach the lower extreme of the air quality distribution, as
15    indicated by notable widening of confidence bounds, is intrinsically related to data density and is
16    not necessarily indicative of an absence of (Vassociated health effects.
17           As discussed in section 2.2.1.6 above, Silverman and Ito (2010) evaluated the
18    concentration-response relationship for Os and asthma hospital admissions in New York City
19    using a co-pollutant model that also included PM2 5.  The authors concluded that a linear
20    relationship between 63 and hospital admissions is a reasonable approximation of the
21    concentration-response function throughout much of the range of ambient Os concentrations.
22    Based on the 95%  confidence intervals around the concentration-response function (US EPA,
23    2012a, Figure 6-15), the most precise estimates of this relationship were reported for average
24    (i.e., averaged across monitors in the study area) 8-hour daily maximum Os concentrations
25    around 40 ppb (the 50th percentile of the distribution was 41 ppb), with confidence intervals
26    becoming progressively wider at average 8-hour daily maximum O3 concentrations below and
27    above this concentration. As noted above, this broadening of the confidence intervals is likely
28    due to the decreasing amounts of data at the lower and upper ends of the air quality distribution.
29
      7In addition, the study by Strickland et al. (2010) evaluated the concentration-response relationship with asthma
      emergency department visits in Atlanta. To the extent it is judged appropriate, based on advice received from
      CASAC on this first draft Policy Assessment, the second draft Policy Assessment could also consider the
      concentration-response relationship reported in the study by Strickland et al. (2010) within the context of the current
      O3 NAAQS in a manner analogous to our consideration in this section of the study by Silverman and Ito (2010).
                                                    4-17

-------
1    Table 4-1. Average Ozone Design Values in U.S. Cities where Epidemiologic Studies have
2    Evaluated Associations between Short-Term Os and Respiratory-Related Hospital Admissions
3    (HA), Emergency Department (ED) Visits, and Mortality
Study
All-year
Strickland et al.
(2010)
Medina-Ramon
et al. (2006)
Katsouyanni et
al. (2009)
Katsouyanni et
al. (2009)
Warm Season
Ito et al. (2007)
Darrow et al.
(2011)
Tolbert et al.
(2007)
Strickland et al.
(2010)
Silverman and
Ito (2010)
Silverman and
Ito (2010)
Medina-Ramon
et al. (2006)
Katsouyanni et
al. (2009)
Zanobetti and
Schwartz (2008)
Katsouyanni et
al. (2009)
Katsouyanni et
al. (2009)
Lin et al. (2008)
Mar and Koenig
(2009)
Mar and Koenig
(2009)
Endpoint

Asthma ED
COPD HA
Respiratory
HA
Respiratory
mortality

Asthma ED
Respiratory
ED
Respiratory
ED
Asthma ED
Asthma
HA
Asthma
HA
COPD HA
Respiratory
HA
Respiratory
mortality
Respiratory
mortality
Respiratory
mortality
Respiratory
HA
Asthma ED
Asthma ED
Location

Atlanta
36 U.S. cities
14 U.S. cities
90 U.S. cities

New York City
Atlanta
Atlanta
Atlanta
New York City
New York City
36 U.S. cities
14 U.S. cities
48 U.S. cities
90 U.S. cities
90 U.S. cities
New York State
(11 regions)
Seattle, WA
Seattle, WA
Age

Children
65+
65+
All

All
All
All
Children
6 to 18
All
65+
65+
All
All
75+

18+
<18
Lag

0-2
0-1
0-1
DLO-
2

0-1
1
0-2
0-2
0-1
0-1
0-1
0-1
0-3
DLO-
2
DLO-
2

2
0
Avg
Time

8-h max
8-h max
1-hmax
1-hmax

8-h max
8-h max
8-h max
8-h max
8-h max
8-h max
8-h max
1-hmax
8-h max
1-hmax
1-hmax

8-h max
8-h max
% Increase
(95% CI)

6.38(3.19,9.57)
0.24 (-0.78,
1.21)
2.38 (0.00, 4.89)
2.54 (-3.32,
8.79)

16.9(10.9,23.4)
2.08(1.25,2.91)
3.90 (2.70, 5.20)
8.43 (4.42, 12.7)
28.2(15.3,41.5)
12.5 (8.27, 16.7)
1.63 (0.48, 2.85)
2. 14 (-0.63,
4.97)
2.51(1.14,3.89)
4.40 (-2. 10,
11.3)
4.07 (-4.23,
13.0)

19.1(3.00,40.5)
33.1(3.00,68.5)
Design
Values
(ppb)8

107
94
92
89

112
107
107
107
104
104
94
92
92
89
89
83
71
71
     8For each study city and each study year, 4th highest 8-hour daily maximum O3 concentrations were identified.  To
     provide some perspective on whether these cities would likely have met or violated the current O3 NAAQS during
     the study period, these concentrations were averaged over the years of the study, resulting in the study average
     design value. For multi-city studies, the median of the distribution of average design values was identified. Design
     values for each of the cities included in these multi-city studies are included in Wells et al. (2012).
                                                        4-18

-------
 1          While we recognize that there is no single air quality concentration that unambiguously
 2    identifies the portion of the concentration-response function in which it is appropriate to place
 3    the greatest confidence, we note that confidence intervals are smallest around 40 ppb (averaged
 4    across monitors), become somewhat wider at 63 concentrations below about 30 ppb (31  ppb was
 5    the 25th percentile concentration reported by Silverman and Ito) and above about 50 ppb (53 ppb
 6    was the 75th percentile reported in the study), and become dramatically wider below about 20
 7    ppb (24 ppb was the 10th percentile reported in the study) and above about 70 ppb (68 ppb was
 8    the 90th percentile reported in the study). In light of this, we reach the preliminary conclusion
 9    that confidence is greatest that a linear concentration-response relationship exists for 63  and
10    asthma hospital admissions in New York City for average 8-hour daily maximum Os
11    concentrations around 40 ppb, with confidence in the nature of the concentration-response
12    relationship decreasing notably for concentrations lower than approximately 20 ppb.
13          As noted above, the 8-hour 63 concentrations reported in the analysis by Silverman and
14    Ito (2010) were based on averaging across multiple monitors in the study area. Because
15    attainment of the current standard is based on the annual 4th highest 8-hour daily maximum Oj,
16    concentration measured at the monitor recording the highest such concentration, the average 8-
17    hour concentrations reported by Silverman and Ito (2010) are not directly comparable to the level
18    of the current 63  standard.  Therefore, it is not clear from the data presented in the study whether
19    particular average 8-hour daily maximum O3 concentrations, based on averaging across
20    monitors, would reflect ambient Os concentrations likely to be above or below those allowed by
21    the current 63 standard. To gain insight into the relationship between average 8-hour daily
22    maximum O3 concentrations (i.e.,  based on averaging across monitors) and the highest 8-hour
23    daily maximum concentrations (i.e., from the individual monitor recording the highest such
24    concentration), we have considered available air quality information from EPA's AQS for the
25    monitors used in the study by Silverman and Ito (2010) over the time period of the study (Wells
26    et al., 2012, section 1.3.3; Figure 4-2, below).
27          In considering the range of average Os concentrations for which we have the greatest
28    confidence in the concentration-response relationship, we first note that for the New York City
29    study area over the period of the study, 742 days had average (i.e., averaged across monitors) 8-
30    hour daily maximum Os concentrations around 40 ppb (i.e., 30 to 50 ppb) (Wells et al., 2012).
31    On these days with average 8-hour daily maximum 63 concentrations from 30 to 50 ppb, the
32    highest 8-hour daily maximum Os concentrations (i.e., from the monitors recording the highest
33    such concentrations) ranged from  approximately 55 ppb (i.e., for average concentrations from 30
34    to 35 ppb) to just below 75 ppb (i.e., 72 ppb for average concentrations from 45 to 50 ppb)
35    (Figure 4-2). In addition, we  note that 996 days had average 8-hour daily maximum Os

                                                   4-19

-------
 1
 2
 3
 4
 5
 6
 7
 8
 9
10
11
12
13
14
     concentrations from 20 to 50 ppb (Wells et al., 2012).  The highest 8-hour daily maximum Os
     concentrations on these days ranged from approximately 50 ppb (i.e., 49 ppb for average
     concentrations from 20 to 25 ppb) to just below 75 ppb (Figure 4-2).
            In considering this analysis within the context of the current standard we note that, of the
     days with average 8-hour daily maximum 63 concentrations from approximately 20 to 50 ppb
     (i.e., corresponding to the majority of the range over which we have reasonable confidence in the
     linearity of the concentration-response relationship), virtually none had highest 8-hour
     concentrations (i.e., from the highest monitor) greater than 75 ppb.  In light of this, we note that
     the linear relationship between ambient Os concentrations and asthma hospital admissions in
     New York City, as reported by Silverman and Ito (2010), persisted for 63 air quality
     concentrations likely to have been well below those allowed by the current O3 standard.
Figure 4-2. Average Versus Highest 8-hour Daily Maximum
City, 1999-2006
                                                                Concentrations in New York
                                                                              25+
                    Irttercept = 6.42
                    Slope = 1,08
                    R-square = 0.96
                          20
15
                              40        60        80
                               Average Concentration (ppb)
                                                                 100
120
                                                   4-20

-------
 1    Long-term exposure studies
 2
 3
 4
 5
 6
 7
 9
10
11
12
       In considering the available epidemiologic evidence for associations with long-term 63
concentrations within the context of considering the adequacy of the current O3 standard, as with
the short-term studies discussed above, we take into account Os design values in the locations
where U.S. studies have reported associations with respiratory effects, including new onset
asthma, asthma hospital admissions, and respiratory mortality.  These design values are
presented in Table 4-2,  below. The approach to identifying design values is discussed in Wells
etal. (2012).
Table 4-2. Ozone Design Values in U.S. Cities where Epidemiologic Studies have Evaluated
Associations between Long-Term Os and Respiratory Effects
Study
Moore et al. (2008)
Islam et al. (2008)
Islam et al. (2009)
Salam et a. (2009)
Meng etal. (2010)
Jerrett et al. (2009)
Zanobetti and
Schwartz (20 11)
Lin et al. (2008)
Location
Southern
California
Communities
12 California
Communities
12 California
Communities
12 California
Communities
San Joaquin
Valley
96 U.S.
MSAs
105 U.S.
cities
Communities
across New
York State
Reported Results
Increases in quarterly average O3 concentrations were
associated with increases in hospitalization for asthma
Apparent genetic protective effect against developing new
onset asthma disappeared in high O3 communities
Apparent genetic susceptibility to developing new onset
asthma was accentuated in high O3 communities
Apparent genetic protective effect against developing new
onset asthma for children in high O3 communities
Positive and statistically significant association between
annual average O3 and asthma emergency department visits
or hospitalizations
Positive and statistically significant association between
long-term O3 and respiratory mortality
Positive and statistically significant association between
summer-time average O3 and mortality in people with
COPD
Positive and statistically significant associations between
asthma hospital admissions long-term O3 concentrations
Design
Value
(ppb)9
178
123
123
123
10010
91
88
83
13
      9For each study city and each study year, 4th-highest 8-hour daily maximum O3 concentrations were identified. To
      provide some perspective on whether these cities would likely have met or violated the current O3 NAAQS during
      the study period, these concentrations were averaged over the years of the study, resulting in the study average
      design value. For multi-city studies, the median of the distribution of design values was identified.  Design values
      for each of the cities included in these multi-city studies are included in Wells et al. (2012).
      10The study period for the study by Meng et al. (2010) was November 1999 to September 2000. The design value
      reported in Table 4-2 reflects all 3-year periods that include 1999 and 2000.
                                                       4-21

-------
 1          With regard to the long-term studies in Table 4-2, we note that 50th percentiles of the
 2    distributions of Os design values across cities ranged from 83 to 178 ppb.  These median
 3    concentrations indicate that most of the cities included in available U.S. multi-city studies of
 4    long-term 63 would likely not have met the current 63 standard over the study periods. In
 5    considering the broader distributions of design values across study cities, we note that even in the
 6    study with the lowest 63 concentrations, the 25th percentile of the design values was above the
 7    level of the current O3 standard (i.e., 79 ppb for the study by Lin et al., 2008) (Wells et al., 2012).
 8    In light of this, we note that the large majority of the cities included in U.S. multi-city studies of
 9    long-term (Vrelated respiratory effects would likely not have met the current 63 standard during
10    the study periods. While U.S. epidemiologic studies provide strong support for the association
11    between long-term Os concentrations and clearly adverse respiratory effects, consideration of the
12    63 design values for the cities included in these studies does not provide insight into the
13    appropriateness of the degree of public health protection provided by the current Os standard.
14          Beyond looking at design values in  long-term exposure studies, we note that several U.S.
15    epidemiologic studies provide information  about annual or seasonal mean 8-hour average
16    concentrations used to differentiate low and high 63 communities or that characterized the range
17    of Os concentrations over which positive concentration-response relationships were observed.
18    As discussed above in chapter 2, section 2.3, key new studies include longitudinal studies of
19    serious effects  such as new-onset asthma and first asthma hospital admission, as well as cross-
20    sectional studies of current asthma and respiratory school absences.
21          For example, recent longitudinal studies from the CHS provide evidence for gene-
22    environment interactions in effects on new-onset asthma by indicating that the lower risks
23    associated with specific genetic variants are found in children who live in lower Os communities.
24    In two studies (Islam et al., 2008; Islam et al., 2009), low ozone communities were defined by an
25    annual mean 8-hour average 63 concentrations of 38.4 ppb. In another study (Salam et al.,
26    2009), the median annual mean 8-hour average concentration used differentiate high and low O3
27    communities was 50 ppb. A cross-sectional study (Wenton et al., 2009) from the CHS provides
28    evidence of gene-environment interactions  in effects on respiratory school absences indicating
29    that the lower risks associated with specific genetic variants are found in children who live in
30    lower O?, communities, in which a median annual mean 8-hour average Os concentration of 46.9
31    ppb was used to differentiate high and low  63 communities.
32          Evidence associating long-term 63 exposure to first asthma hospital admission in a
33    concentration-response relationship is provided in a retrospective cohort study (Lin et al., 2008b)
34    that followed a birth cohort of more than 1.2 million babies born in New York State to first
35    asthma admission. A positive concentration-response relationship between exposure to 63 and

                                                   4-22

-------
 1    childhood asthma hospital admissions was observed across the annual mean 8-hour 63
 2    concentrations which ranged from 37.51 to 47.78 ppb. In a cross-sectional analysis, Akinbami et
 3    al. (2010) examined the association between chronic exposure to Os and asthma outcomes in a
 4    national sample of children ages 3-17 years living in U.S. metropolitan areas.  A positive
 5    association for both currently having asthma and for having at least 1 asthma attack in the
 6    previous year was observed, with a median value for 12-month average 63 levels of 39.5 ppb and
 7    an IQR of 35.9-43.7 ppb, with a positive concentration-response relationship apparent from the
 8    lowest quartile to the highest.
 9           Such air quality information can help inform consideration of the adequacy of the current
10    primary 63 standard, or lower 8-hour average levels that may be considered, in providing
11    protection to children against the effects of long-term O3 exposures. We plan to conduct
12    additional air quality analyses in the second draft PA, relating annual or seasonal 8-hour average
13    63 concentrations in areas across the country with 4th-highest 8-hour daily maximum 63
14    concentrations, to further inform such considerations.
15           4.2.1.3 Total mortality
16           In addition to the morbidity studies discussed above, we consider studies linking short-
17    term Oj, concentrations to all-cause total mortality.11 As is the case with most of the U.S.
18    epidemiologic studies of respiratory-related hospital admissions, emergency department visits,
19    and mortality (see above), U.S. epidemiologic studies of total mortality have not been conducted
20    in locations likely to have met the current 63 standard during the study periods (Table 4-3).  The
21    medians of the distributions of average O3 design values ranged from 88 to 96 ppb.
22
      1 lrThe evidence linking long-term O3 concentrations with mortality was judged in the ISA to be suggestive of a
      causal relationship (US EPA, 2012a, section 2.5.2). Therefore, unlike for morbidity, our consideration of mortality
      focuses on studies of short-term O3 concentrations.
                                                    4-23

-------
 1    Table 4-3. Ozone Design Values in U.S. Cities where Epidemiologic Studies have Evaluated
 2    Associations between Short-Term Os and Non-Accidental Mortality
Study
All-year
Schwartz (2005)
Bell et al. (2007)
Bell and Dominici
(2008)
Katsouyanni et al.
(2009)
Bell et al. (2004)
Warm Season
Schwartz (2005)
Zanobetti and
Schwartz (2008)
Zanobetti and
Schwartz (2008)
Medina-Ramon and
Schwartz (2008)
Franklin and
Schwartz (2008)
Katsouyanni et al.
(2009)
Bell et al. (2004)
Location

14 U.S. cities
98 U.S.
communities
98 U.S. communities
APHENA-U.S. (90
cities)
95 U.S. communities

14 U.S. cities
48 U.S. cities
48 U.S. cities
48 U.S. cities
18 U.S. communities
APHENA-U.S.
95 U.S. communities
Lag

0
0-1
0-6
DL (0-2)
0-6

0
0
0-3
0-2
0
DL (0-2)
0-6
Averaging
Time

1-h max
24-h
24-h avg
1-h max
24-h avg

1-h max
8-h max
8-h max
8-h max
24-h avg
1-h max
24-h avg
% Increase
(95% CI)

0.76(0.13,1.40)
0.64 (0.34, 0.92)
1.04(0.56,1.55)
3.02(1.10,4.89)
1.04(0.54,1.55)

1.00(0.30,1.80)
1.51(1.14,1.87)
1.60(0.84,2.33)
1.96(1.14,2.82)
1.79(0.90,2.68)
3.83(1.90,5.79)
0.78(0.26,1.30)
Median
Design
Value (ppb)

96
89
89
89
88

96
92
92
92
91
89
88
 5          However, several epidemiologic studies have evaluated the nature of the concentration-
 6    response relationship with mortality and the possibility that a threshold concentration exists,
 7    below which Os is no longer associated with mortality (US EPA, 2012a, section 6.6.2.4).  In one
 8    of the largest U.S. studies to address this issue, Bell et al. (2006) used different statistical
 9    approaches to evaluate the potential for a threshold in the Os-mortality relationship in the 98
10    cities in  the NMMAPS dataset.  As discussed above (section 2.2.2), the authors reported positive
11    and statistically significant associations with mortality in a variety of restricted analyses,
12    including analyses restricted to days with 24-hour average 63 concentrations (i.e., averaged
13    across monitors in cities with multiple monitors) below 60, 55, 50, 45, 40, 35, and 30 ppb. In
14    these restricted analyses Os  effect estimates were of similar magnitude and had similar statistical
15    precision.  In the analysis restricted to days with 24-hour average  63 concentrations below 25
16    ppb, the Os effect estimate was similar in magnitude to the effect estimates resulting from
17    analyses with the higher cutoffs, but had somewhat lower statistical precision, with the estimate
18    approaching statistical significance (i.e., based on observation of Figure 2 in Bell et al., 2006).
                                                    4-24

-------
 1    In analyses restricted to days with lower 24-hour average 63 concentrations (i.e., below 20 and
 2    15 ppb), effect estimates were similar in magnitude to analyses with higher cutoffs, but with
 3    notably less statistical precision (i.e., confidence intervals included no (^-associated mortality)
 4    (i.e., based on observation of Figure 2 in Bell et al., 2006).  Ozone was no longer positively
 5    associated with mortality when the analysis was restricted to days with 24-hour Os
 6    concentrations below 10 ppb.
 7           In considering the results of Bell et al. (2006) within the context of the adequacy of the
 8    current 63 standard we note that, as with the study by Silverman and Ito (2010) discussed above,
 9    it is not appropriate to compare Os concentrations based on averaging across monitors (i.e.,
10    which in the case of the study by Bell reflect 24-hour concentrations averaged across monitors
11    for cities with multiple monitors) directly to the level of the current standard (for which
12    attainment is based on the 4th highest 8-hour daily maximum concentration from the  single
13    monitor recording the highest such concentration).  Therefore, we have used EPA's AQS to
14    relate average 24-hour Os concentrations in the 98 study cities (i.e., averaged across monitors in
15    cities with multiple monitors) to the highest 8-hour daily maximum concentrations (i.e., from the
16    individual monitors in each city recording the highest such concentrations) (Wells et al., 2012).12
17           Specifically, we first used AQS to identify average 24-hour 63 concentrations in each of
18    the 98 study cities for each day during the study period. We then used the cutoff values
19    identified by Bell to restrict the distribution of average 24-hour concentrations. Specifically, for
20    each city we restricted the distribution of 63 air quality to days with average 24-hour 63
21    concentrations at or below 60, 55, 50, 45, 40, 35, 30, 25, 20, 15,  10, and 5 ppb (i.e., the cutoffs
22    evaluated by Bell). From these restricted air quality distributions, we then identified the highest
23    8-hour 63 concentrations (i.e., from a single monitor) in each city for each year.  These highest
24    8-hour daily maximum Os concentrations were then used to provide insight into whether the
25    restricted air quality distributions used by Bell et al. (2006) would have reflected 63
26    concentrations likely to have been allowed by the current O3 standard.
27           In considering these restricted analyses, we first focus on the lowest cutoff value that
28    resulted in a positive and statistically significant association with mortality in Bell et al. (2006)
29    (i.e., 30 ppb).  In considering only days  with average 24-hour 63 concentrations at or below 30
30    ppb, we note that the 50th percentile of the distribution of highest 8-hour daily maximum O3
31    concentrations13 was approximately 70 ppb (25th percentile was approximately 64 ppb; 75th
      12As noted in Wells et al. (2012), the MSA was used as a surrogate for the study area boundaries. We recognize that
      this does not exactly match the study area definitions used by Bell et al. (2006). Further refinement of this approach
      may be explored in the 2nd draft Policy Assessment.
      13Percentiles were identified from the distribution of the study period averages of highest 8-hour daily maximum O3
      concentrations within each city.
                                                     4-25

-------
 1    percentile was approximately 78 ppb).  In light of this, we note that ambient 63 concentrations
 2    across most of the cities evaluated by Bell et al. (2006) would likely have been allowed by the
 3    current Os standard for the restricted analysis with a 30 ppb cutoff (Wells et al., 2012).
 4           In further considering the restricted analyses by Bell et al. (2006), we also consider lower
 5    cutoff values.  Specifically, we consider the analysis with a 24-hour average cutoff of 25 ppb,
 6    which resulted in a positive association with mortality that approached statistical significance
 7    (i.e., based on confidence intervals in figure 2 of Bell et al., 2006), and the analysis with a 24-
 8    hour average cutoff of 15 ppb, which resulted in a positive association with mortality but with
 9    notably wider confidence intervals (The 15 ppb cutoff was the lowest cutoff value that resulted
10    in a positive association with mortality).14
11           In considering days with average 24-hour Os concentrations at or below 25 ppb, we note
12    that the 50th percentile of the distribution of highest 8-hour daily maximum 63 concentrations
13    (i.e., the distribution across cities) was approximately 60 ppb (25th percentile was approximately
14    55 ppb; 75th percentile was approximately 67 ppb). In addition, cities had fewer than one day per
15    year, on average, with highest 8-hour daily maximum O3 concentrations greater than 75 ppb (i.e.,
16    average was 0.4 days per year).  In light of this, we note that ambient Os concentrations across
17    most of the cities evaluated by Bell et al. (2006) would likely have been well below those
18    allowed by the current Os standard for the restricted analysis with a 25 ppb  cutoff (Wells et al.,
19    2012).
20           In considering the days with average 24-hour Os concentrations at or below 15 ppb, we
21    note that the 50th percentile of the distribution of highest 8-hour daily maximum Os
22    concentrations (i.e., the distribution across cities) was approximately 40 ppb (25th percentile was
23    approximately 37 ppb; 75th percentile was approximately 46 ppb).  In addition, all cities had far
24    fewer than one day per year, on average, with highest 8-hour daily maximum 63  concentrations
25    greater than 75 ppb  (i.e., average across cities was 0.008 days per year). In light  of this, we note
26    that ambient Os concentrations across all of the cities evaluated by Bell et al. (2006) would likely
27    have been well-below those  allowed by the current 63 standard  for the restricted  analysis with a
28    15 ppb cutoff (Wells et al., 2012).
29           In considering the restricted analyses by Bell et al.  (2006) within the context of the
30    adequacy of the current Os standard, we note that Bell reported  positive and statistically
31    significant associations between 63 and mortality in analyses restricted to 63 concentrations that
32    would likely have been allowed by the current Os standard in most of the cities evaluated.  In
      14As discussed above, the decreasing precision in effect estimates as concentrations approach the lower extreme of
      the air quality distribution, as indicated by notable widening of confidence bounds, is intrinsically related to data
      density and is not necessarily indicative of an absence of O3-associated health effects.
                                                     4-26

-------
 1    addition, positive, though not statistically significant associations were reported in analyses
 2    restricted to lower Os concentrations, likely to have been well-below those allowed by the
 3    current standard across most or all cities.
 4           4.2.1.4 Other health endpoints
 5           In addition to our above consideration of the health effects judged in the ISA to be caused
 6    or likely caused by O3 exposures, we also note that evidence continues to emerge for a broader
 7    range of CVrelated health effects.  Specifically, we note the ISA conclusions that the evidence is
 8    "suggestive of a causal relationship" between short-term 63 exposures and cardiovascular effects
 9    and central nervous system effects and between long-term Os exposures and cardiovascular
10    effects, reproductive and developmental effects, central nervous system effects, and mortality
11    (US EPA, 201 Ib, Table 2-1). Although the bodies of evidence for these CVrelated effects have
12    generally expanded since the last review, important uncertainties remain regarding the
13    contribution of 63 to such effects and the extent to which such effects can result from exposures
14    to 63 concentrations that would be allowed by the current standard.  Therefore, while these
15    studies provide important information on the breadth of health effects potentially  attributable to
16    63 exposures, they provide little additional information to inform a judgment as to the adequacy
17    of the public health protection provided by the current O3 standard.
18           4.2.1.5 Integrated consideration of health evidence
19           There is much new evidence available in this review that supports and builds upon key
20    health-related conclusions drawn in previous reviews of the O?, standard, including important
21    new controlled human exposure studies and  a large number of new epidemiologic studies,
22    including new epidemiologic studies of effects associated with short-term as well as long-term
23    63 exposures. The staff believes that this body of scientific  evidence is very robust, recognizing
24    that it includes large numbers of various types of studies, including toxicological  studies,
25    controlled human exposure studies, and community epidemiological studies, that  provide
26    consistent and coherent evidence of an array of (Vrelated effects associated with short-term
27    exposures including respiratory morbidity effects and mortality effects, both respiratory and total
28    non-accidental mortality, as well as evidence associating long-term Os exposures  with
29    respiratory morbidity and mortality effects, and new onset asthma. Further, the available
30    evidence provides increased confidence that (1) the full range of (Vrelated respiratory effects,
31    including emergency department visits, hospital admissions,  and respiratory mortality, are
32    caused by short-term O3 exposures; and that (2) total non-accidental  mortality is likely to be
33    caused by short-term exposures.
34           In particular, we note that recent controlled human exposure  and epidemiologic studies
35    reinforce the conclusion that the types of (Vrelated effects considered in previous reviews
                                                    4-27

-------
 1    extend to 63 concentrations that would be allowed by the current standard. In looking broadly at
 2    the available evidence from controlled human exposure, epidemiologic, and animal toxicological
 3    studies, we note that the controlled human exposure studies have reported a variety of health
 4    effects, including lung function decrements, respiratory symptoms and pulmonary inflammation
 5    in healthy subjects, following exposures to Os concentrations (i.e., 60-70 ppb) below the level of
 6    the  current 63 standard.  With respect to epidemiologic studies, a limited number have been
 7    conducted at ambient O3 concentrations below those allowed by the current standard,  or have
 8    examined subsets of data that include only days with ambient Os concentrations below the level
 9    of the current standard and continue to report associations with respiratory mortality.
10    Importantly,  there are also many epidemiologic studies done in areas that likely would not have
11    met the current standard but which nonetheless report statistically significant associations that
12    generally extend down to ambient 63 concentrations that are below the level of the current
13    standard; in most studies such effects have not shown evidence for a threshold across  the range
14    of 63 concentrations typically observed in the United States during the 63 season. These include
15    the most severe O3-related effects (i.e., emergency department visits, hospital admissions,
16    premature mortality) as well as a variety of other respiratory effects  (i.e., lung function
17    decrements, respiratory symptoms, and airway inflammation). The evidence from controlled
18    human exposure and epidemiologic studies is  supported by a very large body of toxicological
19    studies which provides additional biological plausibility and coherence to those results.  In this
20    section, we draw preliminary conclusions regarding the extent to which this evidence  supports or
21    calls into question the adequacy of the public health protection provided by the current Os
22    standard.
23          The clearest evidence that the health effects discussed above can plausibly result from
24    exposures to 63 concentrations below those allowed by the current 63 standard comes from a
25    number of new controlled human exposure studies reporting Os-induced lung function
26    decrements, respiratory symptoms, and airway inflammation following exposures of healthy
27    adults to 63 concentrations below 75 ppb. As discussed above, controlled human exposure
28    studies that evaluate only healthy, nonasthmatic subjects likely underestimate the effects of Os
29    exposure on  asthmatics and other susceptible populations. Therefore, relative to the healthy,
30    nonasthmatic subjects used in most controlled human exposure studies, a greater proportion of
31    people with asthma may be affected, and those who are affected may have larger responses
32    following exposures to 63 concentrations below 75 ppb.  This suggests that the lowest observed
33    effects levels demonstrated in controlled human exposure studies that use only healthy subjects
34    (i.e., 60 ppb) may not reflect the lowest levels at which people with asthma or other lung diseases
35    can respond. In support of this, as discussed above we note that epidemiologic studies have
                                                   4-28

-------
 1    reported (Vassociated lung function decrements in children and outdoor workers at 63
 2    concentrations below 60 ppb.
 3           A limited number of epidemiologic studies report that O3-related mortality, hospital
 4    admissions, and emergency department visits occur in areas with ambient Os concentrations
 5    below those allowed by the current 63 standard. This includes studies of short-term 63
 6    concentrations reporting Os-related asthma emergency department visits in children and adults in
 7    Seattle, a location that would likely have met the current Os standard during the study period
 8    (Mar and Koenig, 2009); a study reporting relatively high confidence in a linear concentration-
 9    response relationship between short-term Os concentrations and asthma hospital admissions in
10    New York City at ambient 63 concentrations likely to have been allowed by the current standard
11    (Silverman and Ito, 2010); and a multi-city study reporting positive and statistically significant
12    associations between short-term Os concentrations and mortality in analyses restricted to ambient
13    63 concentrations likely to have been allowed by the current standard (Bell et al., 2006).
14    Moreover, many short-term exposure epidemiologic studies done in areas that likely would not
15    have met the current standard report statistically significant associations that generally extend
16    down to ambient 63 concentrations that are below the level of the current standard. With respect
17    to epidemiologic evidence regarding the effects of long-term exposures to  Os, we note that while
18    U.S. epidemiologic studies provide strong support for the association between long-term 63
19    concentrations and clearly adverse respiratory effects, consideration of the O3 design values for
20    the cities included in these studies does not provide insight into the appropriateness of the degree
21    of public health protection provided by the current 63 standard.
22           In considering the overall body of evidence for Os-related respiratory effects and
23    mortality, including evidence for such effects following exposures to 63 concentrations below
24    those that would be allowed by the current Os standard, we first conclude that the available
25    evidence provides strong support for a standard at least as protective as the current 63  standard.
26    In addition, we reach the further preliminary conclusion that the  available evidence calls into
27    question the adequacy of the current standard and provides support for  considering potential
28    alternative standards to increase public health protection against  the effects related to short-term
29    Os exposures, especially for at-risk groups. This preliminary conclusion places considerable
30    weight on the array of Os-related respiratory effects that have been reported following exposures
31    to short-term 63 concentrations below the level of the current standard. In emphasizing such
32    effects, this preliminary conclusion also places considerable weight on  the plausibility of the
33    linkages between the body of evidence for respiratory effects and mortality following exposures
34    to O3 concentrations below those allowed by the current O3 standard and the broader body  of
35    experimental and epidemiologic evidence for Os-related respiratory  effects and mortality

                                                    4-29

-------
 1    reported following exposures to higher concentrations.  In reaching this preliminary conclusion,
 2    we acknowledge that uncertainties persist in the health evidence; however, in staff s view the
 3    broad array of health effects reported following short-term exposures to Os concentrations below
 4    those allowed by the current standard (i.e., respiratory effects and mortality), combined with the
 5    plausible linkages between these effects and the much larger body of epidemiologic and
 6    controlled human  exposure evidence at higher 63 concentrations, supports the appropriateness of
 7    revising the current O3 standard in order to increase public health protection, particularly for
 8    people with asthma, children and other at-risk groups.
 9    4.2.2  Exposure- and Risk-Based Considerations
10          In this section, we consider the following question:
11       •  To what extent do the Os  exposure and risk analyses presented in the first draft
12          RE A support or call into question the adequacy of the public health protection
13          provided by the current Os primary standard?
14          In addressing this question, we consider the first draft REA assessments of (Vrelated
15    exposures and risks associated with recent air quality and with air quality  adjusted to simulate
16    just meeting the current 63 standard. More specifically, we consider estimates from the REA of
17    the number of people in at-risk populations with O3 exposures above health benchmark levels
18    (section 4.2.2.1); estimates of the number of people in at-risk populations  with impaired lung
19    function resulting from exposures to 63 (section 4.2.2.2);15 and estimates of the potential
20    magnitude in the population of premature mortality and selected morbidity effects (section
21    4.2.2.3).  Section 4.2.2.4 provides an integrated consideration of the exposure  and risk
22    information within the context of the adequacy of the current 63 standard.
23          4.2.2.1 Exposures above health benchmark concentrations
24          As part of the last 63 NAAQS review, EPA conducted exposure analyses for the general
25    population, all school-age children (ages  5-18),  active school-age children, and asthmatic school-
26    age children (US EPA, 2007a, b).  Exposure estimates were generated for 12 urban areas for
27    recent years of air quality and for just meeting the existing 8-hr standard and several alternative
28    8-hr standards.
29                 The exposure analysis conducted in the first draft REA in the current review
30    builds upon the methodology and lessons learned from the exposure analyses conducted in
31    previous reviews,  as well as information provided in the third draft ISA (US EPA, 2012a). In the
32    second draft REA, EPA will be conducting exposure modeling for 16 urban areas located across
             15As noted in section 4.2.2.2, quantitative estimates of O3-induced lung function impairment are being
      developed in parallel with this first draft Policy Assessment. Therefore, these estimates will be considered in the
      second draft Policy Assessment.
                                                    4-30

-------
 1    the U.S. (US EPA, 2012b). In the first draft REA, results are presented for four of these areas,
 2    Atlanta, Denver, Los Angeles, and Philadelphia.
 3                 Population exposures to ambient O3 levels were modeled using the Air Pollutants
 4    Exposure (APEX) model (U.S. EPA, 2012b,c).  Exposure estimates were developed for Os
 5    concentrations in recent years, based on 2006 to 2010 ambient air quality measurements, and for
 6    Os concentrations adjusted to simulate just meeting the current 8-hr Os standard.  Exposures
 7    were modeled for 1) the general population, 2) school-age children (ages 5-18), and 3) asthmatic
 8    school-age children. The strong emphasis on children reflects the finding of the last Os NAAQS
 9    review (EPA, 2007a) and the ISA (EPA, 2012a, Chapter 8) that children are an important at-risk
10    group.
11                 Benchmark levels used in the first draft REA were 60, 70, and 80 ppb, 8-hour
12    average. These levels are based on the evidence from controlled human exposures studies that
13    were conducted using 6.6 hour exposures of healthy adults while at intermittent, moderate
14    exertion.  Although the analysis of these exposures was conducted using three discrete
15    benchmark levels, the concept is more appropriately viewed as a continuum, with greater
16    confidence and less uncertainty about the existence of health effects at the upper end and less
17    confidence and greater uncertainty as one considers increasingly lower 63 exposure levels.
18    Estimating exposures to ambient Os concentrations at and above these benchmark levels, while
19    at moderate or greater exertion, was intended to provide some perspective on the  public health
20    impacts of (Vrelated health effects that have been  demonstrated in controlled human exposure
21    studies, but that cannot currently be evaluated in quantitative risk assessments, such as lung
22    inflammation, increased airway responsiveness, and decreased resistance to infection. As
23    discussed in detail in the ISA (US EPA, 2012a,  section 6.2), and above (sections 2.2.1 and 4.2.1),
24    the 80 ppb benchmark represents an exposure concentration at which there is a substantial
25    amount of human clinical evidence demonstrating a range of (Vrelated effects in healthy
26    exercising adults, including lung function decrements, respiratory symptoms, airway
27    inflammation, impaired lung host defense, and airway hyperresponsiveness. The 70 ppb
28    benchmark reflects an 63 exposure concentration that has been reported to result in lung function
29    decrements and respiratory symptoms in healthy exercising adults. As discussed  above (sections
30    2.2.1 and 4.2.1), the 60 ppb benchmark reflects  consideration of the lowest exposure
31    concentration for which CVinduced respiratory effects have been reported in controlled human
32    exposure studies of healthy exercising adults. Effects at both of these levels reflect the broader
33    range of effects reported following exposures to 80 ppb 63 that have been reported in healthy
34    adults.  And the effects reported in healthy adults at 60 and 70 ppb O3 could result in
35    underestimating the health impacts of Os exposures, as discussed above (section 2.4)

                                                  4-31

-------
 1          As discussed above in section 3, the first draft REA (US EPA, 2012b, section 5)
 2    estimated the number and percentage of school-age children who experience at least one 8-hour
 3    average Os exposure above each benchmark concentration while at the same time engaged in
 4    activities resulting in moderate or greater exertion.  Across the years included in the analysis
 5    (2006-2010), the pattern of exposures differed. For example, in the worst Os year (2006), the
 6    percentage of children exposed to 8-hour 63 concentrations at or above 60 ppb while at moderate
 7    or greater exertion ranged from 30 to 37% across the 4 study areas; the percentage exposed to 8-
 8    hour concentrations at or above 70 ppb ranged from 10 to 21%; and the percentage exposed to 8-
 9    hour concentrations at or above 80 ppb ranged from 1 to 10%. In the year with the lowest 63
10    concentrations (2009), the percentage of children exposed to 8-hour Os concentrations at or
11    above 60 ppb while at moderate or greater exertion ranged from 9 to 32%; the percentage
12    exposed to 8-hour concentrations at or above 70 ppb ranged from 1 to 15%; and the percentage
13    exposed to 8-hour concentrations at or above 80 ppb ranged from 0 to 5%.
14          In specifically considering these exposure results within the context of the adequacy of
15    the current standard, we consider estimated exposures associated with recent unadjusted air
16    quality and with air quality adjusted to  simulate just meeting the current 63 standard. With
17    unadjusted air quality, the average (i.e., average across years 2006 to 2010) percentages of
18    school age children estimated to experience one or more exposures per year to 8-hour 63
19    concentrations above 60, 70, or 80 ppb, while at moderate or greater exertion, were as follows
20    (US EPA, 2012b, Figures 5-1 to 5-15, Tables 4-5 to 4-6, and section 9.1):
21       •  For Denver, approximately 20% above 60 ppb (corresponding to  109,000 children), 4%
22          above 70 ppb (corresponding to approximately 22,000 children), and 0.4% above 80 ppb
23          (corresponding to approximately 2,200 children)
24
25       •  For Atlanta, approximately 22% above 60 ppb (corresponding to  189,000 children), 9%
26          above 70 ppb (corresponding to approximately 75,000 children), and 3% above 80 ppb
27          (corresponding to approximately 24,000 children)
28
29       •  For Philadelphia, approximately 26% above 60 ppb (corresponding to 297,000 children),
30          10% above 70 ppb (corresponding to 117,000 children), and 2% above 80 ppb
31          (corresponding to approximately 28,000 children)
32
33       •  For Los Angeles, approximately 32% above 60 ppb (corresponding to 1,150,000
34          children),  15% above 70 ppb (corresponding to 559,000 children), and 6% above 80 ppb
35          (corresponding to approximately 218,000 children)
36
37          When air quality was adjusted to simulate just meeting the current 63 standard, these
38    estimated benchmark exceedances decreased.  Specifically, for adjusted air quality, the average
                                                  4-32

-------
 1    percentages of school age children estimated to experience one or more exposures per year to 8-
 2    hour Os concentrations above 60, 70, or 80 ppb, while at moderate or greater exertion, were as
 3    follows16 (US EPA, 2012b, Figures 5-1 to 5-15, Tables 4-5 to 4-6):
 4       •  For Denver, approximately 9-11% above 60 ppb (corresponding to approximately
 5          51,000-61,000 children), 0.6-0.7% above 70 ppb (corresponding to approximately 3,300-
 6          4,200 children), and 0.0% above 80 ppb
 7
 8       •  For Atlanta, approximately 8-10% above 60 ppb (corresponding to approximately
 9          72,000-83,000 children), 1% above 70 ppb (corresponding to approximately 9,600-
10          11,200 children), and 0.1% above 80 ppb (corresponding to approximately 700-900
11          children)
12
13       •  For Philadelphia, approximately 9-13% above 60 ppb (corresponding to approximately
14          105,000-145,000 children), 1-2% above 70 ppb (corresponding to approximately 10,000-
15          26,000 children), and 0.0-0.1% above 80 ppb (corresponding to approximately 500-1,600
16          children)
17
18       •  For Los Angeles, approximately 4% above 60 ppb (corresponding to approximately
19          126,000-148,000 children), 0.2-0.3% above 70 ppb (corresponding to approximately
20          7,500-11,300 children), and 0.0% above 80 ppb
21
22          When considering exposures above benchmark concentrations in asthmatic children at
23    moderate or greater exertion, the results were similar in term of numbers, with decreased
24    exposures estimated for air quality adjusted to simulate just meeting the current Os standard (US
25    EPA, 2012b, Tables 5-11 and 5-12). For example, when air quality was adjusted to simulate just
26    meeting the current Os standard, the numbers of asthmatic children estimated to be exposed one
27    or more times per year to 8-hour Os concentrations at or above 60 ppb, while at moderate
28    exertion, ranged from approximately 5,000 to 17,000 across these four cities; the numbers of
29    asthmatic children estimated to be exposed to Os concentrations at or above 70 ppb ranged from
30    approximately 300 to 3,000 across cities; and the numbers of asthmatic children  estimated to be
31    exposed to O3 concentrations at or above 80 ppb ranged from 0 to approximately 200 across
32    cities (US EPA, 2012b, Tables 5-11 and 5-12).
33          4.2.2.2 Estimates of Os-induced lung function impairment
34                 As noted in the first draft of the RE A (US EPA, 2012b), quantitative estimates of
35    lung function impairment are under development and will be released with this first draft PA.
36    Therefore, estimates of lung function impairment will be considered in the second draft PA.
            16The ranges presented represent the range of estimates across each of the 3 -year periods for which air
      quality was adjusted to simulate just meeting the current O3 NAAQS (i.e., 2006-2008, 2008-2010).
                                                  4-33

-------
 1          4.2.2.3 Estimates of O3-associated mortality and morbidity
 2          The first draft REA evaluated risks of mortality and morbidity from short-term exposures
 3    to Os based on application of concentration-response functions derived from epidemiology
 4    studies.  The analysis included both a set of urban area case studies and a national scale
 5    assessment.  The urban case study analyses evaluated mortality and morbidity risks, including
 6    emergency department visits, hospitalizations, and respiratory symptoms associated with recent
 7    63 concentrations (2006-2010) and with 63 concentrations adjusted to simulate just meeting the
 8    current Os standard. Mortality and hospital admissions (HA) were evaluated in 12 urban areas,
 9    while emergency department visits and respiratory symptoms were evaluated in a subset of areas.
10    These 12 urban areas were: Atlanta, GA; Baltimore, MD; Boston, MA; Cleveland, OH; Denver,
11    CO; Detroit,  MI; Houston, TX; Los Angeles, CA; New York, NY; Philadelphia, PA;
12    Sacramento,  CA; and St. Louis, MO.  The urban case study analyses focus on risk estimates for
13    the middle year of each three-year attainment simulation period (2006-2008 and 2008-2010) in
14    order to provide estimates of risk for a year with generally higher Os concentrations (2007) and a
15    year with generally lower Os concentrations (2009).
16          As discussed above (section 1.3.4), in past reviews of the Os NAAQS, Os-related health
17    risks have been estimated using non-threshold concentration-response functions. The use of
18    non-threshold functions was consistent with the lack of a discernible threshold in the
19    concentration-response relationship. Further, in past reviews health risks were estimated for Os
20    concentrations above estimates of policy-relevant background concentrations (referred to as
21    North American background in the third draft ISA and in Chapter 1, section 1.3.4).  The
22    estimation of risks above background concentrations, rather than estimation of total risks, was
23    employed in  past reviews to focus policy decisions on the portion of risk attributable to ambient
24    Os that can be controlled either through regulation of domestic sources or through international
25    agreements within North America.
26          Consistent with the approach in the last review, the first draft REA employed continuous
27    non-threshold concentration-response functions relating ozone exposures to health effect
28    incidence. The use of non-threshold functions reflects the conclusion reached in the ISA that the
29    available evidence supports a linear, no-threshold concentration-response relationship across the
30    range of daily Os concentrations commonly observed in the United States during the Os  season
31    (US EPA, 2012a, section 2.5.4.4). However, there are also some key differences in the approach
32    taken in the first draft REA, compared to the approach taken in the previous review.
33          For example, in contrast to the approach used in the last review, the first draft REA has
34    estimated total risks attributable to Os exposure, not risks in excess of background
35    concentrations. In taking this approach, the REA noted the advice of CAS AC members, who
                                                   4-34

-------
 1    recommended in the last review that EPA move away from using background in calculating
 2    risks (Henderson, 2007).  This approach recognizes that health risks result from Os exposures,
 3    regardless of the source of the 03.
 4           In estimating total Os-related health risks, the REA concluded that the approach most
 5    consistent with the statistical models reported in the epidemiological studies is to apply the
 6    concentration-response functions to all ozone concentrations down to zero. However, consistent
 7    with the conclusions of the ISA that the available evidence indicates less certainty in the shape of
 8    the concentration-response curve at the lower end of the distribution of ambient 63
 9    concentrations, the REA also recognized that confidence in the nature of the concentration-
10    response function and the magnitude of the risks associated with very low concentrations of
11    ozone is reduced because there are few ozone measurements at the lowest levels in many of the
12    urban areas included in the studies. Specifically, the REA noted that estimates of risk associated
13    with Os concentrations below the lowest measured level (LML) for the underlying
14    epidemiological study would be associated with reduced confidence  since these estimates
15    involve applying the concentration-response  function outside of the range of data used in its
16    derivation. In light of this, the REA has characterized mortality risks in excess of lowest
17    measured Oj concentrations17 as well as total risks associated with Oj concentrations down to
18    zero (US EPA, 2012b, sections 7.3.3 and 8.1.1.4).18  In considering these different approaches,
19    the REA concluded that the two sets of estimates provide a reasonable bound on estimated total
20    risks, reflecting uncertainties about the concentration-response functions below the lowest ozone
21    concentrations evaluated  in the studies.
22           In the remainder of this section, we consider the first draft REA estimates of Os-related
23    health risks within  the context of considering the adequacy of the current standard. Specifically,
24    we consider risk estimates for all-cause mortality and respiratory morbidity,  which includes
25    respiratory-related  hospital admissions, emergency department visits, and symptoms.
26           All-cause mortality
27           As noted above, the REA estimated Os-associated mortality in a national assessment and
28    in 12 specific urban areas for 2006-2008 and 2008-2010 air quality and, in the case of the 12
29    urban areas, for air quality adjusted to simulate just meeting the current standard.  Mortality
30    estimates were based on concentration-response relationships from two studies (Zanobetti and
       As discussed above (section 3.2.1), due to data limitations, the REA did not identify the actual LMLs from the
      epidemiologic studies used in the risk assessment. Rather, as a surrogate for the study-based LMLs, the REA used
      the lowest O3 concentrations from the composite monitor O3 distributions used to model health risks (US EPA,
      2012b, section 7.1.1).
      18For morbidity endpoints, risks were estimated down to the LML but not for total O3 concentrations down to zero
      (US EPA, 2012b, chapter 7).
                                                    4-35

-------
 1    Schwartz, 2008; Bell et al., 2004).  Estimates of mortality attributable to short-term Os exposures
 2    under recent conditions varied widely across urban study areas, reflecting differences in ambient
 3    Os concentrations and populations, as well as differences in city-specific effect estimates. The
 4    estimates based on Zanobetti and Schwartz (2008) showed the largest Os-associated mortality
 5    risks in Boston, Detroit, Los Angeles, and New York, while the estimates based on Bell et al
 6    (2004) showed the largest risks in Atlanta, Boston, Houston, Los Angeles, and New York (US
 7    EPA, 2012b, section 9.3).
 8          In the national analysis of Os-attributable mortality for the years 2006 to 2008, the REA
 9    estimated 18,000 Os-attributable deaths based on O3 effect estimates from Bell et al. (2004)
10    (corresponding to approximately 1.9% of total  mortality) and 15,000 Os-attributable deaths
11    based on Os effect estimates from Zanobetti and Schwartz (2008) (corresponding to
12    approximately 2.5% of total mortality) (US EPA, 2012b, chapter 8, Tables 1.2 and 1.3).  Of these
13    Os-attributable deaths, the first draft REA estimated that 85-90% occur in locations where the
14    seasonal average 8-hr daily maximum or 8-hr daily mean (10am-6pm) Os concentration is
15    greater than 40 ppb, corresponding to 4th high  8-hour daily maximum Os concentrations ranging
16    from approximately 50 ppb to 100 ppb.  In considering the potential implications of estimates of
17    Os-attributable mortality for evaluation of the adequacy of the current O3 standard, we consider
18    the analyses in the 12 urban study cities, for which the first draft REA analyzed both unadjusted
19    Os air quality and Os air quality adjusted to simulate just meeting the current standard.
20          Across the 12 study cities, and using estimates based on both Bell et al. (2004) and
21    Zanobetti and Schwartz (2008), Os-attributable mortality risk estimates for unadjusted air quality
22    ranged from approximately 20 to 930 deaths, accounting for approximately 0.5 to 4.9% of total
23    baseline all-cause mortality (for 2007 air quality). When risk estimates were focused on  Os
24    concentrations  above the LML, risk estimates were somewhat smaller (i.e., approximately 10 to
25    730 deaths across the different cities). For 2009 unadjusted air quality, the Os-attributable
26    mortality risk estimates ranged from 20 to approximately 980 deaths across the 12 cities,
27    accounting for  approximately 0.6 to 4.3% of total baseline all-cause mortality.  When risk
28    estimates were focused on Os concentrations above the LML, these risk estimates ranged from
29    approximately  10 to 780 deaths across cities, accounting for approximately 0.4 to  3.0% of total
30    baseline all-cause mortality.  Mortality estimates based on O3 effect estimates from Bell et al.
31    (2004) were generally larger than estimates based on effect estimates from Zanobetti and
                                                   4-36

-------
 1    Schwartz (2008), likely due to the larger effect estimates reported by Bell and to the longer 63

 2    season modeled for the study by Bell (US EPA, 2012b, Table 7-4).

 3          Upon simulating just meeting the current standard, estimates of (Vattributable mortality

 4    decreased across the 12 study cities, as noted in Table 4-1 of the first draft REA (US EPA,

 5    2012b, section 7.5).  Specifically, using effect estimates from both epidemiologic studies, that

 6    the estimated reductions in CVattributable mortality upon simulation of just meeting the current

 7    O3 standard were approximately 10 to 50% for 2007 air quality and approximately 0.1 to 35%

 8    for 2009 air quality.  Although this suggests the potential for important risk reductions as

 9    precursor emissions are reduced to meet the current standard, particularly when these

10    percentages are considered within the context of national estimates of CVattributable mortality,

11    estimates also suggest that substantial (Vattributable risks will remain after meeting the current

12    standard. With regard to the (Vattributable mortality risks estimated to remain after air quality

13    was adjusted to simulate just meeting the current 63 standard, we specifically note the following:

14       •  Using effect estimates from Zanobetti and Schwartz (2008) for the 2007 simulation year,
15          the REA estimated that the range of remaining (Vattributable deaths in the 12 U.S. study
16          cities was approximately 20-850, based on estimates with no 63 concentration cutoff, and
17          10-630, based on estimates  down to the LML.  The cities with the largest remaining 
-------
 1       •  Using effect estimates from Bell et al. (2004) for the 2009 simulation year, the REA
 2          estimated that the range of remaining CVattributable deaths in the 12 U.S. study cities
 3          was approximately 30-820, based on estimates with no O3 concentration cutoff, and 20-
 4          630, based on estimates down to the LML. The cities with the largest remaining CV
 5          attributable mortality risks were Los Angeles [821 (no cutoff), 628 (LML)], New York
 6          City [764 (no cutoff), 576 (LML)], Atlanta [364 (no cutoff), 315 (LML)], Boston [369
 7          (no cutoff), 250 (LML)], and Houston [272 (no cutoff), 211 (LML)].
 9          Respiratory Morbidity

10          The REA also estimated CVattributable respiratory hospital admissions, emergency
11    department visits, and asthma exacerbations for subsets of the 12 cities. These estimates were
12    based on concentration-response relationships from several available epidemiologic studies,
13    using different statistical approaches (US EPA, 2012b, section 7.5, Table 7-4). With regard to
14    these estimates, we specifically note the following:
15
16    •   In Atlanta for 2007 unadjusted air quality, using different studies and statistical models, the
17       REA estimated approximately 3,000 to 6,000 respiratory emergency department visits
18       attributable to O3. Upon simulating just meeting the current O3 standard, the REA estimated
19       that CVattributable emergency department visits decreased by approximately 20%, but that
20       approximately  2,000 to 5,000 (Vattributable emergency department visits remained. For
21       2009, the REA estimated that approximately 3,000 to 7,000 (Vattributable emergency
22       department visits remained (US EPA, 2012b, Table 7-21).
23
24    •   In New York City for 2007 unadjusted air quality, using different statistical models, the REA
25       estimated approximately 7,000 to 11,000 asthma emergency department visits attributable to
26       (V  Upon simulating just meeting the current 63 standard, the REA estimated that (V
27       attributable emergency department visits decreased by approximately 10%, but that
28       approximately  6,000 to 10,000 CVattributable asthma emergency department visits
29       remained.  (US EPA, 2012b, Table 7-21). For 2009, the REA estimated that approximately
30       8,000 to 13,000 (Vattributable emergency department visits remained (US EPA, 2012b,
31       Table 7-21).
32
33    •   In New York City for 2007 unadjusted air quality, using different statistical models, the REA
34       estimated approximately 500 to 700 (Vattributable asthma hospital admissions. Upon
35       simulating just meeting the current O3 standard, the REA estimated that (Vattributable
36       hospital admissions decreased by approximately 10%, but that approximately 500 to 600 CV
37       attributable asthma hospital admissions remained.  For 2009, the REA estimated that
38       approximately  600  to 800 CVattributable asthma hospital admissions remained (US EPA,
39       2012b, Tables  7-22 and 7-23).
40
41    •   Across the 12 urban study cities for 2007 unadjusted air quality,  the REA estimated up to
42       approximately  100  CVattributatble hospital admissions for respiratory causes. Upon

                                                  4-38

-------
 1       simulating just meeting the current 63 standard, the REA estimated that CVattributable
 2       hospital admissions decreased by approximately 10 to 40% across cities, but that up to
 3       approximately 60 CVattributable hospital admissions remained. For 2009, the REA
 4       estimated that up to approximately 250 (Vattributable hospital admissions remained (US
 5       EPA, 2012b, Tables 7-22 and 7-23).
 6
 7    •   In Boston for 2007 unadjusted air quality, the REA estimated approximately 54,000 incidents
 8       of CVattributable wheezing and approximately 20,000 to 30,000 incidents of CVattributable
 9       chest tightness or shortness of breath among asthmatics. Upon simulating just meeting the
10       current 63 standard, the REA estimated that the number of such (Vattributable events
11       decreased by approximately 8%, but that approximately 50,000 incidents of (Vattributable
12       wheezing and approximately 18,000 to 27,000 incidents of chest tightness or shortness of
13       breath remained for Boston (US EPA, 2012b, Table 7-24).
14
15          4.2.2.4 Integrated consideration of the exposure and risk information
16          In this  section, we revisit the following question:
17       •  To what extent do the Os exposure and risk analyses presented in the first draft
18          REA (US EPA, 2012b) support or call into question the adequacy of the public
19          health protection provided by the current Os primary standard?
20          In considering this question, we first note that the REA is currently in draft form and that
21    the final REA will include additional analyses and information to inform a decision  regarding the
22    adequacy of the current 63 NAAQs.  Specifically, we note that future drafts of the REA will
23    include quantitative estimates of O3-induced lung function decrements, estimates of repeated
24    exposures to Os concentrations above health benchmarks, estimates of risks associated with
25    long-term 63 concentrations, an alternative model-based approach to adjusting air quality to
26    simulate just meeting the current standard, and any additional changes judged appropriate in light
27    of comments and recommendations from CASAC and the public.  Therefore, while this section
28    considers the available exposure and risk information within the context of the adequacy of the
29    current Os standard, we also acknowledge that additional information will be available in future
30    drafts to more completely inform consideration of this  question.
31          In first considering estimated exposure results within the context of the adequacy of the
32    current standard we note that adjusting air quality concentrations in order to simulate just
33    meeting the current Os standard resulted in substantial  decreases in the estimated numbers of
34    exposures at or above  benchmark concentrations.  This was particularly notable for the 80 ppb
35    benchmark concentration, for which estimated exposures were almost eliminated in  all four
36    cities. Nonetheless, we note that when simulating just meeting the current Os standard,
37    substantial numbers of children and asthmatic children are estimated to experience one or more
38    Os concentrations per  year above the 60 ppb benchmark, and in some cases above the  70 ppb

                                                  4-39

-------
 1    benchmark. Exposures at and above these lower benchmark levels have been demonstrated to
 2    decrease lung function, increase respiratory symptoms, and increase airway inflammation in
 3    healthy adults. Given that children and asthmatics are likely to experience larger respiratory
 4    responses than the healthy adults on which the estimates at the benchmark concentrations were
 5    based, this analysis indicates that important Os-related health risks will remain in areas that
 6    would meet the current standard.
 7          Further, we note that the potential for serious adverse events (e.g., hospitalizations)
 8    following exposures of at-risk individuals to 63  concentrations above these benchmark
 9    concentrations is likely related to the frequency  of exposures, in addition to the exposure
10    concentrations.  We recognize that the first draft REA does not characterize the occurrence of
11    repeated exposures above benchmark concentrations. To the extent that such information is
12    available in the second draft REA, we will take it into account in reaching conclusions in the
13    second draft PA.
14          In next considering the risk results for mortality  and morbidity within the context of
15    considering the adequacy of the current O3 standard, we note that while the current standard was
16    estimated to reduce Os-attributable mortality by up to approximately 20% and Os-attributable
17    morbidity by up to approximately 30%, depending on the location and time frame simulated,
18    substantial risks were estimated to remain in areas that just met the Os standard. This includes
19    estimates of up to hundreds  of Os-attributable deaths per year and thousands of Os-attributable
20    hospitalizations and emergency department visits per year in  some cities. These risk estimates,
21    combined with the exposure results discussed above, suggest the potential for substantial Os-
22    related health risks that could reasonably be judged to be important from a public health
23    perspective, even in locations that meet the current 63 standard.
24          We also recognize that there is a broader array of (Vrelated adverse health outcomes  for
25    which risk estimates could not be quantified (that are part of a broader "pyramid of effects")
26    and that the scope of the  first draft REA was limited to just a sample  of urban areas and to some,
27    but not all, at-risk populations, leading to a limited estimation of public health impacts associated
28    with Os exposures across the country.
29          In light of the above  considerations, we reach the preliminary conclusion that the
30    available exposure and risk information from the first draft REA supports the available health
31    evidence discussed above (section 4.2.1) and that, at a minimum, exposure and risk results
32    support the appropriateness  of considering a range of potential alternative standards that would
33    increase public health protection against respiratory effects and mortality.  We acknowledge that
34    these preliminary conclusions  are based  on a set of draft analyses and that our conclusions in  the
                                                   4-40

-------
 1   final PA with regard to the current standard will be informed by the exposure and risk analyses
 2   included in the final REA.
 3   4.2.3  CASAC Advice from Previous Review
 4          In the last review of the 63 NAAQS, CASAC stated the following in its letter to the
 5   Administrator: "the CASAC unanimously recommends that the current primary ozone standard
 6   be revised and that the level  that should be considered for the revised standard be from 0.060 to
 7   0.070 ppm" [60 to 70 ppb] (Henderson, 2006c, p. 5). This recommendation followed from
 8   CASAC's more general recommendation that the then current standard, with a level of 0.08 ppm
 9   (effectively 0.084 ppb), needed to be made substantially more protective of human health,
10   particularly for at-risk subpopulations. In a subsequent letter sent specifically to offer advice to
11   aid the Administrator and Agency staff in developing the 63 proposal, the CASAC reiterated that
12   the Panel members "were unanimous in recommending that the level of the current primary
13   ozone standard should be lowered from 0.08 ppm to no greater than 0.070 ppm" (Henderson,
14   2007, p. 2). Further, the CASAC Panel expressed the view that the 2006 Criteria Document and
15   2007 Staff Paper, together with the information in its earlier letter, provided "overwhelming
16   scientific evidence for this recommendation" (Henderson, 2007, p. 2).  In expressing these views
17   and recommendations, the Panel emphasized the Clean Air Act requirement that the primary
18   standard must be set to protect the public health with an adequate margin of safety (Henderson,
19   2007).
20          Following the 2008 decision to revise the primary 63 standard by setting the level at
21   0.075 ppm (75 ppb), CASAC raised serious questions as to whether the standard met the
22   requirements of the CAA. In April 2008, the members of the CASAC Ozone Review Panel sent
23   a letter to EPA stating "[I]n our most-recent letters to you on this subject—dated October 2006
24   and March 2007—the CASAC unanimously recommended selection of an 8-hour average Ozone
25   NAAQS within the range of 0.060 to 0.070 parts per million [60 to 70 ppb] for the primary
26   (human health-based) Ozone NAAQS" (Henderson, 2008). The letter continued:
27          The CASAC now wishes to convey, by means of this letter, its additional,
28          unsolicited advice with regard to the primary and secondary Ozone NAAQS. In
29          doing so,  the participating members of the  CASAC Ozone Review Panel are
30          unanimous in  strongly urging you or your successor as EPA Administrator to
31          ensure that these recommendations be considered during the next review cycle for
32          the Ozone NAAQS that will begin next year (Henderson, 2008).
33          Moreover, the CASAC Panel noted that "numerous medical organizations and public
34   health groups have also expressed their support of these CASAC recommendations"
35   (Henderson, 2008). The letter further stated the following strong, unanimous view:
                                                 4-41

-------
 1          [The CASAC did] not endorse the new primary ozone standard as being sufficient
 2          protective of public health.  The CASAC—as the Agency's statutorily-established
 3          science advisory committee for advising you on the national ambient air quality
 4          standards—unanimously recommended decreasing the primary standard to within
 5          the range of 0.060-0.070 ppm [60 to 70 ppb].  It is the Committee's consensus
 6          scientific opinion that your decision to set the primary ozone standard above this
 7          range fails to satisfy the explicit stipulations of the Clean Air Act that you ensure
 8          an adequate margin of safety for all individuals, including sensitive populations
 9          (Henderson, 2008).
10          As discussed above (section 1.2.3), in 2010 the Administrator proposed to reconsider and
11    revise parts of the 2008 final rule. With regard to the primary standard, she proposed to revise
12    the level to within the range of 60 to 70 ppb (FR 75  2938). This proposal was based on the
13    scientific and technical record from the 2008 rulemaking, including public comments and
14    CASAC advice and recommendations.  In response  to EPA's request for additional advice,
15    CASAC again reaffirmed their conclusion that "the  evidence from controlled human and
16    epidemiological studies strongly supports the selection of a new primary ozone standard within
17    the 60 - 70 ppb range for an 8-hour averaging time" (Samet, 2011). As requested by EPA,
18    CASAC's  advice and recommendations were based on the scientific and technical record from
19    the 2008 rulemaking. In  considering this record for the 2008 rulemaking, CASAC stated the
20    following to summarize the basis for their conclusions (Samet, 2011, pp. ii to iii):

21              •   The evidence available on dose-response for effects of ozone shows
22                 associations extending to levels within the range of concentrations
23                 currently experienced in the United States.
24
25              •   There is scientific certainty that 6.6-hour exposures with exercise of
26                 young, healthy, non-smoking adult volunteers to concentrations > 80 ppb
27                 cause clinically relevant decrements of lung function.
28
29              •   Some healthy individuals have been shown to have clinically relevant
30                 responses, even at 60 ppb.
31
32              •   Since the majority of clinical studies involve young, healthy adult
33                 populations, less is known about health effects in such potentially ozone
34                 sensitive populations as the elderly, children and those with
35                 cardiopulmonary disease. For these susceptible groups, decrements in
36                 lung function may be greater than in  healthy volunteers  and are likely to
37                 have a greater clinical significance.
38
39              •   Children and adults with asthma are at increased risk of acute
40                 exacerbations on or shortly after days when elevated ozone concentrations
                                                   4-42

-------
 1                 occur, even when exposures do not exceed the NAAQS concentration of
 2                 75 ppb.
 O
 4              •  Large segments of the population fall into what EPA terms a "sensitive
 5                 population group," i.e., those at increased risk because they are more
 6                 intrinsically susceptible (children, the elderly, and individuals with chronic
 7                 lung disease) and those who are more vulnerable due to increased
 8                 exposure because they work outside or live in areas that are more polluted
 9                 than the mean levels in their communities.
10          More specifically, with respect to evidence from epidemiologic studies, CAS AC stated
11    "[W]hile epidemiological studies are inherently more uncertain as exposures and risk estimates
12    decrease (due to the greater potential for biases to dominate small effect estimates), specific
13    evidence in the literature does not suggest that our confidence on the specific attribution of the
14    estimated effects of ozone on health outcomes differs over the proposed range of 60-70 ppb."
15    (Samet, 2011, p.10).
16          In reaching staff conclusions in future drafts of the PA, in addition to taking note of this
17    advice provided by CAS AC in the last review, which was based on the scientific evidence and
18    exposure/risk information available in the last review, we will consider advice and
19    recommendations from CASAC as part of the current review (e.g., following their review of this
20    first draft PA), which is based on an updated body of scientific evidence and exposure/risk
21    information.
22    4.2.4  Preliminary Staff Conclusions on the Adequacy of the Current Os Standard
23          In this section, we present staffs preliminary conclusions regarding the adequacy of the
24    public health protection provided by the current 8-hour Os primary standard.  In discussing these
25    preliminary conclusions, we revisit the following overarching question for this chapter:
26       •  To what extent does the available scientific evidence and exposure/risk information
27          support or call into question the adequacy of the public health protection afforded
28          by the current Os primary standard?
29          As discussed above, in addressing this question we have considered the available
30    scientific evidence assessed in the ISA, as discussed above in Chapter 2 and considered in
31    section 4.2.1, the available  exposure and risk information assessed in the REA, as discussed
32    above in Chapter 3 and considered in section 4.2.2, and the advice and recommendations
33    received from CASAC during the last review of the 63 NAAQS, including advice received
34    following the proposal to reconsider the 2008 decision, as discussed above in section 4.2.3.
35          With regard to the scientific evidence related to short-term O3 exposures as considered
36    above (section 4.2.1), we reach the preliminary conclusion that the available evidence clearly

                                                   4-43

-------
 1    calls into question the adequacy of the current standard and provides strong support for
 2    considering potential alternative standards to increase public health protection, especially for at-
 3    risk groups.  This preliminary conclusion places considerable weight on the array of Os-related
 4    respiratory effects that have been reported following short-term exposures to Os concentrations
 5    below the level of the current standard, including clear evidence from controlled human exposure
 6    studies of lung function decrements, respiratory symptoms and pulmonary inflammation, as well
 7    as evidence of clearly adverse effects from epidemiologic studies, including  respiratory hospital
 8    admissions and emergency department visits, and premature mortality.  Staff believes that this
 9    body of scientific evidence is very robust, recognizing that it includes large numbers of various
10    types of studies, including toxicological studies, controlled human exposure studies, and
11    community epidemiological studies, that provide consistent and coherent evidence of a causal
12    relationship between short-term Os exposures and an array of respiratory  morbidity and mortality
13    effects, especially for at-risk populations. Moreover, the evidence supports a likely causal
14    relationship between short-term 63 exposures and non-accidental and cardiopulmonary
15    mortality. In emphasizing such effects, this preliminary conclusion also places considerable
16    weight on the plausibility of the linkages between the body of evidence for respiratory effects
17    and mortality following short-term exposures to 63 concentrations below those allowed by the
18    current Os standard and the broader body of experimental and epidemiologic evidence for Os-
19    related respiratory effects and mortality reported following exposures to higher concentrations.
20    In reaching this preliminary conclusion, we acknowledge that uncertainties persist in the health
21    evidence; however, in staffs view the broad array of health effects reported following exposures
22    to Os concentrations below those allowed by the current standard (i.e., respiratory effects and
23    mortality), combined with the plausible linkages between these effects and the much larger body
24    of epidemiologic and controlled human exposure evidence at higher Os concentrations, supports
25    the appropriateness of considering revising the current 63 standard in order to increase public
26    health protection against adverse health effects from short-term O3 exposures, particularly for
27    children, older adults, people with asthma, and for other at-risk groups.
28           With regard to the scientific evidence related to long-term Os exposures as discussed
29    above (section 4.2.1), we note that while Os-related effects have also been reported following
30    long-term exposures, available studies have not been conducted in areas that would likely have
31    met the current 8-hour standard.  We also note that epidemiologic associations between long-
32    term ambient 63  concentrations and respiratory effects have been reported in locations with
33    ambient Os concentrations above those allowed by the current short-term Os standard, and  that in
34    the absence of discernible thresholds in such associations, there is uncertainty in the extent to
35    which such associations would persist at lower Os concentrations.  Further, we note that
36    respiratory-related  effects reported following long-term Os exposures in animals have been
                                                    4-44

-------
 1    reported following repeated exposures to 63 concentrations well above the level of the current 63
 2    standard. In light of the above considerations we reach the preliminary conclusion that there is
 3    clear support for retaining at least the level of protection against adverse health effects associated
 4    with long-term 63 exposures afforded by the current standard, but that the evidence does not
 5    provide clear support for reaching a conclusion regarding the appropriateness of increasing
 6    public health protection against health effects related to long-term 63 exposures beyond that
 7    afforded by the current standard.
 8           With regard to the exposure and risk information related to short-term exposures as
 9    discussed above (section 4.2.2), we reach the preliminary conclusion that the available exposure
10    and risk information from the first draft REA supports the available health evidence and that, at a
11    minimum, exposure and risk results support the appropriateness of considering a range of
12    potential alternative standards that would increase public health protection against adverse
13    respiratory effects and mortality related to short-term 63 exposures.
14           With regard to CASAC advice (section 4.2.3), we note that the CASAC O3 Panel has
15    repeatedly recommended setting the level of the 8-hour O3 standard no higher than 70 ppb,
16    within a range of 60 to 70 ppb, which is below the level of the current standard (i.e., 0.075 ppm
17    or 75 ppb). Since this advice was provided, based on evidence available in the  last review, the
18    evidence for adverse health effects following short-term exposures to Os concentrations below
19    75 ppb has become even stronger, with the addition of several controlled human exposure and
20    epidemiologic studies conducted at relatively low 63 concentrations.  Given this, we note that, at
21    a minimum, nothing in the recent evidence would contradict CASAC's previous advice and that,
22    in fact, recent evidence provides stronger support for that advice.
23           In light of all of the above considerations, staff reaches the preliminary conclusion that
24    the body of information now available supports consideration of revising the current 8-hour 63
25    primary standard, so as to afford greater public health protection against the adverse health
26    effects of short-term O?, exposures, especially to at-risk groups, and that it does not support
27    retention of the current standard.  In so doing, we also recognize that consideration should be
28    given to the extent which such a revised standard would also provide appropriate protection
29    against the adverse health effects of long-term 63 exposures.
30
                                                    4-45

-------
 1    4.3    ADDITIONAL ANALYSES TO INFORM SECOND DRAFT PA
 2          Given our preliminary conclusion that the body of available evidence and information
 3    supports consideration of revising the current 8-hour primary 63 standard so as to afford greater
 4    public health protection from the adverse health effects of short-term Os exposures, we next
 5    consider the following overarching question:
 6       •  What additional analyses would be appropriate to help inform consideration of
 7          potential alternative standards in the second draft of the PA?
 8          In posing this question, it is not our purpose to draw preliminary staff conclusions in this
 9    first draft PA about a range of potential alternative standards that would be appropriate for
10    consideration by the Administrator in this review.  Such preliminary staff conclusions would
11    clearly be premature at this time prior to completion of the ISA and prior to conducting
12    additional exposure and risk analyses, as well as further analyses of air quality information from
13    epidemiologic studies, so as to translate such information into a basis for considering what
14    potential alternative standards would be appropriate for consideration.  Rather, such preliminary
15    staff conclusions will be developed in the second draft PA based on the assessment of scientific
16    information in the final ISA, the results of additional exposure and risk analyses in the second
17    draft Health REA, further analyses of the epidemiologic evidence in the context of the entire
18    body of available evidence  on Os-related health effects, and CAS AC advice  and public
19    comment.
20          In addressing this overarching question, we consider more specifically the following two
21    questions:
22       •  Beyond the exposure and risk analyses of air quality adjusted to simulate just
23          meeting the current standard in the first draft REA, what range of alternative  Os
24          levels would be appropriate for further exposure and risk analyses in the second
25          draft REA?

26       •  What approaches are appropriate to use in translating information from
27          epidemiologic studies into a basis for identifying potential alternative standards for
28          consideration in the second draft PA?
29    These questions are addressed below in sections  4.3.1  and 4.3.2, respectively.
30    4.3.1  Additional Exposure and Risk Analyses
31          In considering the first question posed above, we are specifically considering alternative
32    air quality scenarios defined in terms of simulations of just meeting short-term standards that
33    would increase public health protection against the effects of short-term Os exposures relative to
34    the degree of protection provided by the  current primary 63 standard.  In identifying a range of

                                                  4-46

-------
 1    alternative scenarios that would be appropriate for further exposure and risk analyses, we note
 2    that to fully define such alternatives, all the elements of a standard need to be specified,
 3    including indicator, averaging time, form, and level.  In this review, the newly available evidence
 4    provides a basis for considering alternative averaging times and levels, while the evidence
 5    continues to support the current Os indicator and provides no basis to focus consideration on
 6    alternative forms of the primary 63  standard at this time. We note that consideration may also be
 7    given in the second draft REA to assessing the degree of protection that such alternative
 8    scenarios may afford against the effects of long-term Os exposures.
 9           In identifying a range of alternative scenarios that would be appropriate for quantitative
10    exposure and risk analyses in the second draft of the REA, we recognize that decisions on the
11    specific set of alternatives to be analyzed depends in  part on initial results from such analyses
12    that may inform how broad a set of alternatives it may be useful to include.  In identifying such
13    alternatives in this first draft document, we are soliciting CAS AC advice and public comment on
14    such a range to further inform selection of the specific alternative scenarios that will be analyzed
15    in the second draft REA.
16           Given the evidence for Os-related effects following both short-term and long-term Os
17    exposures discussed above in chapter 2 and more fully in the ISA (US EPA, 2012a, chapters 6
18    and 7), section 4.3.1.1 below considers both short- and long-term averaging times. Section
19    4.3.1.2 considers available evidence and information in identifying a range of levels that may be
20    appropriate to define alternative air quality scenarios that would be appropriate for further
21    exposure and risk analyses in the second draft REA.
22           4.3.1.1 Averaging time
23           In considering both short- and long-term averaging times, we first note that because Os-
24    related effects have been reported following both short-term and long-term 63 exposures, an
25    important consideration in the current review is the extent to which different O3 standards would
26    be expected to provide appropriate protection against both short-term and long-term Os
27    exposures. In considering the potential appropriateness of short- and long-term averaging times,
28    we consider the available scientific  evidence and air quality information.
29           As discussed above in detail (section 4.2.1), a number of studies have reported O3-related
30    health effects following short-term exposures to Os concentrations below the level of the current
31    8-hour standard (75 ppb) or to distributions of 63 concentrations that would be allowed by the
32    current standard, which support the appropriateness of analyzing alternative scenarios that would
33    increase public health protection against short-term Os exposures. This includes controlled
34    human exposure studies that have reported respiratory effects in healthy adults following short-
                                                    4-47

-------
 1    term (hours) exposures to 63 concentrations as low as 60 ppb; epidemiologic studies reporting
 2    associations between short-term (minutes to hours) ambient Os concentrations allowed by the
 3    current standard and lung function decrements or respiratory symptoms; and epidemiologic
 4    studies reporting associations between short-term (8-hour, 24-hour) ambient 63 concentrations
 5    allowed by the current standard and respiratory-related hospital admissions, respiratory-related
 6    emergency department visits, and respiratory and total mortality.
 7           In addition, while a number of studies support the appropriateness of an Os standard that
 8    protects public health against long-term 63 exposures, the large majority of the cities included in
 9    U.S. multi-city studies of long-term Os-related respiratory effects would likely not have met the
10    current 8-hour 63 standard during the study periods.  Specifically, as discussed in more detail
11    above (section 4.2.1), epidemiologic associations between long-term ambient O3 concentrations
12    and respiratory effects have been reported in locations with 4th-highest 8-hour daily maximum Oj
13    concentrations well above the level of the current 8-hour 63 standard.  Also, respiratory-related
14    effects reported following long-term Os exposures  in animals have been reported following
15    repeated exposures to Os concentrations well above the level of the current Os standard (section
16    4.2.1, above).
17           In light of the above considerations, we reach the preliminary conclusion that, while the
18    available evidence provides strong support for considering the degree of public health protection
19    provided against both short-term and long-term Os exposures, the evidence provides little
20    support for the need to increase public health protection beyond that provided by the current 8-
21    hour standard specifically against long-term Os exposures. Nonetheless, in light of the
22    possibility that Os-related health effects following long-term exposures could persist at lower Os
23    concentrations than may occur  in areas that meet the current standard, we have also considered
24    the extent to which just meeting the current 8-hour standard would be expected to reduce long-
25    term ambient 63  concentrations in areas other than where the currently available studies were
26    conducted. As noted above in section 4.2.1.2, several long-term exposure studies provide
27    information on Os concentrations that have been associated with respiratory effects.  These
28    studies may further inform consideration of the relationships between short- and long-term
29    concentrations beyond the considerations presented here. We plan to conduct additional air
30    quality analyses that will help more fully inform this consideration in the second draft PA.
31           When considering the extent to which available air quality information suggests that
32    meeting the current 8-hour standard could also reduce long-term 63 concentrations, we note that
33    when ambient Os concentrations were adjusted to simulate just meeting the  current standard in
34    the 12 cities evaluated in the first draft REA, estimated long-term Os concentrations also
35    decreased  (Wells et al., 2012, Table 2-6).  Specifically, as illustrated in Table 4-4 below, when

                                                   4-48

-------
 1   air quality was adjusted to simulate just meeting the current O3 standard, the 4th-highest 8-hour
 2   daily maximum O3 concentrations decreased by approximately 10 to 30% across the 12 urban
 3   study locations.  Somewhat smaller decreases in long-term O3 concentrations were also
 4   estimated, based on two different long-term O3 metrics including the seasonal averages of 1-hour
 5   and 8-hour daily maximum O3 concentrations. Although we acknowledge that these initial
 6   simulations done in support of the first draft REA are not intended to reflect specific strategies
 7   for reducing ambient O3 concentrations, they suggest that reductions in ambient O3 implemented
 8   to meet an  8-hour O3 standard would also reduce long-term O3 concentrations.  Further, we note
 9   modeling approaches that are planned for use in the second draft REA will provide more relevant
10   analyses to further inform this consideration in the second draft PA.
11
12   Table 4-4.  Average Estimated Percent Decrease in Maximum and Seasonal O3 Concentrations
13   with Simulation  of Just Meeting the Current Standard (2006-2008) Using Quadratic Rollback19
Urban Area
Atlanta
Baltimore
Boston
Cleveland
Denver
Detroit
Houston
Los Angeles
New York
Philadelphia
Sacramento
St. Louis
4th-highest 8-hour daily
maximum
22
18
9
12
13
8
17
38
17
19
26
12
Seasonal Average 1-
hour maximum
17
15
7
9
12
7
13
32
13
15
21
10
Seasonal Average 8-
hour maximum
15
13
7
8
11
6
11
29
12
13
19
9
     19Averages were calculated from Table 2-6 in Wells et al., 2012.
                                                  4-49

-------
 1
 2          In light of the above considerations, and based on the available scientific evidence and air
 3    quality information, we reach the preliminary conclusion that a standard with a short-term
 4    averaging time, specifically the current 8-hour averaging time, can be an appropriate approach to
 5    providing adequate protection against both short-term and long-term O3 exposures.  In support of
 6    this preliminary conclusion, we note that (1) O3-related health effects have been reported in a
 7    number of studies following short-term O3 exposures below those allowed by the current
 8    standard; (2) while O3-related health effects have also been reported following long-term
 9    exposures, the O3 concentrations in such studies were above those allowed by the current O3
10    standard; and (3) to the extent that O3-related health effects reported following long-term
11    exposures could persist at O3 concentrations below those allowed by the current standard,
12    potential alternative 8-hour O3 standards would be expected to further decrease long-term
13    concentrations, beyond the decreases estimated for the current standard.
14          4.3.1.2 Alternative Levels
15          In considering alternative levels that would be appropriate for defining air quality
16    scenarios for further exposure and risk analyses in the second draft REA, we focus on levels in
17    conjunction with the same averaging time  (8-hour), form (the 3-year average of the annual 4th-
18    highest daily 8-hour maximum) and indicator (O3) as the current O3 standard.  In considering the
19    available evidence to help inform identification of such alternative levels,  we focus in particular
20    on controlled human exposure studies, which provide the clearest evidence for respiratory effects
21    following exposures to  specific O3 concentrations, and also consider epidemiologic studies
22    reporting associations with O3 concentrations below those allowed by the  current O3 standard.
23          As discussed in  detail above (section 4.2.1.1), controlled human exposure studies of O3-
24    induced lung function decrements, respiratory symptoms, and airway inflammation provide
25    strong evidence that exposures to O3 concentrations below the level of the current O3 standard
26    can impair respiratory functioning, resulting in respiratory effects that could be clinically
27    significant, particularly for members of at-risk populations (e.g., people with asthma). In
28    particular, with regard to these studies we note the following:
29    •  Controlled human exposure studies have reported consistent decrements in group mean lung
30       function in healthy adults following 6.6-hour exposures to O3 concentrations ranging from 60
31       ppb to  120 ppb (US EPA, 2012a, Figure  6-1).  Average decrements were approximately 3%
32       following exposures to 60 ppb O3 and 6% following exposures to 70 ppb O3, with larger
33       decrements at higher exposure levels.  Group mean  decrements reported in these studies have
34       generally been statistically significant. Lung function decrements have not been reported to
35       be statistically significant in controlled human exposure studies following exposures to 40
36       ppb O3 (US EPA, 201 Ib, section 6.2.1.1).

                                                   4-50

-------
 1    •   Following 6.6-hour exposures to an average 63 concentration of 60 ppb, the proportion of
 2       healthy adult study subjects with FEVi decrements at or above 10% (i.e., a decrement that is
 3       potentially adverse for asthmatics (Henderson, 2006)) ranged from 3% to 20% across studies.
 4       When the results from such studies were combined, the ISA notes that 10% of subjects
 5       exposed to an average Os concentration of 60 ppb experienced such FEVi decrements (US
 6       EPA, 201 Ib, section 6.2.1.1).  In the one study that evaluated the exposure-response
 7       relationship  (Schelegle et al., 2009), the percentage of subjects experiencing FEVi
 8       decrements greater than or equal to 10% increased with increasing 63 exposure
 9       concentrations (i.e., 0 for filtered air controls; 16% for 60 ppb; 19% for 70 ppb; 29% for 80
10       ppb) (section 4,2,1,1 above).

11    •   Controlled human exposure studies have reported increased respiratory symptoms in healthy
12       adults following exposures to O3 concentrations at or above 60 ppb.  Of the studies that
13       evaluated respiratory symptoms following exposures to 60 ppb 63, two reported trends
14       towards increased symptoms (Adams, 2006; Schelegle et al., 2009).  One of these (Adams,
15       2006) reported that the Os-induced increase was statistically different from initial symptoms,
16       though not from filtered air controls. An increase in respiratory symptoms has been reported
17       to be statistically different from filtered air controls following exposure to 63 concentrations
18       at or above 70 ppb (Schelegle et al., 2009).

19    •   The only controlled human exposure study to have evaluated airway inflammation following
20       exposures below the level of the current Os standard (Kim et al., 2011) reported a statistically
21       significant increase in neutrophilic inflammation following 6.6 hour exposures to 60 ppb Os
22       in healthy adults with intermittent moderate exertion.
23

24          In addition,  we note that, as discussed in detail above (section 4.2.1.1), epidemiologic
25    studies also provide evidence for Os-related respiratory effects in analyses limited to Os
26    concentrations below the level of the current 63 standard.  In particular, with regard to these
27    studies we note the following:

28    •   Two epidemiologic studies have reported associations with lung function decrements in
29       children in analyses restricted to relatively low O?, concentrations.  Spektor et al. (1988)
30       reported that the association was statistically significant when all hourly Os concentrations
31       were below 60 ppb and Brunekreef et al. (1994) reported a positive, but non-significant
32       decrement for analyses restricted to 63 concentrations (10 minutes to 2.4 hours) below 61
33       ppb.

34    •   A small number of studies have reported statistically significant associations in outdoor
35       workers, including one study where the highest 8-hour Os concentration was 65 ppb (Chan
36       and Wu, 2005) and one study restricted to maximum hourly concentrations below 40 ppb
37       (Brauer et al., 1996).

38    •   An epidemiologic study reported a statistically significant association with respiratory
39       symptoms in asthmatic children in a location with a maximum 1-hour Os concentration of 70
40       ppb (Rabinovitch et al., 2004).
                                                   4-51

-------
 1    •  An epidemiologic study (Mar and Koenig, 2009) reported positive and statistically
 2       significant associations with asthma emergency department visits in both children and adults
 3       in Seattle. The average (i.e., over the years of the study) O3 design value for Seattle during
 4       the study period was 71 ppb.
 5    •  A large multi-city epidemiologic study (Bell et al., 2006) reported that 63 mortality effect
 6       estimates were positive and statistically significant for restricted distributions of 63
 7       concentrations that would likely have been below those allowed by the current Os standard
 8       over the study period in most of the cities evaluated.  As discussed in section 4.2.1.3 above,
 9       the 50th percentiles of the distributions of highest 8-hour daily maximum Os concentrations
10       corresponding to these effect estimates were as low as approximately 70 ppb, and 25th
11       percentiles were as low as approximately 64 ppb.  In addition, this study reported that Os-
12       related mortality effect estimates were positive and approached statistical  significance for O3
13       concentrations that would likely have been well below those allowed by the current 63
14       standard over the study period in most of the cities evaluated. The 50th percentile of the
15       distribution of highest 8-hour daily maximum Os concentrations corresponding to this effect
16       estimate was approximately 60 ppb, and the 25th percentile was approximately 55 ppb.
17
18          Beyond the evidence-based considerations outlined above, we also note, as discussed
19    above in section 4.2.3, that CASAC has repeatedly recommended that EPA consider setting the
20    level of the primary Os standard within the range of 60 to70 ppb (Henderson, 2006; Henderson,
21    2007; Henderson, 2008; Samet, 2011).  This advice has been provided based on the evidence that
22    was available in the last review of the Os  NAAQS.
23          Based on the above considerations, in staffs view, the newly available evidence in this
24    review provides increased support for conducting further exposure and risk analyses of
25    alternative levels in the range of 60 to 70  ppb. In addition, it is our view that the new evidence,
26    when considered in the context of the evidence available in the last review, provides support for
27    conducting further exposure and risk analyses of air quality scenarios that extend to a level
28    somewhat below 60 ppb. We recognize increasing uncertainty in interpreting the epidemiologic
29    evidence at such lower levels.
30          As a basis for these views, as outlined above, we particularly note that, following
31    exposures of healthy adults to 60 ppb O3, controlled human exposure studies have reported
32    statistically significant group mean lung function decrements, a statistically significant increase
33    in airway inflammation, and trends towards increased respiratory symptoms.  In addition, we
34    note that when data were combined from  available controlled human exposure studies of Os-
35    induced lung function decrements, 10% of healthy adults exposed to 60 ppb Os experienced lung
36    function decrements that could be adverse for asthmatics. Given that these results were reported
37    in healthy adults, we note that they likely underestimate the magnitude and seriousness of such
38    effects for at-risk groups.  Therefore, we believe that the reporting of multiple respiratory effects

                                                   4-52

-------
 1    in healthy adults following exposures to 60 ppb 63, combined with the possibility that larger
 2    effects could be observed in asthmatics and members of other at-risk groups, supports the
 3    appropriateness of including in further exposure and risk analyses an alternative scenario at a
 4    level somewhat below 60 ppb. In our view, evidence from epidemiologic panel studies reporting
 5    associations with lung function decrements in children and outdoor workers in analyses restricted
 6    to low 63 concentrations supports analyses of such a level, although because these studies
 7    evaluated a variety of averaging times (i.e., minutes to hours) they do not provide clear support
 8    for analyzing any one specific lower level.
 9          Having preliminarily concluded that the health effects evidence provides support for
10    conducting additional exposure and risk analyses for alternative levels somewhat below 60 ppb,
11    we note that depending on how far below 60 ppb one considers, such levels may be close to peak
12    North American background (NAB) concentrations that occur infrequently at some high
13    elevation sites in the western United States or may even approach seasonal mean background
14    concentrations at such sites.  As discussed above (section 1.3.4), estimates of summertime NAB
15    concentrations were generally below about 35 ppb in the eastern United States and below 40 ppb
16    in California, areas where large populations reside and where CVrelated health risks are
17    estimated to be highest (US EPA, 2012b). In considering this information, we recognize that
18    EPA's Rule on Treatment of Data Influenced by Exceptional Events grants authority to exclude
19    air quality monitoring data from regulatory determinations if a state adequately demonstrates that
20    an exceptional event has caused an exceedance or violation of a NAAQS, and that section 179B
21    of the CAA provides for treatment of air quality data from international transport when
22    emissions emanating from outside of the United States have caused an exceedance or violation of
23    a NAAQS (section 1.3.4, above).
24    4.3.2  Approaches to Translating Epidemiologic Evidence
25          We next consider the second question posed at the beginning of section 4.3:
26       •  What approaches are appropriate to use in translating information from
27          epidemiologic studies into a basis for identifying potential alternative standards for
28          consideration in the second draft PA?
29          As an initial matter, while available epidemiologic studies provide strong support for the
30    conclusion that (Vrelated morbidity and mortality associations extend to 63 concentrations
31    below the level of the current Os standard (section 4.2.1), it is a challenge to translate this
32    epidemiologic evidence into the basis for identifying specific alternative  standard levels that
33    would be appropriate for consideration in this review. In particular, we note that such studies do
34    not provide evidence for a discernible population threshold, below which it can be concluded
35    with confidence that (Vrelated effects do not occur, though confidence in the nature of the
                                                   4-53

-------
 1    concentration-response relationship decreases at the low end of the distribution of 63
 2    concentrations (US EPA, 201 Ib, section 2.5.4.4).  As a result, any approach to using these
 3    studies to inform decisions about potential alternative standard levels requires judgments about
 4    how to weigh confidence in (Vrelated effects over the distributions of 63 concentrations.
 5          To help inform consideration of potential alternative standard levels in the second draft
 6    PA, we are soliciting CASAC advice and public comment on approaches that may be appropriate
 7    for interpreting and translating epidemiologic evidence into the basis for such considerations.  To
 8    facilitate such advice and comment, we discuss approaches to using analyses of concentration-
 9    response relationships  as well as analyses of restricted air quality distributions to inform
10    consideration of potential alternative standard levels.  Based on comments and advice received,
11    the second draft of the PA may expand (e.g., to include additional studies that have characterized
12    concentration-response relationships) and/or modify the approach to inform staffs preliminary
13    conclusions regarding the potential alternative standards levels that are appropriate for
14    consideration by the Administrator. In presenting these approaches, we use the information
15    reported in the studies  by Silverman and Ito (2010) and the study by Bell et al. (2006) as
16    examples of applying such approaches in single- and multi-city studies, respectively.
17          We first consider the 63 air quality in the study location evaluated by Silverman and Ito
18    (2010), which evaluated the concentration-response relationship between average 8-hour Os
19    concentrations (i.e., averaged across monitors) and asthma hospital admissions in New York City
20    (US EPA, 2012a, Figure 6-15), as discussed in more detail above (section 4.2.1.2). While we
21    recognize that there is no single air quality concentration that uniquely identifies the portion of
22    the concentration-response function in which it is appropriate to place the greatest confidence,
23    we note that confidence intervals are smallest around average 8-hour 63 concentrations around
24    40 ppb (i.e., approximately 30 to 50 ppb), become somewhat wider at Os concentrations below
25    about 30 ppb, and become notably wider below about 20 ppb.  In light of this, it may be
26    reasonable to conclude that confidence is greatest that a linear concentration-response
27    relationship exists for Os and asthma hospital admissions in New York City for average 8-hour
28    daily maximum Os concentrations ranging from about 30 to 50 ppb, with somewhat lower
29    confidence in the nature of the concentration-response relationship below 30 ppb, and more
30    notably decreasing at concentrations below approximately 20 ppb.
31          As discussed above (section 4.2.1.2), the 8-hour Os concentrations reported in the
32    analysis by Silverman  and Ito (2010) were based on averaging across multiple monitors in the
33    study area while attainment of the current Oj standard is based on the annual 4th-highest 8-hour
34    daily maximum Os concentration measured at the monitor recording the highest such
35    concentration.  Therefore, the average 8-hour concentrations reported by  Silverman and Ito

                                                   4-54

-------
 1    (2010) are not directly comparable to the level of the current 63 standard.  To gain insight into
 2    the relationship between average 8-hour daily maximum Os concentrations (i.e., based on
 3    averaging across monitors) and the highest 8-hour daily maximum concentrations (i.e., from the
 4    individual monitor recording the highest such concentration),20 we have considered available air
 5    quality information from EPA's AQS for the monitors used in the study by Silverman and Ito
 6    (2010) over the time period of the study (Figure 4-2 above, Wells et al., 2012).
 7           In considering the range of average Os concentrations for which we have the greatest
 8    confidence  in the concentration-response relationship, we first note that for the New York City
 9    study area over the period of the study, 742 days had average (i.e., averaged across monitors) 8-
10    hour daily maximum 63 concentrations from 30 to 50 ppb (Wells et al., 2012). On these days,
11    the highest  8-hour daily maximum Oj concentrations (i.e., from the monitors recording the
12    highest such concentrations) ranged from approximately 55 ppb (i.e., for average concentrations
13    from 30 to 35 ppb) to just below 75 ppb (i.e., 72 ppb for average concentrations from 45 to 50
14    ppb) (Figure 4-2).  In addition, we note that 996 days had average 8-hour daily maximum Os
15    concentrations from 20 to 50 ppb (Wells et al., 2012). The highest 8-hour daily maximum Os
16    concentrations on these days extended down to approximately 50 ppb (i.e., 49 ppb for average
17    concentrations from 20 to 25 ppb) (Figure 4-2).
18           In considering these results within the context of the questions to be addressed in the
19    second draft PA regarding potential alternative standards, we note that at ambient Os
20    concentrations for which it may be reasonable to conclude that we have the highest confidence in
21    the linear nature of the concentration-response relationship (i.e., 30 to 50 ppb, 8-hour average),
22    the highest  8-hour daily maximum Os concentrations (i.e., from the monitors recording the
23    highest such concentrations) ranged from approximately 55 to 75 ppb.  In addition, in
24    considering the average 8-hour concentrations just above those where confidence intervals
25    around the concentration-response function become notably wider (i.e., about 20 ppb, 8-hour
26    average), we note that the highest 8-hour daily maximum O3 concentrations were approximately
27    50 ppb (section 4.2.1.2, above). As noted above,  the 8-hour daily maximum concentrations
28    identified here are maximum values, not 4th-highest 8-hour daily maximum values, such that they
29    do not reflect the form of the current standard and should not be considered as levels that
30    correspond to a standard.
      20 As noted above (section 4.2.1.3), in preliminary analyses to inform consideration of the adequacy of the current
      O3 standard, we characterized the relationships between average and highest 8-hour daily maximum concentrations.
      We recognize that, to the extent these types of analyses are carried forward into the second draft PA for informing
      consideration of potential alternative standards, we would also consider the relationships between average and 4th-
      highest 8-hour daily maximum concentrations.
                                                    4-55

-------
 1           We next consider the 63 air quality in the study locations evaluated by Bell et al. (2006)
 2    which, as discussed in more detail above (section 4.2.1.3), evaluated associations between
 3    mortality and average 24-hour O?, concentrations (i.e., averaged across monitors) in 98 U.S.
 4    cities, in a series of analyses where air quality was restricted to concentrations below
 5    progressively lower 24-hour average cut points. In considering these restricted analyses, we note
 6    that 63 effect estimates were similar in magnitude for analyses with cut points ranging from 60
 7    ppb down to 15 ppb. In the Bell et al. analyses, associations with cut points down to 30 ppb
 8    were  statistically significant, associations with a cut point of 25 ppb approached statistical
 9    significance, and associations with cut points of 20 and 15 ppb lost statistical significance.
10    Further, statistical precision decreased as cut points decreased, with a notable widening of
11    confidence intervals for the analyses with cut points at and below 20 ppb.21
12           As with the study by Silverman and Ito (2010), it is not appropriate to compare the 24-
13    hour average 63 concentrations reported by Bell et al. (2006) directly to 8-hour maximum levels
14    or to the level of the current standard.  Therefore, as discussed above (section 4.2.1.3), we have
15    used EPA's AQS to relate average 24-hour Os concentrations in each of the 98 study cities (i.e.,
16    averaged across monitors in cities with multiple monitors) to the highest 8-hour daily maximum
17    concentrations (i.e., from the individual monitors in each  city recording the highest such
18    concentrations) (Wells et al., 2012). In considering these analyses, we focused on 24-hour
19    average cut points of 30, 25, and 20 ppb so as to consider the range over which the associations
20    went  from being statistically significant to  not being significant, and the statistical precision
21    notably decreased.  In considering only days with average 24-hour 63 concentrations at or below
22    the 30 ppb cut point, the 50th percentile of the distribution of highest 8-hour daily maximum Oj,
23    concentrations (i.e., the  distribution across  cities22) was approximately 70 ppb and the 25th
24    percentile of the distribution was approximately 65 ppb (Wells et al., 2012).23  In considering
25    only days with average 24-hour Oj concentrations at or below the 25 ppb cut point, the 50th
26    percentile of the distribution of highest 8-hour daily maximum O?, concentrations (i.e., the
27    distribution across cities) was approximately 60 ppb and the 25th percentile of the distribution
28    was approximately 55 ppb (Wells et al., 2012). Further, in considering only days with average
      21 As discussed above, the decreasing precision in effect estimates as concentrations approach the lower extreme of
      the air quality distribution, as indicated by notable widening of confidence bounds, is intrinsically related to data
      density and is not necessarily indicative of an absence of O3-associated health effects.
      22Percentiles were identified from the distribution of the study period averages of highest 8-hour daily maximum O3
      concentrations within each city.
      23 As noted above (section 4.2.1.3), 50th and 25th percentiles were identified from the distribution of the study period
      averages of highest 8-hour daily maximum O3 concentrations within each city. Identification of the 50th percentile of
      this distribution across cities reflects consideration of an estimate of the central tendency of the distribution.
      Identification of the 25th percentile reflects consideration of the lower part of the distribution where the data density
      is appreciably less.
                                                      4-56

-------
 1    24-hour 63 concentrations at or below the 20 ppb cut point, the 50th percentile of the distribution
 2    of highest 8-hour daily maximum Oj concentrations was approximately 50 ppb and the 25th
 3    percentile was approximately 45 ppb (Wells et al., 2012).

 4           In considering the above analyses within the context of the questions to be addressed in
 5    the second draft PA regarding potential alternative standards, we note the following:

 6       •   In the single-city study by Silverman and Ito, at Os concentrations for which we have the
 7           highest confidence in the linear nature of the concentration-response relationship (i.e., 30
 8           to 50 ppb averaged across monitors), the highest 8-hour daily maximum 63
 9           concentrations (i.e., from the monitors recording the highest such concentrations) ranged
10           from approximately 55 to 75 ppb.  Confidence intervals become notably wider when the
11           highest 8-hour daily maximum concentrations are below about 50 ppb.  In our view,
12           judgments regarding the extent to which this type  of analysis supports consideration of
13           alternative 8-hour standard levels within or below this range of 63 concentrations could
14           depend on the weight placed on the concentration-response relationship over different
15           parts of the distribution of Os concentrations and more generally on the relevance of this
16           air quality information for informing such judgments.
17

18       •   In the multi-city study by Bell et al., Os  effect estimates were similar in magnitude and
19           were relatively statistically precise for distributions  of 63 concentrations characterized by
20           50th percentile concentrations (of 8-hour daily maximum concentrations across the 98
21           cities) at or above 60 ppb and 25th percentile concentrations at or above 55 ppb. Effect
22           estimates remained positive but lost statistical significant for distributions of Os
23           concentrations characterized by lower 50th and 25th percentile concentrations. In our
24           view, judgments regarding the extent to which this type of analysis supports
25           consideration of alternative 8-hour standard levels at concentrations identified by  such
26           percentiles of air quality distributions across cities in multi-city studies could depend on
27           the weight placed on this type of cut point analysis in general and on specific percentiles
28           within such distributions.
29

30           As noted above, our purpose in presenting the approaches discussed above is to solicit
31    CASAC advice and public comment on these approaches to translating epidemiologic evidence
32    into the basis for identifying potential alternative standards  that would be appropriate for
33    consideration in the second draft PA.  Based on such advice and comment, the second draft PA
34    may expand (e.g., to include additional studies that have characterized concentration-response
35    relationships) and/or modify the approaches outlined above to help inform staffs preliminary
36    conclusions regarding the potential alternative standards levels that are appropriate for
37    consideration by the Administrator.
                                                    4-57

-------
 1   4.4    SUMMARY OF STAFF CONCLUSIONS ON THE PRIMARY O3 STANDARD

 2                [To be added in the second draft Policy Assessment.]
 O

 4   4.5    KEY AREAS FOR FUTURE RESEARCH AND DATA COLLECTION

 5                [To be added in the second draft Policy Assessment.]

 6

 7   4.6    REFERENCES

 8   Adams, W. C. (2006a) Comparison of chamber 6.6 hour exposures to 0.04-0.08 ppm ozone via
 9        square-wave and triangular profiles on pulmonary responses. Inhalation Toxicol. 18: 127-
10        136. http://dx.doi.org/10.1080/08958370500306107

11   Adams, WC. (2006b). Human pulmonary responses with 30-minute time intervals of exercise
12        and rest when exposed for 8 hours to 0.12 ppm ozone via square-wave and acute triangular
13        profiles. Inhal Toxicol 18: 413-422. http://dx.doi.org/10.1080/08958370600563599

14   Adams, WC. (2002). Comparison of chamber and face-mask 6.6-hour exposures to ozone on
15        pulmonary function and symptoms responses. Inhal Toxicol 14: 745-764.
16        http://dx.doi.org/10.1080/08958370290084610

17   Akinbami, LJ; Lynch, CD; Parker, JD; Woodruff, TJ. (2010). The association between childhood
18        asthma prevalence and monitored air pollutants in metropolitan areas, United States, 2001-
19        2004. Environ Res  110:  294-301. http://dx.doi.Org/10.1016/j.envres.2010.01.001

20   Alexis, NE; Lay, JC; Hazucha, M; Harris, B; Hernandez, ML; Bromberg, PA; Kehrl, H; Diaz-
21        Sanchez, D; Kim, C; Devlin, RB; Peden, DB. (2010). Low-level ozone exposure induces
22        airways inflammation and modifies cell surface phenotypes in healthy humans. Inhal
23        Toxicol 22: 593-600.

24   ATS (American Thoracic Society). (2000). What constitutes an adverse health effect of air
25        pollution? Am. J. Respir. Crit. Care Med. 161: 665-673.

26   ATS (American Thoracic Society). (1985) Guidelines as to what constitutes an adverse
27        respiratory health effect, with special reference to epidemiological studies of air pollution.
28        Am. Rev. Respir. Dis. 131: 666-668.

29   Bell, ML; Dominici, F. (2008). Effect modification by community characteristics on the short-
30        term effects of ozone exposure and mortality in 98 US communities. Am J Epidemiol  167:
31        986-997. http://dx.doi.org/10.1093/aje/kwm396

                                                 4-58

-------
 1   Bell, ML; Kim, JY; Dominici, F. (2007). Potential confounding of particulate matter on the
 2        short-term association between ozone and mortality in multisite time-series studies.
 3        Environ Health Perspect 115: 1591-1595. http://dx.doi.org/10.1289/ehp.10108

 4   Bell, ML; Peng, RD; Dominici, F. (2006). The exposure-response curve for ozone and risk of
 5        mortality and the adequacy of current ozone regulations. Environ Health Perspect 114: 532-
 6        536.

 7   Bell, M. L.; McDermott, A.; Zeger, S. L.; Samet, J. M.; Dominici, F. (2004) Ozone and short-
 8        term mortality in 95 US urban communities, 1987-2000. JAMA J. Am. Med. Assoc. 292:
 9        2372-2378.

10   Brauer, M; Blair, J; Vedal, S. (1996). Effect of ambient ozone exposure on lung function in farm
11        workers. Am J Respir Crit Care Med 154: 981-987.

12   Brunekreef, B; Hoek, G; Breugelmans, O; Leentvaar, M. (1994). Respiratory effects of low-level
13        photochemical air pollution in amateur cyclists. Am J Respir Crit Care Med 150: 962-966.

14   Chan, CC; Wu, TH. (2005). Effects of ambient ozone exposure on mail carriers' peak expiratory
15        flow rates. Environ Health Perspect 113: 735-738.  http://dx.doi.org/10.1289/ehp.7636

16   Darrow, LA; Klein, M; Sarnat, JA; Mulholland, JA; Strickland, MJ; Sarnat, SE; Russell, AG;
17        Tolbert, PE. (201 la). The use of alternative pollutant metrics in time-series studies of
18        ambient air pollution and respiratory emergency department visits. J Expo Sci Environ
19        Epidemiol 21: 10-19. http://dx.doi.org/10.1038/ies.2009.49

20   Devlin, RB; Mcdonnell, WF; Mann, R; Becker, S; House, DE; Schreinemachers, D; Koren, HS.
21        (1991). Exposure of humans to ambient levels of ozone for 6.6 hours causes cellular and
22        biochemical changes in the lung. Am J Respir Cell Mol Biol 4: 72-81.

23   Dry den, DM; Spooner, CH; Stickland, MK; Vandermeer, B; Tjosvold, L; Bialy, L; Wong, K;
24        Rowe, BH. (2010). Exercise-induced bronchoconstriction and  asthma. (AHRQ Publication
25        No.  10-E001). Rockville, MD: Agency for Healthcare Research and Quality.

26   Franklin, M; Schwartz, J. (2008). The impact of secondary particles on the association between
27        ambient ozone and mortality. Environ Health Perspect 116: 453-458.
28        http://dx.doi.org/10.1289/ehp.10777

29   Henderson, R. (2008) Letter from CASAC Chairman Rogene Henderson to EPA Administrator
30        Stephen Johnson. April 7, 2008, EPA-CASAC-08-001.
31   Henderson, R. (2007) Letter from CASAC Chairman Rogene Henderson to EPA Administrator
32        Stephen Johnson. March 26, 2007, EPA-CASAC-07-002.
                                                  4-59

-------
 1   Henderson, R. (2006) Letter from CASAC Chairman Rogene Henderson to EPA Administrator
 2         Stephen Johnson. October 24, 2006, EPA-CASAC-07-001.
 3   Holz, O; Torres, RA; Timm, P; Mucke, M; Richter, K; Koschyk, S; Magnussen, H. (1999).
 4         Ozone-induced airway inflammatory changes differ between individuals and are
 5         reproducible. Am J Respir Crit Care Med 159: 776-784.

 6   Hoppe, P; Praml, G; Rabe, G; Lindner, J; Fruhmann, G; Kessel, R. (1995). Environmental ozone
 7         field study on pulmonary and subjective responses of assumed risk groups. Environ Res 71:
 8         109-121.

 9   Horstman, DH; Ball, BA; Brown, J; Gerrity, T; Folinsbee, LJ. (1995). Comparison of pulmonary
10         responses of asthmatic and nonasthmatic subjects performing light exercise while exposed
11         to a low level of ozone. Toxicol Ind Health 11: 369-385.

12   Islam, T; Berhane, K; McConnell, R; Gauderman, WJ; Avol, E; Peters, JM; Gilliland, FD.
13         (2009). Glutathione-S-transferase (GST) PI, GSTM1, exercise, ozone and asthma
14         incidence in school children. Thorax 64: 197-202.
15         http://dx.doi.org/10.1136/thx.2008.099366

16   Islam, T; McConnell, R;  Gauderman, WJ; Avol, E; Peters, JM; Gilliland, FD. (2008). Ozone,
17         oxidant defense genes and risk of asthma during adolescence. Am J Respir Crit Care Med
18         177: 388-395. http://dx.doi.org/10.1164/rccm.200706-863OC

19   Ito, K; Thurston, GD; Silverman, RA. (2007b). Characterization of PM2.5, gaseous pollutants,
20         and meteorological interactions in the context of time-series health effects models. J Expo
21         Sci Environ Epidemiol 17: S45-S60.

22   Jerrett, M; Burnett, RT; Pope, CA, III; Ito, K; Thurston, G; Krewski, D; Shi, Y; Calle, E; Thun,
23         M. (2009). Long-term ozone exposure and mortality. N Engl J Med 360:  1085-1095.
24         http://dx.doi.org/10.1056/NEJMoa0803894

25   Katsouyanni, K; Samet, JM; Anderson, HR; Atkinson, R; Le Tertre, A; Medina, S; Samoli, E;
26         Touloumi, G; Burnett, RT; Krewski, D; Ramsay, T; Dominici, F; Peng, RD; Schwartz, J;
27         Zanobetti, A. (2009). Air pollution and health: A European and North American approach
28         (APHENA). (Research Report 142). Boston, MA: Health Effects Institute.
29         http://pubs.healtheffects.org/view.php?id=327

30   Kim, CS; Alexis, NE; Rappold, AG; Kehrl, H; Hazucha, MJ; Lay, JC; Schmitt, MT; Case, M;
31         Devlin, RB; Peden, DB; Diaz-Sanchez, D. (2011). Lung function and inflammatory
32         responses in healthy young adults exposed to 0.06 ppm ozone for 6.6 hours. Am J Respir
33         Crit Care Med 183: 1215-1221. http://dx.doi.org/10.1164/rccm.201011-1813OC
                                                 4-60

-------
 1   Lin, S; Bell, EM; Liu, W; Walker, RJ; Kim, NK; Hwang, SA. (2008a). Ambient ozone
 2         concentration and hospital admissions due to childhood respiratory diseases in New York
 3         State, 1991-2001. Environ Res 108: 42-47. http://dx.doi.Org/10.1016/i.envres.2008.06.007

 4   Lin, S; Liu, X; Le, LH; Hwang, SA. (2008b). Chronic exposure to ambient ozone and asthma
 5         hospital admissions among children. Environ Health Perspect 116: 1725-1730.
 6         http://dx.doi.org/10.1289/ehp. 11184

 7   Mar, TF; Koenig, JQ. (2009). Relationship between visits to emergency departments for asthma
 8         and ozone exposure in greater Seattle, Washington. Ann Allergy Asthma Immunol 103:
 9         474-479.

10   Medina-Ramon, M; Schwartz, J. (2008). Who is more vulnerable to die from ozone air
11         pollution? Epidemiology 19: 672-679.

12   Medina-Ramon, M; Zanobetti, A; Schwartz, J. (2006). The effect of ozone and PM10 on hospital
13         admissions for pneumonia and chronic obstructive pulmonary disease:  A national multicity
14         study. Am J Epidemiol 163: 579-588.  http://dx.doi.org/10.1093/aie/kwi078

15   Meng, YY; Rull, RP; Wilhelm, M; Lombard!, C; Balmes, J; Ritz, B. (2010).  Outdoor air
16         pollution and uncontrolled asthma in the San Joaquin Valley, California. J Epidemiol
17         Community Health 64: 142-147. http://dx.doi.org/10.1136/iech.2008.083576

18   Moore, K; Neugebauer, R; Lurmann, F; Hall, J; Brajer, V; Alcorn, S;  Tager,  I.  (2008). Ambient
19         ozone concentrations cause increased hospitalizations for asthma in children:  An 18-year
20         study in Southern California. Environ  Health Perspect 116:  1063-1070.
21         http://dx.doi.org/10.1289/ehp.10497

22   Peden, DB; Boehlecke, B; Horstman, D; Devlin, R. (1997). Prolonged acute  exposure to 0.16
23         ppm ozone induces eosinophilic airway inflammation in asthmatic subjects with allergies. J
24         Allergy Clin Immunol 100: 802-808.

25   Rabinovitch, N; Zhang, LN; Murphy, JR; Vedal, S; Dutton, SJ; Gelfand, EW. (2004). Effects of
26         wintertime ambient air pollutants on asthma exacerbations in urban minority children with
27         moderate to severe disease. J Allergy Clin Immunol 114: 1131-1137.
28         http://dx.doi.0rg/10.1016/i.iaci.2004.08.026

29   Salam, MT;  Islam, T; Gauderman, WJ; Gilliland, FD. (2009). Roles of arginase variants, atopy,
30         and ozone in childhood asthma. J Allergy Clin Immunol 123: 596-602.
31         http://dx.doi.0rg/10.1016/i.iaci.2008.12.020
32   Samet, J.M.  Clean Air Scientific Advisory Committee (CASAC) Response to Charge Questions
33         on the Reconsideration of the 2008 Ozone National Ambient Air Quality Standards.  EPA-
34         CASAC-11-004. March 30, 2011.  Available online at:

                                                  4-61

-------
 1         http://vosetnite.epa.gOv/sab/sabproduct.nsf/0/F08BEB48Cl 139E2A8525785E006909AC/$
 2         File/EPA-CASAC-ll-004-unsigned+.pdf
 3    Schelegle, ES; Morales, CA; Walby, WF; Marion, S; Allen, RP. (2009). 6.6-hour inhalation of
 4         ozone concentrations from 60 to 87 parts per billion in healthy humans. Am J Respir Crit
 5         Care Med 180: 265-272. http://dx.doi.org/10.1164/rccm.200809-1484OC
 6    Schwartz, J. (2005a). How sensitive is the association between ozone and daily deaths to control
 7         for temperature? Am J Respir Crit Care Med 171: 627-631.
 8    Schwartz, J. (2005b). Who is sensitive to extremes of temperature? A case-only analysis.
 9         Epidemiology 16: 67-72. http://dx.doi.org/10.1097/01.ede.0000147114.25957.71
10    Silverman, RA; Ito, K. (2010). Age-related association of fine particles and ozone with severe
11         acute asthma in New York City. J Allergy Clin Immunol  125: 367-373.
12         http://dx.doi.0rg/10.1016/i.iaci.2009.10.061
13    Smith, RL; Xu, B; Switzer, P. (2009b). Reassessing the relationship between ozone and short-
14         term mortality in U.S. urban communities. Inhal Toxicol 21: 37-61.
15         http://dx.doi.org/10.1080/08958370903161612

16    Spektor, DM; Lippmann, M; Lioy, PJ; Thurston, GD; Citak, K; James, DJ; Bock, N; Speizer,
17         FE; Hayes, C. (1988a). Effects  of ambient ozone on respiratory function in active, normal
18         children. Am Rev Respir Dis 137: 313-320.

19    Spektor, DM; Lippmann, M; Thurston, GD; Lioy, PJ; Stecko, J; O'Connor, G; Garshick, E;
20         Speizer, FE; Hayes, C. (1988b). Effects of ambient ozone on respiratory function in healthy
21         adults exercising outdoors. Am Rev Respir Dis 138: 821-828.

22    Strickland, MJ; Darrow, LA; Klein, M; Flanders, WD; Sarnat, JA; Waller, LA; Sarnat, SE;
23         Mulholland, JA; Tolbert, PE. (2010). Short-term associations between  ambient air
24         pollutants and pediatric asthma emergency department visits.  Am J Respir Crit Care Med
25         182: 307-316. http://dx.doi.org/10.1164/rccm.200908-1201OC

26    Tolbert, PE; Klein, M; Peel, JL;  Sarnat, SE; Sarnat, JA. (2007). Multipollutant modeling issues
27         in a study of ambient air quality and emergency department visits in Atlanta. J Expo Sci
28         EnvironEpidemiol 17:  S29-S35. http://dx.doi.org/10.1038/si.ies.7500625

29    US EPA (2012a). Integrated Science Assessment of Ozone and Related Photochemical Oxidants
30         (Third External Review Draft). U.S. Environmental Protection Agency, Washington, DC.
31         EPA/600/R-10/076C. June 2012. Available at:
32         http://www.epa.gOv/ttn/naaqs/standards/ozone/s o3 2008 isa.html

33    US EPA (2012b). Health Risk and Exposure Assessment for Ozone, First External Review
34         Draft. Office of Air Quality Planning and Standards, Research Triangle Park, NC, 27711.
                                                  4-62

-------
 1         EPA 452/P-12-001.  July 2012. Available at:
 2         http://www.epa.gov/ttn/naaqs/standards/ozone/s o3  2008 rea.html
 3
 4   Wenten, M; Gauderman, WJ; Berhane, K; Lin, PC; Peters, J; Gilliland, FD. (2009). Functional
 5         variants in the catalase and myeloperoxidase genes, ambient air pollution, and respiratory-
 6         related school absences: An example of epistasis in gene-environment interactions. Am J
 7         Epidemiol 170: 1494-1501. http://dx.doi.org/10.1093/aie/kwp310
 8   Zanobetti, A; Schwartz, J. (2011). Ozone and survival in four cohorts with potentially
 9         predisposing diseases. Am J Respir Crit Care Med 184: 836-841.
10         http://dx.doi.org/10.1164/rccm.201102-0227OC
11   Zanobetti, A; Schwartz, J. (2008a). Is there adaptation in the ozone mortality relationship: A
12         multi-city case-crossover analysis.  Environ Health 7: 22. http://dx.doi.org/10.1186/1476-
13         069X-7-22
14   Zanobetti, A; Schwartz, J. (2008b). Mortality displacement in the association of ozone with
15         mortality: An analysis of 48 cities in the United States. Am J Respir Crit Care Med 177:
16         184-189. http://dx.doi.org/10.1164/rccm.200706-823OC
17
                                                   4-63

-------
 1         5. CONSIDERATION OF THE WELFARE EFFECTS EVIDENCE

 2          In this chapter, we pose the following overarching question:
 3       "To what extent has scientific information become available that alters or
 4          substantiates our understanding of the welfare effects that occur following
 5          exposures to Os and our understanding of the biologically relevant Os exposures
 6          at which such effects occur?
 7
 8          To inform our consideration of this issue, we consider the weight-of-evidence
 9    conclusions from the ISA (section 5.1); the scientific evidence regarding the mechanisms
10    governing plant responses to 63 exposures (section 5.2); the scientific evidence linking 63
11    exposures to effects on vegetation (section 5.3); the evidence available regarding the biologically
12    relevant aspects of Os exposures important in inducing effects on vegetation (section 5.4); the
13    evidence linking (Vrelated effects on vegetation to those  at the community and whole
14    ecosystem level (section 5.5); the adversity of (Vrelated effects on vegetation and ecosystems in
15    the context of public welfare (section 5.6); and the evidence concerning other welfare effects,
16    such as (Vrelated effects on climate and ultraviolet (UV-B) radiation (section 5.7).

17    5.1    WEIGHT-OF-EVIDENCE CHARACTERIZATION IN THE ISA
18          Since the conclusion of the last review, the Agency has developed a more formal
19    framework for reaching causal inferences from the body of scientific evidence. This framework
20    provides the basis for a robust, consistent, and transparent process for evaluating the scientific
21    evidence, including uncertainties in the evidence, and drawing conclusions and causal judgments
22    regarding air pollution-related welfare effects.  The causality framework and the approach to
23    characterizing the weight of evidence are discussed briefly below (section 5.1.1) and are
24    described in more detail in the ISA (US EPA, 2012a, Preamble). The ISA weight-of-evidence
25    conclusions for Os are summarized in section 5.1.2.
26    5.1.1  Approach to Characterizing the Weight of Evidence
27          Characterization of the weight of evidence in the ISA is based on the evaluation and
28    synthesis of evidence from across scientific disciplines. The relative importance of different
29    types of evidence varies by  pollutant or assessment, as does the availability of different types of
30    evidence for causality determination. Evidence on welfare effects may be drawn from a variety
31    of experimental approaches (e.g., greenhouse, laboratory,  field) and numerous disciplines (e.g.,
32    community  ecology, biogeochemistry and paleological/historical reconstructions) (US EPA
                                                   5-1

-------
 1    2012a, Preamble). Each of these types of studies has strengths and limitations, as discussed
 2    briefly below and in more detail in the ISA (US EPA, 2012a, Preamble).
 3          For ecological effects assessment, both laboratory and field studies (including field
 4    experiments and observational studies) can provide useful data for causality determination.
 5    Because conditions can be controlled in laboratory studies, responses may be less variable and
 6    smaller differences easier to detect. However, the control conditions may limit the range of
 7    responses (e.g., animals may not be able to seek alternative food sources), so they may not reflect
 8    responses that would occur in the natural environment. In addition, larger-scale processes are
 9    difficult to reproduce in the laboratory (US EPA, 2012a, Preamble).
10          Field observational studies  measure biological changes in uncontrolled situations and
11    describe an association between a disturbance and an ecological effect. Field data can provide
12    important information for assessments of multiple stressors or where site-specific factors
13    significantly influence exposure. They are also often useful for analyses of larger geographic
14    scales and higher levels of biological organization.  However, because conditions are not
15    controlled, variability is expected to be higher and differences harder to detect. Field surveys are
16    most useful for linking stressors with effects when stressor and effect levels are measured
17    concurrently.  The presence of confounding factors  can make it difficult to attribute observed
18    effects to specific stressors (US EPA, 2012a, Preamble).
19          Intermediate between laboratory  and field are studies that use environmental media
20    collected from the field to examine response in the laboratory and experiments that  are
21    performed in the natural environment while controlling for some environmental conditions (i.e.,
22    mesocosm studies). This type of study in manipulated natural environments can be considered a
23    hybrid between a field experiment  and laboratory study since some aspects are performed under
24    controlled conditions but others are not.  They make it possible to observe community and/or
25    ecosystem dynamics and provide strong  evidence for causality when combined with findings of
26    studies that have been made under  more  controlled conditions (US EPA, 2012a, Preamble).
27          In considering available evidence from each of these types of studies, the Os ISA draws
28    conclusions within the context of a causality framework with a five-level hierarchy.  This
29    framework is used to classify the overall weight of evidence into one of the following categories:
30    causal relationship, likely to be a causal relationship, suggestive of a causal relationship,
31    inadequate to infer a causal relationship, and not likely to be a causal relationship (US EPA
32    2012a, Preamble  Table 2).  In making such judgments regarding causality, the ISA evaluates
33    several aspects of the evidence including the consistency of effects across studies, the coherence
                                                    5-2

-------
 1    of the evidence across different types of studies, the strength of reported associations,1 and the
 2    biological plausibility of a causal relationship (US EPA, 2012a, Preamble, Table 1). Confidence
 3    increases that Os exposures cause a given welfare effect as the number of consistently supportive
 4    studies increases, as the coherence of the evidence across different types of studies increases, as
 5    the strength of the relationship with Os increases, and as the support for biological plausibility
 6    increases. The ISA also evaluates evidence related to concentration-response and exposure-
 7    response relationships in order to inform conclusions on the concentrations at which effects are
 8    present. Considerations related to weight-of-evidence conclusions and concentration- and
 9    exposure-response relationships are discussed in more detail in the Preamble to the 63 ISA (US
10    EPA, 2012a).

11    5.1.2   Weight-of-Evidence Conclusions for Os

12           Applying the causality framework to 63, the ISA draws the following weight-of-evidence
13    conclusions:

14       •   "Evidence is sufficient to conclude that there is a causal relationship between ambient O3
15           exposure and the occurrence of (Vinduced visible foliar injury on sensitive vegetation
16           across the U.S." (US EPA 2012a, section 9.4.2.2)

17       •   "Evidence is sufficient to conclude that there is a causal relationship between ambient Os
18           exposure and reduced growth of native woody and herbaceous vegetation." (US EPA,
19           2012a, section 9.4.3.2)
20       •   "Evidence is sufficient to infer that there is a causal relationship between Os exposure
21           and reduced productivity, and a likely to be a causal relationship between 63 exposure
22           and reduced carbon sequestration in terrestrial ecosystems." (US EPA, 2012a,  section
23           9.4.3.5)
24       •   "Evidence is sufficient to conclude that there is a causal relationship between Os
25           exposure and reduced yield and quality of agricultural crops." (US EPA, 2012a, section
26           9.4.4.3)
27       •   "Evidence is sufficient to conclude that there is likely to be a causal relationship
28           between 63 exposure and the alteration of ecosystem water cycling."  (US EPA, 2012a,
29           section 9.4.5.1)
30       •   "Evidence is sufficient to infer that there is a causal relationship  between Os exposure
31           and the alteration of below-ground biogeochemical cycles." (US EPA, 2012a, section
32           9.4.6.6)
33       •   "Evidence is sufficient to conclude that there is likely to be a causal relationship
34           between 63 exposure and the alteration of community composition of some ecosystems."
35           (US EPA, 2012a, section 9.4.7.).
36       •   "Evidence supports a causal relationship between changes in tropospheric Os
37           concentrations and radiative forcing." (US EPA, 2012a, section 10.5)
      lrThe strength of the association refers to the magnitude of the association and its statistical precision.

                                                    5-3

-------
 1       •   "Evidence indicates that there is likely to be a causal relationship between changes in
 2           tropospheric O?, concentrations and effects on climate." (US EPA, 2012a, section 10.5)
 3
 4           Given the weight-of-evidence conclusions in the ISA, staffs consideration of welfare
 5    effects linked to 63 exposures focuses on those vegetation and ecosystem-level effects where
 6    evidence is sufficient to conclude a causal relationship, i.e., visible foliar injury (section 5.3.1),
 7    reduced growth and productivity in forest trees and yields of agricultural crops (section 5.3.2),
 8    alteration of below-ground biogeochemical cycling (section 5.5.4) and radiative forcing (5.7.1).
 9    Other effects where evidence is sufficient to conclude a likely to be causal relationship linked to
10    cumulative 63 exposures are also discussed below, such as reduced carbon sequestration in
11    terrestrial ecosystems (section 5.5.2), alteration of ecosystem water cycling (section 5.5.3), the
12    alteration of community composition of some ecosystems (section 5.5.5) and Os  impacts on
13    climate (5.7.1). While not as strongly supported by the evidence, a number of other important
14    welfare effects will be discussed, including Os impacts on insects and other wildlife (section
15    5.5.6) and UVB radiation (5.7.2) (US EPA, 2012a).
16           Figure 5-1 below (US EPA, 2012a, Figure 9-1) is a simplified illustrative diagram of the
17    major pathway through which 63 enters plants and the major endpoints 63 may affect.  There is
18    evidence that the effects observed across this continuum are related to one another; effects of Os
19    at lower levels of organization, such  as the leaf of an individual plant, can result in effects at
20    higher levels. Ozone enters leaves through stomata and can alter stomatal conductance and
21    disrupt CC>2 fixation. These effects can change rates of leaf gas exchange, growth and
22    reproduction at the individual plant level and result in changes in ecosystems, such as
23    productivity, carbon storage, water cycling, nutrient cycling, and community composition (US
24    EPA, 2012a, Section 2.6).  This pathway forms the framework for the discussion of Os-related
25    effects on vegetation and ecosystems, as presented below in sections 5.2 through 5.5.
                                                    5-4

-------
                           03 exposure
                       03 uptake & physiology (Fig 9-2)
                       •Antioxidant metabolism up-regulated
                       •Decreased photosynthesis
                       •Decreased stomatal conductance
                       or sluggish stomatal response
                                                              D
                                                              I
                                                              Q3_

                                                              o
                                                             w

                                                              CD
                                                                        Affected ecosystem services
                                                                        •Decreased productivity
                                                                        •Decreased C sequestration
                                                                        •Altered water cycling (Fig 9-7)
                                                                        •Altered community composition
                                                                        (i.e., plant, insects microbe)
Effects on leaves
•Visible leaf injury
•Altered leaf production
•Altered leaf chemical composition
  Plant growth (Fig 9.8)
  •Decreased biomass accumulation
  •Altered reproduction
  •Altered carbon allocation
  •Altered crop quality
                      Belowground processes (Fig 9.8)
                      •Altered litter production and decomposition
                      •Altered soil carbon and nutrient cycling
                      •Altered soil fauna and microbial communities
      Figure 5-1     An illustrative diagram of the major pathway through which O$
                     enters plants and the major endpoints that O3 may affect in plants
                     and ecosystems. (Figure 9-1 ISA)
 2    5.2    MECHANISMS GOVERNING PLANT RESPONSE TO O3 EXPOSURES

 3          As seen in Figure 5-1 above, (Vinduced impacts on plants occur first at the molecular,

 4    biochemical, and physiological levels. What happens at these levels determines whether these

 5    impacts are translated into effects at higher levels of biological organization at the whole leaf,
 6    plant, community and ecosystem levels. Thus, while the policy relevance of the scientific

 7    understanding regarding the mechanisms governing plant response to O?, is not as obvious, it

 8    adds to the biological plausibility and coherence of the weight-of-evidence for (^-induced
 9    effects at the higher vegetation and ecosystem levels and further informs the  interpretation of

10    predictions of risk associated with vegetation response at ambient O?, exposures. Therefore, a

11    brief discussion of the state of the science regarding the mechanisms of 63 impacts at the

12    molecular, biochemical, and physiological levels is included here.
                                                    5-5

-------
 1          While the fundamental conclusions regarding the mechanisms of (Vinduced vegetation
 2    response have not changed substantially since the completion of the 2006 AQCD, scientific
 3    understanding has continued to expand. Thus, for many of the topics covered under this section,
 4    information from the 2006 Os AQCD is still valid and has been summarized along with more
 5    recent findings.  One significant new body of research has provided important new insights on
 6    the changes in gene expression in plants exposed to elevated 63, due in part to the advent of new
 7    technologies. However, because these studies often use model organisms or mutants or
 8    transgenic plants and exposure conditions that may  not reflect ambient field conditions (US EPA,
 9    2012a, section 9.3.1), additional work remains to elucidate whether these plant responses are
10    transferable to other plant species exposed to more realistic ambient conditions (US EPA, 2012a,
11    section 9.3.6). These new findings continue to refine and enhance our understanding of how
12    exposure to and uptake of 63 can initiate a cascade of responses within the plant, that, upon
13    reaching sufficient magnitude , lead to an array of whole plant effects, including those
14    considered injury and/or damage.
15          The remainder of this section briefly describes the complex cascade of reactions and
16    processes that link ambient 63 to injury and/or damage at the whole plant level. While there is
17    some overlap between the different organizational levels due to multiple interactions and
18    feedbacks, the latest science will be discussed in the following sections: uptake of 63 into the
19    leaf (section 5.2.1); cellular to systemic response (section 5.2.2); detoxification (section 5.2.3)
20    and effects on plant metabolism (section 5.2.4).
21    5.2.1     Ozone Uptake into the Leaf
22          Ozone enters the plant through openings in the leaves called stomata.  Because O3 does
23    not typically penetrate the leafs cuticle, it must reach the stomatal openings in the leaf for
24    absorption to occur.  The movement of 63 and other gases such as CC>2 into and out of leaves is
25    controlled by stomatal guard cells that regulate the size of the stomatal apertures.  These guard
26    cells respond to  a variety of internal species-specific factors as well as external site specific
27    environmental factors such as light,  temperature, humidity, CC>2 concentration, soil fertility,
28    water  status, and in some cases, the presence of air pollutants, including Os. These modifying
29    factors (see also discussion of modifying factors in section 5.3.3) produce stomatal conductances
30    that vary between leaves of the same plant, individuals and genotypes within a species as well as
31    diurnally and seasonally. Environmental conditions which promote high rates of gas exchange
32    will favor the uptake of the pollutant by the leaf, while other factors such as boundary layer
33    resistance and the size of the stomatal aperture may limit uptake (US EPA, 2012a).
34          Ozone-induced changes in stomatal conductance have been reviewed in detail in previous
35    Os AQCDs. The findings summarized in those documents demonstrate that stomatal conductance
                                                   5-6

-------
 1    is often reduced in plants exposed to 63. Results from recent studies support this understanding
 2    (US EPA, 2012a, section 9.3.2 and 9.3.6).  The reduction in stomatal conductance can occur as
 3    the result of either direct impacts of Os on the stomatal complex which causes closure or from a
 4    number of indirect effects of 63 on cellular functions, including a decrease in carboxylation
 5    efficiency that leads to a buildup of CC>2 in the substomatal cavity.  Impacts on the stomatal
 6    complex is important not only because of its affect on a plant's response to 63 but because it can
 7    also compromise the ability of the stomata to respond to other environmental stimuli, including
 8    light, CC>2 concentration and drought.  Stomatal sluggishness has been described as a delay in
 9    stomatal response to changing environmental conditions in sensitive species exposed to higher
10    concentrations and/or longer-term Os exposures (US EPA, 2012a, section, 9.3.2). Such stomatal
11    sluggishness has been suggested as a possible mechanism for Os-induced changes in plant water
12    use efficiency (see section 5.5.3 below).
13    5.2.2  Cellular to Systemic Response
14          After 63 enters the leaf, it can react with a variety of biochemical compounds, either that
15    are exposed to the air or after it is solubilized into the water lining the cell wall. Experimental
16    evidence suggests that there are likely several different mechanisms by which the plant detects
17    the presence of 63 or its breakdown products, including a change in the redox state of the plant
18    and the oxidation of sensitive molecules. One known signaling molecule is the reactive oxygen
19    species (ROS) hydrogen peroxide (H2O2). The presence of higher-than-normal levels of H2O2
20    within the leaf is a potential trigger for a set of metabolic reactions that include those typical of
21    the well documented "wounding" response or pathogen defense pathway generated by cutting of
22    the leaf or by pathogen/insect attack.  Ethylene is another compound produced when plants are
23    subjected to biotic or abiotic stressors  (US EPA, 2012a, section 9.3.3). Calcium and protein
24    kinases are likely involved in relaying information about  the presence of the stressor to the
25    nucleus and other cellular compartments as a first step in  determining whether and how the plant
26    will respond to the stress (US EPA, 2012a, section 9.3.3).
27          The advent of DNA microarray technology has allowed for the study of gene expression
28    in cells on a large scale, providing a comprehensive picture of simultaneous alterations in gene
29    expression.  These studies have provided more insight into the  complex interactions between
30    molecules, how those interactions lead to the communication of information in the cell (or
31    between neighboring cells), and which role these interactions play in determining tolerance or
32    sensitivity and how a plant may respond to stresses such as 63 (US EPA, 2012a. section 9.3.3).
33          Genes involved in plant defense,  signaling and those  associated with the synthesis of
34    plant hormones and secondary metabolism were generally upregulated, while those related to
35    photosynthesis and general metabolism were typically downregulated in Os-treated plants.
                                                    5-7

-------
 1    Proteome studies support these results by demonstrating concomitant increases or decreases in
 2    the proteins encoded by these genes (US EPA, 2012a, section 9.3.6). Gene and protein
 3    expression patterns generally differ between Os-sensitive and tolerant plants, which could result
 4    from differential uptake or detoxification of 63 or from differential regulation of the
 5    transcriptome and proteome (US EPA, 2012a. section 9.3.3).  Finally, plant hormones, including
 6    ethylene (ET), salic acid (SA), and jasmonic acid (JA), can play an important role in determining
 7    plant response to O3, as demonstrated in many studies. More recent studies have supported these
 8    conclusions (US EPA, 2012a, section 9.3.3).
 9          While much work remains to be done to better elucidate how plants detect Os, what
10    determines their sensitivity to the pollutant and how they might respond to it, it is clear that the
11    mechanism for O3 detection and signal transduction is very complex. Many of the
12    phytohormones and other signaling molecules thought to be involved in these processes are
13    interactive and depend upon a variety of other factors, which could be either internal or external
14    to the plant.  This results in a highly dynamic  and complex system, capable of resulting in a
15    spectrum of plant sensitivity to oxidative stress and generating a variety of plant responses to that
16    stress (US EPA, 2012a, section 9.3.3). These mechanisms may vary by species or
17    developmental stage of the plant or may co-exist and be activated by different exposure
18    conditions.
19    5.2.3 Detoxification
20          Ozone injury will not occur if (1) the rate and  amount of Os uptake is small enough for
21    the plant to detoxify or metabolize 63 or its metabolites or (2) the plant is able to repair or
22    compensate for the O3 impacts. Because plants are exposed to an oxidizing environment on a
23    continual basis, and because many reactions that are part of the basic metabolic processes, such
24    as photosynthesis and respiration, generate ROS, there are extensive and complex mechanisms in
25    place to detoxify these oxidizing radicals, including both enzymes and metabolites, which are
26    located in several locations in the cell and also in the apoplast of the cell (US EPA, 2012a,
27    section 9.3.4).
28          Antioxidant metabolites and enzymes located in the apoplast, including ascorbate, are
29    thought to form a first line of defense by detoxifying Os and/or the ROS that are formed as
30    breakdown products of Os (US EPA, 2012a, section 9.3.2). However, the pattern of changes in
31    the amounts of these antioxidants, including ascorbate, varies greatly among  different species
32    and conditions, and the redox buffering capacity of the apoplast is far less than that of the
33    cytoplasm, as it lacks the regeneration systems necessary to retain a reduced pool of antioxidants.
34    Thus, it is not only the quantity and types of antioxidant enzymes and metabolites present but
35    also the cellular ability to regenerate those antioxidants efficiently that play a large role in
                                                    5-8

-------
 1    determining the plant's ability to effectively protect itself from sustained exposure to oxidative
 2    stress (US EPA, 2012a, section 9.3.4).
 3          In spite of the new research, however, it is still not clear as to what extent detoxification
 4    protects against Os injury. It is likely that the role of antioxidants and their interaction with other
 5    plant responses to 63, such as the activation of signal transduction pathways, is far more
 6    complex than is currently understood.  In addition, it has been hypothesized that alterations in
 7    carbon metabolism would be necessary to supply the needed reducing power for antioxidant
 8    regeneration (Dizengremel et al., 2008) (US EPA, 2012a, section 9.3.4). Thus, the 2006 AQCD
 9    conclusions regarding the need for further investigation on these interactions, including whether
10    generation of these antioxidants in response to (Vinduced stress potentially diverts resources
11    and energy away from other vital uses, remain appropriate.
12          Once 63 injury has occurred in leaf tissue, some plants are able to repair or compensate
13    for the impacts. In general, plants have a variety of compensatory mechanisms for low levels of
14    stress including reallocation of resources, changes in root/shoot ratio, production of new tissue,
15    and/or biochemical shifts, such as increased photosynthetic capacity in new foliage and changes
16    in respiration rates, indicating possible repair or replacement of damaged membranes or
17    enzymes. Since these mechanisms are genetically determined, not all plants have the same
18    complement of compensatory mechanisms or degree of tolerance, and these may vary over the
19    life of the plant as not all stages of a plant's development are equally sensitive to 03. At higher
20    levels or over longer periods of 63 stress, some of these compensatory mechanisms, such as a
21    reallocation of resources  away from storage in the roots in favor of leaves or shoots, could occur
22    at a cost to the  overall health of the plant. However, it is not yet clear to what degree or how the
23    use of plant resources for repair or compensatory processes affects the overall carbohydrate
24    budget or subsequent plant response to 63 or other stresses (US EPA, 1996, US EPA, 2006).
25    5.2.4  Effects on Plant  Metabolism
26          Ozone inhibits photosynthesis, the process by which plants produce energy rich
27    compounds (e.g., carbohydrates) in the leaves.  This impairment can result from direct impact to
28    chloroplast function and/or (Vinduced stomatal closure resulting in reduced uptake of CC>2. The
29    2012 ISA states that "[t]he 2006  O3 AQCD described the mechanism by which plant exposure to
30    Os reduces the  quantity of the central carboxylating enzyme, Rubisco, that plays an important
31    role in the production of  carbohydrates within the chloroplast. Recent studies, including those
32    evaluating Os induced changes in the transcriptome and proteome of several different species,
33    confirm these findings" (US EPA, 2012a, section 9.3.5). While several measures of the light
34    reactions of photosynthesis are sensitive to exposure to O3, photosynthetic carbon assimilation is
35    generally considered to be more affected by pollutant exposure, resulting in an overall decline in
                                                   5-9

-------
 1    photosynthesis.  Experimental evidence suggests that both decreases in Rubisco synthesis and
 2    enhanced degradation of the protein contribute to the measured reduction in its quantity (US
 3    EPA, 2006). As discussed in the 2006 AQCD, several studies have found that O3 has a greater
 4    effect as leaves age, with the greatest impact of 63 occurring on the oldest leaves. The loss of
 5    this key enzyme as a function of increasing O3 exposure is also linked to an early senescence or a
 6    speeding up of normal development leading to senescence (US EPA, 2006).
 7           Due to its central importance, any decrease in Rubisco may have severe consequences for
 8    the plant's productivity, including reductions in biomass and yield. Proteomics studies have also
 9    confirmed the effects of O3 on proteins involved in carbon assimilation (US EPA, 2012a, section
10    9.3.5). If total plant photosynthesis is sufficiently reduced, the plant will respond by reallocating
11    the remaining carbohydrate at the level of the whole organism, typically away from the roots and
12    into above ground vegetative components (US EPA, 2012a, section 9.4.3).

13    5.3     EFFECTS ON VEGETATION
14           Ozone injury at the cellular level can accumulate sufficiently to induce effects at the level
15    of a whole leaf or plant. These larger scale effects can include: visible foliar injury and/or
16    premature senescence (section 5.3.1); reduced carbohydrate production, carbohydrate
17    reallocation, reduced growth, reduced reproduction and yield, and reduced plant vigor (section
18    5.3.2).  Section 5.3.3 then describes factors that are known to modify these responses. Much of
19    what is now known about these O3-related effects, as summarized below, is based on research
20    that was available in the 1997 and 2008 reviews.  Studies available in this rulemaking continue
21    to support and expand this knowledge (US EPA, 2012a).
22    5.3.1 Visible Foliar Injury and Biomonitoring
23           Cellular injury to leaves due to exposure to O3 can and often does become visible. While
24    both visible and non-visible injury can be significant to the plant, this section focuses on visible
25    foliar injury for  several reasons: it is often associated with additional public welfare impacts such
26    as impaired aesthetics in specially designated protected areas (such as federal Class I areas) and
27    impaired marketability of O3-sensitive ornamental species; it is easily observed and measured in
28    the field, making some species useful as indicators in biomonitoring networks.  Thus, the
29    remainder of this section uses the terms injury or foliar injury to refer to visible foliar injury.
30           Acute foliar injury usually appears within 24 hours after exposure to O3 and, depending
31    on species, can occur under a range of exposures and durations, whereas chronic foliar  injury
32    may be mild to severe. In  some  cases, cell death or premature leaf senescence may occur.
33    Because the significance of O3-induced injury at the leaf and whole plant levels depends on how
                                                   5-10

-------
 1    much of the total leaf area of the plant has been affected, as well as the plant's age, size,
 2    developmental stage, and degree of functional redundancy among the existing leaf area, it is not
 3    presently possible to determine, with consistency across species and environments, what degree
 4    of injury at the leaf level has significance to the vigor of the whole plant (72 FR 37886/7).
 5          Typical visible injury symptoms on broad-leaved plants include stippling, flecking,
 6    surface bleaching, bifacial necrosis, pigmentation (e.g., bronzing), chlorosis, and/or premature
 7    senescence. Typical visible injury symptoms for conifers include chlorotic banding, tip burn,
 8    flecking, chlorotic mottling, and/or premature senescence of needles. Although common patterns
 9    of injury develop within a species, these foliar lesions can vary considerably between and within
10    taxonomic groups.  Several pictorial atlases and guides have been published, providing details on
11    diagnosis and identification of O3-induced visible foliar injury on many plant species throughout
12    North America (Flagler, 1998; NAPAP, 1987)  and Europe (Innes et al., 2001; Sanchez et al.,
13    2001) (US EPA, 2012a, section 9.4.2).
14          The use of sensitive plants as biological indicators to detect phytotoxic levels of 63 is a
15    longstanding and effective methodology (Chappelka and Samuelson, 1998; Manning and Krupa,
16    1992). Each bioindicator exhibits typical Os injury symptoms when exposed under appropriate
17    conditions. These symptoms are considered diagnostic as they have been verified in exposure-
18    response studies under experimental conditions.  Since the 2006 Os AQCD, new sensitive plant
19    species have been identified from field surveys and verified in controlled exposure studies and
20    several multiple-year field surveys have also been conducted at National Wildlife Refuges in
21    Maine, Michigan, New Jersey, and South Carolina (US EPA, 2012a, section 9.4.2).
22          The USD A Forest Service through the Forest Health Monitoring Program (FHM) (1990 -
23    2001) and currently the Forest Inventory and Analysis (FIA) Program has been collecting data
24    regarding the incidence and severity of visible foliar injury on a variety of (Vsensitive plant
25    species throughout the U.S. (Coulston et al., 2003; Smith et al., 2003; Campbell et al. 2007).
26    The plots where these data are taken are known as biosites. These biosites are located
27    throughout the country, and analysis of visible  foliar injury within these sites follows a set of
28    established protocols. For more details, see http://www.nrs.fs.fed.us/fia/topics/ozone/ (USDA,
29    2011). The network has provided evidence of Os concentrations high enough to induce visible
30    symptoms on sensitive vegetation. From  repeated observations and measurements made over a
31    number of years, specific patterns of areas experiencing visible Os injury symptoms can be
32    identified (US EPA, 2012a, section 9.4.2).
33          When considering the use of visible foliar injury information in a policy context, there
34    are two important caveats. First, visible foliar injury occurs only when sensitive plants are
35    exposed to elevated Os concentrations in a predisposing environment. A major modifying factor
                                                   5-11

-------
 1    for O3-induced visible foliar injury is the amount of soil moisture available to a plant during the
 2    year that the visible foliar injury is being assessed.  This is because lack of soil moisture
 3    generally decreases stomatal conductance of plants and, therefore, limits the amount of O3
 4    entering the leaf that can cause injury (Matyssek et al., 2006; Panek, 2004; Grulke et al., 2003a;
 5    Panek and Goldstein, 2001; Temple et al.,  1992; Temple et al., 1988). Consequently, many
 6    studies have shown that dry periods in local areas tend to decrease the incidence and  severity of
 7    O3-induced visible foliar injury;  therefore, the incidence of visible foliar injury is not always
 8    higher in years and areas with higher O3, especially with co-occurring drought (Smith et al.,
 9    2003).  Other factors such as leaf age influence the severity of symptom expression with older
10    leaves showing greater injury severity as a result of greater seasonal  exposure (Zhang et al.,
11    2010) (US EPA, 2012, section 9.4.2).  Thus, the degree and extent of visible foliar injury
12    development varies from year to year and site to site (Orendovici-Best et al., 2008; Chappelka et
13    al., 2007; Smith et al., 2003), even among co-members of a population exposed to similar O3
14    levels, due to the influence of co-occurring environmental and genetic factors. Second, direct
15    links between O3-induced visible foliar injury symptoms and other adverse effects are also not
16    always found, so that it is not always a reliable indicator of the potential for other negative plant
17    effects occurring (US EPA, 2012, section 9.4.2), because other effects (e.g., biomass loss) have
18    been reported with and without the presence of visible injury.
19           As stated above, however, the presence of O3-induced visible symptoms alone can
20    represent an important adverse impact to the public welfare.  Specifically, it can reduce the
21    appearance and market value of  ornamentals (such as petunia, geranium, and poinsettia) used in
22    urban landscapes, and affect the  aesthetic value  of scenic vistas in protected natural areas such as
23    national parks and wilderness areas. Many businesses rely on healthy-looking vegetation for
24    their livelihoods (e.g., horticulturalists, landscapers, Christmas tree growers, farmers of leafy
25    crops),  and a variety of ornamental species have been listed as sensitive to O3 (Abt Associates
26    Inc., 1995). Though not quantified, there is likely some level of economic impact to businesses
27    and homeowners from O3-related injury on sensitive ornamental species due to the cost
28    associated with more frequent replacement and/or increased maintenance (fertilizer or pesticide
29    application). In addition, because O3 not only results in discoloration of leaves but can  also lead
30    to more rapid senescence (early  shedding of leaves) there potentially could be some lost tourism
31    revenue at sites where fall foliage is less available or attractive (72 FR 37887).
32           In summary, visible foliar injury resulting from exposure to O3 has been well
33    characterized and documented over several decades of research on many tree, shrub,  herbaceous,
34    and crop species (US EPA, 2012a, 2006, 1996, 1984, 1978a). Ozone-induced visible foliar
35    injury symptoms on certain bioindicator plant species are considered diagnostic as they have

                                                   5-12

-------
 1    been verified experimentally in exposure-response studies, using exposure methodologies such
 2    as continuous stirred tank reactors (CSTRs), OTCs, and free-air fumigation. Experimental
 3    evidence has clearly established a consistent association of visible injury with O?, exposure, with
 4    greater exposure often resulting in greater and more prevalent injury. Since the 2006 63 AQCD,
 5    results of several multi-year field surveys of Os-induced visible foliar injury at National Wildlife
 6    Refuges in Maine, Michigan, New Jersey, and South Carolina have been published. New
 7    sensitive species showing visible foliar injury continue to be identified from field surveys and
 8    verified in controlled exposure studies (US EPA, 2012a, section 9.4.2).
 9    5.3.2  Growth, Productivity and Carbon Storage
10          Studies published since the 2008 review continue to support the conclusions of previous
11    AQCDs (US EPA, 1996 and 2006) that there is strong and consistent evidence that ambient
12    concentrations of 63 decrease photosynthesis and growth in numerous plant species across the
13    U.S. (US EPA, 2012a, section 9.4.3; section 5.2.4 above).  Regarding photosynthesis, one recent
14    meta-analysis of 55 studies (Wittig et al., 2007) reported that current 63 concentrations in the
15    northern hemisphere are decreasing stomatal conductance by 13% and photosynthesis by 11%
16    across tree species, and that decreases in photosynthesis are consistent with the cumulative
17    uptake of 63 into the leaf (US EPA, 2012a, section 9.4.3).  As discussed in section 5.2.4 above
18    and in the ISA (US EPA, 2012a, section 9.3.5), there are numerous mechanisms that provide the
19    biological plausibility for O?, leading to decreases in photosynthesis.
20          As expected, reductions in photosynthesis are often followed by observed reductions in
21    growth.  In a recent study, McLaughlin et al., (2007a) investigated the effects of ambient 63 on
22    tree growth at forest sites in the southern Appalachian Mountains. The authors reported that the
23    cumulative effects of ambient levels of Os decreased seasonal stem growth by 30-50% for most
24    tree species in a high 63 year, compared to a low 63 year.  Another field- based gradient study
25    documented Os-induced growth effects on eastern cottonwood.  This study placed eastern
26    cottonwood (Populus deltoides) saplings at sites along a continuum of ambient Os exposures that
27    gradually increased from urban to rural areas in the New York City area (Gregg et al., 2003).
28    Eastern cottonwood is a fast growing Os-sensitive tree species that is important ecologically
29    along streams and commercially for pulpwood, furniture manufacturing, and as a possible new
30    source for energy biomass (Burns and Hankola, 1990). Gregg et al. (2003) found that the
31    cottonwood saplings grown in urban New York City grew faster than saplings grown in
32    downwind rural areas. Because these saplings were grown in pots with carefully controlled soil
33    nutrient and moisture levels, the authors were able to control for most of the differences between
34    sites. After carefully considering these and other  factors, the authors concluded the primary
35    explanation for the difference in growth was the gradient of cumulative 63 exposures that

                                                   5-13

-------
 1   increased as one moved downwind from urban to less urban and more rural sites.  It was
 2   determined that the lower Os exposure within the city center was due to NOX titration reactions
 3   which removed Os from the ambient air. This study found results of a similar magnitude as
 4   those previously seen in OTC studies but without the use of chambers or other fumigation
 5   methods (Gregg et al., 2003). The authors were able to reproduce the growth responses observed
 6   in the field in a companion OTC experiment, confirming Os as the stressor inducing the growth
 7   loss response (US EPA, 2006).
 8          A set of field-based studies published since the 2006 AQCD employed the modified Free
 9   Air CC>2 Enrichment (FACE) methodology to expose vegetation in a forest in Wisconsin to
10   elevated 63 without the use of chambers.  This exposure method more closely replicates
11   conditions in the field than do OTCs. Over the first seven years of stand development at the
12   Aspen FACE site, Kubiske et al. (2006) observed that elevated Os decreased tree heights,
13   diameters, and main stem volumes in the aspen community by 11, 16, and 20%, respectively.  In
14   addition, Kubiske et al. (2007) reported that elevated Os may change intra- and inter-species
15   competition. For example, Os treatments increased the rate of conversion from a mixed aspen-
16   birch community to a birch dominated community.
17          In previous AQCDs, the majority of evidence of 63 growth effects on trees and other
18   vegetation and yield loss in crops came from open-top-chamber (OTC) studies such as those
19   conducted at the NUEERL-WED research site in Corvallis, Oregon and those conducted through
20   the National Crop Loss Assessment Network (NCLAN) program, respectively. Despite the
21   robustness of the concentration-response (C-R) functions developed using OTC, the debate over
22   their applicability in the field has persisted.
23          To further evaluate this issue, EPA staff identified two FACE projects which had
24   sufficient data that could support an examination of the predictive capability of C-R functions
25   generated in OTCs when applied under field exposure  conditions. Specifically, the AspenFACE
26   and SoyFACE projects were conducted over multiple years and the hourly O?, data were
27   available. There are necessary differences in the modeling of exposure-response in annual plants
28   such as  soybean, and in perennial plants such as aspen trees, when exposure takes place over
29   multiple years.  These differences and the needed adjustments to the comparative analyses are
30   taken into account in the analyses and discussed  in the ISA (US EPA, 2012a, section 9.6.3). The
31   EPA staff first conducted a comparison between yield of soybean as predicted by the composite
32   function three-step process (ISA, section 9.6.2) using NCLAN data, and observations of yield in
33   SoyFACE.  The median composite function for relative yield was derived for the NCLAN
34   soybean Weibull functions for non-droughted studies, and comparisons between the predictions
35   of the median composite and SoyFACE observations were conducted. For the years 2007 and

                                                 5-14

-------
 1   2008, SoyFACE yield data were available for 7 and 6 genotypes, respectively. Those data were
 2   used to compare the relative change in yield observed in SoyFACE in a given year between
 3   ambient Os and elevated Os, versus the relative change in yield predicted by the NCLAN-based
 4   median composite function between those same two values of 63 exposure. Since comparisons
 5   of absolute values were thought to also be of interest, the predictive functions were also scaled to
 6   the observed data.

 7          As discussed in section 9.6.3.2 of the ISA (US EPA, 2012a), the agreement between
 8   predictions based on NCLAN data and SoyFACE observations was notably close in single-year
 9   comparisons. Together with the very high agreement between median composite models for
10   NCLAN and SoyFACE, it provides very strong mutual confirmation of those two projects'
11   results with respect to the response of yield of soybeans to Os exposure.  It is readily apparent
12   from these results that the methodology described in Section 9.6.2 of the ISA for obtaining
13   predictions of yield or yield loss from NCLAN data is strongly validated by SoyFACE results.
14   As described in section 9.2 of the ISA, the exposure technologies used in the two projects were
15   in sharp contrast, specifically with respect to the balance each achieved between control of
16   potential interacting factors or confounders, and fidelity to natural conditions. The comparisons
17   that EPA conducted therefore demonstrate that the methodology used in developing the
18   composite functions is resistant to the influence of nuisance variables and that predictions are
19   reliable.  They may also suggest that the aspects in which the two exposure technologies differ
20   have less influence on exposure-response than initially supposed.  These results are also in
21   agreement with comparative studies reviewed in section  9.2.6 of the ISA (US EPA, 2012a)

22          In a second analysis, EPA staff compared relative and absolute above-ground biomass
23   loss at ambient and elevated exposures observed in AspenFACE for aspen with the
24   NHEERL/WED-OTC C-R function derived predictions.  EPA found that effects on biomass
25   accumulation in aspen during the first seven years closely agreed with the exposure-response
26   function  based on data from earlier OTC experiments (US EPA, 2012a, section 9.6.3).  As in the
27   comparisons between NCLAN and SoyFACE, the agreement between predictions based on
28   NHEERL/WED data and Aspen FACE observations was very  close.  The results of the two
29   projects strongly reinforce each other with respect to the response of aspen biomass to 63
30   exposure. The methodology used for obtaining the median composite function is shown to be
31   capable of deriving a predictive model despite potential confounders, and despite the added
32   measurement error that is expected from calculating biomass using allometric equations. In
33   addition, the function based on one year of growth was shown  to be applicable to subsequent
34   years (US EPA, 2012a, section 9.6.3.2).
                                                  5-15

-------
 1          As discussed in section 9.6.3.2 of the ISA (US EPA, 2012a), the results of experiments
 2    that used different exposure methodologies, different genotypes, locations, and durations
 3    converged to the same values of response to Os exposure for each of two very dissimilar plant
 4    species, and predictions based on the earlier experiments were validated by the data from current
 5    ones. However, in these comparisons, the process used in establishing predictive functions
 6    involved aggregating data over variables such as time, locations, and genotypes,  and the use of a
 7    robust statistic (quartiles) for that aggregation.  The validating data, from SoyFACE and Aspen
 8    FACE, were in turn aggregated over the same variables.  The accuracy of predictions is not
 9    expected to be conserved for individual values of those variables over which aggregation
10    occurred.  For example, the predicted values for soybean, based on data for five genotypes, are
11    not expected to be valid for each genotype separately. As shown in the validation, however,
12    aggregation that occurred over different values of the same variable did not affect accuracy:
13    composite functions based on one set of genotypes were predictive for another set, as long as
14    medians were used for both sets.

15          Further support for the coherence of 63 growth effects across numerous studies and
16    species that used a variety of experimental techniques comes from a recent meta-analysis of peer
17    reviewed studies from the last 40 years (Wittig et al., 2009). This meta-analysis  found that
18    current ambient 63 concentrations as reported in those studies (with  an average of 40 ppb)
19    significantly decreased annual total biomass growth  (7%) across 263 studies (US EPA, 2012a,
20    section 9.4.3).  The  decreased growth effect was reported to be greater (11 to 17%) in elevated
21    O3 exposures (with  an average 97 ppb) (Wittig et al., 2009) (US EPA, 2012a, section 9.4.3).

22          Thus, on the basis of the combined evidence  from the SoyFACE and AspenFACE/OTC
23    comparisons and the results of the field-based studies discussed above, the ISA concludes that
24    additional compelling and important support is provided for the continued use of the C-R
25    functions derived using OTC from the NCLAN and NHEERL-WED studies to estimate risk to
26    crops and tree seedlings under ambient field exposure conditions.  These results strengthen our
27    understanding of Os effects on studied crop and forest species and make a significant
28    contribution to the coherence of the weight of evidence available in this review.  They also
29    further demonstrate the relevance of the knowledge gained from trees grown in OTC and provide
30    additional evidence that Os-induced effects observed in chambers  can be anticipated to occur in
31    unmanaged systems in the field.
32          In addition to reduced growth, some plants adapt to Os-induced reduction in
33    photosynthesis by reallocating the remaining carbohydrate at the level of the whole organism.
34    Many studies have demonstrated that Os reduces C allocation to roots (US EPA,  2012a, section
                                                  5-16

-------
 1    9.4.3). For example, in one meta-analysis, Grantz et al. (2006) estimated the effect of 63 on the
 2    root: shoot allometric coefficient (k), the ratio between the relative growth rate of the root and
 3    shoot. The results showed that Os reduced the root:shoot allometric coefficient by 5.6%, and the
 4    largest decline of the root: shoot allometric coefficient was observed in slow-growing plants.
 5    Further, Vollsnes et al. (2010) studied the in vivo root development of subterranean clover
 6    (Trifolium subterraneum) before, during and after short-term Os exposure.  It was found that Os
 7    reduced root tip formation, root elongation, the total root length, and the ratios between below-
 8    and above-ground growth within one week after exposure. Those effects persisted for up to three
 9    weeks; however, biomass and biomass ratios were not significantly altered  at the harvest five
10    weeks after exposure.
11          When  fewer carbohydrates are present in the roots, less energy is available for root-
12    related functions such as acquisition of water and nutrients.  Thus, by inhibiting photosynthesis
13    and the amount of carbohydrates available for transfer to the roots, 63 can impact below ground
14    processes. These below-ground changes could signal a shift in nutrient cycling with significance
15    at the ecosystem level (Young and Sanzone, 2002) (see discussion in  section 5.5.4 below).
16    These below ground effects have been documented in the field.  Data from  a long-studied
17    pollution  gradient in the San Bernardino Mountains of southern California suggest that Os
18    substantially reduces root growth in natural stands of ponderosa pine (Pinusponderosd). Root
19    growth in mature trees was decreased at least 87 percent in a high-pollution site as compared to a
20    low-pollution  site (Grulke et al., 1998), and a similar pattern was found in a separate study with
21    whole-tree harvest along this gradient (Grulke and Balduman, 1999).  Though effects on other
22    ecosystem components were not examined, a reduction of root growth of this magnitude could
23    have significant implications for the below-ground communities at those sites. Therefore,
24    studies that examine only above-ground vegetative components may miss important (Vinduced
2 5    change s b el ow ground.
26          Trees and other perennials, in addition to cumulating the effects of O3 exposures over the
27    annual growing season, can also cumulate effects across multiple years. It has been reported that
28    effects can "carry over" from one year to another (US EPA, 2006). Growth affected by a
29    reduction in carbohydrate storage in one year may result in the limitation of growth in the
30    following year (Andersen, et al., 1997) (US EPA, 2012a, section 9.4.8). Carry-over effects have
31    been documented in the growth of some tree seedlings and in roots. More recent studies at the
32    Aspen FACE  site have also reported effects on paper birch and aspen that likely have
33    implications for the subsequent growing season.  Specifically, scientists found that elevated 63
34    decreased birch seed weight, germination, and bud starch levels as well as aspen bud  size.  The
35    effects on birch seeds could lead to a  negative impact on species regeneration, while the bud

                                                   5-17

-------
 1    effects may have been related to the observed delay in spring leaf development and have the
 2    potential to alter carbon metabolism of overwintering buds, which may have carry-over effects in
 3    the following year (US EPA, 2012a, section 9.4.3). However, because most studies on the
 4    effects of 63 on growth do not take into account the possibility of carry-over effects on growth in
 5    subsequent years, the true implication of these annual biomass losses may be missed. It is likely
 6    that under ambient exposure conditions, some sensitive trees and perennial plants could
 7    experience compounded impacts that result from multiple year exposures.
 8          The detrimental effect of 63 on crop production has been recognized since the 1960s, and
 9    a large body of research has subsequently stemmed from those initial findings. Previous Os
10    AQCDs have extensively reviewed this body of literature (US EPA, 2006).  Current 63
11    concentrations across the U. S. are high enough to cause yield loss for a variety of agricultural
12    crops including, but not limited to, soybean, wheat, potato, watermelon, beans, turnip, onion,
13    lettuce, and tomato (US EPA, 2012a, section 9.4.4).
14          New research is beginning to consider the mechanism of damage caused by prolonged,
15    lower O3 concentrations (so-called chronic exposure) compared to short, very high O3
16    concentration (so-called acute exposure).  Both types of Os exposure cause damage to
17    agricultural crops, but through very different mechanisms. Historically, most research on the
18    mechanism of Os damage used acute exposure studies. During the last decade, it has become
19    clear that the cellular and biochemical processes involved in the  response to acute Os exposure
20    are not involved in response to chronic 63 exposure, even though both cause yield loss in
21    agriculturally important crops (US EPA, 2012a, section 9.4.4). In addition, recent research has
22    highlighted the effects of Os on crop quality. Increasing Os concentration decreases nutritive
23    quality of grasses, decreases macro- and micro-nutrient concentrations in fruits and vegetable
24    crops, and decreases cotton fiber quality.  These areas of research require further investigation to
25    determine  mechanisms and exposure-response relationships (US EPA, 2012a, section 9.4.4).
26    5.3.3  Factors that Modify Plant Response to O3
27          A plant's response to Os  is influenced by the many ambient biotic and abiotic factors in
28    its milieu,  such as insects, pathogens, root microbes and fungi, temperature, water and nutrient
29    availability, and other air pollutants, as well as elevated CC>2. These factors, when present, can
30    potentially exacerbate or mitigate the effects of 03. Thus, observed plant response to Os in the
31    field represents an integration of the plant's response to all the biotic and  abiotic factors to which
32    it is exposed. Because these interactions are species specific and dependent on the particular
33    circumstances of the interact on(s), it is impossible to incorporate the nearly infinite number of
34    possible combinations in any quantitative way into an assessment of vegetation risk. As will be
35    discussed in section 5.4 below, it is also not possible to fully account for these interactions in an
                                                   5-18

-------
 1    exposure index. However, in interpreting the range of risks experienced by vegetation across the
 2    country, it is helpful to understand how plant response to Os can be modified by other biotic and
 3    abiotic factors and how plant response to other factors can be altered by exposure to 03.
 4    Therefore, some of the most important modifying factors are summarized here.
 5          Regarding biotic factors, the most influential is actually within the plant itself: genes.
 6    Plant response to Os is determined by the set of genes that are directly related to oxidant stress
 7    and to an unknown number of genes that are not specifically related to oxidants, but instead
 8    control leaf and cell wall thickness, stomatal conductance, and the internal architecture of the air
 9    spaces (US EPA, 2012a, section 9.4.8). The genetic makeup of each plant therefore confers an
10    inherent sensitivity or tolerance to 63 exposure.
11          Plant response to Os can also be influenced by the presence and type of a disease
12    outbreak or insect infestation.  Conversely, a plant's response to 63 can also alter the degree or
13    severity  of the disease or pest attack.  Ozone can also directly affect the disease or pest species,
14    and the interaction between 63, a plant, and a pest or pathogen, may influence the response of the
15    target host species to O3 (US EPA, 2012a, section 9.4.8; see also discussion in section 5.5.6
16    below).
17          In contrast, there are also mutually beneficial  relationships or symbioses involving higher
18    plants and bacteria or fungi. These include (1) the nitrogen-fixing species Rhizobium and
19    Frcmkia that nodulate the roots of legumes and alder and (2) the mycorrhizae that infect the roots
20    of many crop and tree species, all of which may be affected by exposure of the host plants to Os.
21    In some  cases, these interactions may offer some protection to the host plant from other stresses.
22    In addition, O3 has also been shown to alter soil fauna communities (US EPA,  2012a, section
23    9.4.8; see also discussion in section 5.5.4 below).
24            Intra- and inter-specific competition are also important factors in determining vegetation
25    response to 63.  Plant competition involves the ability of individual plants to acquire the
26    environmental resources needed for growth and development: light, water, nutrients, and space.
27    Intra-specific competition involves individuals of the same species, typically in monoculture
28    crop situations, while inter-specific competition refers to the interference exerted by individuals
29    of different species on each other when they are in a mixed  culture (US EPA, 2012a, section
30    9.4.8).
31          Physical or abiotic factors also play a large role in influencing plant response to Os.
32    These can include temperature, light, relative humidity, soil moisture  conditions and the presence
33    of other  pollutants.  However, the nature of these interactions appears to be complicated, largely
34    species-specific and depends to some extent upon the sequence in which the stressors occur and

                                                    5-19

-------
 1    how closely a plant is to its optimal growing conditions when the change or stress occurs. Most
 2    studies of stressor interactions have just included Os and one other stressor at a time (US EPA,
 3    2012a, section 9.4.8). However, it should be realized that there are situations or scenarios in
 4    which multiple biotic and abiotic conditions change together. For example, under climate
 5    change, shifts in temperature, precipitation, CC>2, Os and insects and/or disease incidence are
 6    expected to be interrelated. Additional research into multiple interactions would help our
 7    understanding of how plants integrate these stressors.

 8    5.4     BIOLOGICALLY RELEVANT EXPOSURE INDICES
 9           The CAA requires that national ambient air quality standards be based on air quality
10    criteria, which "shall accurately reflect the latest scientific knowledge useful in indicating the
11    kind and  extent of all identifiable effects on public health or welfare which may be expected
12    from the presence of such pollutant in the ambient air, in varying quantities." It further states
13    that the criteria "... shall include information on (A) those variable factors (including
14    atmospheric conditions) which of themselves or in combination with other factors, may alter the
15    effects on public health or welfare...." Thus, in order to ensure that national ambient air quality
16    standards provide the requisite degree of public health and welfare protection, it is important that
17    the standard that is set by EPA would control the aspects of pollutant exposure that are known,
18    based  on  available information, to induce the adverse health and/or welfare response. This is
19    done by reviewing the most recent available science that elucidates the aspects of pollutant
20    exposures that are important in inducing an adverse response.  This section summarizes the
21    vegetation exposure and effects science that forms the basis for our understanding of the variable
22    aspects of ambient 63 exposures that have biological relevance (section 5.4.1), exposure indices
23    or metrics that have been designed to reflect or incorporate this science, making them
24    biologically relevant (5.4.2), and past and current  conclusions based on the exposure and effects
25    science, as well as policy considerations, regarding appropriate indices to consider in the context
26    of national standards in past and current reviews (5.4.3).
27    5.4.1   Exposure Factors that Influence Plant Response
28           There  is an extensive body of vegetation effects research from which various factors with
29    known or suspected bearing on the exposure-response relationship have been identified. These
30    factors include concentration, time of day, respite  time, frequency of peak occurrence, plant
31    phenology, predisposition, among others (US EPA, 2012a, section 9.5.2). Much of the  work
32    identifying the aspects of exposure important for plant response had been done by the mid-1990s
33    and was summarized in the 1996 Criteria Document.  In particular, the importance of the
34    duration of the exposure (5.4.1.1) and the relatively greater importance of higher concentrations

                                                    5-20

-------
 1    over lower (5.4.1.2) in determining plant response to Os have been well documented. These
 2    conclusions have been reiterated in the current ISA which states that "[t]he attributes of exposure
 3    indices that are most relevant to plant damage are the weighting of Os concentrations and the
 4    daily and seasonal time-periods" (US EPA, 2012a, section 9.5.1).
 5          5.4.1.1 Relevant exposure periods: diurnal and seasonal
 6          With respect to the importance of exposure duration in explaining plant response, the
 7    1996 AQCD stated, "when O3 effects are the primary cause of variation in plant response, plants
 8    from replicate studies of varying duration showed greater reductions in yield or growth when
 9    exposed for the longer duration" and "the mean exposure index of unspecified duration could not
10    account for the year-to-year variation in response" (US EPA, 1996, pg. 5-96).  Further, "because
11    the mean exposure index treats all concentrations equally and does not specifically include an
12    exposure duration  component, the use of a mean exposure index for characterizing plant
13    exposures appears inappropriate for relating exposure with vegetation effects" (US EPA, 1996,
14    pg. 5-88).  As discussed below,  there is both a diurnal and seasonal component to biologically
15    relevant exposures.
16          Diurnal. The conditions for uptake of 63 into the plant occur mainly during the daytime
17    hours. In general, plants have the highest stomatal conductance during the daytime and in many
18    areas atmospheric  turbulent mixing is greatest during the day as well (Uddling et al., 2010; US
19    EPA, 2006). This  is because the high-temperature and high-light conditions that typically
20    promote the formation of tropospheric Os also promote physiological activity in vegetation (US
21    EPA, 2012a, section 9.5.3).
22          In addition to daytime uptake, several studies have also reported Os uptake at night. As a
23    result, EPA has considered the appropriateness of including nighttime hours in an exposure
24    index, both in the context of the last and current reviews. However, typically, the rate  of
25    stomatal conductance at night is much lower than during the day (Caird et al., 2007).
26    Atmospheric turbulence at night is also often low, which results in stable boundary layers and
27    unfavorable conditions for Os uptake into vegetation (Finkelstein et al., 2000). Notable
28    exceptions to maximum daytime conductance are cacti  and other plants with crassulacean acid
29    metabolism (CAM photosynthesis) which only open their stomata at night. Due to a paucity of
30    information on exposure-response relationships in species with CAM photosynthesis, this section
31    will focus on plants with C3  and C4 photosynthesis, which generally have maximum stomatal
32    conductance during the daytime (US EPA, 2012a, section 9.5.3).
33          For significant nocturnal stomatal flux and Os effects to occur, the right combination of
34    specific conditions must exist.  In particular, a susceptible plant with nocturnal stomatal

                                                   5-21

-------
 1    conductance and low defenses must be growing in an area with relatively high night-time 63
 2    concentrations and appreciable nocturnal atmospheric turbulence. It is unclear how many areas
 3    there are in the U.S. where these conditions occur. It may be possible that these conditions exist
 4    in mountainous areas of southern California, front-range of Colorado (Turnipseed et al., 2009)
 5    and the Great Smoky Mountains of North Carolina and Tennessee. More information is needed
 6    in locations with high night-time 63 to assess the local 63 patterns, micrometeorology and
 7    responses of potentially vulnerable plant species. Work by Mereu et al. (2009) in Italy on
 8    Mediterranean species indicated that nocturnal uptake was from 10 to 18% of total daily uptake
 9    during a weak drought and up to 24%  as the drought became more pronounced. The proportion
10    of night-time uptake was greater during the drought due to decreases in daytime stomatal
11    conductance (Mereu et al., 2009). These studies show that effects can be seen with night-time
12    exposures to 63, but when atmospheric conditions are stable at night, it is uncertain how these
13    exposures may affect plants and trees with complex canopies in the field (US EPA, 2012a,
14    section 9.5.3). Therefore, due to the substantial uncertainties that remain regarding the extent
15    and importance of nocturnal exposures, EPA continues to focus on exposures occurring during
16    daytime/daylight hours.
17           Seasonal. Vegetation across the U.S. has widely varying periods of physiological activity
18    during the year due to variability in climate and phenology.  In order for a particular plant to be
19    vulnerable to O3 pollution, it must have foliage and be physiologically active at the time of
20    exposure. Annual crops are typically grown for periods of two to three months. In contrast,
21    perennial species may be photosynthetically active longer (up to 12 months each year for some
22    species) depending on the species and where it is grown.  In general, the period of maximum
23    physiological  activity and thus, potential Os uptake for vegetation coincides with some or all of
24    the intra-annual period defined as the 63 season, which varies on a state-by-state basis (US EPA,
25    2012a, Figure 3-24). This is because the high-temperature and high-light conditions that
26    typically promote the formation of tropospheric Os also promote physiological activity in
27    vegetation.  There are very limited exceptions to this pattern where Os can form in the winter in
28    areas in the western U.S. with intense  natural gas exploration (Pinto, 2009), but this is typically
29    when plants are dormant and there is little chance of 63 uptake. The selection of any single
30    window of time for a national standard to consider hourly O3 concentrations represents a
31    compromise, given the significant variability in growth patterns and lengths of growing season
32    among the wide range of vegetation species that may experience adverse effects associated with
33    O3 exposure (US EPA, 2012a, section 9.5.3).
34           Various intra-annual averaging and accumulation time periods have been considered for
35    the protection of vegetation.  The 2007 proposal for the secondary Os standard (72 FR 37899)

                                                   5-22

-------
 1   proposed to use the maximum consecutive 3-month period within the 63 season.  Alternatively,
 2   the U.S. Forest Service and federal land managers have used a 24-h W126 accumulated for
 3   6 months from April through September (FLAG, 2000). However, some monitors in the U.S. are
 4   operational for as little as four months and would not have enough data for a 6-month seasonal
 5   window. The exposure period in the vast majority of Os exposure studies conducted in the U.S.
 6   has been much shorter than 6 months.  Most of the crop studies done through NCLAN had
 7   exposures less than three months with an average of 77 days. Open-top chamber studies of tree
 8   seedlings, compiled by the EPA, had an average exposure of just over three months  or 99 days.
 9   In more recent FACE experiments, SoyFACE exposed soybeans for an average of approximately
10   120 days per year, and the Aspen FACE experiment exposed trees to an average of
11   approximately 145 days per year of elevated Os, which included the entire growing season at
12   those particular sites. Despite the possibility that plants may be exposed to ambient 63 longer
13   than 3 months in some locations, there is a lack of exposure experiments conducted for longer
14   than 3 months (US EPA, 2012a, section 9.5.3).
15          In an analysis conducted by ORD/NCEA to test the importance of going beyond the
16   previously recommended 3 month period to capture cumulative 63 exposures over a longer
17   growing season, the 3- and 6-month maximum W126 values were calculated for over 1,200 AQS
18   and CASTNET EPA monitoring sites for the years 2008-2009.  This analysis found that these
19   two accumulation periods resulted in highly correlated metrics (US EPA, 2012a, section 9.5.2,
20   Figure 9-13). The two accumulation periods were centered on the yearly maximum for each
21   monitoring site, and it is possible that this correlation would be weaker if the two periods were
22   not temporally aligned. In the U.S., W126 cumulated over 3 months and W126 cumulated over
23   6 months are proxies of one another, as long as the period in which daily W126 is accumulated
24   corresponds to the seasonal maximum. Therefore, it is expected that either statistic will predict
25   vegetation response equally well. In other words, the strength of the correlation between
26   maximum 3-month W126 and maximum 6-month W126 is such that there is no material
27   difference in their predictive value for vegetation response. (US EPA, 2012a, section 9.5.3)
28          5.4.1.2 Relevant  exposure concentrations: differential weighting
29          It has been clearly shown in the scientific literature that, all else being equal, plants
30   respond more to higher concentrations, though there continues to be no evidence of an exposure
31   threshold for vegetation effects (US EPA, 2012a, section 9.5.3).  Regarding the relative
32   importance of higher concentrations than lower in determining plant response, the 1996 Criteria
33   Document concluded that "the ultimate impact of long-term exposures to Os on crops and
34   seedling biomass response depends on the integration of repeated peak concentrations during the
35   growth of the plant" and further that, "at this time, exposure indices that weight the hourly 63

                                                  5-23

-------
 1    concentrations differentially appear to be the best candidates for relating exposure with predicted
 2    plant response" (US EPA, 1996, pgs.  5-136). While the significant role of peak 63
 3    concentrations was established based  on several experimental studies previously described in the
 4    1996 AQCD, several more recent studies (Oksanen and Holopainen,  2001; Yun and Laurence,
 5    1999; Nussbaum et al., 1995) have added further support for the important role that peak
 6    concentrations, as well as the pattern of occurrence, plays in plant response to 63 (US EPA,
 7    2012a, section 9.5.3).  For example, Oksanen and Holopainen (2001) found that the peak
 8    concentrations and the shape of the Os exposure (i.e., duration of the event) were important
 9    determinants of foliar injury in European white birch saplings, though growth reductions were
10    found to be more related to total cumulative exposure (US EPA, 2012a, section 9.5.3).
11    5.4.2. Biologically Relevant Exposure Metrics
12          In conjunction with the extensive body of vegetation effects research on plant response to
13    Os, "mathematical approaches for summarizing ambient air quality information in biologically
14    meaningful forms for 63 vegetation effects assessment purposes have been explored for more
15    than 80 years" (US EPA, 2012a, section 9.5.2).  A large set of indices have been developed that
16    use a variety of functions to weight factors that have been shown to influence vegetation
17    exposure-response relationships (US EPA, 2012a, section 9.5.2).  As with any summary
18    statistic, exposure indices retain information on some, but not all, characteristics of the original
19    observations. As discussed in the ISA, several indices have been developed to attempt to
20    incorporate some of the biological, environmental, and exposure factors that influence the
21    magnitude of the biological response and contribute to observed variability  (US EPA, 2012a,
22    section 9.5.2).  The resulting indices were evaluated by ranking them according to the goodness-
23    of-fit of a regression model of growth or yield response (Lee et al., 1989).  For example, in an
24    analysis using the National Crop Loss Assessment Network (NCLAN) data, Lee et al. (1988)
25    found several  indices which cumulated and weighted higher concentrations (e.g., W126, SUM06,
26    SUM08, and AOT40) performed very well.  Among this group of indices, no one index had
27    consistently better fits than the other indices across all studies and species (Heagle et al., 1994b;
28    Lefohn et al.,  1988; Musselman et al., 1988). Lee et al. (1988) also found that adding phenology
29    weighting to the index somewhat improved the performance of the indices.  However, because it
30    required data on species and site conditions, it was not considered appropriate for general use.
31    Based on the above science, the attributes of exposure indices that are most relevant to plant
32    damage are the cumulation and weighting of hourly Os concentrations and the diurnal and
33    seasonal time-periods (US EPA, 2012a, section 9.5.2).
34          In past reviews, EPA staff has compared the appropriateness of using different
35    cumulative, peak weighted metrics in the context of selecting a form for a secondary national

                                                  5-24

-------
 1    ambient air quality standard for 63. In particular, the 1996 Staff Paper compared three of these
 2    forms, the SUM06, AOT06, and W126 exposure metrics (US EPA, 1996, pp. 223-227).  Below
 3    are the definitions of the three cumulative index forms previously evaluated in the NAAQS
 4    context:
 5              •   SUM06: Sum of all hourly 63 concentrations greater than or equal to 0.06 ppm
 6                 observed during a specified daily and seasonal time window.
 7              •   AOT06: Sum of the differences between hourly Os concentrations greater than
 8                 the specified threshold of 0.06 ppm during a specified daily and seasonal time
 9                 window.
10              •   W126: Sigmoidally weighted sum of all hourly O3 concentrations observed
11                 during a specified daily and seasonal time window (Lefohn et al., 1988; Lefohn
12                 and Runeckles, 1987).
13          Following completion of the Staff Paper, a consensus workshop was convened to
14    consider the science supporting a range of topics, including biologically relevant indices. In
15    comparing the SUM06 to the W126 form, there was  a consensus that the cumulative
16    concentration-weighted indices being considered were equally capable of predicting plant
17    response (Heck and Cowling, 1997).
18          Based on both its own staff analysis and the workshop consensus statement, the EPA
19    proposed as one alternative in the 1997 review to use the SUM06 form of the secondary NAAQS
20    for 63 to protect vegetation from damage (72 FR 37818).  In the subsequent 2008 review, the
21    EPA staff decided to revisit its selection of the SUM06 form by again comparing the different
22    relevant features of the SUM06 form to the W126 form (U.S. EPA, 2007b, pp. 8-21/22).  On the
23    basis of that comparison which took into account both science and policy considerations relevant
24    in that review, the EPA proposed to use the W126 form as the most appropriate to use in that
25    context.
26          In its discussion of biologically relevant exposure indices, the current ISA continues this
27    comparison and further elaborates  by  stating, "[i]t should be noted that there are some important
28    differences between the SUM06 and W126. When considering the response of vegetation to Os
29    exposures represented by the threshold (e.g., SUM06) and non-threshold (e.g., W126) indices,
30    the W126 metric does not have a cut-off in the weighting scheme as does SUM06 and thus it
31    includes consideration of potentially damaging exposures below 60 ppb. The W126 metric also
32    adds increasing weight to hourly concentrations from about 40 ppb to about 100 ppb (Lefohn et
33    al., 1988; Lefohn and Runeckles, 1987)." This is unlike cut-off metrics such as the SUM06
34    where all concentrations below 0.06 ppm are given a weight of zero and concentrations at or

                                                  5-25

-------
 1    above 0.06 ppm are treated equally and given a weight of 1.0. This is an important feature of the
 2    W126 since as hourly concentrations become higher, they become increasingly likely to
 3    overwhelm plant defenses and are known to be more detrimental to vegetation (US EPA, 2012a,
 4    section 9.5.3).
 5           Given the above, EPA staff concludes that it is appropriate to continue to focus on the
 6    W126 form in this review.  The equation for the W126 (75 FR 2999 and US EPA, 2012a, section
 7    9.5.2, equation 9-1) is provided here. This daily ozone index is defined as follows:
               daily W126 =  7, wc- Q > where d = hourly OB at hour i, and wc =
                                                                               4403e
                                                                                    -126C
10    The daily index values are then summed over each month within the 63 season, and the annual
1 1    highest consecutive three month sum is determined.
12          In addition to the development of exposure metrics, research has and continues to be
13    conducted on the development of flux-based models. These models attempt to calculate the Os
14    concentration from the atmosphere that enters the leaf (i.e., flux or deposition).  Interest has been
15    increasing in recent years, particularly in Europe, in using mathematically tractable flux models
16    for Os assessments at the regional, national, and European scale (Matyssek et al., 2008; Paoletti
17    and Manning, 2007; Emberson et al., 2000b; Emberson et al., 2000a) (US EPA, 2012a, section
18    9.5.5).  Clearly, from a theoretical perspective, a measure of plant Os uptake or dose from
19    ambient air (either rate of uptake or cumulative seasonal uptake) might seem more desirable as a
20    better predictor of O3 damage to plants than an exposure index and therefore more useful in
21    improving risk assessment. This is because an uptake estimate would have to integrate all those
22    environmental factors that influence stomatal conductance, including but not limited to
23    temperature, humidity, and soil water status (see discussion in section 5.3.3 above).
24          However, there are still important limitations with this approach that make it
25    inappropriate to consider in the context of a national standard for the U.S. First, while some
26    efforts have been made in the U.S. to calculate Os flux into leaves and canopies (Turnipseed et
27    al., 2009; Uddling et al., 2009; Bergweiler et al., 2008; Hogg et al., 2007; Grulke et al., 2004;
28    Grantz et al., 1997; Grantz et al., 1995), little information has been published relating these
29    fluxes to effects on vegetation. At this time, dose-response relationships have not been
30    developed for these flux calculations for plants growing in the U.S.,  and large amounts of data on
3 1    the physiology of each plant species and the local growing conditions for the growing range of
32    each plant species, including climate patterns, would be required for flux calculations. The

                                                   5-26

-------
 1    models have to also distinguish between stomatal and non-stomatal components of 63 deposition
 2    to adequately estimate actual concentration reaching the target tissue of a plant to elicit a
 3    response (Uddling et al., 2009) (US EPA, 2012a, section 9.5.4).
 4           The lack of data in the U. S. and the lack of understanding of detoxification processes
 5    have made this technique less viable for vulnerability and risk assessments in the U.S. Thus,
 6    EPA staff recognizes that the selection of an appropriate form of exposure index that can be
 7    nationally applied will necessarily represent a simplification of the multiple factors that can
 8    potentially affect specific plant response across the wide variety of species and
 9    ecosystems/conditions that occur within the U.S.
10    5.4.3   Conclusions
11           The 2012 ISA (US EPA, 2012a, section 9.5.5) states that the main conclusions from the
12    1996 and 2006 Os AQCDs regarding a biologically relevant Os exposure index are still valid.
13    These key conclusions can be restated as follows:
14       •   Os effects in plants are cumulative;
15       •   higher 63 concentrations appear to be more important than lower concentrations in
16           eliciting a response;
17       •   plant sensitivity to Os varies with time of day  and plant development stage; and
18       •   quantifying exposure with indices that accumulate the Os hourly concentrations and
19           preferentially weight the higher concentrations improves the explanatory power of
20           exposure/response models for growth and yield, over using indices based on mean
21           and peak exposure values.
22    5.5     ECOSYSTEM-LEVEL IMPACTS
23           Ecosystems are comprised of complex assemblages of organisms and the physical
24    environment with which they interact. Each level of organization within an ecosystem has
25    functional and structural characteristics. At the ecosystem level, functional characteristics
26    include, but are not limited to, energy flow; nutrient, hydrologic, and biogeochemical cycling;
27    and maintenance of food chains. The sum of the functions carried out by ecosystem components
28    provides many benefits to humankind, as in the case of forest ecosystems (Smith, 1992).  Some
29    of these benefits, also termed "ecosystem goods and services," include food, fiber production,
30    aesthetics, genetic diversity, maintenance of water quality, air quality, and climate, and energy
31    exchange.  A conceptual framework for discussing the effects of stressors, including air
32    pollutants such as Os, on ecosystems was developed by the EPA Science Advisory Board (Young
33    and Sanzone, 2002). In this report, the authors identify  six essential ecological attributes (EEAs)
34    of ecosystems including landscape condition, biotic condition, chemical/physical condition,
35    ecological processes, hydrology/geomorphology, and natural  disturbance regime. Each EEA is
                                                  5-27

-------
 1    depicted as one of six triangles that together build a hexagon. On the outside of each triangle is a
 2    list of stressors that can act on the EEA. Tropospheric Os is listed as a stressor of both biotic
 3    condition and the chemical/physical condition of ecosystems. As each EEA is linked to all the
 4    others, it is clearly envisioned in this framework that Os could either directly or indirectly impact
 5    all of the EEAs associated with an ecosystem that is being stressed by Os.
 6          Vegetation often plays an influential role in defining the structure and function of an
 7    ecosystem, as evidenced by the use of dominant vegetation forms to classify many types of
 8    natural ecosystems, e.g., tundra, wetland, deciduous forest, and conifer forest. Plants
 9    simultaneously inhabit both above- and below-ground environments, integrating and influencing
10    key ecosystem cycles of energy, water, and nutrients. When a sufficient number of individual
11    plants within a community have been affected, O3-related effects can be propagated up to
12    ecosystem-level effects. Thus, through its impact on vegetation, Os can be an important
13    ecosystem stressor.
14    5.5.1  Overview: Ecosystem Scale, Structure,  Function, and Services
15          For this assessment, "ecosystem" is defined as the interactive system formed from all
16    living organisms and their abiotic (physical and chemical) environment within a given area
17    (TPCC, 2007a). The boundaries of what could be  called an ecosystem are somewhat arbitrary,
18    depending on the focus of interest or study. Thus, the extent of an ecosystem may range from
19    very small spatial scales or levels of biological organization to, ultimately, the entire Earth
20    (IPCC, 2007a).  All ecosystems, regardless of size or complexity, have interactions and physical
21    exchanges between biota and abiotic factors: this includes both structural (e.g., soil type and food
22    web trophic levels) and functional  (e.g., energy flow, decomposition, nitrification) attributes (US
23    EPA, 2012a, section 9.4.1).
24          Ecosystems can be described, in part, by their structure, i.e., the number and type of
25    species present. Structure may refer to a variety of measurements including the species richness,
26    abundance, community composition  and biodiversity as well as landscape attributes.
27    Competition among and within species and their tolerance to environmental  stressors are key
28    elements of survivorship.  When environmental conditions are shifted, for example, by the
29    presence of anthropogenic air pollution, these competitive relationships may change, and
30    tolerance to stress may be exceeded.  Ecosystems may also be defined on a functional basis.
31    "Function" refers to the suite of processes and  interactions among the ecosystem components
32    and their environment that involve nutrient and energy flow as well as other attributes including
33    water dynamics and the flux  of trace gases.  Plants, via such processes as photosynthesis,
34    respiration, C allocation, nutrient uptake and evaporation, affect energy flow, C, nutrient cycling
35    and water cycling. The energy accumulated and stored by vegetation (via photosynthetic C
                                                   5-28

-------
 1    capture) is available to other organisms. Energy moves from one organism to another through
 2    food webs, until it is ultimately released as heat. Nutrients and water can be recycled. Air
 3    pollution alters the function of ecosystems when elemental cycles or the energy flow are altered.
 4    This alteration can also be manifested in changes in the biotic composition of ecosystems (US
 5    EPA, 2012a, 9.4.1).
 6           There are at least three levels of ecosystem response to pollutants: (1) the individual
 7    organism and its environment; (2) the population and its environment; and (3) the biological
 8    community composed of many species and their environment (Billings,  1978). Individual
 9    organisms within a population vary in their ability to withstand the stress of environmental
10    change. The response of individual organisms within a population is based on their genetic
11    constitution, stage of growth at time of exposure to stress, and the microhabitat in which they are
12    growing (Levine and Pinto, 1998). The stress range within which organisms can exist and
13    function determines the ability of the population to survive (US EPA, 2012a, 9.4.1).
14           Ecosystem structure and function may be translated into ecosystem services. Ecosystem
15    services are the benefits people obtain from ecosystems (UNEP, 2003).  Ecosystems provide
16    many goods and services that are of vital importance for the functioning of the biosphere and
17    provide the basis for the  delivery of tangible benefits to human society.  Hassan et al. (2005)
18    define these benefits to include supporting, provisioning, regulating, and cultural services (US
19    EPA, 2012a, 9.4.1).
20           Supporting services are necessary for the production of all other  ecosystem  services.
21    Some examples include biomass production, production of atmospheric  62, soil formation and
22    retention, nutrient cycling,  water cycling, and provisioning of habitat. Biodiversity is a
23    supporting service that is increasingly recognized to sustain many of the goods and services that
24    humans enjoy from ecosystems. These provide a basis for three higher-level categories of
25    services. Provisioning services may include products (Gitay et al., 2001), i.e., food (including
26    game, roots, seeds, nuts and other fruit, spices, fodder), water, fiber (including wood, textiles),
27    and medicinal and cosmetic products (such as aromatic plants, pigments).  Regulating services
28    that are of paramount importance for human society may include (1) C sequestration, (2) climate
29    and water regulation, (3) protection from natural hazards such as floods, avalanches, or rock-fall,
30    (4) water and air purification, and (5) disease and pest regulation. Cultural services satisfy
31    human spiritual and aesthetic appreciation of ecosystems and their components, including
32    recreational and other nonmaterial benefits (US EPA, 2012a, 9.4.1).
33           Collectively these examples suggest that 63 is an important stressor in natural
34    ecosystems, but ecosystem changes, including those for which O3 was a contributor, are difficult
35    to evaluate in natural settings, due to the complexity of interactions, the  number of potential
                                                   5-29

-------
 1    confounders, and the large spatial and temporal scales. In most ecosystem level studies, only a
 2    few ecosystem components are examined and characterized for Os effects. Thus, the full extent
 3    of ecosystem changes in these ecosystems is not fully understood. Clearly, there is a need for
 4    highly integrated ecosystem studies that specifically investigate the effect of 63 on ecosystem
 5    structure and function in order to fully understand the extent to which Os is affecting ecosystem
 6    services.
 7          In the sections that follow, available information on the ecosystem-level effects of Os on
 8    productivity and C sequestration (5.5.2), water cycling (5.5.3), below-ground processes (5.5.4),
 9    community composition (5.5.5), and insects and wildlife (5.5.6) are considered and, where
10    appropriate, put in the context of ecosystem services (US EPA, 2012a, 9.4.1).
11    5.5.2  Productivity and Carbon Sequestration
12          During the last NAAQS review, there were limited studies that investigated the effect of
13    63 exposure on ecosystem productivity and C sequestration. Both experimental and modeling
14    studies have provided new information on effects of Os exposure at the stand or site level, i.e., at
15    the local scale (US EPA, 2012a, section 9.4.3).
16          Recent field-based studies from long-term FACE experiments, such as Aspen FACE,
17    SoyFACE and the Kranzberg Forest (Germany),  provide more evidence of the association of 63
18    exposure and reduced productivity at the ecosystem level of organization. Studies at the leaf and
19    plant scales show that Os decreased photosynthesis and plant growth, which provides coherence
20    and biological plausibility for the decrease in ecosystem productivity (US EPA, 2012a, section
21    9.4.3).
22          For example, the above- and below-ground biomass and net primary production (NPP)
23    were measured at the Aspen FACE site after 7 years of Os exposure. Elevated Os caused 23, 13
24    and 14% reductions in total biomass relative to the control in the aspen, aspen-birch and aspen-
25    maple communities, respectively (King et al., 2005).  At the Kranzberg Forest FACE experiment
26    in Germany, Oj, reduced annual volume growth by 9.5 m3/ha in a mixed mature stand of Norway
27    spruce and European beech (Pretzsch et al., 2010). At the grassland FACE experiment at Alp
28    Flix, Switzerland, Os reduced the seasonal mean  rates of ecosystem respiration and total carbon
29    gain (GPP) by 8% but had no significant impacts on aboveground dry matter productivity or
30    growing season net ecosystem production (NEP) (Volk et al., 2011). Ozone  also altered C
31    accumulation and turnover in soil, as discussed in Section 9.4.6 (US EPA, 2012a, section 9.4.3).
32          Changes in forest stand productivity under elevated Os were assessed by several model
33    studies. TREGRO (Table 9-2) has been widely used to simulate the effects of 63 on the growth
34    of several species in different regions in the U.S.  Hogsett et al. (2008) used TREGRO to

                                                  5-30

-------
 1    evaluate the effectiveness of various forms and levels of air quality standards for protecting tree
 2    growth in the San Bernardino Mountains of California. They found that Os exposures at the
 3    Crestline site resulted in a mean 20.9% biomass reduction from 1980 to 1985 and 10.3% biomass
 4    reduction from 1995 to 2000, compared to the "background" 63 concentrations (63
 5    concentration in Crook County, Oregon).  The level of vegetation protection projected was
 6    different depending on the air quality scenarios under consideration (US EPA, 2012a, section
 7    9.4.3).
 8          The model ZELIG is a forest succession gap model which has been used to evaluate the
 9    dynamics of natural stand succession.  Combining TREGRO with ZELIG, Weinstein et al.
10    (2005) simulated the effects of different O3 levels (0.5, 1.5,  1.75,  and 2 times [x] ambient) on the
11    growth and competitive interactions of white fir and ponderosa pine at three sites in California:
12    Lassen National Park, Yosemite National Park, and Crestline. Their results suggested that Os
13    had little impact on white fir but greatly reduced the growth of ponderosa pine (US EPA, 2012a,
14    section 9.4.3). To evaluate the influence of interspecies competition on Os effects, the linked
15    TREGRO and ZELIG modeling system was used to predict the effects of Os over 100 years on
16    the basal area of species in a Liriodendron tulipifera-dominated forest in the Great Smoky
17    Mountains National Park (Weinstein et al., 2001).  Ambient Os was predicted to decrease
18    individual tree C budget by 28% and reduce the basal area of L. tulipifera by 10%, whereas a
19    1.5 x-ambient exposure was  predicted to cause a 42% decrease in the individual tree C budget
20    and a 30% reduction in basal area. Individual tree C balance for Acer rubrum decreased 14% and
21    23% under ambient and 1.5x-ambient exposure, respectively. Prunus serotina was predicted to
22    have less than a 2% decrease in tree C balance in all scenarios, but its basal area was greatly
23    altered by the Os effects on the other tree  species. Basal area of A. rubrum and P. serotina was
24    predicted to increase for some years but then decrease by up to 30%, depending  on the scenario
25    (US EPA, 2012a,  section 9.4.3).
26          The effects of O3 on stand productivity and dynamics were also studied by other tree
27    growth or stand models, such as ECOPHYS, INTRASTAND and LINKAGES. ECOPHYS is a
28    functional-structural tree growth model.  The model used the linear relationship between the
29    maximum capacity of carboxylation and Os dose to predict the relative effect of Os on leaf
30    photosynthesis (Martin et al., 2001).  Simulations with ECOPHYS found that Os decreased stem
31    dry matter production, stem diameter and leaf dry matter production, induced earlier leaf
32    abscission, and inhibited root growth (Martin et al., 2001). INTRASTAND is an hourly time
33    step model for forest  stand carbon and water budgets. LINKAGES is a monthly time step model
34    simulating forest growth and community dynamics. Linking INTRASTAND with LINKAGES,
35    Hanson et al. (2005) found that a simulated  increase in Os concentration in 2100 (a mean 20-ppb

                                                  5-31

-------
 1    increase over the current O3 concentration) yields a 35% loss of net ecosystem C exchange
 2    (NEE) with respect to the current conditions (174 g C/m2/year) (US EPA, 2012a, section 9.4.3).
 3          Since the publication of the 2006 O3 AQCD, there is additional evidence suggesting that
 4    O3 exposure alters ecosystem productivity and biogeochemical cycling at the regional scale, i.e.,
 5    at scales ranging from watershed to subcontinental divisions and at continental and global scales.
 6    Most of those studies were conducted by using process-based ecosystem models.  In one
 7    example, Ollinger et al. (1997a) simulated the effect of O3 on hardwood forest productivity of 64
 8    hardwood sites in the northeastern U.S. with PnET-il.  Their simulations indicated that O3 caused
 9    a 3-16% reduction in NPP from 1987 to 1992 (US EPA, 2012a, section 9.4.3).
10          Felzer et al. (2004) developed TEM 4.3 (US EPA,  2012a, Table 9-2) to simulate the
11    effects of Os on plant growth and estimated effects of O3 on NPP and C sequestration of
12    deciduous trees, conifers and crops in the conterminous U.S.  The results indicated that O3
13    reduced NPP and C sequestration in the U.S. with the largest decreases (over 13% in some
14    locations) in NPP occurring in the Midwest agricultural lands during the mid-summer. DLEM
15    was developed to simulate the detrimental effect of O3 on  ecosystems and has been used to
16    examine the O3 damage on NPP and C sequestration in Great Smoky Mountains National Park
17    (Zhang et al., 2007), grassland ecosystems and terrestrial ecosystems in China (Ren et al., 2007b;
18    Ren et al., 2007a) (US EPA,  2012a, section 9.4.3).
19          Instead of using AOT40 as their O3 exposure metric as PnET, TEM and DLEM did, Sitch
20    et al. (2007) incorporated a different O3 metric named CUOt (cumulative stomatal uptake of Os),
21    derived from Pleijel et al. (2004a), into the MOSES-TRIFFID coupled model (Table 9-2). In the
22    model, reduced ecosystem C uptake due to O3 damage results in additional CO2 accumulation in
23    the atmosphere and an indirect radiative forcing of climate change. Their simulations indicated
24    that the indirect radiative forcing caused by O3 (0.62-1.09  W/m2) could have even greater impact
25    on global warming than the direct radiative forcing of O3 (0.89 W/m2) (Sitch et al., 2007) (US
26    EPA, 2012a,  section 9.4.3).
27          Results from the various model studies are difficult to compare because of the various
28    spatial and temporal scales used. However, all the studies  showed that O3 exposure decreased
29    ecosystem productivity and C sequestration.  These results are consistent and coherent with
30    experimental results obtained from studies at the leaf, plant and ecosystem scales (Sitch et al.,
31    2007; Felzer et al., 2005). Many of the models use the same underlying function to  simulate the
32    effect of Os exposure to C uptake.  Therefore, it is not surprising that the results are  similar.
33    While these models can be improved and more evaluation with experimental data can be done,
34    these models represent the state of the science for estimating the effect of O3 exposure on
35    productivity and C sequestration (US EPA, 2012a, section 9.4.3).
                                                  5-32

-------
 1          Although O3 generally causes negative effects on plant growth, the magnitude of the
 2    response varies among plant communities. For example, O3 had little impact on white fir but
 3    greatly reduced growth of ponderosa pine in southern California (Weinstein et al., 2005).  Ozone
 4    decreased net primary production (NPP) of most forest types in the Mid-Atlantic region but had
 5    small impacts on spruce-fir forest (Pan et al., 2009) (US EPA, 2012a, section 9.4.3).
 6          In addition to plant growth, other indicators that are typically estimated by model studies
 7    include net ecosystem CC>2 exchange (NEE), C sequestration, and crop yield. Model simulations
 8    consistently found that O3 exposure caused negative impacts on these indicators, but the severity
 9    of these impacts was influenced by multiple  interactions of biological and environmental factors
10    such as N deposition, elevated CC>2 and land use history. Model simulations suggested that 63
11    partially offset the growth stimulation caused by elevated CO2 and N deposition in both
12    Northeast- and Mid-Atlantic-region forest ecosystems of the U.S.  (Pan et al., 2009; Ollinger et
13    al., 2002) (US EPA, 2012a, section 9.4.3).
14          Results across different ecosystem models, such as TREGRO, PnET, TEM and DLEM,
15    are consistent with the FACE experimental evidence, which show that O3 reduced productivity of
16    various ecosystems. Productivity is measured by various metrics such as GPP, NPP, NEP, NCE,
17    NEE and individual tree biomass gain. All these metrics indicate a decrease in CC>2 fixation by
18    the systems that were studied (US EPA, 2012a, section 9.4.3).
19    5.5.3  Water Cycling
20          Ozone can affect water use in plants and ecosystems through several mechanisms
21    including damage to stomatal functioning and loss of leaf area.  However, there is not a clear
22    consensus on the nature of leaf-level stomatal conductance response to O3 exposure.  When
23    measured under steady-state high light conditions, leaf-level stomatal conductance is often found
24    to be reduced when plants are exposed to O3. In contrast, measurements of stomatal conductance
25    under dynamic light and VPD conditions, which are more typical in the field indicate sluggish
26    responses under elevated O3 exposure, which could potentially lead to increased water loss from
27    vegetation in some situations (US EPA, 2012a, section 9.4.5).
28          Field studies by (McLaughlin et al., 2007a; 2007b) provided valuable insight into the
29    possible consequences of stomatal sluggishness for ecosystem water cycling. McLaughlin et al.
30    (2007a,b) indicated that O3 increased water use in a mixed deciduous forest  in eastern Tennessee.
31    McLaughlin et al. (2007a,b) found that O3, with daily maximum levels ranging from 69.2 to 82.9
32    ppb, reduced stem growth by 30-50% in the high-O3 year 2002. The decrease in growth rate was
33    caused in part by amplification of diurnal cycles of water loss and recovery.  Peak hourly  O3
34    exposure increased the rate of water loss through transpiration as indicated by the increased stem

                                                  5-33

-------
 1    sap flow. The authors suggested that a potential mechanism for the increased sap flow could be
 2    altered stomatal regulation from Os exposure, but this was inferred through sap flow
 3    measurements and was not directly measured.  The increased canopy water loss resulted in
 4    higher water uptake by the trees as reflected in the reduced soil moisture in the rooting zone.
 5    The change in tree water use led to further impacts on the hydrological cycle at the landscape
 6    level.  Increased water use under high 63 exposure was reported to reduce late-season modeled
 7    streamflow in three forested watersheds in eastern Tennessee (McLaughlin et al., 2007b).
 8          In addition to the impacts on stomatal performance, (Vinduced physiological changes,
 9    such as reduced leaf area index and accelerated leaf senescence  could alter water use efficiency.
10    Elevated 63 could also affect evapotranspiration by altering tree crown interception of
11    precipitation. Although evidence was limited (based on a few field and modeling studies),
12    findings showed an association between Os exposure and alteration of water use and cycling in
13    vegetation and  at the watershed level (US EPA, 2012a, section 9.4.5).
14    5.5.4  Below-ground Processes
15          Since the 2006 Os AQCD, more evidence has shown that although the responses are
16    often site specific, 63 altered the quality and quantity of litter input to soil, microbial community
17    composition, and C and nutrient cycling. Biogeochemical cycling of below-ground processes is
18    fueled by C input from plants.  Studies at the leaf and plant level have provided biologically
19    plausible mechanisms, such as reduced photosynthetic rates, increased metabolic cost, and
20    reduced root C  allocation for the association of Os exposure and the alteration of below-ground
21    processes (US EPA, 2012a, section 9.4.6).
22          Results from Aspen FACE and other experimental studies consistently found that O?,
23    reduced litter production and altered C chemistry, such as soluble sugars, soluble phenolics,
24    condensed tannins, lignin, and macro/micro nutrient concentration in litter (Parsons et al., 2008;
25    Kasurinen et al., 2006; Liu et al., 2005). Under elevated 63, the changes in substrate quality and
26    quantity could alter microbial metabolism and therefore soil C and nutrient cycling. Several
27    studies indicated that Os suppressed soil enzyme activities (Pritsch et al., 2009; Chung et al.,
28    2006). However, the impact of 63 on litter decomposition was inconsistent and varied among
29    species,  sites and exposure length.  Similarly, Os had inconsistent impacts on dynamics of micro
30    and macro nutrients (US EPA, 2012a, section 9.4.6).
31          Studies from the Aspen FACE experiment suggested that the response of below-ground C
32    cycle to 63 exposure,  such as litter decomposition, soil respiration and soil C content, changed
33    over time. For example, in the early part of the experiment (1998-2003), Os had no impact on
34    soil respiration but reduced the formation rates of total soil C under elevated CC>2.  However,

                                                   5-34

-------
 1    after 10-11 years of exposure, 63 was found to increase soil respiration but have no significant
 2    impact on soil C formation under elevated CC>2 (US EPA, 2012a, section 9.4.6).
 3    5.5.5  Community Composition
 4          In the 2006 63 AQCD, the impact of 63 exposure on species competition and community
 5    composition was assessed. Ozone was found to cause a significant decline in ponderosa and
 6    Jeffrey pine in the San Bernardino Mountains  in southern California. Ozone exposure also
 7    tended to shift the grass-legume mixtures in favor of grass species  (US EPA, 2006). Since the
 8    2006 O3 AQCD, more evidence has shown that Os exposure changed the competitive
 9    interactions and could lead to loss of Os-sensitive species or genotypes.  Studies at plant level
10    found that the severity of Os damage on growth, reproduction and foliar injury varied  among
11    species, which provided the biological plausibility for the alteration of community composition.
12    Additionally, research since the last review has shown that Os can  alter community composition
13    and diversity of soil microbial communities (US EPA, 2012a, section 9.4.7).
14          The decline of conifer forests under Os exposure was continually observed in several
15    regions.  Ozone damage was  believed to be an important causal factor in the dramatic decline of
16    sacred fir in the valley of Mexico (de Lourdes de Bauer and Hernandez-Tejeda, 2007), as well as
17    cembran pine in southern France and the Carpathian Mountains (Wieser et al., 2006).  Results
18    from the Aspen FACE site indicated that Os could alter community composition of broadleaf
19    forests as well. At the Aspen FACE site, O3 reduced growth and increased mortality of a
20    sensitive aspen clone, while the Os-tolerant clone emerged as the dominant clone in the pure
21    aspen community.  In the mixed aspen-birch and aspen-maple communities, Os reduced the
22    competitive capacity  of aspen compared to birch and maple (Kubiske et al., 2007) (US EPA,
23    2012a, section 9.4.7).
24          The tendency for Os exposure to shift the biomass of grass-legume mixtures in favor of
25    grass species was reported in the 2006 Os AQCD and has been generally confirmed by recent
26    studies. However, in a high elevation mature/species-rich grass-legume pasture, O3 fumigation
27    showed no significant impact on community composition (Bassin et al., 2007b) (US EPA, 2012a,
28    section 9.4.7).
29          Ozone exposure altered the community composition of not only plant species,  but also
30    microorganisms. The shift in community composition of bacteria and fungi has been observed in
31    both natural and agricultural ecosystems, although no general patterns could be identified
32    (Kanerva et al., 2008; Morsky et al., 2008; Kasurinen et al., 2005)  (US EPA, 2012a, section
33    9.4.7).
                                                  5-35

-------
 1    5.5.6  Insects and Other Wildlife
 2          Recent information on 63 effects on insects and other wildlife is limited to a few species,
 3    and there is no consensus on how these organisms respond to elevated O3.  Studies published
 4    since the last review show impacts of elevated Os on both species-level responses (reproduction,
 5    growth, feeding behavior) and community and ecosystem-level responses (population growth,
 6    abundance, shift in community structure) in some insects and soil fauna.  Changes in
 7    ecologically important behaviors such as feeding and thermoregulation have recently been
 8    observed with 63 exposure in amphibians and reptiles, however, these responses occur at
 9    concentrations of Os much higher than ambient levels.
10          Recent information available since the last review considers the effects of Os on chemical
11    signaling in insect and wildlife interactions. Specifically, studies on Os effects on pollination
12    and seed dispersal, defenses against herbivory and predator-prey  interactions all consider the
13    ability of Os to alter the chemical signature of VOCs emitted during these pheromone-mediated
14    events. The effects of 63 on chemical signaling between plants, herbivores and pollinators as
15    well as interactions between multiple trophic levels comprise an emerging area of study that may
16    result in further elucidation of Os effects at the species, community and ecosystem levels (US
17    EPA, 2012a, section 9.4.9).

18    5.6    CONSIDERATIONS REGARDING ADVERSITY IN PUBLIC WELFARE
19          CONTEXT
20          The 2008 final rule recognized that the statute requires that a secondary standard be
21    protective against  only those known or anticipated 63 effects that are "adverse" to the public
22    welfare, not all identifiable Os-induced effects.  Unlike the use of the terms adverse, injury or
23    damage in the scientific literature, in the NAAQS policy context, these terms have been
24    interpreted in a particular way.  Specifically, O3-induced "injury" to vegetation has been defined
25    as encompassing all plant reactions, including reversible changes or changes in plant metabolism
26    (e.g., altered photosynthetic rate), altered plant quality or reduced growth that does not impair the
27    intended use or value of the plant.  In contrast, "damage" has been defined to include only those
28    injury effects that reach sufficient magnitude as to also reduce or impair the intended use or
29    value of the plant to the public and thus potentially become adverse to the public welfare.
30    Examples of vegetation effects that have been classified as damage include reductions in
31    aesthetic values (e.g., foliar injury in ornamental species) as well as losses in terms of weight,
32    number, or size of harvestable plant parts. For example, biomass loss in tree species can be
33    considered damage or adverse to the public welfare if it includes  slower growth in species
34    harvested for timber or other fiber uses. In the context of evaluating effects on single plants or

                                                   5-36

-------
 1    species grown in monocultures such as managed forests, this construct continues to remain
 2    useful (73 FR 16492/96).
 3          However, given the increasing scientific literature linking Oj effects on plants or species
 4    to effects at the community or ecosystem level, as discussed above in section 5.5, a more
 5    expansive construct or paradigm of what constitutes 63 "damage" beyond that of the individual
 6    or species level is appropriate. A number of broader paradigms have been discussed in the
 7    literature (72 FR 37890), and in the 2008 review, the Administrator expressed support for using
 8    such a broader paradigm (73 FR 16492/96).
 9          Since the 2008 63 review, EPA's approach to assessing adversity to the public welfare
10    has continued to evolve. In particular, the concept of ecosystem services has been incorporated
11    into this broader paradigm. An extensive look at the range of services than can be provided by
12    ecosystems is described in the Millennium Ecosystem Assessment (MEA, 2005) (see discussion
13    in section 5.5 above). Ecosystem services can be generally defined as the benefits that
14    individuals and organizations obtain from ecosystems.  EPA has defined ecological goods and
15    services as the "outputs of ecological functions or processes that directly or indirectly contribute
16    to social welfare or have the potential to do  so in the future. Some outputs may be bought and
17    sold, but most are not marketed" (77 FR 20232).  The most recent secondary NAAQS reviews
18    have recognized that changes in ecosystem services may be used to aid in characterizing a
19    known or anticipated adverse effect to public welfare and that an evaluation of adversity to the
20    public welfare might consider the likelihood, type, magnitude, and spatial scale of the effect, as
21    well as the potential for recovery and any uncertainties relating to these conditions (US EPA,
22    2009, Appendix 8; 77 FR 20231). In the context  of this review, ecosystem services are being
23    evaluated and assessed in the REA to determine the possible benefits received from ecosystem
24    resources and how those benefits might be expected to change under different air quality
25    scenarios (US EPA, 2012c, chapter 6).
26          In addition, to the above considerations regarding adversity, in the last review the
27    Administrator also recognized that the public welfare significance of (Vinduced effects on
28    sensitive vegetation growing within the U.S. can vary, depending on the nature of the effect, the
29    intended use of the sensitive plants or ecosystems, and the types of environments in which the
30    sensitive vegetation and ecosystems are located.  Any given O3-related effect on vegetation and
31    ecosystems (e.g., biomass loss, foliar injury), therefore, may be judged to have a different degree
32    of impact on the public depending, for example, on whether that effect occurs in a Class I area, a
33    city park, or commercial cropland.  In the 2008 review, the Administrator judged it appropriate
34    that this variation in the significance of Os-related vegetation effects should be taken into
                                                   5-37

-------
 1    consideration in judging the level of ambient 63 that is requisite to protect the public welfare
 2    from any known or anticipated adverse effects (73 FR 16496).

 3    5.7    CONSIDERATIONS REGARDING OTHER WELFARE EFFECTS
 4    5.7.1  Ozone Effects on Climate Change
 5          Tropospheric Os is a major greenhouse gas, third in importance after CC>2 and CH4.
 6    While the developed world has successfully reduced emissions of Os precursors in recent
 7    decades, many developing countries have experienced large increases in precursor emissions and
 8    these trends are expected to continue, at least in the near term. Projections of radiative forcing
 9    due to changing 63 over the 21st century show wide variation, due in large part to the uncertainty
10    of future emissions of source gases. In the near-term (2000-2030), projections of O3 radiative
11    forcing range from near zero to +0.3 W/m2, depending on the emissions scenario (Stevenson et
12    al., 2006).  Reduction of tropospheric 63 concentrations could therefore provide an important
13    means to slow climate change in addition to the added benefit of improving surface air quality
14    (US EPA, 2012a, section 10.5).
15          It is clear that increases in tropospheric Os lead to warming.  However the precursors of
16    63 also have competing effects on the greenhouse gas CH4, complicating emissions reduction
17    strategies.  A decrease in CO or VOC emissions would enhance OH concentrations, shortening
18    the lifetime of CH4, while a decrease in NOx emissions could depress OH concentrations in
19    certain regions and lengthen the CH4 lifetime (US EPA, 2012a, section 10.5).
20          Abatement of CH4 emissions would likely provide the most straightforward means to
21    address climate change since CH4 is itself an important precursor of background O3 (West et al.,
22    2007; West et al., 2006; Fiore et al., 2002).  A reduction of CH4 emissions would also improve
23    air quality in its own right. A set of global abatement measures identified by West and Fiore
24    (2005) could reduce CH4 emissions by 10% at a cost savings, decrease background Os by about
25    1 ppb in the Northern Hemisphere summer, and lead to a global net cooling of 0.12 W/m2. West
26    et al. (2007) explored further the benefits of CH4 abatement, finding that a 20% reduction in
27    global CH4 emissions would  lead to greater cooling per unit reduction in surface Os, compared to
28    20% reductions in VOCs or CO (US EPA, 2012a, section 10.5).
29          Important uncertainties remain regarding the effect of tropospheric Os on future climate
30    change.  To address these uncertainties, further research is needed to: (1) improve knowledge of
31    the natural atmosphere; (2) interpret observed trends of Os in the free troposphere and  remote
32    regions;  (3) improve understanding of the CHj budget, especially emissions from wetlands and
33    agricultural sources, (4) understand the relationship between regional O3 radiative forcing and
                                                  5-38

-------
 1    regional climate change; and (5) determine the optimal mix of emissions reductions that would
 2    act to limit future climate change (US EPA, 2012a, section 10.5).
 3           The IPCC has estimated the effect of the tropospheric O3 change since preindustrial times
 4    on climate has been estimated to be about 25-40% of the anthropogenic CC>2 effect and about
 5    75% of the anthropogenic CH4 effect.  There are large uncertainties in the radiative forcing
 6    estimate attributed to tropospheric Os, making the effect of tropospheric Os on climate more
 7    uncertain than the effect of the long-lived greenhouse gases (US EPA, 2012a, section 10.5).
 8           Radiative forcing does not take into account the climate feedbacks that could amplify or
 9    dampen the actual surface temperature response.  Quantifying the change in surface temperature
10    requires a complex climate simulation in which all important feedbacks and interactions are
11    accounted for. As these processes are not well understood or easily modeled, the surface
12    temperature response to a given radiative forcing is highly uncertain and can vary greatly among
13    models and from region to region within the same model (US EPA, 2012a, section 10.5).
14    5.7.2.  UV-B Radiation Effects on Ecosystems and Materials
15           UV radiation emitted from the Sun contains sufficient energy when it reaches the Earth to
16    break (photolyze) chemical bonds in molecules, thereby leading to damaging effects on living
17    organisms and materials. Atmospheric 63 plays a crucial role in reducing exposure to solar UV
18    radiation at the Earth's surface. Ozone in the stratosphere is responsible for the majority of this
19    shielding effect, as approximately 90% of total atmospheric Os is located there over mid-
20    latitudes. Ozone in the troposphere provides supplemental shielding of radiation in the
21    wavelength band from 280-315 nm, referred to as UV-B radiation (US EPA, 2012a, section
22    10.4.1). UV-B radiation has important effects on ecosystems and is associated with materials
23    damage.  There is a lack of published studies that critically examine the incremental welfare
24    effects  (adverse or beneficial) attributable specifically t3o changes in UV-B exposure resulting
25    from perturbations in tropospheric Os concentrations. While the effects are expected to be small,
26    they cannot yet be critically  assessed within reasonable uncertainty (US EPA, 2012a, section
27    10.4.4,  10.5).
28
29
30
31
                                                   5-39

-------
 1    5.8   REFERENCES

 2    Abt Associates, Inc. (1995). Ozone NAAQS benefits analysis: California crops. Report to U.S.
 3         EPA, July.

 4    Andersen, CP; Wilson, R; Plocher, M; Hogsett, WE. (1997). Carry-over effects of ozone on root
 5         growth and carbohydrate concentrations of ponderosa pine seedlings. Tree Physiol  17: 805-
 6         811.

 7    Bassin, S; Volk, M; Suter, M; Buchmann, N; Fuhrer, J.  (2007). Nitrogen deposition but not
 8         ozone affects productivity and community composition of subalpine grassland after 3 yr of
 9         treatment. NewPhytol 175: 523-534. http://dx.doi.Org/10.llll/j.1469-8137.2007.02140.x

10    Bergweiler, CJ; Manning, WJ; Chevone, BI. (2008). Seasonal and diurnal gas exchange
11         differences in ozone-sensitive common milkweed (Asclepias syriaca L.) in relation to
12         ozone uptake. Environ Pollut 152: 403-415. http://dx.doi.Org/10.1016/j.envpol.2007.06.019

13    Billings, WD. (1978). Plants and the ecosystem. Belmont, CA: Wadsworth Publishing Company,
14         Inc.

15    Burns, R. M., B. H. Honkala, tech. coords. (1990). Silvics of North America: 1. Conifers; 2.
16         Hardwoods. Agriculture Handbook 654. U.S. Department of Agriculture, Forest Service,
17         Washington, DC. vol.2,  877 p.

18    Caird, MA; Richards, JH; Donovan, LA. (2007). Nighttime stomatal conductance and
19         transpiration in C-3 and  C-4 plants. Plant Physiol 143: 4-10.
20         http://dx.doi.org/10.1104/pp.106.092940

21    Campbell, SJ; Wanek, R; Coulston, JW. (2007). Ozone injury in west coast forests: 6 years of
22         monitoring - Introduction. Portland, OR: U.S. Department of Agriculture.

23    Chappelka, AH; Samuelson, LJ. (1998). Ambient ozone effects on forest trees of the eastern
24         United States: A review [Review]. New Phytol 139:91-108.
25         http://dx.doi.0rg/10.1046/j.1469-8137.1998.00166.x

26    Chappelka, AH; Somers, GL;  Renfro, JR. (2007). Temporal patterns of foliar ozone symptoms
27         on tall milkweed (Asclepias exaltata L) in Great Smoky Mountains National Park. Environ
28         Pollut 149: 358-365. http://dx.doi.Org/10.1016/j.envpol.2007.05.015

29    Chung, HG; Zak,  DR; Lilleskov, EA. (2006). Fungal community composition and metabolism
30         under elevated CO2 and  O3. Oecologia 147: 143-154. http://dx.doi.org/10.1007/s00442-
31         005-0249-3

32    Coulston, JW; Smith, GC; Smith, WD. (2003). Regional assessment of ozone sensitive tree
33         species using bioindicator plants. Environ Monit Assess 83:  113-127.

                                                  5-40

-------
 1    de Lourdes de Bauer, M; Hernandez-Tejeda, T. (2007). A review of ozone-induced effects on the
 2         forests of central Mexico [Review]. Environ Pollut 147: 446-453.

 3    Dizengremel, P; Le Thiec, D; Bagard, M; Jolivet, Y. (2008). Ozone risk assessment for plants:
 4         Central role of metabolism-dependent changes in reducing power. Environ Pollut 156: 11-
 5         15. http://dx.doi.0rg/10.1016/j.envpol.2007.12.024

 6    Emberson, L; Ashmore, MR; Cambridge, HM; Simpson, D; Tuovinen, JP. (2000a). Modelling
 7         stomatal ozone flux across Europe. Environ Pollut 109: 403-413.
 8         http://dx.doi.org/10.1016/80269-7491(00)00043-9

 9    Emberson, LD; Wieser, G; Ashmore, MR. (2000b). Modelling of stomatal conductance and
10         ozone flux of Norway spruce: Comparison with field data. Environ Pollut 109: 393-402.
11         http://dx.doi.org/10.1016/S0269-7491(00)00042-7

12    Federal Register 2007. National Ambient Air Quality Standards for Ozone; Proposed Rule. 40
13         CFR 50; Federal Register 72: 37818

14    Federal Register 2008. National Ambient Air Quality Standards for Ozone; Final Rule. 40
15         CFRparts 50 and 58; Federal Register 73:16436

16    Felzer, B; Kicklighter, D; Melillo, J; Wang, C; Xhuang, Q;  Prinn, R. (2004). Effects of ozone on
17         net primary  production and carbon sequestration in the conterminous United States using a
18         biogeochemistry model. Tellus B Chem Phys Meteorol 56: 230-248.
19         http://dx.doi.0rg/10.llll/j.1600-0889.2004.00097.x

20    Felzer, B; Reilly,  J; Melillo, J; Kicklighter, D; Sarofim, M;  Wang, C; Prinn, R; Zhuang, Q.
21         (2005). Future effects of ozone on carbon sequestration and climate change policy using a
22         global biogeochemical  model. Clim Change 73: 345-373.
23         http://dx.doi.org/10.1007/sl0584-005-6776-4

24    Finkelstein, PL; Ellestad, TG; Clarke, JF; Meyers, TP; Schwede, DB; Hebert, EO; Neal, JA.
25         (2000). Ozone and sulfur dioxide dry deposition to forests: Observations and model
26         evaluation. J Geophys Res 105: 15365-15377.

27    Fiore, AM; Jacob, DJ; Field, BD; Streets, DG; Fernandes, SD; Jang, C. (2002). Linking ozone
28         pollution and climate change: The case for controlling methane. Geophys Res Lett 29:
29         1919. http://dx.doi.org/10.1029/2002GL015601

30    Flagler, RB. (1998). Recognition of air pollution injury to vegetation: A pictorial atlas (2nd ed.).
31         Pittsburgh, PA: Air & Waste Management Association.

32    Gitay, H; Brown,  S; Easterling, W; Jallow, B. (2001). Ecosystems and their goods and services.
33         In Climate change 2001: Impacts,  adaptation and vulnerability: Contribution of Working
                                                  5-41

-------
 1         Group II to the third assessment report of the Intergovernmental Panel on Climate Change.
 2         Cambridge, United Kingdom: Cambridge University Press.

 3    Grantz, DA; Gunn, S; Vu, HB. (2006). 63 impacts on plant development: A meta-analysis of
 4         root/shoot allocation and growth. Plant Cell Environ 29: 1193-1209.
 5         http://dx.doi.0rg/10.llll/j.1365-3040.2006.01521.x

 6    Grantz, DA; Zhang, XJ; Massman, W; Delany, A; Pederson, R. (1997). Ozone deposition to a
 7         cotton (Gossypium hirsutum L) field:  Stomatal and surface wetness effects during the
 8         California Ozone Deposition experiment. Agr Forest Meteorol 85: 19-31.
 9         http://dx.doi.org/10.1016/80168-1923(96)02396-9

10    Grantz, DA; Zhang, XJ; Massman, WJ; Den Hartog, G; Neumann, HH; Pederson, JR. (1995).
11         Effects of stomatal conductance and surface wetness on ozone deposition in field-grown
12         grape. Atmos Environ 29: 3189-3198. http://dx.doi.org/10.1016/1352-2310(95)00129-M

13    Gregg, JW; Jones, CG; Dawson, TE. (2003). Urbanization effects on tree growth in the vicinity
14         of New York City [Letter]. Nature 424: 183-187. http://dx.doi.org/10.1038/nature01728

15    Grulke, N. E.;  Balduman, L. (1999) Deciduous conifers: high N deposition and O3 exposure
16         effects on growth and biomass allocation in ponderosa pine. Water Air Soil Pollut.  116:
17         235-248.

18    Grulke, NE; Alonso, R; Nguyen, T; Cascio, C; Dobrowolski, W. (2004). Stomata open at night
19         in pole-sized and mature ponderosa pine: Implications for O3 exposure metrics. Tree
20         Physiol 24: 1001-1010.

21    Grulke, NE; Johnson, R; Esperanza, A; Jones, D; Nguyen, T; Posch, S; Tausz, M. (2003).
22         Canopy transpiration of Jeffrey pine in mesic and xeric microsites: O3 uptake and injury
23         response. Trees Struct Funct 17: 292-298.

24    Grulke, N. E.;  Andersen, C. P.; Fenn, M. E.; Miller, P. R. (1998). Ozone exposure and nitrogen
25         deposition lowers root biomass of ponderosa pine in the San Bernardino Mountains,
26         California. Environ. Pollut.  103: 63-73.

27    Hanson, PJ; Wullschleger, SD; Norby, RJ; Tschaplinski, TJ; Gunderson, CA. (2005). Importance
28         of changing CO2, temperature, precipitation, and ozone on carbon and water cycles of an
29         upland-oak forest: incorporating experimental results into model simulations. Global
30         ChangeBiol 11: 1402-1423. http://dx.doi.Org/10.llll/j.1365-2486.2005.00991.x

31    Hassan, R; Scholes, R;  Ash, N. (2005). Ecosystems and human well-being: Current state and
32         trends. Washington, DC: Island Press.

33    Heagle, AS; Miller, JE; Sherrill, DE. (1994). A white clover system to estimate effects of
34         tropospheric ozone on plants. J Environ Qual 23: 613-621.
                                                  5-42

-------
 1   Heck, WW; Cowling, EB. (1997).The need for a long term cumulative secondary ozone standard
 2         - An ecological perspective. EM January: 23-33.

 3   Hogg, A; Uddling, J; Ellsworth, D; Carroll, MA; Pressley, S; Lamb, B; Vogel, C. (2007).
 4         Stomatal and non-stomatal fluxes of ozone to a northern mixed hardwood forest. Tellus B
 5         Chem Phys Meteorol 59:  514-525. http://dx.doi.Org/10.llll/j.1600-0889.2007.00269.x

 6   Hogsett, WE; Tingey, DT; Lee, EH; Beedlow, PA; Andersen, CP. (2008). An approach for
 7         evaluating the effectiveness of various ozone Air Quality Standards for protecting trees.
 8         Environ Manage 41: 937-948. http://dx.doi.org/10.1007/s00267-007-9057-3

 9    Innes, JL;  Skelly, JM; Schaub, M. (2001). Ozon, Laubholz- und Krautpflanzen: Ein Fuhrer zum
10         Bestimmen von Ozonsymptomen. Bern, Switzerland: Paul Haupt Publishers.

11   IPCC (Intergovernmental Panel on Climate Change). (2007). Climate change 2007: Impacts,
12         adaptation and vulnerability. Cambridge, UK: Cambridge University Press.

13   Kanerva, T; Palojarvi, A; Ramo, K; Manninen, S. (2008). Changes in soil microbial community
14         structure under elevated tropospheric 63 and CC>2.  Soil Biol Biochem 40: 2502-2510.
15         http://dx.doi.0rg/10.1016/j.soilbio.2008.06.007

16   Kasurinen, A; Riikonen, J; Oksanen, E; Vapaavuori, E; Holopainen, T. (2006). Chemical
17         composition and decomposition of silver birch leaf litter produced under elevated CC>2 and
18         O3. Plant Soil 282: 261-280. http://dx.doi.org/10.1007/slll04-005-6026-6

19   Kasurinen, A; Keinanen, MM; Kaipainen, S; Nilsson, LO; Vapaavuori, E; Kontro, MH;
20         Holopainen, T. (2005). Below-ground responses of silver birch trees exposed to elevated
21         CC>2 and Os levels during three growing seasons. Global Change Biol 11: 1167-1179.
22         http://dx.doi.0rg/10.llll/j.1365-2486.2005.00970.x

23   King, JS; Kubiske, ME; Pregitzer, KS; Hendrey, GR;  McDonald, EP; Giardina, CP; Quinn, VS;
24         Karnosky, DF. (2005). Tropospheric 63 compromises net primary production in young
25         stands of trembling aspen, paper birch and sugar maple in response to elevated atmospheric
26         CO2.  NewPhytol 168: 623-635. http://dx.doi.Org/10.llll/j.1469-8137.2005.01557.x

27   Kubiske, ME; Quinn, VS; Heilman, WE; McDonald, EP; Marquardt, PE; Teclaw, RM; Friend,
28         AL; Karnoskey, DF. (2006). Interannual climatic variation mediates elevated CO2 and O3
29         effects on forest growth. Global Change Biol 12: 1054-1068.
30         http://dx.doi.org/10.! 111/j. 1365-2486.2006.01152.x

31   Kubiske, ME; Quinn, VS; Marquardt, PE; Karnosky, DF. (2007). Effects of elevated
32         atmospheric CO2 and/or O3 on intra- and interspecific competitive ability of aspen. Plant
33         Biol (Stuttg) 9: 342-355. http://dx.doi.org/10.1055/s-2006-924760
                                                  5-43

-------
 1    Lee, EH; Tingey, DT; Hogsett, WE. (1988a). Evaluation of ozone-exposure indices for relating
 2         exposure to plant production and for estimating agricultural losses. (EPA/600/3-88/039).
 3         Washington, DC: U.S. Environmental Protection Agency.

 4    Lee, EH; Tingey, DT; Hogsett, WE. (1988b). Evaluation of ozone exposure indices in exposure-
 5         response modeling. Environ Pollut 53: 43-62. http://dx.doi.org/10.1016/0269-
 6         7491(88)90024-3

 7    Lee, EH; Tingey, DT; Hogsett, WE. (1989). Interrelation of experimental exposure and ambient
 8         air quality data for comparison of ozone exposure indices and estimating agricultural
 9         losses. (EPA/600/3-89/047). Corvallis, OR: U.S. Environmental Protection Agency.

10    Lefohn, AS; Laurence, JA; Kohut, RJ. (1988). A comparison of indices that describe the
11         relationship between exposure to ozone and reduction in the yield of agricultural crops.
12         Atmos Environ 22: 1229-1240. http://dx.doi.org/10.1016/0004-6981(88)90353-8

13    Lefohn, AS; Runeckles, VC. (1987). Establishing standards to protect vegetation - ozone
14         exposure/dose considerations. Atmos Environ 21: 561-568. http://dx.doi.org/10.1016/0004-
15         6981(87)90038-2

16    Levine, JS; Pinto, JP. (1998). The production of CO by biomass burning. In MAK Khalil; JP
17         Pinto; MJ Shearer (Eds.), Atmospheric carbon monoxide and its environmental effects:
18         Proceedings of the international conference; December 1997; Portland, Oregon (pp. 251-
19         256). Research Triangle Park, NC: U.S. Environmental Protection Agency, Office of
20         Research and Development.

21    Liu, L; King, J; Giardina, C. (2005). Effects of elevated concentrations  of atmospheric CO2 and
22         tropospheric Os on leaf litter production and chemistry in trembling aspen and paper birch
23         communities. Tree Physiol 25: 1511-1522.

24    Manning, WJ; Krupa, SV. (1992). Experimental methodology for studying the effects of ozone
25         on crops and trees. In AS Lefohn (Ed.), Surface level ozone exposures and their effects on
26         vegetation (pp. 93-156). Chelsea, MI: Lewis Publishers.

27    Martin, MJ; Host, GE; Lenz, KE; Isebrands, JG. (2001). Simulating the growth response of
28         aspen to elevated ozone: A mechanistic approach to scaling a leaf-level model of ozone
29         effects on photosynthesis to a complex canopy architecture. Environ Pollut 115: 425-436.

30    Matyssek, R; Le Thiec, D; Low, M; Dizengremel, P; Nunn, AJ; Haberle, KH. (2006).
31         Interactions between drought and Os stress in forest trees. Plant Biol (Stuttg) 8: 11-17.
32         http://dx.doi.org/10.1055/s-2005-873025
                                                  5-44

-------
 1    Matyssek, R; Sandermann, H; Wieser, G; Booker, F; Cieslik, S; Musselman, R; Ernst, D. (2008).
 2         The challenge of making ozone risk assessment for forest trees more mechanistic. Environ
 3         Pollut 156: 567-582. http://dx.doi.Org/10.1016/j.envpol.2008.04.017

 4    McLaughlin, SB; Nosal, M; Wullschleger, SD; Sun, G. (2007a). Interactive effects of ozone and
 5         climate on tree growth and water use in a southern Appalachian forest in the USA. New
 6         Phytol 174: 109-124. http://dx.doi.Org/10.llll/j.1469-8137.2007.02018.x

 7    McLaughlin, SB; Wullschleger, SD; Sun, G; Nosal, M. (2007b). Interactive effects of ozone and
 8         climate on water use, soil moisture content and streamflow in a southern Appalachian
 9         forest in the USA. New Phytol 174: 125-136. http://dx.doi.org/10.! 11 l/j.1469-
10         8137.2007.01970.x

11    Mereu, S; Gerosa, G; Finco, A; Fusaro, L; Muys, B; Manes, F. (2009). Improved sapflow
12         methodology reveals considerable night-time ozone uptake by Mediterranean species.
13         Biogeosciences 6: 3151-3162. http://dx.doi.org/10.5194/bg-6-3151-2009

14    Millennium Ecosystem Assessment (MEA). (2005). Ecosystems and Human Well-being:
15         CurrentState and Trends, Volume 1. R. Hassan, R. Scholes, and N. Ash, eds.
16         Washington,DCisland Press. Available at
17         http://www.millenniumassessment.Org/documents/document.766.aspx.pdfhttp://www.mille
18         nniumassessment.org/documents/document.766.aspx.pdf.

19    Morsky, SK; Haapala, JK; Rinnan, R; Tiiva, P; Saarnio, S;  Silvola, J; Holopainen, T;
20         Martikainen, PJ. (2008). Long-term ozone effects on vegetation, microbial community and
21         methane dynamics of boreal peatland microcosms in open-field conditions. Global Change
22         Biol 14: 1891-1903. http://dx.doi.Org/10.llll/j.1365-2486.2008.01615.x

23    Musselman, RC; McCool, PM; Younglove, T. (1988). Selecting ozone exposure statistics for
24         determining crop yield loss  from air pollutants. Environ Pollut 53: 63-78.
25         http://dx.doi.org/10.1016/0269-7491(88)90025-5

26    NAPAP (National Acid Precipitation Assessment Program). (1987). Diagnosing injury to
27         Eastern forest trees: A manual for identifying damage caused by air pollution, pathogens,
28         insects, and abiotic  stresses. University Park, PA: Pennsylvania State University.

29    Nussbaum, S; Geissmann, M; Fuhrer, J. (1995). Ozone exposure-response relationships for
30         mixtures of perennial ryegrass and white clover depend on ozone exposure patterns. Atmos
31         Environ 29: 989-995. http://dx.doi.org/10.1016/1352-2310(94)00368-U

32    Oksanen, E; Holopainen, T. (2001). Responses of two birch (Betula pendula Roth) clones to
33         different ozone profiles with similar AOT40 exposure. Atmos Environ 35:  5245-5254.
34         http://dx.doi.org/10.1016/S1352-2310(01)00346-6
                                                  5-45

-------
 1    Ollinger, SV; Aber, JD; Reich, PB. (1997). Simulating ozone effects on forest productivity:
 2         Interactions among leaf- and stand-level processes. Ecol Appl 123: 351-358.

 3    Ollinger, SV; Aber, JD; Reich, PB; Freuder, RJ. (2002). Interactive effects of nitrogen
 4         deposition, tropospheric ozone, elevated CC>2 and land use history on the carbon dynamics
 5         of northern hardwood forests. Global Change Biol 8: 545-562.
 6         http://dx.doi.0rg/10.1046/j.1365-2486.2002.00482.x

 7    Orendovici-Best, T; Skelly, JM; Davis, DD; Ferdinand, JA;  Savage, JE; Stevenson, RE. (2008).
 8         Ozone uptake (flux) as it relates to ozone-induced foliar symptoms of Prunus serotina and
 9         Populus maximowizii x trichocarpa. Environ Pollut 151: 79-92.
10         http://dx.doi.0rg/10.1016/j.envpol.2007.03.003

11    Pan, YD; Birdsey, R; Horn, J; McCullough, K. (2009). Separating effects of changes in
12         atmospheric composition, climate and land-use on carbon sequestration of US Mid-Atlantic
13         temperate forests. For Ecol Manage 259: 151-164.
14         http://dx.doi.0rg/10.1016/j.foreco.2009.09.049

15    Panek, JA. (2004). Ozone uptake, water loss and carbon exchange dynamics in annually drought-
16         stressed Pinus ponderosa forests: Measured trends and parameters for uptake modeling.
17         Tree Physiol 24: 277-290.

18    Panek, JA; Goldstein, AH. (2001). Responses of stomatal conductance to drought in ponderosa
19         pine: Implications for carbon and ozone uptake. Tree Physiol 21: 337-344.

20    Paoletti, E; Manning, WJ. (2007).  Toward a biologically significant and usable standard for
21         ozone that will also protect plants. Environ Pollut 150: 85-95.
22         http://dx.doi.0rg/10.1016/j.envpol.2007.06.037

23    Parsons, WFJ; Bockheim, JG; Lindroth, RL. (2008). Independent, interactive, and species-
24         specific responses of leaf litter decomposition to elevated  CO2 and Os in a northern
25         hardwood forest. Ecosystems 11: 505-519.

26    Pinto, J. (2009). Wyoming winter smog. Nat Geosci 2: 88-90. http://dx.doi.org/10.1038/ngeo430

27    Pleijel, H; Danielsson, H; Ojanpera, K; De Temmerman, L; Hogy, P; Badiani, M; Karlsson, PE.
28         (2004). Relationships between ozone exposure and yield loss in European wheat and
29         potato—a comparison of concentration- and flux-based exposure indices. Atmos Environ
30         38:2259-2269.

31    Pretzsch, H; Dieler, J; Matyssek, R; Wipfler, P. (2010). Tree and stand growth of mature Norway
32         spruce and European beech under long-term ozone fumigation. Environ Pollut 158:  1061-
33         1070. http://dx.doi.0rg/10.1016/j.envpol.2009.07.035
                                                  5-46

-------
 1    Pritsch, K; Esperschuetz, J; Haesler, F; Raidl, S; Winkler, B; Schloter, M. (2009). Structure and
 2         activities of ectomycorrhizal and microbial communities in the rhizosphere of Fagus
 3         sylvatica under ozone and pathogen stress in a lysimeter study. Plant Soil 323: 97-109.
 4         http://dx.doi.org/10.1007/slll04-009-9972-6

 5    Ren, W; Tian, H; Chen, G; Liu, M; Zhang, C; Chappelka, AH; Pan, S. (2007a). Influence of
 6         ozone pollution and climate variability on net primary productivity and carbon storage in
 7         China's grassland ecosystems from 1961 to 2000. Environ Pollut 149: 327-335.
 8         http://dx.doi.0rg/10.1016/j.envpol.2007.05.029

 9    Ren, W; Tian, HQ; Liu, ML; Zhang, C;  Chen, GS; Pan, SF; Felzer, B; Xu, XF. (2007b). Effects
10         of tropospheric ozone pollution on net primary productivity and carbon storage in
11         terrestrial ecosystems of China. J Geophys Res 112: D22S09.
12         http://dx.doi.org/10.1029/2007jd008521

13    Sanchez, MJS; Pena, GS; Lorente, VC;  Gallego, TM; Albert, JC. (2001). La contamination
14         atmosferica en los bosques: Guia para la identificacion de dafios visibles causados por
15         Ozono. Madrid, Spain: Ministerio de Medio Ambiente.
16         http://www.ceam. es/VentaLibros/guia_O3/index. htm

17    Sitch, S; Cox, PM; Collins, WJ; Huntingford, C. (2007). Indirect radiative forcing of climate
18         change through ozone effects on the land-carbon sink. Nature 448: 791-794.
19         http://dx.doi.org/10.1038/nature06059

20    Smith, W. H. (1992). Air pollution effects on ecosystem processes. In: Barker, J. R.; Tingey, D.
21         T., eds. Air pollution effects on biodiversity. Van Nostrand Reinhold; pp. 234-260.

22    Smith, G; Coulston, J; Jepsen, E; Prichard, T. (2003). A national ozone biomonitoring program:
23         Results from field surveys of ozone sensitive plants in northeastern forests (1994-2000).
24         Environ Monit Assess 87: 271-291.

25    Stevenson, DS; Dentener, FJ; Schultz, MG; Ellingsen, K; Van Noije, TPC; Wild, O; Zeng, G;
26         Amann, M; Atherton, CS; Bell, N; Bergmann, DJ; Bey, I; Butler, T; Cofala, J; Collins, WJ;
27         Derwent, RG; Doherty, RM; Drevet, J; Eskes, HJ; Fiore, AM; Gauss, M; Hauglustaine,
28         DA; Horowitz, LW; Isaksen, ISA; Krol, MC; Lamarque, JF; Lawrence, MG; Montanaro,
29         V; Muller, JF; Pitari, G; Prather, MJ; Pyle, JA; Rast, S; Rodriguez, JM; Sanderson, MG;
30         Savage, NH; Shindell, DT; Strahan, SE; Sudo, K; Szopa, S. (2006). Multimodel ensemble
31         simulations of present-day and near-future tropospheric ozone. J Geophys Res 111:
32         D08301. http://dx.doi.org/10.1029/2005JD006338

33    Temple, PJ; Kupper, RS; Lennox,  RW; Rohr, K. (1988). Injury and yield responses of
34         differentially irrigated cotton to ozone. Agron  J 80: 751-755.
35         http://dx.doi.org/10.2134/agroni 1988.0002196200800005001 Ix
                                                   5-47

-------
 1    Temple, PJ; Riechers, GH; Miller, PR. (1992). Foliar injury responses of ponderosa pine
 2         seedlings to ozone, wet and dry acidic deposition, and drought. Environ Exp Bot 32: 101-
 3         113. http://dx.doi.org/10.1016/0098-8472(92)90035-Z

 4    Turnipseed, AA; Burns, SP; Moore, DJP; Hu, J; Guenther, AB; Monson, RK. (2009). Controls
 5         over ozone deposition to a high elevation subalpine forest. Agr Forest Meteorol 149: 1447-
 6         1459. http://dx.doi.0rg/10.1016/j.agrformet.2009.04.001

 7    Uddling, J; Hogg, AJ; Teclaw, RM; Carroll, MA; Ellsworth, DS. (2010). Stomatal uptake of O3
 8         in aspen and aspen-birch forests under free-air CO2 and O3 enrichment. Environ Pollut
 9         158: 2023-2031. http://dx.doi.Org/10.1016/j.envpol.2009.12.001

10    Uddling, J; Teclaw, RM; Pregitzer, KS; Ellsworth, DS. (2009). Leaf and canopy conductance in
11         aspen and aspen-birch forests under free-air enrichment of carbon dioxide and ozone. Tree
12         Physiol 29: 1367-1380. http://dx.doi.org/10.1093/treephys/tpp070

13    UNEP (United Nations Environment Programme). (2003). Millennium Ecosystem Assessment:
14         Ecosystems and human well-being: A framework for assessment. Washington, DC: Island
15         Press.

16    US EPA (2012a). Integrated Science Assessment of Ozone and Related Photochemical Oxidants
17         (Third External Review Draft). U.S. Environmental Protection Agency, Washington, DC.
18         EPA/600/R-10/076C. June 2012.  Available at:
19         http://www.epa.gov/ttn/naaqs/standards/ozone/s_o3_2008_isa.html

20    US EPA (2012b). Health Risk and Exposure Assessment for Ozone, First External Review
21         Draft.  Office of Air Quality Planning and Standards, Research Triangle Park, NC, 27711.
22         EPA 452/P-12-001. July 2012.  Available at:
23         http://www.epa.gov/ttn/naaqs/standards/ozone/s_o3_2008_rea.html

24    US EPA (2012c). Welfare Risk and Exposure Assessment for Ozone, First External Review
25         Draft.  Office of Air Quality Planning and Standards, Research Triangle Park, NC, 27711.
26         EPA 452/P-12-004. July 2012.  Available at:
27         http://www.epa.gov/ttn/naaqs/standards/ozone/s_o3_2008_rea.html

28    US EPA (2009). Risk and Exposure Assessment for the Review of the Secondary national
29         Ambient Air Quality Standards for Oxides of Nitrogen and  Oxides of Sulfur.
30         EPA/452/R/09/008a. U. S environmental Protection Agency, Office of Air Quality planning
31         and Standards, Health and  Environmental Impacts Division, Research Triangle Park, NC.

32    US EPA (2006). Air Quality Criteria for Ozone and Related Photochemical Oxidants (2006
33         Final).  U.S. Environmental Protection Agency, Washington, DC.  EPA/600/R-05/004aF-
34         cF. March 2006. Available at:
35         http://www.epa.gOv/ttn/naaqs/standards/ozone/s _o3_cr_cd. html

                                                  5-48

-------
 1    US EPA (1996). Air quality criteria for ozone and related photochemical oxidants [EPA Report].
 2         (EPA/600/P-93/004AF).  U.S. Environmental Protection Agency, Research Triangle Park,
 3         NC.

 4    US EPA (1984). Air quality criteria for ozone and other photochemical oxidants, volume III of V
 5         (review draft) [EPA Report]. (EPA-600/8-84-020A3). Research Triangle Park, NC.
 6         http://www.ntis.gov/search/product.aspx? ABBR=PB85126050

 7    US EPA (1978). Air quality criteria for ozone and other photochemical oxidants [EPA Report].
 8         (EPA/600/8-78/004). Washington, DC.

 9    US Forest Service, National Park Service, and U.S. Fish and Wildlife Service. (2000). Federal
10         land managers air quality related values work group (FLAG): Phase I report. Denver, CO:
11         U.S. Forest Service, http://www.nature.nps.gov/air/permits/flag/index.cfm

12    Volk, M; Obrist, D; Novak, K; Giger, R; Bassin, S; Fuhrer, J. (2011). Subalpine grassland
13         carbon dioxide fluxes indicate substantial carbon losses under increased nitrogen
14         deposition, but not at elevated ozone concentration. Global Change Biol 17: 366-376.
15         http://dx.doi.0rg/10.llll/j.1365-2486.2010.02228.x

16    Vollsnes, AV; Kruse, OMO; Eriksen, AB; Oxaal, U; Futsaether, CM. (2010). In vivo root
17         growth dynamics of ozone exposed Trifolium subterraneum. Environ Exp Bot 69:  183-188.
18         http://dx.doi.0rg/10.1016/j.envexpbot.2010.03.007

19    Weinstein, DA; Gollands, B; Retzlaff, WA. (2001). The effects of ozone on a lower slope forest
20         of the Great Smoky Mountain National Park: Simulations linking an individual tree model
21         to a stand model. Forest Sci 47:  29-42.

22    Weinstein, DA; Laurence, JA; Retzlaff, WA; Kern, JS; Lee, EH; Hogsett, WE; Weber, J. (2005).
23         Predicting the effects of tropospheric ozone on regional productivity of ponderosa  pine and
24         white fir. For Ecol Manage 205: 73-89. http://dx.doi.Org/10.1016/j.foreco.2004.10.007

25    West, JJ; Fiore, AM. (2005). Management of tropospheric ozone by reducing methane
26         emissions. Environ Sci Technol 39: 4685-4691.

27    West, JJ; Fiore, AM; Horowitz, LW; Mauzerall, DL. (2006). Global health benefits of mitigating
28         ozone pollution with methane emission controls. PNAS 103: 3988-3993.
29         http://dx.doi.org/10.1073/pnas.0600201103

30    West, JJ; Fiore, AM; Naik, V; Horowitz, LW; Schwarzkopf, MD; Mauzerall, DL. (2007). Ozone
31         air quality and radiative forcing consequences of changes in ozone precursor emissions.
32         Geophys Res Lett 34: L06806. http://dx.doi.org/10.1029/2006GL029173
                                                  5-49

-------
 1   Wieser, G; Manning, WJ; Tausz, M; Bytnerowicz, A. (2006). Evidence for potential impacts of
 2         ozone on Pinus cembra L. at mountain sites in Europe: An overview. Environ Pollut 139:
 3         53-58. http://dx.doi.0rg/10.1016/j.envpol.2005.04.037

 4   Wittig, VE; Ainsworth, EA; Long, SP. (2007). To what extent do current and projected increases
 5         in surface ozone affect photosynthesis and stomatal conductance of trees? A meta-analytic
 6         review of the last 3 decades of experiments [Review]. Plant Cell Environ 30: 1150-1162.
 7         http://dx.doi.0rg/10.llll/j.1365-3040.2007.01717.x

 8   Wittig, VE; Ainsworth, EA; Naidu, SL; Karnosky, DF; Long, SP. (2009). Quantifying the impact
 9         of current and future tropospheric ozone on tree biomass, growth, physiology and
10         biochemistry: A quantitative meta-analysis. Global Change Biol 15: 396-424.
11         http://dx.doi.org/10.! 111/j. 1365-2486.2008.01774.X

12   Young, T. F.; Sanzone, S., eds. (2002). A framework for assessing and reporting on ecological
13         condition: an SAB report. Washington, DC: U.S. Environmental Protection Agency,
14         Science Advisory Board; report no. EPASAB- EPEC-02-009. Available:
15         http://www.epa.gov/sab/pdf/epec02009.pdf [9 December, 2003].

16   Yun, SC; Laurence, JA. (1999). The response of sensitive and tolerant clones of Populus
17         tremuloides to dynamic ozone exposure under controlled environmental conditions. New
18         Phytol 143:305-313.

19   Zhang, C; Tian, HQ; Chappelka, AH; Ren, W; Chen, H; Pan, SF; Liu, ML; Styers, DM; Chen,
20         GS; Wang, YH. (2007). Impacts of climatic and atmospheric changes on  carbon dynamics
21         in the Great Smoky Mountains National Park. Environ Pollut 149: 336-347.
22         http://dx.doi.0rg/10.1016/j.envpol.2007.05.028

23   Zhang, J; Schaub, M; Ferdinand, JA; Skelly, JM; Steiner, KC; Savage, JE. (2010). Leafage
24         affects the responses of foliar injury and gas exchange to tropospheric ozone in Prunus
25         serotina seedlings. Environ Pollut 158: 2627-2634.
26         http://dx.doi.0rg/10.1016/j.envpol.2010.05.003

27

28

29
                                                  5-50

-------
 1       6.     CONSIDERATION OF EXPOSURE AND RISK ASSESSMENTS

 2          In order to put (Vrelated vegetation and ecosystems effects information discussed in the
 3    ISA into the context of welfare risks at the national scale, the first draft Welfare REA has
 4    performed analyses to estimate Os exposures to vegetation and ecosystems and (^-associated
 5    risks related to these welfare effects across the United States, including impacts in federal Class I
 6    case study areas (US EPA, 2012b). The first draft REA has estimated the Os exposures and
 7    welfare risks that are associated with recent air quality and with air quality adjusted to simulate
 8    just meeting the current O3 standard (US EPA, 2012b).  The second draft and final REA will also
 9    include estimates of Os exposures and risks associated with air quality adjusted to simulate just
10    meeting potential alternative 63 standards.1
11          The first draft REA has identified the following goals for the exposure and risk
12    assessments: (1) to provide estimates of the ecological effects of O3 exposure across a range of
13    environments;  (2) to provide estimates  of ecological effects within selected case study areas; (3)
14    to provide estimates of the effects of 63 exposure on specific urban and non-urban ecosystem
15    services based  on the causal ecological effects; and (4) to develop a better understanding of the
16    response of ecological systems and ecosystem services to changing levels of Os exposure to
17    inform the PA  regarding alternative standards that might be considered. This quantitative
18    exposure and risk assessment builds on the approach used and lessons learned in the last Os risk
19    assessment and focuses on improving the characterization of the overall confidence in the risk
20    estimates, including related uncertainties, by incorporating a number of enhancements, in terms
21    of both the methods and data used in the analyses. This assessment considers a variety of welfare
22    endpoints for which, in staff s judgment, there is adequate information to develop quantitative
23    risk estimates that can meaningfully inform the review of the secondary Os NAAQS (US EPA,
24    2012c, section  1.2).
25          Air quality inputs to the exposure and risk assessments include (1) ambient monitoring
26    data for all Air Quality Systems (AQS) monitors in the U.S. for several relevant metrics for
27    2006-2010; (2) a CMAQ model results for 2007; and (3) a national-scale spatial surface
28    generated by fusing 2006-2008 measured ambient Os data for all AQS monitors in the U.S. with
29    the 2007 CMAQ model simulation for  the three-year period of 2006-2008 and for each
30    individual year within that period. The  generated national-scale spatial surface provides
31    estimates of W126 concentrations throughout the U.S. for 2006-2008 and of scenarios of just
      1 Air quality simulations are meant to provide perspective on the O3-associated exposures and welfare risks under
      different air quality scenarios. These simulations do not reflect any consideration of specific control programs or
      strategies designed to achieve the reductions in emissions required to meet the specified standards. Further, these
      simulations do not represent predictions of when, whether, or how areas might meet the specified standards.
                                                    6-1

-------
 1    meeting the current O3 standard of 0.075 ppm. The 2006-2008 W126 national-scale "fused"
 2    spatial surface reflects changes in the distribution of Os air quality estimated to occur when an
 3    area just meets the current Os NAAQS.
 4           Simulation of just meeting the current Os standard is accomplished in the first draft REA
 5    using a quadratic rollback method, similar to that used in the previous risk and exposure analysis
 6    for the 2008 O3 NAAQS review (U. S. EPA, 2007a,b,c). In evaluating just meeting the current
 7    standard, the first draft REA focuses on air quality changes that are likely to occur as the U.S.
 8    puts in place programs to meet the standard. As  such, the REA uses U.S. background
 9    concentrations as a floor for the quadratic rollback.  The first draft REA also explores alternative
10    simulation approaches based on modeled sensitivities of 63 to U.S. emissions. These alternative
11    approaches, which will be evaluated more fully in the second draft REA,2 will remove the need
12    for imposing a specific floor to prevent adjustments beyond those likely to occur due to U.S.
13    emissions reductions.
14           In selecting welfare effects endpoints on which to base estimates of (Vrelated impacts,
15    the first draft REA considers the weight-of-evidence conclusions from the ISA. Specifically, the
16    first draft REA notes the ISA conclusions that there is a causal relationship between Os exposure
17    and visible foliar injury, reduced growth and productivity and alteration of below-ground
18    biogeochemical cycles and likely  to be a causal relationship between Os exposure and reduced
19    carbon sequestration, alteration of terrestrial ecosystem water cycling, and community
20    composition.  In light of these conclusions, the first draft REA estimates vegetation effects in
21    terms of relative biomass loss following cumulative, seasonal Os exposures, and provides
22    preliminary estimates of the impacts of relative biomass loss on several ecosystem services
23    associated with forest ecosystems. The first draft REA also provides information on the
24    geographic extent of vegetation species that are sensitive to visible foliar injury, although
25    estimates of the magnitude of that injury under recent ozone conditions and after just meeting the
26    current standard will not be available until the second draft REA.  The remainder of this chapter
27    discusses the assessment of Os exposures (section 6.1) and preliminary Os-associated vegetation
28    risks (section 6.2) for recent 63 air quality concentrations and for  63 air quality adjusted to
29    simulate just meeting the current 8-hour Os standard.
      In the second draft, the REA will evaluate approaches for simulating attainment of current and alternative standards
      that are based on modeling the response of O3 concentrations to reductions in anthropogenic NOx and VOC
      emissions, using the Higher-order Decoupled Direct Method (HDDM) capabilities in the Community Multi-scale
      Air Quality (CMAQ) model. This modeling incorporates all known emissions, including emissions from non-
      anthropogenic sources and anthropogenic emissions from sources in and outside of the U.S. As a result, the need to
      specify values for U.S. background is not necessary, as it is incorporated in the modeling directly. The evaluation of
      this new approach is presented in Chapter 4 of the draft REA (US EPA, 2012b) and in Simon et al. (2012).
                                                     6-2

-------
 1           Two additional analyses have also been conducted. These assessments include (1) a
 2    simulation of changes in urban forest structure, functions and values using the i-Tree version 4.0
 3    (i.e., i-Tree Eco, previously known as UFORE. The Urban Forest Effects (UFORE) which
 4    contains protocols to measure and monitor urban forests as well as estimate ecosystem functions
 5    and values and (2) a simulation of market outcomes under alternative O3 concentrations using the
 6    Forest and Agricultural Sector Optimization Model with Greenhouse Gases (FASOMGHG).3
 7    These results, along with assessments of visible foliar injury and newly incorporated additional
 8    federal Class 1 case study and critical habitat area assessments in the second draft REA will be
 9    fully considered in the second draft PA.

10    6.1     EXPOSURE

11    6.1.1   Monitoring and Air Quality
12           State and local monitoring agencies operate O3 monitoring sites at various locations,
13    depending on the  size of the area and typical peak O3 concentrations (US EPA, 2012a, sections
14    3.5.6.1, 3.7.4).  In 2010, there were 1,250 State and local O3 monitors reporting concentrations to
15    EPA (US EPA, 2012a, Figures 3-21 and 3-22).  Ozone monitors are only required in
16    Metropolitan Statistical Areas (MSAs) with populations over 350,000.4 Since O3 concentrations
17    decrease significantly in the colder parts of the year in many areas, O3 is required to be
18    monitored only during the "O3 season," which varies by state (US EPA, 2012a, section 3.5.6 and
19    Figure 3-20).5 In 2010, there were approximately 112  monitoring sites being operated in rural
20    areas. These sites included 15 National Core (NCore) monitors, 80 Clean Air Status and Trends
21    Network (CASTNET) monitors, and 17 Portable O3  Monitoring Systems (POMS) network
22    monitors operated by the National Park Service  (NPS). The locations of these monitors are
23    shown in Figure 6-1 (US  EPA, 2012c, figure 4-2).
24           As discussed in chapter 1 above and in chapter 3 of the ISA, rural monitoring sites tend to
25    be less directly affected by anthropogenic pollution sources than urban sites, although rural sites
26    can be affected by transport of O3 or O3 precursors from upwind urban areas and by local
27    anthropogenic sources such as motor vehicles, power generation, biomass combustion, or oil and
28    gas operations (US EPA,  2012a, section 3.6.2).  In addition, O3 tends to persist longer in rural
      3These analyses are included in a revised version of chapter 6 of the Welfare REA and its associated appendices that
      is being released in parallel with this first draft Policy Assessment.
      4The current monitor and probe siting requirements have an urban focus and do not address siting in non-urban, rural
      areas.  States may operate O3 monitors in non-urban or rural areas to meet other objectives (e.g., support for
      research studies of atmospheric chemistry or ecosystem impacts).
      5 Some States and Territories operate O3 monitors year-round, including Arizona, California, Hawaii, Louisiana,
      Nevada, New Mexico, Puerto Rico, Texas, American Samoa, Guam and the Virgin Islands.
                                                     6-3

-------
 1   than in urban areas due to lower rates of chemical scavenging in non-urban environments. At
 2   higher elevations, increased Os concentrations can also result from stratospheric intrusions (US
 3   EPA, 2012a, sections 3.4, 3.6.2).  As a result, Os concentrations measured in some rural sites can
 4   be higher than those measured in nearby urban areas (US EPA, 2012a, section 3.6.2) and the ISA
 5   concludes that cumulative exposures for humans and vegetation in rural areas can be substantial,
 6   and they are often higher than cumulative exposures in urban areas (US EPA, 2012a, section
 7   3.7.5).
                                                        9  mnmm
                                                        *  eiuiHer
                  IAHIIP   iffl«
10   Figure 6-1    U.S. Rural NCore, CASTNET and NFS POMS sites in 2010 (U.S. EPA,
11                 2012a, Figure 3-22)
12
13   6.1.2  Approach to Estimating Exposures
14          Recent Os air quality information from both monitored and modeled sources is used to
15   assess current exposures, as well as to estimate the relative change in exposure resulting from
16   adjusting Os concentrations to simulate just meeting the current Os standard. The air quality
17   monitoring data used to inform the first draft Cb Risk and Exposure Assessments are hourly Cb
18   concentrations collected between 1/1/2006 and 12/31/2010 from all US monitors meeting EPA's
                                                  6-4

-------
 1    siting, method, and quality assurance criteria and extracted from EPA's AQS database6. These
 2    exposures are assessed both in terms of the current secondary 8-hour average form and in terms
 3    of the cumulative concentration weighted W126 exposure metric as it was proposed in 2007 and
 4    2010 and which EPA staff continues to conclude is appropriate for consideration in this review
 5    (Chapter 5 above).
 6           In addition to the selection of a the W126 metric, which uses a logistic weighting
 7    function to place less emphasis on exposure to low concentrations and more emphasis on
 8    exposure to high concentrations (Lefohn et al,  1988), as the form of the standard (US EPA,
 9    2012a, section 9.5.2, equation 9-1), EPA staff has again revisited the appropriateness of the
10    diurnal and seasonal exposure periods (averaging times) defined in previous reviews.  As
11    discussed above in chapter 5, staff has again concluded that a 12 hour diurnal (defined as 8:00
12    am to 8:00 pm local time) and maximum consecutive Smonth seasonal exposure period remain
13    appropriate.
14    6.1.3   Exposure Surface Generation
15           Due to the fact that the Os monitoring network is limited in some parts of the country,
16    especially in rural areas, there was a need to generate a national 63 exposure surface.  The
17    welfare risk and exposure assessment therefore analyzed a national-scale spatial surface of W126
18    for the three-year period of 2006-2008 and for each individual year: 2006, 2007 and 2008. This
19    analysis employed a data fusion approach to take advantage of the accuracy of monitor
20    observations and the comprehensive spatial information of the CMAQ modeling system to create
21    a national-scale "fused" spatial  surface of seasonal average 63. The spatial surface is created by
22    fusing 2006-2008 measured ambient O3 data for all AQS monitors in the U.S. with the 2007
23    CMAQ model simulation run for a 12 km gridded domain, using the EPA's Model Attainment
24    Test Software (MATS; Abt Associates, 2010), which employs the enhanced Voronoi Neighbor
25    Averaging (eVNA) technique (Timin et al., 2010) enhanced with information on the spatial
26    gradient of O?, provided by CMAQ results.  The generated national-scale spatial surface provides
27    estimates of W126 concentrations throughout the U.S. for 2006-2008  and for scenarios of just
28    meeting the current Os standard of 0.075 ppm.  The ambient 2006-2008 W126 national-scale
29    "fused" spatial surface is shown in Figure 6-2. More detail on the ambient measurements and the
30    2007 CMAQ model simulation, as well as the spatial fusion technique, can be found in Wells et
31    al. (2012).
      EPA's AQS database is a state-of-the-art repository for many types of air quality and related monitoring data. AQS
      contains monitoring data for the six criteria pollutants dating back to the 1970's, as well as more recent additions
      such as air toxics, meteorology, and quality assurance data. At present, AQS receives O3 monitoring data collected
      hourly from over 1,300 monitors, and quality assured by one of over 100 state, local, or tribal air quality monitoring
      agencies.
                                                    6-5

-------
                                  Recent Ambient Conditions
            | 10.01 -15.00
            | 15.01 -20.00
              20.01 -25.00
            | 25.01 -30.00
          ^H 30-01 -35.00
            • >35
 2
 3   Figure 6-2.   National Ozone Exposure Surface of Recent (2006 - 2008) Ambient W126

 4          To generate a national-scale spatial surface that represents 2006-2008 W126
 5   concentrations after simulating just meeting the current NAAQS, the spatial surface for 2006-
 6   2008 air quality was adjusted to reflect the rolled-back W126 monitor concentrations at non-
 7   attaining monitors in urban areas. To do this, the rolled-back W126 monitor values were inserted
 8   into the spatial surface at the monitor locations and the W126 surface was smoothed using the
 9   Voronoi Neighbor Averaging (VNA) spatial averaging technique to minimize any sharp
10   gradients between the national-scale spatial surface that represents 2006-2008 W126
11   concentrations and the rolled-back W126 monitor concentrations.
12          Figure 6-3 shows the national-scale 2006-2008 W126 surface that reflects simulation of
13   just meeting the current standard of 0.075 ppm. The state of California was most affected by the
14   rollback, with average changes in W126 of around 20 ppm-hours. Other areas with notable
15   changes include the areas around Atlanta, Charlotte, Denver, Phoenix, and Salt Lake City and the
                                                   6-6

-------
 1
 2
 3
 4
6
7
    area between Washington, D.C., and Boston (all areas that had relatively high 8-hour 63
    concentrations above the current standard).
                                   Current Standard Rollback Scenario
                    0.45-5.00
                    5.01 -10.00
                    10.01 -15.00
                    15.01 -20.00
                    20.01 -25.00
                    25.01 -30.00
                    30.01 -35.00
                    >35
     Figure 6-3.   National Ozone Exposure Surface Simulating Just Meeting the Current
                   Standard
 8   6.2 CONSIDERATION OF O3 RISK ESTIMATES

 9   6.2.1 Approaches to Assessing Risks
10          In the previous review of the secondary standards, the focus of the ecological risk
1 1   assessment was on estimation of changes in tree seedling biomass and agricultural crop yields
12   and resulting impacts on forest and agricultural production as well as qualitative consideration of
13   effects on ecosystem services.  As mentioned above, in this review, we are focusing on those
14   effects determined to have a causal or likely to be a causal relationship to O3 exposure.
                                                   6-7

-------
 1    Therefore, we are continuing to focus on the quantitative assessment of 63 impacts to trees,
 2    including individual tree species relative biomass loss, visible foliar injury, aggregated and
 3    scaled relative biomass loss, modeled impacts to trees in urban settings, and ecosystem case
 4    studies.  We have placed less focus on impacts to agricultural ecosystems given the highly
 5    managed nature of those systems; however, in assessing impacts of Os on ecosystem services
 6    related to changes in forest yields, we include the impacts of 63 on both forest and agricultural
 7    production in order to properly model the interactions between the two production systems in
 8    determining land use decisions. In addition, we are expanding the discussion of ecosystem
 9    services to include a broader array of impacts, both quantitative and qualitative, resulting from
10    known or anticipated effects of Os on ecosystem functions.  EPA has begun using an ecosystem
11    services framework to help inform determinations of the adversity to public welfare associated
12    with changes in ecosystem functions (Rea et al., 2012).
13          These services may be characterized as supporting services that are necessary for all
14    other services (e.g., primary  production); cultural services including existence and bequest
15    values, aesthetic values, and recreation values, among others; provisioning services (e.g., food
16    and timber); and regulating services such as climate regulation or hydrologic cycle (MEA, 2005).
17    The REA identifies those ecosystem services that are associated with the ecological effects
18    caused or likely to be caused by 63 exposure.  For the first draft REA, ecosystem services
19    potentially impacted by O3 were able to be quantified for two classes of ecosystem services
20    including those associated with commercial forestry and those associated with urban trees.
21    These cover only a small portion of the total ecosystem services that might be impacted by 63,
22    and the REA provides contextual information regarding the nature and magnitude of additional
23    ecosystem services that may be impacted by Os exposures, noting that the specific magnitude  of
24    the impact of Os on those ecosystem services is unknown.
25    6.2.2  Risk of Biomass Loss in Individual Sensitive Tree Species
26          The previous O3 AQCDs (US EPA, 1996, 2006) and current third draft ISA (US EPA,
27    2012a) conclude that there is strong and consistent evidence that ambient concentrations of 63
28    decrease photosynthesis and growth in numerous plant species across the U.S. The quantitative
29    exposure-response relationships described in the 2006 63 AQCD have not changed in the current
30    draft ISA, with the exception of the addition of eastern cottonwood, and their continued
31    usefulness and relevance has been further confirmed by the EPA staff analyses comparing the
32    performance and predictive capability of OTC-derived C-R functions using data from FACE
33    studies (US EPA, 2012a, section 9.6.3).  Therefore, EPA staff concludes  that these response
34    functions provide an adequate basis for quantifying biomass loss damages in the risk assessment.

-------
 1          The exposure-response functions summarized in the third draft ISA were computed using
 2   the W126 metric and cumulated over 90 days. The median response functions for relative
 3   biomass loss (RBL) for each of the 11 tree species used in the risk assessment are shown in
 4   Figure 6-4.  Understanding the differences in the shapes of these curves can aid in understanding
 5   the results of analyses presented later in this chapter.  From this figure it is clear that there is
 6   significant variability in 63 sensitivity across studied species. Some species are extremely
 7   sensitive (eastern cottonwood and black cherry), even at very low levels of W126, while others
 8   appear to be relatively insensitive (Douglas fir and Virginia pine). Several species have response
 9   functions with gradual and consistent slopes, resulting in a more or less constant rate of change
10   in RBL over a range of Os exposure consistent with ambient exposure levels. In contrast, other
11   species have response functions that indicate large changes in RBL over a small range of Os
12   concentrations and relatively small changes in RBL over other 63 concentration ranges.

                                              A
13
14
15
16
17
            CO
            0

            CD
            O
        CD
            CXI
            O

            p
            O
                  \
                  0
                     20     40      60
                               W126
                                  80     100
                                                                  Eastern Cottonwood
                                                                  Sugar Maple
                                                                  Tufip Poplar
                                                                  Black Cherry
                                                                  White Pine
                                                             Quaking Aspen
                                                             Douglas Fir
                                                             Ponderosa Pine
                                                             Red Alder

                                                             Red Maple
                                                             Virginia Pine
Figure 6-4.
Relative Biomass Loss Concentration-Response Functions for 11 Tree
species. The vertical dashed lines represent the maximum W126 values for
(A) recent (2006 to 2008) conditions, and (B) just meeting the current
standard.
                                                 6-9

-------
 1
 2          The vertical dashed lines in Figure 6-4 represent the maximum W126 observed across the
 3   U.S. for the recent ambient exposure surface (brown dashed line) and simulated for the just
 4   meeting the current standard scenario (gray dashed line). The RBL value at the point where the
 5   response curve for each species intersects these lines represents the maximum potential RBL
 6   value for that species under the two Os exposure scenarios.
 7          Figure 6-5 zooms in on the portion of Figure 6-4 that includes W126 values from zero up
 8   to a W126 of 30, which is close to the maximum W126 value after simulating just meeting the
 9   current standard.  This figure is important in showing the predicted levels of RBL loss upon just
10   meeting the current secondary standard.  After simulating just meeting the current standard,
11   maximum potential RBL exceeds 20 percent for a number of ecologically and economically
12   important species, including sugar maple, black cherry, tulip poplar,  and white pine.
13
                                                  6-10

-------
 1
 2
 3
 4
 5
 6
 7
 8
 9
10
11
12
13
         CD
             CO
             o
             c\j
             o
             p
             o
                   \
                  0
                               \
                               5
                                                                    Eastern Cottonwood
                                                                    Black Cherry
                                                                         Sugar Maple
                                                                         Tulip Poplar
                                                                         White Pine
                                                                         Quaking Aspen

                                                                         Ponderosa Pine
                                                                         Red Alder
                                                                         Red Maple
                                                                         Virgil
                                                                         Douglas Fir
10    15     20
     W126
25    30
Figure 6-5.   "Blow Up" View of Relative Biomass Loss Concentration-Response
             Functions for 11 Tree species. Gray Horizontal Lines show RBL Levels of
             5%, 10%, 15%, and 20%.
      Using the functions shown in Figure 6-4 with the air quality surfaces presented in Figures
6-2 and 6-3, staff generated median and maximum RBL values across the grid cells within each
species range. Table 6-1 summarizes the results of the analyses for individual tree species under
ambient conditions and simulated to meet the current standard. It is important to note that under
the scenario of just meeting the current standard, taking into account O3 exposures within the
geographic range of each species, maximum RBL values for five species exceedlO%, and for
two species (eastern cottonwood and black cherry) exceed 40%.
                                           6-11

-------
 1   Table 6-1.    Summary of Tree Species Biomass Loss Analyses
Species
Eastern Cottonwood (Populus deltoides)
Black Cherry (Prunus serotina)
Quaking Aspen (Populus tremuloides)
Eastern White Pine (Pinus strobus)
Tulip Poplar (Liriodendron tulipifera)
Ponderosa Pine (Pinus ponder osti)
Red Alder (Alnus rubra)
Sugar Maple (Acer saccharum)
Red Maple (Acer rubrum)
Virginia Pine (Pinus virginiana)
Douglas Fir (Pseudotsuga menzeiesii)
Median RBL
(Ambient)
0.564
0.225
0.039
0.034
0.045
0.038
0.005
0.000
0.009
0.008
0.000
Maximum RBL
(Ambient)
0.999
0.547
0.377
0.226
0.291
0.294
0.118
0.206
0.039
0.018
0.0012
Median RBL
(Current
Standard)
0.464
0.194
0.035
0.028
0.030
0.032
0.005
0.000
0.007
0.006
0.000
Maximum RBL
(Current
Standard)
0.990
0.420
0.189
0.150
0.138
0.130
0.106
0.042
0.028
0.011
0.0006
 3          The first draft REA also expanded on the analysis from the previous review by including
 4   metrics that weight the RBL for individual species by their relative abundance within forest
 5   ecosystems. These metrics provide one measure of the potential importance of species level 63
 6   impacts for overall forest ecosystem health.  However, these importance weighted RBL metrics
 7   are most useful as relative indicators of changes in risk, and not as useful as absolute measures of
 8   ecosystem risk because 1) the units are not straightforward to interpret (as compared to RBL
 9   which is simply a percent of potential biomass), and 2) because we only have C-R functions for
10   11 different tree species, which represents a very small subset of the total plant species in most
11   forest ecosystems, the aggregate 63 impacts  on most forest ecosystems are not well
12   characterized. However, the proportional change in the importance weighted RBL values
13   between just meeting the current standard and meeting potential alternative standards will be
14   useful in evaluating the impacts of moving from the current 8-hour form to the W126 form, as
15   well as evaluating potential alternative levels of the W126 standard. The importance weighted
16   metrics are also useful in allowing us to focus on potential impacts on RBL and associated
17   ecosystem services in geographic areas of interest, including selected federal Class I areas and
                                                  6-12

-------
 1    critical habitats. Analyses of alternative standard levels using these importance weighted RBL
 2    metrics will be included in the second draft REA, and will be evaluated in the second draft PA.
 3    6.2.3  Risk of Foliar Injury Incidence
 4          Visible foliar injury resulting from exposure to 63 has been well characterized and
 5    documented over several decades for many tree, shrub, herbaceous, and crop species (US EPA,
 6    2012a, 2006, 1996, 1984, 1978). Visible foliar injury symptoms are considered diagnostic as
 7    they have been verified experimentally in exposure-response studies using exposure
 8    methodologies such as open top chambers and free-air fumigation (see section 9.2 of the ISA for
 9    more detail on exposure methodologies). Although the majority of (Vinduced visible foliar
10    injury occurrence has been observed on seedlings and small plants, many studies have reported
11    visible injury to mature coniferous trees, primarily in the western U.S. (Arbaugh  et al., 1998) and
12    to mature deciduous trees in eastern North America (Schaub et al., 2005; Vollenweider et al.,
13    2003; Chappelka et al., 1999a; Chappelka et al., 1999b; Somers et al., 1998; Hildebrand et al.,
14    1996).
15          The first draft REA includes a limited screening analysis of risk of visible foliar injury
16    based on an updating of a 2007 analysis done for the National Park Service (Kohut et al,  2007).
17    This screening analysis focused on indicators of foliar risk including two Os exposure indices, a
18    measure of soil moisture, and an indicator of the presence of 63 sensitive species within a park.
19    While this screening analysis provides some limited information regarding parks that may be
20    likely to experience Os related foliar damages, it does not quantify the specific extent of visible
21    foliar injury. The REA for the previous review included additional information regarding visible
22    foliar damage based on the Forest Health Monitoring Program. We plan to update this
23    assessment in the second draft REA based on data we expect to receive from the  U.S. Forest
24    Service, and will consider the results of that updated analysis in the second draft PA.

25    6.3    ECOSYSTEM SERVICES

26          As mentioned above, a number of analyses pertaining to the Os impacts on ecosystem
27    services, i.e. those based on the i-Tree and FASOMGHG models, are being released in parallel
28    with the first draft REA.  The FASOMGHG analyses include estimates of impacts of Os  on
29    yields of Os sensitive commercial timber species and the resulting impact on the  overall value of
30    forest products, as well as providing estimates of the impacts on carbon storage in commercial
31    forest ecosystems. The i-Tree analyses include estimates of the impacts of Os on carbon storage
32    by urban trees, and the impacts of Os on air pollution removals by urban trees. While chapter 6
33    of the first draft REA provides important background discussion and limited preliminary results
34    regarding risk to ecosystem services from O?, exposures, EPA staff feels it is premature to bring
                                                  6-13

-------
 1    that discussion forward into this first draft PA.  The second draft of the PA will include
 2    consideration of the results of the quantitative ecosystem risk analyses, as well as additional
 3    qualitative analyses including additional Class I area case study analyses, which are planned for
 4    incorporation into the second draft REA..

 5    6.4    UNCERTAINTIES IN RISK ASSESSMENTS

 6          In order to better interpret the results of the exposure and risk assessments described in
 7    the REA, and to help inform the Administrator regarding the appropriate weight to put on the
 8    results, a full and indepth discussion of the uncertainties associated with the REA analyses is
 9    important. Because of the preliminary nature of the results presented in this first draft, and
10    because it is unclear at this time how the methodology for adjusting air quality to just attain the
11    current and various alternative standards may change and what analyses will be most relevant to
12    informing the range of options presented for the Administrator's consideration, this first draft PA
13    does not include a discussion of uncertainties. This will be added in the second draft REA and
14    PA.

15    6.5    REFERENCES

16    Abt Associates, Inc. (2010). Model Attainment Test Software (Version 2). Bethesda, MD.
17          Prepared for the U.S. Environmental Protection Agency Office of Air Quality Planning
18          and Standards. Research Triangle Park, NC. Available on the Internet at
19          http://www.epa.gov/scram001/modelingapps.mats.htm.

20    Arbaugh, MJ; Miller, PR; Carroll, JJ; Takemoto, BL; Proctor,  T. (1998). Relationships of ozone
21          exposure to pine injury in the Sierra Nevada and San Bernardino Mountains of
22          California, USA. Environ Pollut 101: 291-301.

23    Benoit, LF; Skelly, JM; Moore, LD; Dochinger, LS. (1982). Radial growth reductions of Pinus
24          strobus L correlated with foliar ozone sensitivity as an indicator of ozone-induced losses
25          in eastern forests. Can J For Res 12: 673-678.

26    Black, VJ; Black, CR; Roberts, JA; Stewart, CA. (2000). Impact of ozone  on the reproductive
27          development of plants. New Phytol 147: 421-447.

28    Chappelka, A; Skelly, J; Somers, G; Renfro, J; Hildebrand, E.  (1999a). Mature black cherry used
29          as a bioindicator of ozone injury. Water Air Soil Pollut 116: 261-266.

30    Chappelka, A; Somers, G; Renfro, J. (1999b). Visible ozone injury on forest trees in Great
31          Smoky Mountains National Park, USA. Water Air Soil Pollut 116:  255-260.
                                                   6-14

-------
 1    Chappelka, AH. (2002). Reproductive development of blackberry (Rubus cuneifolius) as
 2          influenced by ozone. New Phytol 155: 249-255.

 3    Gregg, JW; Jones, CG; Dawson, TE. (2006). Physiological and developmental effects of O3 on
 4          cottonwood growth in urban and rural sites. Ecol Appl 16: 2368-2381.

 5    Hildebrand, E; Skelly, JM; Fredericksen, TS. (1996). Foliar response of ozone-sensitive
 6          hardwood tree species from 1991 to 1993 in the  Shenandoah National Park, Virginia. Can
 7          J For Res 26: 658-669.

 8    Karnosky, DF; Gagnon, ZE; Dickson, RE; Coleman, MD; Lee, EH; Isebrands, JG. (1996).
 9          Changes in growth, leaf abscission, biomass associated with seasonal tropospheric ozone
10          exposures of Populus tremuloides clones and seedlings. Can J For Res 26: 23-37.

11    Kohut, R. (2007). Assessing the risk of foliar injury from ozone on vegetation in parks in the US
12          National Park Service's Vital Signs Network. Environ Pollut 149: 348-357.

13    MEA (Millennium Ecosystem Assessment Board). (2005). Ecosystems and Human Well-being:
14          Current State and Trends, Volume 1. Edited by R. Hassan, R. Scholes, and N. Ash.
15          Washington: Island Press. Available at http://www.millenniumassessment.Org/8
16          documents/document.766.aspx.pdf

17    Peterson, DL; Arbaugh, MJ; Wakefield, VA; Miller, PR. (1987). Evidence of growth reduction
18          in ozone-injured Jeffrey pine (Pinus jeffreyi Grev and Balf) in Sequoia and Kings
19          Canyon National Parks.  J Air Waste Manag Assoc 37: 906-912.

20    Rea et al, (2012). Using Ecosystem Services To Inform  Decisions on U.S. Air Quality Standards.
21          Environ. Sci. Technol.46 (12), 6481-6488.

22    Schaub, M; Skelly, JM; Zhang,  JW; Ferdinand,  JA; Savage, JE; Stevenson, RE; Davis, DD;
23          Steiner, KC. (2005). Physiological and foliar symptom response in the crowns of Prunus
24          serotina, Fraxinus americana and Acer rubrum canopy trees to ambient ozone under
25          forest conditions. Environ Pollut 133: 553-567.

26    Somers, GL; Chappelka, AH; Rosseau, P; Renfiro, JR. (1998). Empirical evidence of growth
27          decline related to visible ozone injury. For Ecol  Manage 104: 129-137.

28    Timin B, Wesson K, Thurman J. Application of Model  and Ambient Data Fusion Techniques to
29          Predict Current and Future Year PM2 5 Concentrations in Unmonitored Areas. (2010). Pp.
30          175-179 in Steyn DG, Rao St (eds). Air Pollution Modeling and Its Application XX.
31          Netherlands: Springer.
                                                 6-15

-------
 1   US EPA (U.S. Environmental Protection Agency). (1978). Air quality criteria for ozone and
 2          other photochemical oxidants (pp. 373). (EPA/600/8-78/004). Washington, DC.

 3   US EPA (U.S. Environmental Protection Agency). (1984). Air quality criteria for ozone and
 4          other photochemical oxidants, Vol. 3 (pp. 405). (EPA/600/8-84/020A). Research Triangle
 5          Park, NC.

 6   US EPA (U.S. Environmental Protection Agency). (1996). Review of national ambient air
 7          quality standards for ozone: Assessment of scientific and technical information: OAQPS
 8          staff paper. (EPA/452/R-96/007). Research Triangle Park, NC.

 9   US EPA (U.S. Environmental Protection Agency). (2006). Air quality criteria for ozone and
10          related photochemical oxidants (pp. 2118). (EPA/600/R-05/004AF). Research Triangle
11          Park, NC.

12   US EPA (U.S. Environmental Protection Agency). (2007). Review of the national ambient air
13          quality standards for ozone: Policy assessment of scientific and technical information:
14          OAQPS staff paper. (EPA/452/R-07/007). July. Research Triangle Park, NC.

15    US EPA (U.S. Environmental Protection Agency). (2009). The Risk and Exposure Assessment
16          for Review of the Secondary National Ambient Air Quality Standards for Oxides of
17          Nitrogen and Oxides of Sulfur. (EPA-452/R-09-008a).. September. Available at
18          http://www.epa.gov/ttn/naaqs/standards/no2so2sec/data/NOxSOxREASep2009MainCont
19          ent.pdf

20   US EPA (U.S. Environmental Protection Agency). (2012a). Integrated Science Assessment for
21          Ozone and Related Photochemical Oxidants (Third External Review Draft). (EPA/600/R-
22          10/076C). July. Research Triangle Park, NC. Available at
23          http://cfpub.epa.gov/ncea/isa/recordisplay.cfm?deid=242490.

24   US EPA (U.S. Environmental Protection Agency). (2012b). Welfare Risk and Exposure
25          Assessment for Ozone (EPA 452/P-12-004). July. Research Triangle Park, NC.  Available
26          at http://www.epa.gov/ttn/naaqs/standards/ozone/s_o3_2008_rea.html
27   Vollenweider, P; Woodcock, H; Kelty, MJ; Hofer, R, -M. (2003). Reduction of stem growth and
28          site dependency of leaf injury in Massachusetts black cherries exhibiting ozone
29          symptoms. Environ Pollut 125: 467-480.
                                                  6-16

-------
 1              7.  PRELIMINARY STAFF CONCLUSIONS REGARDING THE
 2                                SECONDARY O3 NAAQS

 3          The purpose of this chapter is to present staffs preliminary considerations and
 4    conclusions regarding the adequacy and appropriateness of the current secondary 63 standard
 5    and additional analyses that would be appropriate to inform consideration of potential alternative
 6    standards in the second draft PA, including additional exposure and risk analyses in the second
 7    draft Welfare REA. Our preliminary conclusions in this first draft chapter are based on the
 8    assessments and integrative synthesis of information presented in the third draft Os ISA, the
 9    exposure and risk analyses presented in the first draft Welfare REA, and the information
10    presented above in chapters 5 and 6. In the final PA, this chapter will present staffs final
11    considerations and conclusions for the Administrator to consider regarding the adequacy  and
12    appropriateness of the current secondary 63 NAAQS and, if appropriate, staffs considerations
13    and conclusions regarding the range of potential alternative standards that could be supported by
14    the scientific evidence and exposure/risk information available in this review.
15          In this first draft PA, we consider the following overarching questions:
16       •  To what extent do the available scientific evidence and exposure/risk information
17          support or call into question the adequacy and appropriateness of the public welfare
18          protection afforded by the current secondary Os standard?
19
20       •  What additional analyses would be appropriate to help inform  consideration of
21          potential alternative standards in the second draft of the PA?
22
23          In considering our approach to developing preliminary conclusions regarding the
24    adequacy and appropriateness of the secondary 63 standard, we first note that the CAA charges
25    the Administrator with setting secondary NAAQS that are "requisite" (i.e., neither more nor less
26    stringent than necessary) to protect public welfare from any "known or anticipated adverse
27    effects." In light of this requirement, we note that a decision on the adequacy of the  public
28    welfare protection provided by the current Os standard, and by potential alternative standards,
29    will be a public welfare policy judgment in which the Administrator weighs the available
30    evidence, exposure/risk information, and the uncertainties and limitations inherent in that
31    evidence and information.  Therefore, in developing preliminary conclusions in this first draft
32    PA, we are mindful that the Administrator's ultimate judgments on the current and potential
33    alternative standards will most appropriately reflect an interpretation of the available scientific
34    evidence and exposure/risk information that neither overstates nor understates the strengths and
35    limitations of that evidence and information.
                                                   7-1

-------
 1           Section 7.1 discusses the approach to reviewing the 63 NAAQS in the last review and
 2    provides a general overview of the approach being taken in this review. Section 7.2 presents
 3    staffs preliminary considerations and conclusions regarding the adequacy and appropriateness of
 4    the current 8-hour 63 secondary standard. Section 7.3 presents staffs preliminary considerations
 5    and conclusions regarding alternative secondary Os standards appropriate for further analyses in
 6    the second draft REA. In the second draft PA, section 7.4 will present a summary of staffs
 7    conclusions on the current standard and, if appropriate, on alternative standards for the
 8    Administrator's consideration.  Section 7.5 will be added to outline key areas for future research
 9    and data collection to address key uncertainties identified in this review.

10    7.1     APPROACH TO REVIEWING THE SECONDARY O3 NAAQS
11           Welfare effects, as defined in section 302(h) (42 U.S.C. 7602(h))  of the CAA, include,
12    but are not limited to, "effects on soils, water, crops, vegetation, manmade materials, animals,
13    wildlife, weather, visibility, and climate,  damage to and deterioration of property, and  hazards to
14    transportation, as well as effects on economic values and on personal comfort and well-being."
15    As in the last review, this review focuses on effects on Os-sensitive vegetation, including
16    especially  natural vegetation as well as crops, since effects on these public welfare categories are
17    well-studied and currently known to be of most concern at O3 concentrations typically occurring
18    in the U.S. Further, by adversely affecting natural vegetation, Os can adversely affect natural
19    ecosystems and their components (e.g., soils, water, animals, and wildlife) as well as the services
20    that such ecosystems provide.  As discussed above in chapter 5, for other welfare effects
21    categories, insufficient new information was available to inform consideration of the adequacy of
22    the current secondary 63 standard, and they are not discussed further, except in terms of research
23    needs to be added in the next draft of this chapter.
24           This section discusses the approach taken to consider these Os-related welfare effects in
25    reviewing  the secondary O?, standard in the review completed in 2008 (section 7.1.1) and the
26    approach being taken in the current review (section 7.1.2).
27    7.1.1   Approach Taken in 2008 Review
28           In the last review of the Os NAAQS, the Administrator revised the level of the 8-hour Os
29    secondary  standard to 0.075 ppm, making it identical to the revised primary  standard.  The
30    Administrator's final decision on the secondary standard involved making a choice between the
31    two fundamentally different options that had been proposed in 2007. In the 2007 proposal, the
32    Administrator agreed with the conclusions drawn in the 2006 Criteria Document, the 2007 Staff
33    Paper and  by CAS AC that the scientific evidence available in the 2008 review continued to
34    demonstrate  the cumulative nature of Os-induced plant effects and the need to give greater
                                                    7-2

-------
 1    weight to higher concentrations. Thus, the Administrator proposed that a cumulative exposure
 2    index that differentially weights Os concentrations could represent a reasonable policy choice for
 3    a seasonal secondary standard to protect against the effects of Os on vegetation. The
 4    Administrator further agreed with both the 2007 Staff Paper and CAS AC that the most
 5    appropriate cumulative, concentration-weighted form to consider in the 2008 review was the
 6    sigmoidally weighted W126 form, due to the recognition that there is no evidence in the
 7    literature for an exposure threshold that would be  appropriate across all O3-sensitive vegetation
 8    and that this form is unlikely to be significantly influenced by Os air quality within the range of
 9    PRB levels identified in the 2008 review.  Thus, the Administrator proposed as one option to
10    replace the 1997 8-hour average secondary standard form with the cumulative, seasonal W126
11    form. The Administrator also proposed a second option to revise the 1997 secondary standard by
12    making it identical to the 8-hour primary standard proposed in 2007, which was proposed to be
13    within the range of 0.070 to 0.075 ppm. In putting forward such a proposal, as discussed below,
14    the Administrator focused on the decision made in the 1997 review, and the rationale for that
15    decision that made the revised secondary standard identical to the revised primary standard.
16          The 2008 final rule reported that within the Administration at that time there had been a
17    robust discussion of the strengths and weaknesses associated with each option that had been
18    proposed in 2007. The process by which the Administrator reached his final conclusion is
19    described in the 2008 final rule (73 FR 16497).  The rationale for the decision presented in the
20    2008 final rule (73 FR 16499-16500) is described below.
21          In considering the appropriateness of establishing  a new standard defined in terms of a
22    cumulative, seasonal form, or revising the 1997 secondary standard by making it identical to the
23    revised primary standard, the Administrator took into account the approach used by the Agency
24    in the 1997 review, the conclusions of the 2007 Staff Paper, CAS AC advice, and the views of
25    public commenters.  In giving consideration to the approach taken in the 1997 review, the
26    Administrator first considered the 2007 Staff Paper analysis of the projected degree of overlap
27    between counties with air quality expected to meet the revised 8-hour primary standard,  set at a
28    level of 0.075 ppm, and alternative levels of a W126 standard based on currently monitored air
29    quality data. This analysis showed significant overlap between the revised 8-hour primary
30    standard and selected levels of the  W126 standard form being considered, with the degree of
31    overlap between these alternative standards depending greatly on the W126 level selected and
32    the distribution of hourly Oj concentrations within the annual and/or 3-year average period.1 On
33    this basis, as an initial matter, the Administrator concluded that a secondary standard set identical
      1 Prior to publication of the 2008 final rule, EPA did further analysis of the degree of overlap to extend the 2007
      Staff Paper analyses, and that analysis was available in the docket.
                                                    7-3

-------
 1    to the proposed primary standard would provide a significant degree of additional protection for
 2    vegetation as compared to that provided by the then-current 0.084 ppm secondary standard. In
 3    further considering the significant uncertainties that remain in the available body of evidence of
 4    (Vrelated vegetation effects and in the exposure and risk analyses conducted for the 2008
 5    rulemaking, and the difficulty in determining at what point various types of vegetation effects
 6    become adverse for sensitive vegetation and ecosystems, the Administrator focused his
 7    consideration on a level for an alternative W126 standard at the upper end of the proposed range
 8    (i.e., 21 ppm-hours).  The 2007 Staff Paper analysis showed that at that W126 standard level,
 9    there would be essentially no counties with air quality that would be expected both to exceed
10    such an alternative W126 standard and to meet the revised 8-hour primary standard - that is,
11    based on this analysis of currently monitored counties, a W126 standard would be unlikely to
12    provide additional protection in any monitored areas beyond that likely to be provided by the
13    revised primary standard.
14           The Administrator also recognized that the general lack of rural monitoring data made
15    uncertain the degree to which the revised 8-hour standard or an alternative W126 standard would
16    be protective in those areas, and that there would be the potential for not providing the
17    appropriate degree of protection for vegetation in areas with air quality distributions that result in
18    a high cumulative,  seasonal exposure but do not result in high 8-hour average exposures.  While
19    this potential for under-protection using an 8-hour standard was clear, the number and size of
20    areas at issue and the  degree of risk was hard to determine.  However, the Administrator
21    concluded at that time that an 8-hour standard would also tend to avoid the potential for
22    providing more protection than is necessary, a risk that he concluded would arise from moving to
23    a new form for the secondary standard despite significant uncertainty in determining the degree
24    of risk  for any exposure level and the appropriate level of protection, as well as uncertainty in
25    predicting exposure and risk patterns.
26           The Administrator also considered the views and recommendations of CASAC and
27    agreed  that a cumulative, seasonal standard was the most  biologically relevant way to relate
28    exposure to plant growth response. However, as reflected in some public comments,  he also
29    judged that there remained significant uncertainties in determining or quantifying the degree of
30    risk attributable to varying levels of O?, exposure, the degree of protection that any specific
31    cumulative, seasonal  standard would produce, and the associated potential for error in
32    determining the standard that will provide a requisite degree of protection — i.e., sufficient but
33    not more than what is necessary. Given these significant uncertainties, the Administrator
34    concluded at that time that establishing a new secondary standard with a cumulative, seasonal
                                                    7-4

-------
 1    form would result in uncertain benefits beyond those afforded by the revised primary standard
 2    and therefore may be more than necessary to provide the requisite degree of protection.
 3          Based on his consideration of the views discussed above, the Administrator judged in the
 4    2008 rulemaking that the appropriate balance to be drawn was to revise the secondary standard
 5    to be identical in every way to the revised  primary standard.  Specifically, the Administrator
 6    revised the then-current 8-hour average 0.084 ppm secondary standard by making it identical to
 7    the revised 8-hour primary standard set at  a level of 0.075 ppm.  The Administrator believed that
 8    such a standard would be sufficient to protect public welfare from known or anticipated adverse
 9    effects and did not believe that an alternative cumulative, seasonal standard was needed to
10    provide this degree of protection. The Administrator believed that this judgment appropriately
11    considered the requirement for a standard that is neither more nor less stringent than necessary
12    for this purpose.
13    7.1.2  Approach in the Current Review
14          In this review our approach to considering the adequacy and appropriateness of the
15    current secondary  Os standard, and to identifying a range of potential alternative secondary
16    standards for consideration, draws from the approaches used in previous reviews. As discussed
17    above, past approaches have been based most fundamentally on using information from Os
18    vegetation effects  studies and exposure and risk assessments to inform the selection of 63
19    standards that, in the Administrator's judgment, protect the public welfare from any known or
20    anticipated adverse effects. These fundamental considerations again are the basis for our
21    approach in this review. Past approaches have also considered air quality analyses, based on
22    monitored air quality data, of the projected degree of overlap between counties with air quality
23    expected to meet the a secondary standard made identical to the primary standard and counties
24    with  air quality expected to meet alternative levels of a standard  defined in terms of a
25    biologically relevant, cumulative seasonal exposure index. In this review, we consider the
26    implications of such analyses as a basis for reaching conclusions about the adequacy and
27    appropriateness of the current secondary 63 standard.
28          In evaluating whether the current secondary standard is adequate or whether
29    consideration of revisions is appropriate, we are taking a weight-of-evidence approach that
30    considers information across the variety of vegetation-related research areas evaluated in the ISA
31    (e.g., visible foliar injury, tree growth and productivity, and ecosystem effects, including
32    productivity  and carbon storage, below ground processes, water cycling, and community
33    composition).  We are also considering assessments of air quality, exposures, and qualitative and
34    quantitative risks associated with alternative air quality scenarios, which in the first draft REA
35    includes recent air quality and air quality simulated to just meet the current secondary  Os
                                                    7-5

-------
 1    standard. With respect to vegetation effects information, we have considered the conclusions
 2    drawn at the end of the last review based on evidence from chamber, free air, gradient, model
 3    and field-based observation studies for a variety of vegetation effects endpoints in light of more
 4    recent evidence. We place greater weight on U.S. studies due to the often species-, site-, and
 5    climate-specific nature of Os-related vegetation response. With respect to quantitative exposure-
 6    and risk-based considerations, we consider the initial new, updated assessments as described
 7    above in chapter 6 as well as assessments conducted in the last review.
 8           We note that using scientific  and technical information to inform decisions on the
 9    secondary Os standard is complicated by the recognition that no vegetation effects threshold that
10    would apply to all sensitive U.S. vegetation, below which it can be concluded with confidence
11    that O3-related effects  do not occur, can be discerned from the available evidence (US EPA,
12    2012a, 73 FR 16486).  A further complication includes the uncertainties associated with
13    determining at what point the known or anticipated effects become adverse to the public welfare.
14    As a result, any approach to reaching conclusions on the adequacy of the current standard or
15    what alternative standards could be supported by the available  scientific evidence and
16    exposure/risk information requires judgments about how to consider that evidence and
17    information, including consideration of how to weigh associated uncertainties.  Such an approach
18    to considering  the available scientific and technical information is consistent with setting
19    standards that are neither more nor less stringent than necessary, recognizing that a zero-risk
20    standard is not required by the CAA.
21           Our approach to reviewing the Os secondary standard is outlined in Figure 7-1 below,
22    which takes into account the four basic elements of the NAAQS (i.e., indicator, averaging time,
23    form, and level).
                                                    7-6

-------
Figure 7-1.  Overview of Approach to Reviewing the Secondary Os NAAQS
                              Adequacy of Current 8-Hour Secondary 03 Standard
           Evidence-Based Considerations
  > ISA weight-of-evidence conclusions
  > Evidence for vegetation effects from cumulative
  exposures allowed by the current standard (OTC)?
  > Evidence for vegetation effects in locations that would
  likely have met the current standard (field-based)?
                        Exposure-/Risk-Based Considerations
                      > What is the nature, magnitude, and importance
                      of estimated exposures and risks associated with
                      current 03 standard?
                      > Uncertainties in the exposure and risk
                      estimates?
                                               Joes inform atior^
                                               call into question
                                              the adequacy and
                                              appropriateness of
                                                currentS-hour
                                                secondary 03
                                                  standard?
                                            NO
                                                      Considerretaining
                                                       current 8-hour 03
                                                           standard
                                                      LYES
                                  Consider Potential Alternative Standards
            Indicator
    > Support for retaining 03?
              Form
>Supportfor cumulative, seasonal
form?
>Supportfor peak-weighted form?
          Averaging Time
>Supportfor 12 hour diurnal period?
>Supportfor 3-month seasonal
period?
>Supportfor annual vs. 3-year period?
                                                     Level
          > Evidence-based considerations
                   •03 exposure concentrations in OTC and field-based studies
                   •03air quality projections across U.S. and case study locations
                   •Uncertainties in the evidence
          > Exposure and risk considerations
                   •What is the nature, magnitude,  and importance of estimated exposures and risks associated
                   with potential alternative 03 standards?
                   •Uncertainties in the exposure/ risk  assessment?
                            Identify range of potential alternative standards for consideration
                                                  7-7

-------
 1    7.2    PRELIMINARY STAFF CONCLUSIONS REGARDING THE ADEQUACY OF
 2          THE CURRENT SECONDARY O3 NAAQS
 3          As discussed throughout the third draft ISA, the new evidence available in this review
 4    generally continues to support and strengthen key conclusions drawn from the previous reviews
 5    (US EPA, 2012a). Thus, in considering the adequacy of the current secondary O3 standard, we
 6    take into account assessments that supported the last review as well as information and
 7    assessments in the third draft ISA and the first draft REA prepared for this review.
 8          In particular, we focus  primarily on assessments of Os-related vegetation effects that have
 9    been well characterized based  on extensive research and for which the ISA has concluded that
10    there is a causal relationship (section 7.2.1).  Such effects include visible foliar injury and
11    reduced growth and productivity, especially in sensitive vegetation and in natural terrestrial
12    ecosystems.  We also focus on exposure indices that have been considered in past reviews with
13    regard to their appropriateness for use in defining a standard that can provide appropriate
14    protection against such effects (section  7.2.2). In addition, we consider CASAC advice and
15    recommendations from the last review and in conjunction with EPA's 2010 proposed
16    reconsideration  of the 2008 O?, NAAQS (section 7.2.3). Based on these considerations,
17    preliminary staff conclusions are presented in section 7.2.4.
18    7.2.1  Vegetation Evidence-, Exposure-, and Risk-based Considerations
19          7.2.1.1 Visible foliar injury
20          The ISA (chapter 2, section 2.6.1) concludes that visible foliar injury resulting from
21    exposure to 63 has been well characterized and documented over several decades of many
22    species of trees  and other types of vegetation. The ISA concludes that experimental evidence has
23    clearly shown a consistent association of visible injury with 63 exposure, with greater exposure
24    often resulting in greater and more prevalent injury.  As discussed above in chapter 5, new
25    evidence evaluated in the ISA  includes  several multiple-year field surveys at several National
26    Wildlife Refuges in the U.S., controlled exposure studies that have verified such effects, and a
27    multi-year study that estimated the risk of Os-induced visible foliar injury in 244 national parks
28    across the country showing high risk of injury in 27% of the parks (US EPA, 2012a; p. 2-40).
29          While new assessments of visible foliar injury are underway in the first draft REA, as
30    noted above in chapter 6, more complete results will be available for consideration in the next
31    draft of this document.  To inform our preliminary consideration of the adequacy of the current
32    standard, we look to the visible foliar injury assessment done in the last review.  As described in
33    the 2007 Staff Paper, an assessment combining U.S. Forest Service Forest Inventory and
34    Analysis (FIA) biomonitoring  site data  with the county level air quality data for those counties

                                                    7-8

-------
 1    containing the FIA biomonitoring sites was conducted That assessment showed that incidence
 2    of visible foliar injury ranged from 21 to 39% during the four-year period (2001-2004) across all
 3    counties with air quality levels at or below that of the  then-current 8-hour standard of 84 ppb,
 4    and from 11 to 30% across all counties with air quality at or below an alternative 8-hour standard
 5    of 74 ppb, which is just below the level of the current 75  ppb standard. The magnitude of these
 6    percentages suggests that phytotoxic exposures sufficient to induce visible foliar injury would
 7    still occur in many areas upon meeting the current secondary O3 standard. Additionally, the
 8    analysis showed that visible foliar injury occurrence is geographically widespread and is
 9    occurring on a variety of plant species in forested and other natural systems (see 2007 Staff
10    Paper, Figure 7-19 in section 7.6.3.2).
11           As discussed above in chapter 5, although visible foliar injury is a valuable indicator of
12    the presence of phytotoxic concentrations of Os, it is not  always a reliable indicator of other
13    negative effects on vegetation  In some cases, however, visible foliar injury symptoms have
14    been correlated with decreased vegetative growth and impaired reproductive function.
15    Conversely, the lack of visible foliar injury does not always indicate a lack of other adverse
16    vegetative effects.  Nonetheless, EPA has determined in past reviews that the presence of visible
17    foliar injury in and of itself can adversely impact the public welfare. For example, visible foliar
18    injury in national parks and wilderness areas can impact the aesthetic experience for both
19    outdoor enthusiasts and the occasional park visitor. In addition, because these areas are afforded
20    a higher degree of protection, the presence of Os-induced vegetation effects, including visible
21    foliar injury, can take on increased significance. Specifically, federal land managers (FLMs)
22    ".. .have determined that given the high ecological, aesthetic, and intrinsic value of federal lands,
23    all native species are significant and warrant protection"  (NFS, 2000). As a result, FLMs have
24    identified visible foliar injury, along with other (Vinduced vegetation effects, as air quality
25    related values (AQRV) of concern (NFS, 2000). Numerous Os-sensitive species are found on
26    Class I federal lands (2007 Staff Paper, Appendix 7J).  In addition, the presence of visible foliar
27    injury also has the potential to economically impact those who rely on healthy looking vegetation
28    for their livelihood (e.g., horticulturalists, farmers of leafy crops, landscapers, Christmas tree
29    growers). Many ornamental species have been listed as sensitive to 63 (Abt, 1993). Similarly,
30    early senescence of fall foliage could also diminish the time available for viewing fall foliage,
31    important in some regions of the country in drawing tourists. Although data are not available to
32    allow the quantification of these impacts, they are nonetheless important to consider
33    qualitatively.
                                                     7-9

-------
 1          7.2.1.2  Reduced growth and productivity effects
 2          The ISA (chapter 2, section 2.6.2) concludes that ambient Os concentrations have long
 3    been known to cause decreases in photosynthesis rates and plant growth, and that O3-induced
 4    damage at the plant scale may translate to damage at the stand and ecosystem scales, causing
 5    changes in productivity and carbon storage. With regard to natural ecosystems, newly available
 6    studies continue to support this conclusion from the last review and strengthen our understanding
 7    of Os effects in forested ecosystems. Also, with regard to exposure-response relationships, the
 8    ISA concludes that the newly available information in this review on effects of Os on vegetation
 9    has strengthen the assessment of quantitative exposure-response relationships for growth and
10    productivity effects that was presented in the 2006 AQCD (ISA, chapter 2, section 2.6.6.2).
11          To help inform consideration of these effects, the first draft REA presents a preliminary
12    exposure assessment using updated air quality monitoring data (2006-2010) and new modeling
13    approaches to characterize recent Os air quality as well as air quality simulated to just meet the
14    current Os standard, as summarized above in chapter 6.  To inform our preliminary conclusions
15    on the adequacy of the current standard, we first consider the national O3 exposure surface
16    generated in terms of the W126 exposure index from the simulation of just meeting the current
17    standard (chapter 6, Figure 6-3) relative to the range of W126 values that was the focus of EPA
18    and CAS AC in the last review and in the 2010 proposed reconsideration of secondary Os
19    standard set in 2008. In particular, a range of 7 to 21 ppm-hours was proposed in the last review,
20    and a range of 7 to 15 ppm-hours has been repeatedly recommended by CAS AC both during and
21    subsequent to the last review and was proposed by EPA in the 2010 reconsideration notice. As
22    discussed in the 2007 and 2010 proposal notices and in various CAS AC letters to the past and
23    present EPA Administrators, these ranges are based on consideration of the full array of Os-
24    related effects on vegetation,  as summarized above in chapter 5, with particular emphasis on
25    reduced growth and productivity effects in sensitive vegetation and in terrestrial ecosystems.
26          In examining the W126 exposure surface depicted in Figure 6-3, based on a preliminary
27    simulation of just meeting the current standard, we observe that a number of relatively small
28    areas in the east and several somewhat larger areas in the west are estimated to have Os
29    concentrations in the range of 15 to 20 ppm-hour, which is above the range proposed by EPA in
30    2010 and recommended by CAS AC and in the upper part of the range proposed by EPA in 2007.
31    Even higher concentrations, in the range of 20 to 30 ppm-hours, are estimated to occur primarily
32    in two areas in the mountain west2. To the extent that effects judged to  be adverse to public
33    welfare in this review are associated with W126 levels within the previously considered range, it
      2 In at least one of these areas, high episodic peak O3 concentrations have been associated with the occurrence of
      wild fires.
                                                   7-10

-------
 1    is also appropriate to consider areas estimated to have 63 concentrations within this range upon
 2    just meeting the current standard.  In so doing, we observe that very broad areas across the
 3    eastern and western regions of the U.S. are estimated to have Os concentrations within the range
 4    of 10 to 15 ppm-hours, which includes the mid- to upper part of the ranges previously proposed
 5    by EPA and recommended by CASAC. Even broader areas, encompassing most of the rest of
 6    the eastern and western regions, are estimated to have 63 concentrations in the range of 5 to 10
 7    ppm-hours, which includes the lower part of the EPA and CASAC ranges. Thus, this analysis
 8    indicates that many areas are  estimated to have Os concentrations above the ranges of W126
 9    values that were the focus of EPA and CASAC in the last review and in the proposed
10    reconsideration, and that much broader geographic areas are estimated to have concentrations in
11    the upper part of the previously considered range. Based in part on these observations, we reach
12    the preliminary conclusion that 63 concentrations that have been or could be reasonably
13    anticipated to cause adverse effects in sensitive vegetation and natural terrestrial ecosystems are
14    likely to remain in substantial geographical areas across the country upon just meeting the
15    current standard everywhere.3
16           The first draft REA also presents estimates of (Vrelated annual relative biomass loss
17    (RBL) for eleven sensitive tree species, based on the estimated exposure surface simulating just
18    meeting the current standard and on biomass loss concentration-response functions drawn from
19    the ISA based on the W126 metric. These RBL estimates are shown above in chapter 6, Table 6-
20    1. This analysis also reflects  consideration of the actual growing ranges of these tree species in
21    that it only estimates RBL in  each modeled grid cell that is part of each species' growing range.
22    Results presented include median and maximum RBL estimates from the distribution of RBL
23    estimates for all the modeled  grid cells within the species' range. In considering these results,
24    we note that seven of the eleven tree species included in this analysis have maximum annual
25    RBL estimates greater than 10%, with maximum estimates for five of these species (quaking
26    aspen, eastern white pine, tulip poplar, ponderosa pine, and red alder) within the range of
27    approximately 10 to 20% and with much higher estimates for the two most sensitive species
28    (black cherry and eastern cottonwood). Median estimates for these seven species were
29    approximately 3% for four of the species, 19% and 46% for the two most sensitive species, and
30    less than 1% for the other species (red alder).
31           In the last two Os NAAQS reviews, to help inform consideration of what magnitude of
32    annual biomass loss could reasonably by considered to be important from a public welfare
      3 As noted above in chapter 6, alternative modeling approaches to simulating just meeting the current and alternative
      standards will be applied in the second draft REA.  In the second draft PA we will revisit these observations based
      on refined analyses in the second draft REA.
                                                   7-11

-------
 1    perspective, EPA considered the consensus views of a group of experts who convened in 1996 to
 2    provide input and advice to the Agency in its review of the Os NAAQS that was ongoing at that
 3    time. At the 1996 Consensus Workshop, these experts expressed views with regard to
 4    considering a 2% annual biomass loss as significant within the context of multi-year
 5    compounding of effects over the life of the tree. More generally, consensus views were reached
 6    on ranges of 63 concentrations that would be expected to be protective for growth and
 7    productivity effects, including a range equivalent to a W126 range of 7 to 13 ppm-hours for tree
 8    seedlings in natural forest stands, which was an important consideration in the standard level
 9    ranges proposed in the last two reviews (Heck and Cowling, 1997).
10          Based on the above considerations, these analyses provide support for our preliminary
11    conclusion that O3 concentrations that can reasonably be anticipated to cause detrimental growth
12    and productivity effects in sensitive vegetation and natural terrestrial ecosystems are likely to
13    remain upon just meeting the current standard for a number of tree species that are important in
14    natural terrestrial ecosystems.  We recognize that additional exposure and risk assessments are
15    ongoing and will be presented in the second draft REA, which we will consider, together with
16    CAS AC advice and comments, in reaching conclusions on the adequacy of the current standard
17    in the second draft PA.
18    7.2.2  Biologically Relevant Exposure Indices
19          In considering the adequacy of the current secondary O3 standard, the exposure index
20    used to define the standard necessarily is a critical  element to consider.  The ISA (chapter 2,
21    section 2.6.6.1) states that no information newly available in this review alters the basic
22    conclusions with regard to biologically relevant exposure indices in the 2006 and 1996 AQCDs.
23    Once again, based on the current state of knowledge and the best available data assessed in this
24    review, the ISA concludes that exposure indices that cumulate and differentially weight  the
25    higher hourly average concentrations over a season and also include the mid-level values
26    continue to offer the most defensible approach for  use in developing response functions  and in
27    defining indices for vegetation protection (ISA, chapter 2, section 2.6.6.1).
28          In considering the adequacy of the current 8-hour average secondary 63 standard, we
29    note that in the last two Os NAAQS reviews, EPA has recognized that the risk to vegetation
30    comes from cumulative seasonal exposures. Thus, it is clear that the purpose of the secondary
31    63 standard should be to provide an appropriate degree of protection against cumulative,
32    seasonal exposures to Os that are known or anticipated to harm sensitive vegetation or
33    ecosystems.  Further, it is also clear that a cumulative, seasonal form has a distinct advantage
34    over an 8-hour  average form in protecting against such exposures. Such a form is specifically
35    designed to measure directly the kind of Os exposures that can cause harm to vegetation during
                                                   7-12

-------
 1    the growing season.  In contrast, an 8-hour standard does not measure cumulative, seasonal
 2    exposures directly and can only indirectly afford some degree of protection against such
 3    exposures.
 4          To the extent that clear relationships were to exist between 8-hour daily peak Os
 5    concentrations and cumulative, seasonal exposures, the 8-hour form and averaging time would
 6    have the potential to be effective as an indirect surrogate. However, as discussed in the 2007
 7    proposed rule and the 2008 final rule, the evidence shows that there are known types of Os air
 8    quality patterns that can lead to high levels of cumulative, seasonal 63 exposures without the
 9    occurrence of high daily 8-hour peak Os concentrations, such as in rural, high elevation areas.
10    The lack of any clear consistent relationship between 8-hour daily peak 63 concentrations and
11    cumulative, seasonal exposures was shown in the last review (2007 Staff Paper, chapter 7,
12    section 7.5; Figure 7-1 and Appendix 7B).  Based on 3-year average values, it was shown that for
13    sites with a maximum 8-hour average concentration (fourth-highest maximum value) at the level
14    of the current standard, W126 values varied from approximately less than 5 ppm-hours to over
15    20 ppm-hours, a range that goes from below the lower end of CAS AC's recommended range to
16    above the upper end of the recommended range. Thus, an 8-hour standard form cannot be
17    expected to provide a reasonably consistent degree of protection from known or anticipated
18    adverse effects on vegetation associated with cumulative, seasonal 63 exposures in all areas
19    across the country.  Clearly an 8-hour form and averaging time is a very indirect way to
20    characterize biologically relevant exposure patterns, is poorly correlated with such exposure
21    patterns, and therefore is far less likely to identify and protect against the kind of cumulative,
22    seasonal exposure patterns that have been determined to be harmful than would a standard
23    defined in terms of a cumulative, seasonal exposure index.
24          Past reasons for not moving to a cumulative, seasonal form in the last two reviews have
25    not been based on disagreement over the biological relevance of the cumulative, seasonal form
26    or the recognized disadvantages of an 8-hour average standard form in measuring and identifying
27    a specified cumulative, seasonal exposure pattern.  Rather, the reasons for not moving to such a
28    form have been based on concerns over whether there is an adequate scientific basis to identify
29    the nature and magnitude of cumulative, seasonal exposure patterns that the standard should be
30    designed to protect against in light of various uncertainties in the evidence and the lack of rural
31    monitoring data.  This reasoning is built on the implicit presumption that there is less uncertainty
32    associated with specifying an appropriate degree of protection based on an 8-hour average
33    standard than there is in specifying an appropriate degree of protection based on a standard
34    defined in terms of a cumulative, seasonal exposure index.  In considering this presumption in
35    this review, EPA staff finds strong support in the available scientific and technical information

                                                   7-13

-------
 1    for the contrasting view that there is appreciably less uncertainty associated with specifying an
 2    appropriate degree of protection based on a cumulative, seasonal standard than on an 8-hour
 3    standard, recognizing that the standard is intended to provide protection against effects on
 4    vegetation that are inherently related to cumulative, seasonal 63 exposures.
 5          As discussed below, using a standard with a form based on maximum daily 8-hour 63
 6    levels as an indirect surrogate to provide the desired degree of protection from cumulative,
 7    seasonal Os exposures is likely directonally to have a greater risk of under-protection in some
 8    areas and a greater risk of over-protection in some other areas, compared to using a W126
 9    cumulative seasonal form.  Specifying an appropriate degree of protection for sensitive
10    vegetation in terms of a standard form that is defined based on the scientific evidence relating
11    short-term O3 exposures to respiratory-related effects in human populations is not warranted
12    based on the scientific evidence.
13          As discussed above in section 7.1.1, the reasoning that has led to past decisions to set the
14    secondary 63 standard identical to the primary 63 standard in both form and level has most
15    explicitly been based on analyses of the projected degree of overlap between counties with air
16    quality expected to meet the selected short-term primary standard and alternative levels of a
17    cumulative, seasonal standard based on then-current monitored air quality data. Such an analysis
18    in the last review showed essentially complete overlap between the revised 8-hour, 75 ppb
19    primary  standard and the W126 standard level  at the upper end of the proposed range, 21  ppm-
20    hours. That analysis was interpreted as showing that the revised 8-hour primary standard would
21    provide a significant degree of additional protection for vegetation as compared to that provided
22    by the then-current 84 ppb secondary standard.  Further, the conclusion was reached that a W126
23    standard set at 21 ppm-hours would be unlikely to provide additional protection in any monitored
24    areas beyond that likely to be provided by the revised primary standard. However, this line of
25    reasoning is based entirely on available monitoring data and thus does not reflect consideration
26    of areas for which monitoring data are not available, importantly including rural, high elevation
27    areas with natural forested ecosystems.  As noted above, it is reasonable to anticipate that such
28    areas may well have relatively high cumulative, seasonal 63 air quality patterns even as peak 8-
29    hour average Os concentrations may be relatively low.
30          Based on these considerations in the last review, the Administrator concluded that the
31    potential for under-protection from anticipated Os-related effects in such areas was clear,  while
32    noting that the number and size of such areas and the degree of risk of under-protection is hard to
33    determine (73 FR 16500).  Nonetheless, it was reasoned that making the secondary standard
34    identical to the primary standard would "tend to avoid the potential for providing more
35    protection than is necessary, a risk that would arise from moving to a new form for the secondary

                                                   7-14

-------
 1    standard despite significant uncertainty in determining the degree of risk for any exposure level
 2    and the appropriate level of protection, as well as uncertainty in predicting exposure and risk
 3    patterns" (73 FR 16500).

 4          In this review, it is the staffs view that for any given cumulative, seasonal level that is
 5    being considered as providing an appropriate degree of protection, one would expect less risk of
 6    under-protection or overprotection with a standard based on a cumulative, seasonal form as
 7    compared to determining an 8-hour standard that would indirectly provide the desired degree of
 8    cumulative, seasonal protection.  As observed above, based on the available scientific and
 9    technical information, it is more reasonable to conclude that there would be appreciably less
10    uncertainty associated with specifying an appropriate degree of protection against effects on
11    vegetation that are inherently related to cumulative, seasonal O3 exposures by using a standard
12    form that is explicitly cumulative and seasonal in nature.

13    7.2.3  CASAC Views on the Adequacy of the Current Standard

14          During the last review of the 63 NAAQS, CASAC stated the following in its letter to the
15    Administrator:

16          An important difference between the effects of acute exposures to ozone on
17          human health and the effects of ozone exposures on welfare is that vegetation
18          effects are more dependent on the cumulative exposure to, and uptake of, ozone
19          over the course of the entire growing season (defined to be a minimum of at least
20          three months). Therefore, there is a clear need for a secondary standard which is
21          distinctly different from the primary standard in averaging time, level and form.
22          Developing a biologically-relevant ozone air quality index would be directly
23          responsive to the 2004 National Research Council (NRC) recommendations on
24          Air Quality Management in the United States (NAS, 1994) and will help support
25          important new Agency initiatives to enhance ecosystem-related program tracking
26          and accountability.
27          ... [T]he compelling weight of evidence provided in Chapter 7 of the 2nd Draft
28          Ozone Staff Paper results from the convergence of results from many various and
29          disparate assessment methods including chamber and free air exposure, crop yield
30          and tree seedling biomass experimental studies, foliar injury data from
31          biomonitoring plots, and modeled mature tree growth.
32          Based on the Ozone Panel's review of Chapters 7 and 8, the CASAC unanimously
33          agrees that it is not appropriate to try to protect vegetation from the substantial,
34          known or anticipated, direct and/or indirect, adverse effects of ambient ozone by
35          continuing to promulgate identical primary and secondary standards for ozone.
36          Moreover, the members of the  Committee and a substantial  majority of the Ozone
37          Panel agrees with EPA staff conclusions  and encourages the Administrator to
38          establish an alternative cumulative secondary standard for ozone and related
39          photochemical oxidants that is  distinctly  different in averaging time, form and

                                                   7-15

-------
 1          level from the currently existing or potentially revised 8-hour primary standard."
 2          (Henderson, 2006, pp. 5-7).
 3
 4    In a subsequent letter sent to offer advice to aid the Administrator and Agency staff in
 5    developing the 2007 Os proposal, the CASAC reiterated its unanimous support for the
 6    recommendation in the Final Ozone Staff Paper "that protection of managed agricultural crops
 7    and natural terrestrial ecosystems requires a secondary Ozone NAAQS that is substantially
 8    different from the primary ozone standard in averaging time, level and form" (Henderson, 2007,
 9    p. 3).
10          Following the 2008 decision on the Os standards, the members of the CASAC Ozone
11    Review Panel sent a letter to EPA in April 2008 stating "[i]n our most-recent letters to you on
12    this subject - dated October 2006 and March 2007 - ... the Committee recommended an
13    alternative secondary standard of cumulative form that is substantially different from the primary
14    Ozone NAAQS in averaging time, level and form — specifically, the W126 index within the
15    range of 7 to 15 ppm-hours, accumulated over at least the 12 'daylight' hours and the three
16    maximum ozone months of the summer growing season . . .  [t]he CASAC now wishes to
17    convey, by means of this letter, its additional, unsolicited advice with regard to the primary and
18    secondary Ozone NAAQS.  In doing so, the participating members of the CASAC Ozone
19    Review Panel are unanimous in strongly urging you or your successor as EPA Administrator to
20    ensure that these recommendations be considered during the next review cycle for the Ozone
21    NAAQS that will begin next year" (Henderson, 2008). The letter further stated the following
22    views:

23          The CASAC was ... greatly disappointed that you failed to change the form of the
24          secondary standard to make it different from the primary standard. As stated in
25          the preamble to the Final Rule, even in the previous 1996 ozone review, 'there
26          was general agreement between the EPA staff, CASAC, and  the Administrator, ...
27          that a cumulative, seasonal form was more biologically relevant than the previous
28          1-hour and new 8-hour average forms (61 FR 65716)' for the secondary standard.
29          Therefore, in both the previous review and in this review, the Agency staff and its
30          advisors agreed that a change in the form of the secondary standard was
31          scientifically well-justified.
32          Unfortunately, this scientifically-sound approach of using a cumulative exposure
33          index for welfare effects was not adopted, and the default position of using the
34          primary standard for the secondary standard was once again instituted. Keeping
35          the same form for the secondary Ozone NAAQS as for the primary standard is not
36          supported by current  scientific knowledge indicating that different indicator
37          variables are needed to protect vegetation compared to public health. The
38          CASAC was further disappointed that a secondary standard of the W126 form

                                                  7-16

-------
 1          was not considered from within the Committee's previously-recommended range
 2          of 7 to 15 ppm-hours. The CASAC sincerely hopes that, in the next round of
 3          Ozone NAAQS review, the Agency will be able to support and establish a
 4          reasonable and scientifically-defensible cumulative form for the secondary
 5          standard. (Henderson, 2008).
 7          Following EPA's proposed reconsideration of the 2008 NAAQS, CASAC responded to
 8   EPA's request for CASAC's views on the proposal by stating the following:

 9          CASAC also supports EPA's secondary ozone standard as proposed: anew
10          cumulative, seasonal standard expressed as an annual index of the sum of
11          weighted hourly concentrations (i.e., the W126 form), cumulated over 12 hours
12          per day (Sam to 8pm) during the consecutive 3-month period within the ozone
13          season with the maximum index value, set as a level within the range of 7 to [1]5
14          ppm-hours. This W126 metric can be supported as an appropriate option for
15          relating ozone exposure to vegetation responses, such as visible foliar injury and
16          reductions in plant growth. We found the Agency's reasoning, as stated in the
17          Federal Register notice of January 19, 2010, to be supported by the extensive
18          scientific evidence considered in the last review cycle. In choosing the W126
19          form for the secondary standard, the Agency acknowledges the distinction
20          between the effects of acute exposures to ozone on human health and the effects
21          of chronic ozone exposures on welfare, namely that vegetation effects are more
22          dependent on the cumulative exposure to, and uptake of, ozone over the course of
23          the entire growing season (defined to be a minimum of at  least three months). In
24          this proposal, the Agency is responding to the clear need for a secondary standard
25          that is different from the primary standard in averaging time, level and form.
26          (Samet, 2010)
27
28          In reaching staff conclusions in the next draft of the PA, in addition to taking note of this
29   advice provided by CASAC in the last review, we will consider advice and recommendations
30   from CASAC based on their reviews of the first drafts of the Welfare REA and this first draft
31   PA.

32   7.2.4  Preliminary Staff Conclusions on the Current Secondary Os Standard

33          In this section, we present staffs preliminary conclusions  regarding the adequacy and
34   appropriateness of the public welfare protection provided by the current 8-hour secondary 63
35   standard.  In discussing these preliminary conclusions, we address the following questions:

36       •  To what extent does the available scientific evidence and exposure/risk information
37          support or call into question the adequacy of the public welfare protection afforded
38          by the current Os secondary standard?
                                                  7-17

-------
 1       •   To what extent does the available scientific evidence, and in particular vegetation
 2           exposure-response information, support or call into question the appropriateness of
 3           the form and/or averaging time of the current Os secondary standard?
 4           As discussed above, in addressing these questions we have considered the available
 5    scientific evidence assessed in the third draft ISA, as discussed above in chapter 5, and the
 6    available exposure and risk information assessed in the first draft REA, as discussed above in
 7    chapter 6, and considered in section 7.2.1; considerations related to a biologically relevant
 8    exposure index discussed in section 7.2.2 and the advice and recommendations received from
 9    CASAC during the last review and following the proposal to reconsider the 2008 decision, as
10    discussed above in section 7.2.3.
11           With regard to the scientific evidence, we reach the preliminary conclusion that the
12    available evidence clearly calls into question the adequacy and appropriateness of the current
13    standard and provides strong support for considering potential alternative standards to increase
14    public welfare protection, especially for sensitive vegetation and ecosystems occurring in
15    federally protected Class I and similar areas.  This preliminary conclusion places considerable
16    weight on the array of Os-related vegetation and ecosystem level effects that have been reported
17    following cumulative, seasonal exposures to 63 in studies or areas that have air quality that
18    would be allowed by  the current standard, including visible foliar injury and growth and
19    productivity reductions in sensitive trees and natural terrestrial ecosystems, as well as changes in
20    community composition, below ground processes, and water cycling as discussed above in
21    chapter 5. In emphasizing such effects, this preliminary  conclusion also places considerable
22    weight on the linkages between the evidence for these vegetation and ecosystem-level effects
23    and the broader body of evidence available at the molecular, biochemical, and  physiological
24    levels demonstrating  biologically plausible mechanisms  for Os as the causal agent in these higher
25    order effects. In reaching this preliminary conclusion, we acknowledge that uncertainties persist
26    in the welfare evidence; however, in staff s view the broad array of welfare effects reported
27    following cumulative, seasonal exposures to Os in studies or areas with air quality that would be
28    allowed by the current standard, combined with the plausible linkages between these effects and
29    the much larger body of molecular, biochemical, and physiological evidence, supports the
30    appropriateness of considering revising the current secondary O?, standard in order to increase
31    public welfare protection.
32           With regard to the exposure and risk information we reach the preliminary conclusion
33    that the available exposure and risk information from the first draft REA and from assessments
34    in the last review support the appropriateness of considering alternative standards that would
35    increase public welfare protection against vegetation and ecosystem effects.
                                                    7-18

-------
 1          With regard to CASAC advice, we note that the CASAC 63 Panel has repeatedly
 2    recommended setting a secondary standard distinctly different from the primary standard in
 3    averaging time, level and form. Since this advice was provided, based on evidence available in
 4    the last review, the available evidence in this review for the importance of cumulative,
 5    concentration weighted exposures in inducing adverse welfare effects has been confirmed and
 6    strengthened.
 7          Based on the above considerations, staff reaches the preliminary conclusion that the body
 8    of information now available supports consideration of revising the current 8-hour secondary Os
 9    standard, so as to afford greater and more appropriate public welfare  protection by selecting a
10    different form, averaging time and level than that of the primary standard, and that it does not
11    support retention of the current secondary Os standard.

12    7.3    POTENTIAL ALTERNATIVE STANDARDS FOR FURTHER ANALYSIS
13          Given our preliminary conclusion that the body of information now available supports
14    consideration of revising the current 8-hour secondary 63 standard so as to afford greater and
15    more appropriate public welfare protection from the adverse effects of cumulative, seasonal Os
16    exposures, we next consider the following overarching question:
17       •  What additional analyses would be appropriate to help inform consideration of
18          potential alternative standards in the second draft of the PA?
19          In considering this question, we specifically consider additional exposure and risk
20    analyses of alternative standards that would provide increased and more appropriate public
21    welfare protection relative to that provided by the current secondary Os standard. In this first
22    draft PA, the purpose of such considerations is to identify alternative standards appropriate for
23    additional analyses in the second draft REA. In the second draft PA, we will consider the
24    available scientific evidence and exposure/risk information in drawing conclusions about a range
25    of potential alternative standards that would be appropriate for consideration by the
26    Administrator in this review.
27          In identifying alternative standards appropriate for additional  exposure and risk analyses,
28    we note that the scientific evidence can provide meaningful insights into such alternative
29    standards when all the elements of the standard (i.e., indicator, averaging time, form, and level)
30    are considered together. With regard to the 63 indicator, in this review the available evidence
31    continues to support the current Os indicator. Given the available scientific evidence as
32    described in US EPA, 2012a, Chapter 9, the following  sections identify potential alternative
33    cumulative, seasonal standard forms (7.3.1), averaging times (7.3.2), and levels  (7.3.3),  for
34    further analyses.

                                                   7-19

-------
 1    7.3.1  Form
 2          As discussed above, a secondary 63 standard should provide an appropriate degree of
 3    protection against cumulative, seasonal exposures to O3 that are known or anticipated to harm
 4    sensitive vegetation or ecosystems. In considering alternative cumulative, seasonal forms that
 5    have been assessed and evaluated in the ISA, chapter 9, section 9.5.2, we reach the preliminary
 6    conclusion that the W126 exposure index is the best cumulative, seasonal, concentration-
 7    weighted form to consider, consistent with conclusions reached by EPA staff, CASAC, and the
 8    Agency in the last review.
 9          This conclusion is based on considering features of the W126 index that most closely
10    align with our current understanding of the science at this time.  First, the cumulative nature of
11    the W126 is clearly supported by the basic biological understanding of how most plants in the
12    U.S. are most biologically active during the warm season and are exposed to ambient 63
13    throughout this biologically active period.  Second,  it has been clearly  shown in the scientific
14    literature  that, all else being equal, plants respond more to higher concentrations, though there
15    continues to be no evidence of an exposure threshold for vegetation effects.  The W126
16    sigmoidal weighting function reflects both of these understandings, by not including  a threshold
17    below which concentrations are not included and by differentially weighting concentrations to
18    give greater weight to higher concentrations and less weight to lower ones. Specifically, this
19    index assigns increasing weight to each increase in hourly O?, concentrations between 0 and 100
20    ppb, though the sigmoid shape of the weighting scheme illustrates the point that lower
21    concentrations (below 40 ppb) are de-emphasized by assigning a very low weight,  while
22    concentrations between 40 ppb and 100 ppb are given more emphasis by assigning unique and
23    increasing higher weights, and all peaks above 100 ppb receive a weight of 1.
24          By contrast, other cumulative indices such as the SUM06 or AOT40 weight every
25    concentration below a defined threshold (i.e. 60 ppb and 40 ppb, respectively) with a weight of 0
26    and every concentration above that given threshold with a weight of 1, without  any differential
27    weighting of concentrations above the index threshold. Because there is less refinement in these
28    latter weighting schemes (each concentration can only be assigned either a 0 or 1), such
29    threshold weighting schemes are less able to reflect more subtle differences in 63 exposure
30    profiles that could account for observed differences in plant response.
31          Therefore, since the W126 form has a strong biological underpinning, and given the
32    strength of the scientific record, CASAC support, and evidence that it performs consistently well
33    over a wide range of conditions, we again conclude that a secondary standard form defined in
34    terms of the W126 exposure index is best suited and most appropriate to consider for inclusion in
35    additional analyses in the second draft REA on alternative secondary Os standards.
                                                   7-20

-------
 1    7.3.2  Averaging Times
 2          For a standard with a form defined in terms of the W126 exposure index, the NAAQS
 3    element of "averaging time" is appropriately considered in terms of exposure periods - diurnal
 4    and seasonal — over which the index would be summed in any given year.  This element also
 5    reflects consideration of whether the exposure index would be compared to the level of the
 6    standard on an annual basis or averaged over three years, as is the case for most other NAAQS.
 7    These considerations are discussed below.

 8          7.3.2.1 Diurnal exposure period
 9          As discussed above in chapter 5, section 5.4.1.1 and more fully in the ISA, chapter 9,
10    section 9.5.3.2, the diurnal conditions for  maximum uptake of 63 into the plant for the majority
11    of plants, occur mainly during the daytime hours. This is because, in general, (1) plants have the
12    highest stomatal conductance during the daytime; (2) atmospheric turbulent mixing is greatest
13    during the day in many areas; (3) the high temperature and high light conditions that typically
14    promote the formation of tropospheric Os also promote physiological activity in vegetation.
15          In addition to daytime uptake, a number of studies have also reported Os uptake at night
16    in some species. Typically the rate of stomatal conductance at night is much lower than during
17    the day. Several field studies have attempted to quantify night-time O3 uptake with a variety of
18    methods. Across the studies discussed in  the ISA, nocturnal conductance ranged from negligible
19    to 25% of daytime values (US EPA, 2012a, section 9.5.3). In some studies the percent of
20    nocturnal uptake varied by season and drought conditions. However, many of these studies did
21    not link the night-time flux to measured effects on plants. Thus, it is difficult to know whether
22    the impacts on the plant from nocturnal exposures are greater  or less than those from  similar
23    daytime exposures,  and whether or not they should be considered as separate impacts or as
24    additive or synergistic with impacts from  the preceding daytime exposure.
25          In addition to the uncertainties associated with understanding the plant response to
26    nocturnal uptake, there are also uncertainties associated with the extent of its occurrence.  This is
27    because for significant nocturnal stomatal flux and Os effects  to occur, the right combination of
28    specific conditions must exist. In particular, a susceptible plant with nocturnal stomatal
29    conductance and low defenses must be growing in an area with relatively high night-time O3
30    concentrations (often high elevation sites) and appreciable nocturnal atmospheric turbulence. It is
31    unclear how many areas there are in the U.S. where these conditions occur. It may be possible
32    that these conditions exist in mountainous areas of southern California, front-range of Colorado
33    and the Great Smoky Mountains of North Carolina and Tennessee.  More information is needed
                                                   7-21

-------
 1    in locations with high night-time 63 to assess the local 63 patterns, micrometeorology and
 2    responses of potentially vulnerable plant species.
 3          Therefore, due to the substantial uncertainties that remain regarding the importance and
 4    extent of nocturnal exposures associated with plant uptake, and whether and how they might be
 5    incorporated into a national index, EPA staff continues to focus on the 12-hour daytime exposure
 6    period of 8:00 am to 8:00 pm, consistent with CASAC advice in the last review.  In so doing, we
 7    recognize, as did CASAC (Henderson, 2007, p. 3) that in some parts of the country this daytime
 8    period represents a minimum acceptable period and may not include all daytime hours or
 9    exposures of importance to vegetation.  On this basis, we reach the preliminary conclusion that a
10    12-hour diurnal window (8:00 am to 8:00 pm) is appropriate to use in defining alternative W126
11    standards for additional analyses in the second  draft REA
12          7.3.2.2 Seasonal exposure period
13          The selection of any single window of time over which to cumulate Os exposures for a
14    national standard necessarily will represent a balance of factors, given the significant variability
15    in growth patterns and lengths of growing season among the wide range of vegetation species
16    that may experience detrimental effects associated with 63 exposure. Various intra-annual
17    averaging and accumulation time periods have  been considered for the protection of vegetation.
18    In 2007, EPA proposed use of the maximum consecutive 3-month period within the Os season as
19    a surrogate for vegetation growing seasons nationally.  A 3-month exposure period was also
20    supported by CASAC both in advice provided during the last review and on the 2010 proposed
21    reconsideration, as noted above in section 7.2.3. Alternatively, the U.S. Forest Service and
22    federal land managers have used a 24-hour W126 accumulated for 6 months from April through
23    September.

24          As an initial matter, in considering the alternatives of 3- or 6-month seasonal exposure
25    periods, we note that the exposure period in the vast majority of O3 exposure studies conducted
26    in the U.S. has been much shorter than 6 months, ranging from an average of 77 days in most
27    NCLAN crop studies to 145 days in the Aspen FACE experiment which represented the entire
28    growing season at that site. As a result, analyses of effects studies done in terms of the W126
29    exposure index have typically defined the index in terms of a 3-month exposure  period or at least
30    in terms of periods shorter than 6 months.  In addition, the Os season within which Os
31    monitoring is required is shorter than 6 months in many areas in the country.

32          To further help inform consideration  of 3- and 6-month seasonal periods, the ISA
33    presented the results of an analysis conducted by EPA  staff of the relationship between 3- and
34    6-month maximum  W126 values calculated for over 1,200 AQS and CASTNET EPA monitoring
                                                  7-22

-------
 1    sites for the years 2008-2009.  This analysis found that these two accumulation periods resulted
 2    in highly correlated metrics (US EPA, 2012a, Figure 9-13). As discussed in the ISA (section
 3    9.5.3), the two accumulation periods were centered on the yearly maximum for each monitoring
 4    site, and it is possible that this correlation would be weaker if the two periods were not
 5    temporally aligned. In the U.S., W126 cumulated over 3 months and W126 cumulated over 6
 6    months are proxies of one another, as long as the period in which daily W126 is accumulated
 7    corresponds to the seasonal maximum. Therefore, it is expected that either statistic will predict
 8    vegetation response equally well.

 9          Based on the above considerations, we reach the preliminary conclusion that the
10    maximum consecutive 3-month period within the Os growing season for vegetation remains an
11    appropriate and useful  surrogate for seasonal exposures for vegetation throughout the U.S., and
12    that a 3-month seasonal window is appropriate to use in defining alternative W126 standards for
13    additional analyses in the second draft REA.

14          7.3.2.3 Annual or 3-year average
15          In considering whether an annual or 3-year averaging period is more appropriate, we
16    recognize, as was recognized by EPA in the last review and in the 2010 proposed
17    reconsideration, that though most cumulative, seasonal exposure levels of concern for vegetation
18    have been expressed in terms of an annual timeframe, it is also appropriate to consider a 3-year
19    averaging period for purposes of standard stability. In so doing, we note that for certain welfare
20    effects of concern, including visible foliar injury and decreased growth and productivity effects
21    on perennial and annual vegetation, an annual time frame may be a more appropriate period in
22    which to assess what level would provide the requisite degree of protection, while for other
23    welfare effects, such as effects on mature tree biomass loss  and effects at the ecosystem level, a
24    3-year averaging period may also be appropriate.

25          Consistent with past consideration of this issue, we again recognize that should  a 3-year
26    average of the 3-month, 12-hour W126 form be used in defining an alternative secondary O3
27    standard, a potentially lower level should be considered to reduce the potential of adverse
28    impacts to annual species from a single high 63 year that could still occur while attaining a
29    standard on average over 3-years. In considering this issue in the last review, the CASAC Panel
30    concluded that multi-year averaging to promote a "stable" secondary standard is less appropriate
31    for a cumulative, seasonal secondary standard than for a primary standard based on maximum 8-
32    hour concentrations, and further concluded that if multi-year averaging is employed to increase
33    the stability of the secondary standard, the level of the standard should be revised downward to
                                                   7-23

-------
 1    assure that the desired degree of protection is not exceeded in individual years (Henderson, 2007;
 2    p. 3).

 3           In considering whether alternative W126 standards to be analyzed in the second draft
 4    REA should be based on an annual or 3-year average index, we reach the preliminary conclusion
 5    that while a 3-year average may be the most appropriate choice from a standard stability
 6    perspective, it would also be useful to conduct an analysis of the sensitivity of the exposures and
 7    risks to a standard defined in terms of an annual index.  Such a sensitivity analysis may help
 8    inform consideration of the extent to which it might be appropriate to consider lower standard
 9    levels in conjunction with a standard specified in terms of a 3-year average index.

10    7.3.3   Level
11           In our consideration of what alternative range of levels would be appropriate to consider
12    in conducting exposure and risk analyses of alternative  secondary O3  standards in the second
13    draft REA, we build on the preliminary conclusions reached above on the W126 form  and 12-
14    hour and 3-month diurnal and seasonal exposure windows.  In so doing, we give preliminary
15    consideration to the nature and degree of effects of Os to the public welfare,  including what
16    constitutes an adverse effect and the strengths and limitations of the evidence that is available
17    regarding known or anticipated adverse effects from cumulative, seasonal exposures and its
18    usefulness in informing selection of a range of levels appropriate for further analyses.  We also
19    consider CAS AC's views from the last review and the reconsideration of the 2008 NAAQS
20    regarding the strength of the evidence and its adequacy to inform a range of alternative levels
21    that would be appropriate for further analyses.
22           In considering the nature and degree of effects of Os on the public welfare, we  recognize
23    that the significance to the public welfare of (Vinduced effects on sensitive  vegetation growing
24    within the U.S. can vary, depending on the nature of the effect, the intended  use of the sensitive
25    plants or ecosystems, the degree to which the vegetation is managed for certain attributes and
26    uses, and the types of environments in which the sensitive vegetation and ecosystems are located.
27    Any given Os-related effect on vegetation and  ecosystems (e.g., biomass loss, foliar injury),
28    therefore, may be judged to have  a different degree of impact on the public welfare depending,
29    for example, on whether that effect occurs in a Class I area or a city park, or whether the effect
30    occurs in commercial crops or unmanaged forests.  In our view, it is appropriate that this
31    variation in the significance of (Vrelated vegetation effects should be taken into consideration in
32    identifying a range of alternative levels that may be appropriate to consider in additional
33    exposure and risk analyses.  In this regard, we agree with the definition of adversity discussed
34    above in chapter 5 (section 5.6), which draws from the discussion in section IV.A.3 of the 2010

                                                    7-24

-------
 1    proposed reconsideration and in the 2008 rulemaking.  As a result, we conclude that the primary
 2    focus of a secondary O3 standard should be on those effects that occur on sensitive species that
 3    are known to or are likely to occur in federally protected areas such as Class I areas4 or on lands
 4    set aside by States, Tribes and public interest groups to provide similar benefits to the public
 5    welfare, for residents on those lands,  as well as visitors to those areas.  In addition to the primary
 6    focus as described above, we also consider impacts on private lands where there is less
 7    opportunity to offset the degree of damage from O3 exposure, for example in commercial forests
 8    on private lands, trees grown in urban settings, or foliar damage in ornamental plants in
 9    residential areas.
10           We recognize that the same known or  anticipated O3-induced effects may call  for less
11    protection if they occur in other areas or on vegetation used for certain purpose, such as crops
12    grown commerically on privately held lands.  For example, the maintenance of adequate
13    agricultural crop yields is extremely important to the public welfare and is currently achieved
14    through the application of intensive management practices, including plant breeding for tolerance
15    to various environmental factors, extensive application of fertilizer, and irrigation.  These
16    management practices, in conjunction with market forces and government programs, assure an
17    appropriate balance between costs of production and market availability. In light of such
18    intensive management practice there is substantially more uncertainty in characterizing the
19    extent to which O3 exposures are detrimental to commercial crops than for other senstive
20    vegetation. Thus, while research on agricultural crop species remains useful in illuminating
21    mechanisms of action and physiological processes, we consider information from this sector on
22    O3-induced effects less useful in informing judgments  on alternative standards levels that are
23    appropriate for further analysis. With respect to commercial production of commodities, we note
24    that judgments about the extent to which O3-related effects on commercially managed vegetation
25    are adverse from a public welfare perspective are particularly difficult to reach, given that what
26    is known about the relationship between O3 exposures and agricultural crop yield response
27    derives largely from data generated almost 20 years ago. We recognize that there is substantial
28    uncertainty at this time as to whether these data remain relevant to the majority of species and
29    cultivars of crops being grown in the field today.  In addition, the extensive management of such
30    vegetation may to some degree mitigate potential O3-related effects, and the experiments from
31    decades ago do not reflect recent advances in  precision management of agricultural inputs to
32    maximize potential yield. The management practices used on these lands are highly variable and
      4 For example, the level of protection granted by Congress under the Wilderness Act of 1964 for designated
      "wilderness areas" requires that these areas "shall be administered for the use and enjoyment of the American people
      in such manner as will leave them unimpaired for future use as wilderness, and so as to provide for the protection of
      these areas, the preservation of their wilderness charactef (The Wilderness Act, 1964).
                                                    7-25

-------
 1    are designed to achieve optimal yields, taking into consideration various environmental
 2    conditions. Thus, overall we expect that Os exposure would present less risk to the public
 3    welfare from detrimental effects on agricultural crops than would Os exposure to other more
 4    sensitive natural vegetation and ecosystems. We draw a distinction between crops grown on
 5    agricultural lands and commercial forests, which while managed for timber yields are less
 6    amenable to the types of daily and seasonal intensive management practices available to
 7    agricultural farm managers. As such, adverse effects on these commercial forests can still be
 8    substantial and not likely fully captured by markets. In addition, commercial forests are often
 9    used for multiple purposes, providing recreational services as well as timber production.
10           While our focus is appropriately on non-agricultural vegetation, we note that providing
11    adequate protection for other more sensitive natural vegetation and ecosytems should tend to
12    provide protection for agricultural crops as well. Based on this, we conclude that such effects
13    should not beconsidered as a primary basis for identifying alterntative standard levels that are
14    appropriate for further analysis.
15          We also recognize that O3-related effects on sensitive vegetation can occur in other areas
16    that have not been afforded special federal protections, ranging from effects on vegetation
17    growing in residential or commercial settings, such as ornamentals used in urban/suburban
18    landscaping, to vegetation grown in land use categories that are managed for commercial
19    production of commodities such as timber. For vegetation used for residential or commercial
20    ornamental purposes, such as urban/suburban landscaping, we believe that there is not adequate
21    information at this time to identify alternative standard levels for further analysis based
22    specifically on impairment of urban/suburban landscaping and other uses of ornamental
23    vegetation, but we note that alternative secondary standard levels that would provide protection
24    for sensitive natural vegetation and ecosystems would likely also provide some degree of
25    protection for such ornamental vegetation. Furtherr, we recognize there is substantial data
26    available on the number and types of trees growing in urban areas, including urban parks as well
27    as trees planted along roadways and in other public places.  The volume of trees in urban areas
28    can be substantial, and as discussed in the first draft REA, provide important ecosystem services
29    including air pollution removal and carbon storage. As noted in above in chapter 6, the results of
30    the analysis of impacts on urban tree related ecosystem services will be considered in the second
31    draft PA in terms of the relevance of those results for informing potential alternative standards.
32          Based on the above, we find that the type of information most useful in informing the
33    identification of a range of alternative standard levels appropriate for further analysis is
34    appropriately focused on information regarding exposures and responses of sensitive trees and
35    other native species known or anticipated to occur in protected areas such as Class I areas or on

                                                    7-26

-------
 1    lands set aside by States, Tribes and public interest groups to provide similar benefits to the
 2    public welfare, for residents on those lands, as well as visitors to those areas.
 3          With regard to the available evidence, we find that the newly available evidence in this
 4    review is consistent with and reinforces the coherence and strength of the evidence available in
 5    the last review. As summarized above in chapter 5, and evaluated in chapter 9 of the ISA, this
 6    evidence addresses a broad array of Os-induced effects on a variety of tree species across a range
 7    of growth stages (i.e., seedlings, saplings and mature trees) using diverse field-based (e.g. free
 8    air, gradient and ambient) and OTC exposure methods.  It demonstrates that significant numbers
 9    of forest tree species  are potentially experiencing Os-induced stress under levels of ambient air
10    quality that would be allowed under the current secondary 63 standard.
11          In considering the advice of CASAC from the last review and on the proposed
12    recnsideration, we note that in its 2007 letter to the Administrator, the CASAC 63 Panel  agreed
13    with past EPA staff recommendations that the lower bound of the range within which a seasonal
14    W126 63  standard should be considered is approximately 7 ppm-hour. However, "it does not
15    agree with Staffs recommendations that the upper bound of the range should be as high  as 21
16    ppm-hour. Rather, the Panel recommends that the upper bound of the range considered should be
17    no higher than 15 ppm-hour,  which the Panel estimates is approximately equivalent to a  seasonal
18    12-hour SUM06 level of 20 ppm-hour" (Henderson, 2007). The CASAC provided the same
19    advice in its support of the 2010 proposed reconsideration, which proposed a range of levels
20    from 7 to  15 ppm-hours (Samet, 2010).
21          In also considering the recommendations from the 1996 Consensus Workshop, we
22    recognize that the 1997 Workshop Report did not clearly document the basis for its
23    recommendations, which included a consensus recommendation for a range equivalent to a
24    W126 range of 7 to 13 ppm-hours that would be protective for tree seedlings in natural forest
25    stands. While the absence of such documentation calls for caution in placing weight on this
26    recommendation, the recommendation was an important consideration in the standard level
27    ranges proposed in the last two reviews and in the reconsideration.
28          Based on the above considerations, we reach the preliminary conclusion that for a
29    secondary Os standard defined in terms of a W126 index, with 12-hour and 3-month exposure
30    periods, based on either an annual or 3-year average index, it appropriate to further analyze
31    alternative standard levels in the range of 7 to 15  ppm-hours in the second draft REA.
32          While the upper end of this range is lower than the upper end of 21 ppm-hour
33    recommended in the 2007 Staff Paper, this upper level of 21 ppm-hour was originally put
34    forward in the 1997 review in terms of a SUM06 of 25 ppm-hour (W126 of 21 ppm-hour) and

                                                  7-27

-------
 1    was justified on the basis that it was predicted to allow up to 10% biomass loss annually in 50%
 2    of studied commercial crops and tree seedling species. Recognizing the significant uncertainties
 3    that are associated with evaluating effects on commercial crops from a public welfare
 4    perspective, we now believe that commercial crop data are not appropriate to use as a basis for
 5    identiying a range of alternative standard levels for further analysis, such that we conclude that a
 6    level  of 15 ppm-hours is the upper end of the range of levels that is appropriate for further
 7    analyses in the second draft REA.
 8          With regard to the lower end of this range, we acknowledge that growth effects and
 9    visible foliar injury can still occur in sensitive species at levels below 7 ppm-hours. However,
10    we also recognize that significant uncertainties remain regarding the risk of such effects.  For
11    example, we conclude that remaining uncertainties make it difficult to judge the point at which
12    visible foliar injury becomes adverse to the public welfare in various types of specially protected
13    areas. Uncertainties associated with monitoring ambient exposures must be considered in
14    evaluating the strength of predictions regarding the degree of tree seedling growth impairment
15    estimated to occur at varying ambient exposures. These uncertainties add to the  challenge of
16    judging which exposure levels  are expected to be associated with levels of tree seedling growth
17    effects considered adverse to public welfare.

18    7.4    SUMMARY OF STAFF CONCLUSIONS ON THE SECONDARY O3 STANDARD
19          [To be added in the second draft PA.]
20
21    7.5    KEY AREAS FOR FUTURE RESEARCH AND DATA COLLECTION
22          [To be added in the second draft PA.]
23

24    7.6    REFERENCES
25    Federal Register 2007. National Ambient Air Quality Standards for Ozone; Proposed Rule. 40
26         CFR 50; Federal Register 72: 37818
27    Federal Register 2008. National Ambient Air Quality Standards for Ozone; Final Rule. 40 CFR
28         parts 50 and 58; Federal Register 73:16436
29    Heck, WW; Cowling, EB. (1997).The need for a long term cumulative  secondary ozone standard
30         - An ecological perspective. EM January:  23-33.
                                                  7-28

-------
 1   Henderson, R. (2008) Letter from CASAC Chairman Rogene Henderson to EPA Administrator
 2        Stephen Johnson. April 7, 2008, EPA-CASAC-08-001.

 3   Henderson, R. (2007) Letter from CASAC Chairman Rogene Henderson to EPA Administrator
 4        Stephen Johnson. March 26, 2007, EPA-CASAC-07-002.

 5   Henderson, R. (2006) Letter from CASAC Chairman Rogene Henderson to EPA Administrator
 6        Stephen Johnson. October 24, 2006, EPA-CASAC-07-001.

 7   Samet, J.M. Clean Air Scientific Advisory Committee (CASAC) Response to Charge Questions
 8        on the Reconsideration of the 2008 Ozone National Ambient Air Quality Standards.  EPA-
 9        CASAC-11-004. March 30, 2011. Available online at:
10        http://vosemite.epa.gOv/sab/sabproduct.nsf/0/F08BEB48Cl 139E2A8525785E006909AC/$

11   US EPA (2012a).  Integrated Science Assessment of Ozone and Related Photochemical Oxidants
12        (Third External Review Draft). U.S. Environmental Protection Agency, Washington, DC.
13        EPA/600/R-10/076C. June 2012.  Available at:
14        http://www.epa.gov/ttn/naaqs/standards/ozone/s_o3_2008_isa.html

15   US EPA (2012b).  Health Risk and Exposure Assessment for Ozone, First External Review
16        Draft. Office of Air Quality Planning and Standards, Research Triangle Park, NC, 27711.
17        EPA 452/P-12-001. July 2012.  Available at:
18        http://www.epa.gov/ttn/naaqs/standards/ozone/s_o3_2008_rea.html File/EPA-CASAC-11-
19        004-unsigned+.pdf

20   US EPA (2012c).  Welfare Risk and Exposure Assessment for Ozone, First External Review
21        Draft. Office of Air Quality Planning and Standards, Research Triangle Park, NC, 27711.
22        EPA 452/P-12-004. July 2012.  Available at:
23        http://www.epa.gov/ttn/naaqs/standards/ozone/s_o3_2008_rea.html

24   US EPA (2006). Air Quality Criteria for Ozone and Related Photochemical Oxidants (2006
25        Final). U.S.  Environmental Protection Agency, Washington, DC. EPA/600/R-05/004aF-
26        cF. March 2006. Available at:
27        http://www.epa.gov/ttn/naaqs/standards/ozone/s_o3_cr_cd.html

28   US EPA (1996). Air quality criteria for ozone and  related photochemical oxidants [EPA Report].
29        (EPA/600/P-93/004AF).  U.S. Environmental Protection Agency, Research Triangle Park,
30        NC.
                                                 7-29

-------