Potential for Atrazine Use in the Chesapeake Bay
Watershed to Affect Six Federally Listed
Endangered Species: Shortnose Sturgeon
(Acipenser brevirostrum); Dwarf Wedgemussel
(Alasmidonta heterodon); Loggerhead Turtle
(Caretta caretta); Kemp's Ridley Turtle
(Lepidochelys kempii); Leatherback Turtle
(Dermochelys coriacea); and Green Turtle (Chelonia
mydas)
Pesticide Effects Determination
Environmental Fate and Effects Division
Office of Pesticide Programs
Washington, D.C. 20460
August 31,2006
(March 14,2007 - amended during informal consultation with U.S.
Fish and Wildlife Service and National Marine Fisheries Service)
-------
Table of Contents
1.0 Executive Summary 10
2.0 Problem Formulation 12
2.1 Purpose 12
2.2. Scope of Assessment 14
2.2.1. Assessed Uses 14
2.2.2. Chemicals Assessed 14
2.2.3. Species Assessed 16
2.3 Previous Assessments 23
2.4. Characterization of Waters Inhabited by the Six Assessed Species 24
2.5. Action Area 32
2.6. Stressor Source and Distribution 35
2.6.1. Environmental Fate and Transport Summary 35
2.6.2. Use Characterization 36
2.7. Mechanism of Herbicidal Action 40
2.8. Assessment Endpoints and Measures of Ecological Effect 40
2.9. Conceptual Model 43
2.9.1. Risk Hypotheses 43
2.9.2. Diagram 43
3.0 Exposure Assessment 45
3.1. Conceptual Model of Exposure 46
3.2. Use of Modeling to Characterize Potential Exposures to Atrazine in the Chesapeake
Bay Watershed 47
3.2.1. Modeling Approach 47
3.2.2. Model Inputs 55
3.2.3. Model Results 61
3.3. Additional Modeling Exercises Used to Characterize Potential Exposures 63
3.3.1. Residential Uses 63
3.3.2. Additional Characterization of Agricultural Use EECs 67
3.3.3. Specific Characterization for Headwater Streams at Locations of the Dwarf
Wedge- mussel 72
3.3.4. Summary of Alternative Modeling Exercises 79
3.4. Monitoring Data 79
3.4.1. National USGS NAWQA Data 80
3.4.2. USGS Watershed Regression of Pesticides (WARP) Data 84
3.4.3. Regional USGS NAWQA Data 84
3.4.4. Chesapeake Bay Program Data 86
3.4.5. Heidelberg College Data 92
3.4.6. U.S. EPA ORD Great Lakes Program - Lake Michigan Mass Balance Project.. 94
3.4.7. Summary of Open Literature Sources of Monitoring Data for Atrazine 95
3.5 Summary of Modeling vs. Monitoring Data 96
3.6. Oral Exposure to Sea Turtles 97
3.6.1. Dietary Exposure from Contaminated Food Items 97
2
-------
3.6.2. Potential Exposure to Sea Turtles from Water Intake 97
4.0 Effects Assessment 98
4.1. Toxicity Data Used to Evaluate Assessment Endpoints 99
4.2. Toxicity Classification Scheme 101
4.3. Laboratory Effects Data 102
4.3.1. Toxicity to Fish 102
4.3.2. Toxicity to Aquatic Invertebrates 105
4.3.3. Toxicity to Sea Turtles 109
4.4. Terrestrial Plant Toxicity 112
4.5. Aquatic Plant Toxicity Data 114
4.5.1. Laboratory Data 114
4.5.2. Field Data 115
4.6. Community-Level Endpoints: Threshold Concentrations 115
4.7. Use of Probit Slope Response Relationship to Provide Information on the Endangered
Species Levels of Concern 118
4.8. Incident Database Review 119
5.0 Risk Characterization 121
5.1. Direct Effects Assessment 121
5.1.1. Risk Estimation 121
5.1.2. Risk Description, Direct Effects 123
5.1.3. Summary of Direct Effects Conclusions 127
5.2. Indirect Effects 127
5.2.1. Summary of Biological and Ecological Information Used to Evaluate Potential
Indirect Effects of Atrazine 129
5.2.2. Evaluation of the Potential for Atrazine to Induce Indirect Effects on the
Shortnose Sturgeon, Dwarf Wedgemussel, and Sea Turtles from Reduction in
Animal Food Items 132
5.2.3. Summary of Effects Determinations: Indirect Effects from Direct Effects to
Aquatic Animals 140
5.2.4. Evaluation of Potential Indirect Effects to the Six Listed Species from Potential
Effects to Aquatic Plants 140
5.2.5. Potential Indirect Effects to the Listed Species via Direct Effects to Terrestrial
Plants 144
5.2.6. Potential for Atrazine to Affect the Assessed Species via Effects on Riparian
Vegetation in the Chesapeake Bay Watershed 152
6.0 Uncertainties 163
7.0 Conclusions 172
8.0 References 174
3
-------
List of Appendices
Appendix A
Appendix B
Appendix C-l
Appendix C-2
Appendix D
Appendix E
Appendix F
Appendix G
Appendix H
Appendix I
Appendix J
Appendix J-2
Ecological Effects Data
Supporting Information for the Aquatic Community Level of Concern
(LOC) and Method to Apply the LOC to the Exposure Data (Use of
Threshold Concentrations and CASM Model)
Summary of Exposure Modeling Approach and Supporting Information
for Scenario Development
Residential Exposure Assumptions
Profiles of the Six Assessed Species
Incident Database Information
RQ Method and Description of LOCs
Map of Submerged Aquatic Vegetation in the Chesapeake Bay
Jellyfish Count and Biovolume Data in the Chesapeake Bay
Land use data Surrounding Dwarf Wedgemussel Populations and
Analyses of Riparian Areas of Dwarf Wedgemussel Populations on
Maryland's Eastern Shore
Bibliography of ECOTOX Open Literature Not Evaluated
Bibliography of Rejected ECOTOX Open Literature
4
-------
LIST OF TABLES
Table 1.1. Summary of Effects Determinations For Six Listed Species 11
Table 1.2. Identification and Listing Status of Six Listed Species Included in this Assessment 12
Table 2.1. Summary of Formation Pathway of Atrazine Degradates 15
Table 2.2. Summary of Available Degradate Toxicity Data in Birds and Mammals 16
Table 2.3. Location of the Dwarf Wedgemussel in the Chesapeake Bay Watershed 18
Table 2.4. Summary of Location of The Six Assessed Listed Species Within the Chesapeake
Bay 25
Table 2.5. Summary of Assessment Endpoints and Measures of Ecological Effect 42
Table 3.1. Types of Waters Inhabited by the Assessed Species and Data Used in Exposure
Refinement 47
Table 3.2 Label Application Information for the Chesapeake Bay Endangered Species
Assessment 57
Table 3.3. Comparison of Modeled Application Rates and Number of Applications with Typical
Use Data Used in the Triazine Cumulative Risk Assessment 60
Table 3.4. Summary of PRZM/EZAMS Environmental Fate Data Used For Aquatic Exposure
Inputs For Atrazine Endangered Species Assessment for the Chesapeake Bay 61
Table 3.5. Summary of PRZM/EXAMS Output for all Scenarios Modeled for the Atrazine
Endangered Species Assessment for the Chesapeake Bay Watershed using the Standard Water
Body 62
Table 3.6. Summary of the Impact of Variations in Pervious to Impervious Ratio on Predicted
Exposures from the PRZM/EXAMS Residential Scenario (granular) with 1% Overspray 63
Table 3.7. Comparison of Residential Scenario with an Assumption of No Overspray on
Impervious Surface to the Alternate Assumptions of 10% and 1% Overstay to Impervious
Surfaces 64
Table 3.8. Comparison of Residential Scenario with an Assumption of 50% of the Vi acre lot
treated to the Residential Scenario (granular) with an Assumption of 75% and 10% of the ]A acre
lot treated 65
Table 3.9. Percentage of Runoff Resulting from Impervious Surfaces 66
Table 3.10. Summary of the Impact of Runoff Differential on the Predicted Exposures from the
PRZM/EXAMS Residential Scenario (granular) Using the Analysis from Table 3.9 67
Table 3.11. Summary of PRZM/EXAMS Output for all Scenarios Modeled for the Atrazine
Endangered Species Assessment for the Chesapeake Bay Watershed Using a Standard Water
Bod 68
Table 3.12. Comparison of Alternative PRZM Modeling with EEC Generated Using a Static
Water Body 71
Table 3.13. Estimated Flow Rate for Water Bodies Known to be Inhabited by the Dwarf
Wedgemussel 72
Table 3.14. Aquia Creek Flow Information (fit3/sec) 74
5
-------
Table 3.15. South Anna River Flow Information (ft3/sec) 75
Table 3.16. Summary of Alternative PRZM Modeling Using the Index Reservoir and Site
Specific Flow data from Aquia Creek and South Anna River Compared with PRZM EEC
Modeling Using a Static Water Body 77
Table 3.17. Annualized Time Weighted Mean (TWM) Concentration (|ig/L) for the Top Ten
NAWQA Surface Water Sites Ranked by Maximum Concentration Detected 82
Table 3.18. Maximum Concentration (|ig/L) for the Top Ten NAWQA Surface Water Sites
Ranked by Maximum Concentration Detected 83
Table 3.19. Annual Time Weighted Mean and Maximum Concentration from the Top Three
USGS NAWQA Surface Water Sites Located in the Potomac Study Unit 85
Table 3.20. Annual Time Weighted Mean and Maximum Concentration from the Top Three
USGS NAWQA Surface Water Sites Located in the Lower Susquehanna River Study Unit 86
Table 3.21. Annual Time Weighted Mean and Annual Maximum Concentrations from Selected
Sample Locations from the Surface Water Monitoring Data from the Chesapeake Bay Program
88
Table 3.22. Annual Time Weighted Mean and Maximum Concentrations (|ig/L) for Atrazine in
Two Ohio Watersheds from the Heidelberg College Data 92
Table 3.23. Magnitude and Duration Estimates from the 1997 Data from Sandusky Watershed94
Table 3.24. Dietary Exposure Estimations for Sea Turtle 98
Table 4.1. Summary of Toxicity Data Used to Evaluate the Assessment Endpoints for the Six
Assessed Listed Species 101
Table 4.2. Summary of Fish Toxicity Studies Used In Risk Quotient Calculations 103
Table 4.3. Acute and Chronic Aquatic Invertebrate Toxicity Values Used 108
Table 4.4. Summary of Available Acute Oral, Subacute Dietary, Reproduction Toxicity Studies
in Birds, and Available Studies in Reptiles Ill
Table 4.5. Nontarget Terrestrial Plant Seedling Emergence Toxicity (Tier II) 113
Table 4.6. Nontarget Terrestrial Plant Vegetative Vigor Toxicity (Tier II) 113
Table 5.1. Summary of Aquatic RQs to Assess Potential Direct Effects to the Shortnose
Sturgeon 122
Table 5.2. Summary of RQs Used to Estimate Direct Effects to the Dwarf Wedgemussel 122
Table 5.3. Summary of RQs Used to Assess Potential Direct Effects to Sea Turtles 123
Table 5.4. Summary of Direct Effects Determinations to the Six Assessed Listed Species 127
Table 5.5. Summary of Shortnose Sturgeon Dietary Items 130
Table 5.6 Summary of Dietary Items of the Assessed Sea Turtles 132
Table 5.7. Summary of Animal Prey Items of the Six Assessed Species 133
Table 5.8. Summary of RQs Used to Estimated Indirect Effect to Shortnose Sturgeon and Sea
Turtles via Potential Direct Effects on Dietary Items 133
Table 5.9. Summary of RQs Used to Assess Potential Risk to Animal Food Items of the
Shortnose Sturgeon, Dwarf Wedgemussel, and Four Sea Turtles 135
6
-------
Table 5.10. Summary of Indirect Effects Determinations to the Six Assessed Listed Species
Resulting from Direct Effects to Aquatic Animals 140
Table 5.11. Summary of RQs Used to Estimated Indirect Effect to Shortnose Sturgeon, Dwarf
Wedgemussel, and Sea Turtles via Direct Effects on Primary Productivity 141
Table 5.12. Summary of PRZM/EXAMS Output for Corn, Sorghum, and Fallow/Idle Land and
Comparison of Estimated Atrazine Concentrations to Community Level Effects Thresholds.. 142
Table 5.13. Summary of Results of EPA Monitoring Data 143
Table 5.14. Summary of Indirect Effects Determinations to the Six Assessed Listed Species
Resulting from Direct Effects to Aquatic Plants 144
Table 5.15. Semi-quantitative Criteria Related to Riparian Vegetation for Assessing the Health
of Riparian Areas for Supporting Aquatic Habitats 145
Table 5.16. Screening-Level Exposure Estimates for Terrestrial Plants to Atrazine 147
Table 5.17. Nontarget Terrestrial Plant Seedling Emergence Toxicity 148
Table 5.18. Nontarget Terrestrial Plant Vegetative Vigor Toxicity 149
Table 5.19. Conclusions for the Potential of Atrazine to Affect Specific Dwarf Wedgemussel
Populations 159
Table 7.1. Summary of Effects Determinations For Six Listed Species 173
7
-------
LIST OF FIGURES
Figure 2-1. Shortnose Sturgeon Captured Between 1994 and 2003 20
Figure 2-2. Conceptual Model of an Exposure Scenario Representing Headwater Streams in the
Chesapeake Bay Watershed 26
Figure 2-3 Conceptual Model of an Exposure Scenario Representing Mid-Level River Reach
in the Chesapeake Bay Watershed 27
Figure 2-4. Conceptual Model of an Exposure Scenario Representing Major River Reach in the
Chesapeake Bay Watershed 28
Figure 2-5 Conceptual Model of an Exposure Scenario Representing a Mouth of Major River
in the Chesapeake Bay Watershed 29
Figure 2-6. Conceptual Model of an Exposure Scenario Representing a Minor Estuarine Inlet in
the Chesapeake Bay Watershed 30
Figure 2-7. Conceptual Model of an Exposure Scenario Representing the Main Stem of the
Chesapeake Bay 31
Figure 2-8. Action Area Defined by Chesapeake Bay Watershed for the Atrazine Endangered
Species Assessment 34
Figure 2-9 National Distribution of Atrazine Use 37
Figure 2-10. Atrazine Use (from Kaul et al, 2005) in the Immediate Vicinity of the Chesapeake
Bay 39
Figure 2-11. Conceptual Model 44
Figure 3-1. Percentage of Impervious Surfaces in the Chesapeake Bay and its Immediate
Tributaries 52
Figure 3-2 Density of Road, Railways, and Pipelines as Surrogate for Right of Way Density in
Chesapeake Bay Watershed (Action Area) 54
Figure 3-3. Location of NAWQA Surface Water Sites on Aquia Creek Relative to Chesapeake
Bay Program (CBP) Surface Water Sites 78
Figure 3-4. Range of Atrazine Concentrations Detected in the Chesapeake Bay and its
Immediate Tributaries 89
Figure 3-5. Location of Surface Water Monitoring Sites in the Chesapeake Bay and its
Immediate Tributaries 90
Figure 3-6. Location of Maximum Atrazine Detection (ng/1) in the Chesapeake Bay Watershed
in data from the Chesapeake Bay Program 91
Figure 4-1. Range of Aquatic Invertebrate Acute Toxicity Values Reported for Atrazine 106
Figure 4-2. Use of Threshold Concentrations in Endangered Species Assessment Error!
Bookmark not defined.
Figure 5-1. Range of Aquatic Invertebrate Acute Toxicity Values Reported for Atrazine 137
Figure 5-2. Map of Land Cover Data in the Po River Watershed 155
Figure 5-3. Example of a Riparian Area of Norwich Creek 157
8
-------
Figure 5-4. Example of Riparian Area Upstream of the Norwich Creek Site. Presented in Figure
5-3 158
Figure 5-5. Summary of the Potential of Atrazine to Affect the Six Assessed Listed Species via
Riparian Habitat Effects 163
Figure 6-1. Jellyfish Monitoring Data in the Chesapeake Bay 168
9
-------
1.0 Executive Summary
The purpose of this assessment is to make an "effects determination" for six listed species in the
Chesapeake Bay watershed: shortnose sturgeon, dwarf wedgemussel, loggerhead sea turtle,
leatherback sea turtle, Kemp's ridley sea turtle, and green sea turtle. The following assessment
endpoints were evaluated: (1) direct toxic effects on the survival, reproduction, and growth of the
assessed species; (2) indirect effects resulting from reduction of food supply; and (3) indirect
effects resulting from habitat modification. This assessment was completed in accordance with
the U.S. Fish and Wildlife Service (USFWS) and National Marine Fisheries Service (NMFS)
Endangered Species Consultation Handbook (USFWS/NMFS, 1998), the August 5, 2004 Joint
Counterpart Endangered Species Act Section 7 Consultation Regulations specified in 50 CFR
Part 402 (USFWS/NMFS, 2004a; FR 69 47732-47762), and procedures outlined in the Agency's
Overview Document (U.S. EPA, 2004).
Environmental fate and transport models were used to estimate high-end exposure values as a
result of agricultural and non-agricultural atrazine use in accordance with label directions.
Modeling was initially performed using the standard static water body. However, the
environments in which the assessed species are located include primarily flowing water bodies
such as streams and rivers and the main stem of the Chesapeake Bay. Except for short-term
exposures in small, flowing streams and small estuarine inlets, estimated exposures from the
available standard models are not likely to be representative of the types of waters inhabited by
the assessed species. Therefore, additional modeling was used together with available
monitoring data for the purpose of characterizing atrazine exposures in flowing waters. This
analysis shows that peak atrazine concentrations are expected to be approximately 50 |ig/L or
higher, but longer-term (days to weeks) exposures are expected to be in the low |ig/L range.
The assessment endpoints include direct toxic effects on the survival, reproduction, and growth
of the assessed species, as well as indirect effects, such as reduction of the prey base and/or
modification of its habitat. Direct effects are based on toxicity information for surrogate species
(U.S. EPA, 2004). Given that food items and habitat requirements of the assessed species are
dependant on the availability of aquatic invertebrates, aquatic plants, and terrestrial plants (i.e.,
riparian habitat), toxicity information for these taxonomic groups is also discussed. In addition,
indirect effects, via impacts to aquatic plant community structure and function, are also evaluated
based on time-weighted threshold concentrations that correspond to potential aquatic plant
community-level effects.
Risk quotients (RQs) are derived as quantitative estimates of potential high-end risk. Acute and
chronic RQs are compared to the Agency's levels of concern (LOCs) to identify instances where
atrazine use within the action area has the potential to adversely affect the six assessed species
via direct toxicity or indirectly based on direct effects to their food supply (i.e., freshwater
invertebrates) or habitat (i.e., aquatic plants and terrestrial riparian vegetation). When RQs for a
particular type of effect are below LOCs, the potential for adverse effects to the assessed species
is expected to be negligible, leading to a conclusion of "no effect". Where RQs exceed LOCs, a
potential to cause adverse effects is identified, leading to a conclusion of "may affect". If a
determination is made that use of atrazine within the action area "may affect" the assessed
species, additional information is considered to refine the potential for exposure and effects, and
10
-------
the best available information is used to distinguish those actions that "may affect, but are not
likely to adversely affect" from those actions that are "likely to adversely affect" the assessed
species.
A summary of the risk conclusions and effects determinations for the six listed species is
presented in Table 1-1. Further information on the results of the effects determination is
included as part of the Risk Description in Section 5.
Table 1.1. Summary of Effects Determinations For Six Listed Species
Assessment
Endpoint
Species
Effects
Determination
Basis for Determination
Direct effects to
listed species
(Section 5.1)
All six assessed
species
No Effect
No acute or chronic LOCs for endangered species are
exceeded.
Indirect effects to
listed species via
reduction of aauatic
animals as food
supply
(Section 5.2.2.)
Shortnose
sturgeon,
loggerhead turtle,
Kemp's ridley
turtle, green turtle,
leatherback turtle
Not likely to
adversely affect
Acute LOCs are exceeded for some animals that are food
items of the assessed species. However, the low magnitude of
potential effects on any one species, the low number of
dietary species potentially affected (indicated by LOC
exceedances) relative to the number potentially consumed by
the assessed species, and the conservative nature of the EECs
used to derive RQs for organisms in flowing water systems
suggests that the potential effects to the food supply of the
assessed species constitutes an insignificant effect.3
Dwarf
wedgemussel
No effect
No acute or chronic LOCs are exceeded.
Indirect effects to
listed species via
reduction of aauatic
olants as food items
or primary
productivity
(Section 5.2.4.)
All six assessed
species
Not likely to
adversely affect
No known obligate relationship between the assessed species
and any single aquatic plant species exists, and short-term and
long-term atrazine concentrations were estimated to be lower
than established thresholds for community-level effects to
aquatic vegetation.
Indirect effects to
listed species via
direct effects on
riparian areas
required to maintain
acceptable water
quality and spawning
habitat
(Section 5.2.5.)
Shortnose sturgeon
and each of the
four assessed sea
turtles
Not likely to
adversely affect
Acreage of riparian habitat expected to be sensitive to atrazine
is sufficiently low in the Chesapeake Bay watershed such that
potential impacts of atrazine to sensitive riparian buffers are
not expected to result in a measurable effect to the assessed
species that reside in the main stem of the Chesapeake Bay
and the Major river systems. Therefore, potential effects to
riparian areas from use of atrazine are expected to constitute
an insignificant effecta.
Dwarf
wedgemussel
Not likely to
adversely affect
Land cover data surrounding watersheds of dwarf
wedgemussel habitats suggest that riparian area exposure to
atrazine is expected to be minimal and/or that the predominant
riparian area adjacent to waters of dwarf wedgemussel
habitats is not expected to be sensitive to atrazine.
Therefore, potential effects to the dwarf wedgemussel from
effects to riparian areas are expected to constitute an
insignificant effect.3
a Significance of Effect: Insignificant effects are those that cannot be meaningfully measured, detected, or
evaluated in the context of a level of effect where take occurs for even a single individual
11
-------
2.0 Problem Formulation
Problem formulation provides a strategic framework for the risk assessment. By identifying the
important components of the assessed species and ecological stressor, it focuses the assessment
on the most relevant life history stages, habitat components, chemical properties, exposure
routes, and endpoints. This assessment was completed in accordance with the August 5, 2004
Joint Counterpart Endangered Species Act (ESA) Section 7 Consultation Regulations specified
in 50 CFR Part 402 (U.S. FWS/NMFS, 2004a; FR 69 47732-47762). The structure of this risk
assessment is based on guidance contained in U.S. EPA's Guidance for Ecological Risk
Assessment (U.S. EPA, 1998), the Services' Endangered Species Consultation Handbook
(USFWS/NMFS, 1998) and procedures outlined in the Overview Document (U.S. EPA, 2004).
2.1 Purpose
The purpose of this ecological risk assessment is to evaluate the potential direct and indirect
effects resulting from use of the herbicide atrazine (6-chloro-N-ethyl-N-isopropyl-l, 3, 5-
triazine-2, 4-diamine) in the Chesapeake Bay watershed on the survival, growth, and/or
reproduction of the following six federally listed species: (1) shortnose sturgeon (Acipenser
brevirostrum)\ (2) dwarf wedgemussel (Alasmidonta heterodon)\ (3) loggerhead turtle (Caretta
caretta); (4) Kemp's ridley turtle (Lepidochelys kempii)\ (5) green turtle (Chelonia mydas); and
(6) leatherback turtle (Dermochelys coriacea). A summary of the listing status for these species
is provided in Table 1.2, and a brief summary of key biological and ecological components
related to an assessment of these species is provided in Section 2.2.3. This ecological risk
assessment is a component of the settlement for the Natural Resources Defense Council, Civ.
No: 03-CV-02444 RDB (filedMarch 28, 2006). No critical habitat has been designated within
the Chesapeake Bay watershed for the assessed species.
Table 1.2. Identification and Listing Status of Six Listed Species Included in This
Assessment
Species
Status
Date Listed
Listing Agency
Shortnose sturgeon
(Acipenser
brevirostrum)
Endangered
32 FR 4001; 38FR
41370
March 11, 1967
National Marine Fisheries
Service (NMFS)
Dwarf wedgemussel
(Alasmidonta
heterodon)
Endangered
55FR 9447
March 4, 1990
U.S. Fish and Wildlife
Services (USFWS)
Loggerhead turtle
{Caretta caretta)
Threatened
32FR 4001; 43 FR
32800
July 28, 1978
USFWS and NMFS
Kemp's ridley turtle
Lepidochelys kempii
Endangered
35 FR 18319-18322
December 2, 1970
USFWS and NMFS
Green turtle
Chelonia mydas
Endangered
43 FR 32808
July 28, 1978
USFWS and NMFS
Leatherback turtle
Dermochelys coriacea
Endangered
35 FR 8491-8498
June 2, 1970
USFWS and NMFS
12
-------
In this endangered species assessment, direct and indirect effects to the six assessed species are
evaluated in accordance with the screening-level methodology described in the Agency's
Overview document (U.S. EPA, 2004). The indirect effects analysis in this assessment utilizes
more refined data than is generally available for ecological risk assessment. Specifically, a
robust set of microcosm and mesocosm data and aquatic ecosystem models are available for
atrazine that allowed for a refinement of the indirect effects associated with potential aquatic
community-level effects (via aquatic plant community structural change and subsequent habitat
modification). Use of such information is consistent with the guidance provided in the Overview
document (U.S. EPA, 2004), which specifies that "the assessment process may, on a case-by-
case basis, incorporate additional methods, models, and lines of evidence that EPA finds
technically appropriate for risk management objectives" (Section V, page 31 of U.S. EPA, 2004).
As part of the "effects determination", one of the following three conclusions is reached
regarding the potential for atrazine to adversely affect the assessed species:
• "No effect"
• "May affect, but not likely to adversely affect"
• "Likely to adversely affect"
If during the screening-level assessment it is determined that there are no indirect effects, and
LOCs for listed species are not exceeded for direct effects, a "no effect" determination is made
based on atrazine's use within the designated action area. A description of the action area for
the assessed species is provided in Section 2.5.
If a determination is made that use of atrazine within the action area may affect the listed species,
additional information is considered to allow for further refinement and characterization of
exposure and effects. Based on the additional characterization, the best available information is
used to distinguish those actions that may affect, but are "not likely to adversely affect" from
those actions that are "likely to adversely affect" a particular listed species.
The criteria used to make determinations that the effects of an action are not likely to adversely
affect listed resources include the following:
• Significance of Effect: Insignificant effects are those that cannot be meaningfully
measured, detected, or evaluated in the context of a level of effect where take
occurs for even a single individual
o "Take" in this context means to harass or harm
¦ Harm includes significant habitat modification or degradation that
results in death or injury to listed species by significantly impairing
behavioral patterns such as breeding, feeding, or sheltering.
¦ Harass is defined as actions that create the likelihood of injury to
listed species to such an extent as to significantly disrupt normal
behavior patterns which include, but are not limited to, breeding,
feeding, or sheltering.
13
-------
• Likelihood of the Effect Occurring: Discountable effects are those that are
extremely unlikely to occur. Dose-response information is used to estimate the
likelihood of effects.
• Adverse Nature of Effect: Effects that are wholly beneficial without any adverse
effects are not considered adverse. See Assessment Endpoints in Table 2.5.
2.2. Scope of Assessment
2.2.1. Assessed Uses
This risk assessment is for currently registered uses of atrazine. Atrazine is currently registered
as an herbicide in the U.S. to control annual broadleaf and grass weeds in corn, sorghum,
sugarcane, and other crops. In addition to food crops, atrazine is also used on a variety of non-
food crops, forests, residential/industrial uses, golf course turf, recreational areas, and rights-of-
way. Although atrazine is used on a number of commodities and non-agricultural areas, this
assessment addresses atrazine use in the Chesapeake Bay Watershed. Predominant uses in the
Chesapeake Bay include corn, sorghum, residential uses, turf, rights-of-ways, and fallow/idle
land.
Application rates and use patterns for these uses are described in Section 3. Use data considered
in this assessment were obtained from the available labels and from U.S. EPA's Biological and
Economic Analysis Division (BEAD) as discussed in Section 2.6.2.
2.2.2. Chemicals Assessed
This ecological risk assessment includes all potential ecological stressors resulting from the use
of atrazine within the Chesapeake Bay watershed, including atrazine and its potential degradates
of concern. Degradates of concern may include those that are found at significant (>10% by
weight relative to parent) concentrations in available degradation studies and those that are of
toxicological concern. Atrazine degradates and their routes of formation are summarized in
Table 2.1 below.
14
-------
Table 2.1. Summary of Formation Pathway of Atrazine Degradates
Degradate
Formation Pathway
Photolysis
in Water
Photolysis
in Soil
Aerobic
Metabolism in
Soil
Anaerobic
Metabolism in
Soil
Anaerobic
Metabolism in
Water
Deethylatrazine (DEA)
X (18%)a
X (18%)a
X
X
X
Deisopropylatrazine (DIA)
X
X
X
X
X
Diaminochlorotriazine (DACT)
X (15%)a
X
X
X
X
Hydroxyatrazine (HA)
X
X
X
X
Deethylhydroxyatrazine (DEHA)
X
X
Deisopropylhydroxy atrazine
(DIHA)
X
X
a Values in parentheses are percentage of parent formed; only values for major (>10%) degradates are shown. See
U.S. EPA, 2003a for additional discussion on these degradates.
Degradates of atrazine include hydroxyatrazine (HA), deethylatrazine (DEA),
deisopropylatrazine (DIA), and diaminochloroatrazine (DACT). Comparison of available
toxicity information for the degradates of atrazine indicates lesser aquatic toxicity than the parent
for fish, aquatic invertebrates, and aquatic plants. Specifically, the available degradate toxicity
data for HA indicate that it is not toxic to freshwater fish and invertebrates at the limit of its
solubility in water. In addition, no adverse effects were observed in fish or daphnids at DACT
concentrations up to 100 mg/L. Acute toxicity values for DIA are 3- and 36-fold less sensitive
than acute toxicity values for atrazine in fish and daphnids, respectively. In addition, available
aquatic plant degradate toxicity data for HA, DEA, DIA, and DACT report non-definitive EC50
values (i.e., 50% effect was not observed at the highest test concentrations) at concentrations that
are at least 700 times higher than the lowest reported aquatic plant EC50 value for parent atrazine.
Although degradate toxicity data are not available for terrestrial plants, lesser or equivalent
toxicity is assumed, given the available ecotoxicological information for other taxonomic groups
including aquatic plants and the likelihood that the degradates of atrazine may lose efficacy as an
herbicide. Therefore, given the lesser toxicity of the degradates as compared to the parent, and
the relatively small proportion of the degradates expected to be in the environment and available
for exposure relative to atrazine, the focus of this assessment is parent atrazine. Additional
details on available toxicity data for the degradates are provided in Section 4 and Appendix A.
The results of available toxicity data for mixtures of atrazine with other pesticides are presented
in Section A. 6 of Appendix A. According to the available data, other pesticides may combine
with atrazine to produce synergistic, additive, and/or antagonistic toxic effects. Synergistic
effects with atrazine have been demonstrated for a number of organophosphate insecticides
including diazanon, chlorpyrifos, and methyl parathion, as well as herbicides including alachlor.
If chemicals that show synergistic effects with atrazine are present in the environment in
combination with atrazine, the toxicity of atrazine may be increased, offset by other
environmental factors, or even reduced by the presence of antagonistic contaminants if they are
also present in the mixture. The variety of chemical interactions presented in the available data
set suggest that the toxic effect of atrazine, in combination with other pesticides used in the
environment, can be a function of many factors including but not necessarily limited to: (1) the
exposed species, (2) the co-contaminants in the mixture, (3) the ratio of atrazine and co-
contaminant concentrations, (4) differences in the pattern and duration of exposure among
contaminants, and (5) the differential effects of other physical/chemical characteristics of the
15
-------
receiving waters (e.g. organic matter present in sediment and suspended water). Quantitatively
predicting the combined effects of all these variables on mixture toxicity to any given taxa with
confidence is beyond the capabilities of the available data. However, a qualitative discussion of
implications of the available pesticide mixture effects data involving atrazine on the confidence
of risk assessment conclusions for the freshwater mussels is addressed as part of the uncertainty
analysis for this effects determination.
However, DEA has been shown to be of similar toxicity to birds on an acute oral basis compared
with atrazine. Other dealkylatrazine degradates have been shown to be more acutely toxic to
female rats and more developmentally toxic to gestating rat pups than the parent atrazine (Table
2.2 below). Acute avian studies suggest that DIA is less toxic than atrazine to birds on an acute
oral basis. No avian toxicity data for DACT are available; therefore, based on the equivalent
toxicity in mammals, DACT may also be of toxicological concern in birds, which are used as
surrogate species for turtles. For this reason, DACT and DEA were both considered qualitatively
in the risk assessment for sea turtles. Because preliminary analyses described in Section 5
indicate that the degradates would be expected to have negligible impact for sea turtles, aquatic
concentrations of the degradates were not quantified and were instead discussed qualitatively in
the Risk Characterization (Section 5).
Table 2.2. Summary of Avai
able Degradate Toxicity Data in Birds and Mammals
Chemical
Acute Bird LD50
(mg/kg-bw)
Acute Mammal LD50
(female rats; mg/kg-bw)
Mammal Developmental NOAEC
(mg/kg-bw)
Atrazine
940 (MRID 00024721)
1200 (U.S. EPA, 2003a)
200 (U.S. EPA, 2003a)
HA
>2000 (MRID 46500008)
Not available
500 (U.S. EPA, 2003a)
DEA
768 (MRID 46500009)
670 (U.S. EPA, 2003a)
25 (U.S. EPA, 2003a)
DIA
>2000 (MRID 46500007)
810 (U.S. EPA, 2003a)
5 (U.S. EPA, 2003a)
DACT
Not available
Not available
50 (U.S. EPA, 2003a)
2.2.3. Species Assessed
A brief introduction to the six listed species assessed, including a summary of habitat, diet, and
reproduction data relevant to ecological risk assessment for the Chesapeake Bay and its source
waters is presented below. A more comprehensive discussion of the biology and ecology of the
six assessed species is provided in Appendix D and in the Risk Characterization (Section 5).
16
-------
2.2.3.1. Dwarf Wedgemussel
The dwarf wedgemussel is an Atlantic Coast freshwater mussel usually found in sand, firm
muddy sand, and gravel bottoms in rivers of varying sizes with slow to moderate current. To
survive, the dwarf wedgemussel needs silt-free, stable, stream beds and well oxygenated water
(U.S. EPA, 2003b). Host fish (see Appendix D for information on the life cycle of the dwarf
wedgemussel) for populations in the Chesapeake Bay are unknown for this species. The dwarf
wedgemussel filter feeds on suspended detritus, phytoplankton, and zooplankton. Known
locations of the dwarf wedgemussel within the Chesapeake Bay watershed are summarized in
Table 2.3 below (U.S. FWS, 1993; Maryland Department of Natural Resources (MD DNR),
2006; Virginia Department of Game and Inland Fisheries (VA DGIF), 2006
[DWM_locations_distl783. Vector digital data. Acquired August 01, 2006],
17
-------
Table 2.3. Known Locations of Dwarf Wedgemussels in the Chesapeake Bay Watershed
Location
County, State
Description"
Status of Population and Major Threatsb
Tuckahoe Creek Drainage
Norwich Creek
Queen Anne's and Talbot
Counties, Maryland
Headwater
streams
Status: Poor, not reproducing
Threats: Non-point chemical pollution; sedimentation
from agriculture;
population density too low to allow successful
reproduction; residential, highway, or industrial
development
Long Marsh Ditch;
Mason Branch
Queen Anne's/Caroline
Counties, MD
Headwater
streams
Status: Poor, not reproducing
Threats: Non-point chemical pollution,
sedimentation from forestry operations;
sedimentation from agriculture; population density too
low to allow successful reproduction; headwater
channelization and "stream improvement" projects
Mason Branch and Long Marsh Ditch records likely
represent a single population
Potomac River Drainage
Mcintosh Run
Saint Mary's County, Maryland
Headwater
streams
Status: Fair, reproducing
Threats: Residential, highway, or industrial
development
Nanjemoy Creek
Charles County, Maryland
Headwater
streams
Status: Fair, reproducing
Threats: Not listed
Aquia Creek
Stafford County, Virginia
Headwater
streams
Status: Fair to good
Threats: Non-point chemical pollution;
Sedimentation from forestry operations;
Sedimentation from agriculture;
Residential, highway, or industrial development
York River Drainage
South Anna River
Louisa and Hanover Counties,
VA
Headwater
streams, mid-
level reach
Status: Poor
Threats: Sedimentation from forestry operations;
sedimentation from agriculture;
population density too low to allow successful
reproduction; residential, highway, or industrial
development
Po River
Spotsylvania County, VA
Headwater
streams, mid-
level reach
Status: Not listed
Threats: Not listed
Rappahannock River Drainage
Rappahannock
River
Spotsylvania County
Headwater
streams, mid-
level reach
Location data from Virginia Department of Game and
Inland Fisheries, 2006 (DWM_locations_distl783.
Vector digital data. Acquired August 01, 2006.)
Carter Run
Fauquier County
Headwater
streams
Southeast Creek or Corsica River Drainage
Browns branch,
Granny Finley,
Southeast Creek
tributary
Queen Anne's County, MD
Headwater
streams
Location data provided by Maryland Department of
Natural Resources, 2006
Brown's branch, Granny Finley branch, and Southeast
creek tributary records may represent a single
Corsica River
tributary
metapopulation
a See Section 2.4 for description of stream classification
b Status and threats information from U.S. FWS (1993), status and threats were not available for populations not included in U.S.
FWS (1993).
18
-------
2.2.3.2. Shortnose Sturgeon
The most recent documentation of shortnose sturgeon locations are from incidental capture via
the U.S. Fish and Wildlife Service Reward program for Atlantic Sturgeon, which began in 1996
in the Chesapeake Bay and its tributaries. Shortnose sturgeon were primarily captured in the
upper Chesapeake Bay north of Hart-Miller Island (Figure 2-1). However, sturgeon have also
been captured in the lower Susquehanna River, Bohemia River, Potomac River, and Elk River,
south of the Bay Bridge near Kent Island, near Howell Point, near Hoopers Island, and in Fishing
Bay (U.S. EPA, 2003b). Historical records indicate that shortnose sturgeon have also been
documented in the Chesapeake Bay, the Potomac River, near the mouth of the Susquehanna
river, and near the mouth of the James and Rappahannock rivers (U.S. EPA, 2003b; NMFS,
1998).
In many river systems, shortnose sturgeon appear to spend most of their life in their natal
river systems and only occasionally enter higher salinity environments. They are benthic
omnivores and continuously feed on benthic and epibenthic invertebrates, including mollusks,
crustaceans and oligochaete worms (NMFS, 1998; U.S. EPA, 2003b).
Shortnose sturgeon depend on free-flowing rivers and seasonal floods to provide suitable
spawning habitat. For shortnose sturgeon, spawning grounds have been found to consist mainly
of gravel or rubble substrate in regions of fast flow. Flowing water provides oxygen, allows for
the dispersal of eggs, and assists in excluding predators. Seasonal floods scour substrates free of
sand and silt, which might suffocate eggs (U.S. EPA, 2003b).
Shortnose sturgeon spawn in upper, freshwater sections of rivers and feed and overwinter
in both freshwater and saline habitats. In populations that have free access to the total length of a
river (absence of dams), spawning areas are located at the farthest accessible upstream reach of
the river, often just below the fall line (U.S. EPA, 2003b). Tributaries of the Chesapeake Bay
that appear to have suitable spawning habitat for the shortnose sturgeon include the Potomac,
Rappahannock, James, York, Susquehanna, Gunpowder, and Patuxent Rivers (U.S. EPA,
2003b). Other scientists believe that very little if any suitable spawning habitat remains for
shortnose sturgeon because of past sedimentation in tidal freshwater spawning reaches (U.S.
EPA, 2003b).
19
-------
Maryland
All Shortriose Sturgeon Caught Between 1994 and March 2003
=#v<\ rn
Tit
i§ J?
•*& t W¦
Legend
©All Shortnose Sturgeon Capture Sites
it State and U.S. Capitals
~Chesapeake Bay Watershed
f sJ I
^ Delaw-are
V
ugima
Figure 2-1. Shortnose Sturgeon Captured Between 1994 and 2003
(Source: U.S. EPA, 2003b)
20
-------
2.2.3.3.
Loggerhead Turtle
Between 2,000 and 10,000 loggerhead sea turtles enter the Chesapeake Bay each spring/summer
from May to early November (Kimmel, et al, 2006) when the sea temperatures rise to 18-20° C.
The majority are juveniles that utilize the Bay seasonally as a feeding ground
(http://www.chesapeakebay.net/info/seaturtle.cfm; http://www.2fla.com/loggerhead.htm.
accessed May 16, 2006; http://www.fisheries.vims.edu/turtletracking/stsp.html accessed March
10, 2006). They live along the channel edges (17 to 43 feet), forage on the bottom, and appear to
have foraging site fidelity (Byles 1988; Keinath et al. 1987; Kimmel, Driscoll and Brush 2006).
Many loggerheads remain in the Virginia portion of the Bay, where salinities are higher
(http://www.chesapeakebay.net/info/seaturtle.cfm, accessed May 16, 2006). Loggerheads
concentrate their feeding around river mouths and areas of the Bay deeper than 13 feet. The
loggerhead's range within the Bay includes primarily the main body of the Bay, river mouths,
estuarine inlets, and river main stems.
This species is carnivorous throughout its life, and its diet varies by region. Hatchlings eat small
animals living in sea grass mats, which are often distributed along drift lines and eddies.
Juveniles and adults eat a wide variety of prey such as conchs, clams, crabs, horseshoe crabs,
shrimp, sea urchins, sponges, fish, squid, and octopus
(http://www.tpwd.state.tx.us/huntwild/wild/species/endang/animals/reptiles amphibians/logghea
d.phtml).
All of the loggerhead nesting beaches are located outside of the Chesapeake Bay action area. Of
all the sea turtles, loggerheads are known to nest the furthest north on the eastern coast of the
United States. Some nest as far north as Virginia, just outside of the Chesapeake Bay; however,
no available reports have been located to indicate that loggerheads nest within the Chesapeake
Bay action area.
2.2.3.4. Kemp's Ridley Turtle
Kemp's ridley sea turtles are found within the Gulf of Mexico, and up through the Atlantic Coast
of the United States, including the Chesapeake Bay. Developmental habitat for Kemp's ridley
sea turtles, especially for foraging, has been identified within the Chesapeake Bay (Lutz and
Musick 1997). Habitat for the Kemp's ridley turtle, outside of nesting beaches that are located
outside of the Chesapeake Bay action area, includes mostly near shore and inshore waters,
usually less than 50 m deep. Within the Chesapeake Bay, Kemp's ridleys tend to stay in coastal
areas, foraging in beds of eelgrass. The primary range for the Kemp's ridley turtle within the
Bay includes main body of the Bay, river mouths, estuarine inlets, and river main stems.
Kemp's ridleys are primarily carnivorous, with a preference for crab species, especially blue crab
(Ccillinectes sapichis)\ however, they also feed on other crustaceans, mollusks, jellyfish, fish, and
marine plants and algae. While hatchlings and juveniles normally feed at the surface, adult
Kemp's ridleys feed on the bottom of coastal habitats as well.
21
-------
2.2.3.5.
Leatherback
Oceanic jellyfish are the preferred prey of the leatherback sea turtle throughout its life stages.
They also may incidentally ingest algae, vertebrates, and other invertebrate species. The
leatherback will often follow schools of jellyfish floating at the ocean's surface for its food
source; however, prey is also found in benthic habitats, especially near coastal habitats (NMFS
1992a). It has been estimated that hatchling leatherbacks eat approximately their weight in
jellyfish per day for growth and maintenance (Ernst, et al, 1994). Jellyfish that are likely to be
food items for the leatherback in the Chesapeake Bay include pink comb (Beroe ovata) and sea
walnut {Mnemiopsis leidyi), among other species (www.chesapeakebav.net/bavbio.htm. accessed
May 17, 2006 ). There is no specific tracking information on leatherback turtles within the Bay.
Their range may include the main body of the Bay, river mouths, and possibly main stems of
rivers, and estuarine inlets.
Similar to the other sea turtles in the Bay, the preferred nesting beaches for female leatherbacks
are outside the action area of the Chesapeake Bay and are generally high-energy beaches with
proximity to deep water, generally rough seas, and sufficiently-sloped sandy beaches backed
with vegetation (Ernst, et al., 1994; http://www.fws.gov/northflorida/SeaTurtles/Turtle
F act sheet s/1 eatherb ack- sea-turtl e. htm. Accessed April 13, 2006).
2.2.3.6. Green Turtle
Green turtles are distributed worldwide in tropical and subtropical waters; however, a very small
number enter the Chesapeake Bay each summer. Occasional juveniles and adults have been
identified in the Bay (Mansfield and Kimmel, personal communication 2006;
http://www.chesapeakebay.net/info/seaturtle.cfm, accessed May 16, 2006;
http://www.fisheries.vims.edu/turtletracking/stsp.html, accessed May 16, 2006 ). With its rich
food supply and extensive shoals, the Chesapeake Bay provides ideal habitat for juvenile green
turtle development. Many of the turtles remain in the Virginia portion of the Bay where
salinities are higher. In general, green turtles are found in waters inside reefs, bays, and inlets
(except when migrating). The turtles are attracted to lagoons and shoals with an abundance of
marine grass and algae (NMFS 1991a). Like the leatherback, there is no specific tracking
information on greens within the Bay. Their primary range may include the main body of the
Bay, river mouths, and possibly main stems of rivers and estuarine inlets.
Hatchling green turtles eat a variety of plants and animals, but adults feed almost exclusively on
sea grasses (especially Sargassum spp.) and marine algae with small amounts of animal foods
such as sponges, crustaceans, sea urchins, and mollusks. Within the Chesapeake Bay, green
turtle diet items likely include eelgrass, widgeon grass and algae, though there is no available
data on specific gut contents. (www.fws.gov/northflorida/SeaTurtles/Turtle Factsheets/green-
sea-turtle.htm. Accessed April 13, 2006;
www.tpwd.state.tx.us/huntwild/wild/species/endang/animals/reptiles_amphibians/
greentur.phtml, accessed May 16, 2006).
22
-------
As with the other sea turtle species in the Bay, all nesting beaches for the green turtle are located
outside of the Chesapeake Bay action area.
2.3 Previous Assessments
In January 2003, EPA completed a refined risk assessment that evaluated the potential impacts of
atrazine on the environment (USEPA, 2003a). This assessment was based on toxicity data from
laboratories as well as microcosm and mesocosm field studies coupled with exposure data
including model-estimated environmental concentrations and a substantial amount of monitoring
data from freshwater streams, lakes, reservoirs, and estuarine areas. Additionally, incident
reports of adverse effects on aquatic and terrestrial organisms associated with the use of atrazine
were also evaluated. In the refined assessment, risk is described in terms of the likelihood that
concentrations in water bodies (i.e., monitoring sites in lakes/reservoirs, streams, and estuarine
areas) may equal or exceed concentrations shown to cause adverse effects in laboratory and
field-based toxicity studies. The results of the refined aquatic ecological assessment indicated
that exposure to atrazine is likely to result in community-level and population-level effects to
aquatic communities at concentrations greater than or equal to 10-20 |ig/L on a recurrent basis or
over a prolonged period of time.
The results of the ecological assessments for atrazine are fully discussed in the January 31, 2003,
Interim Reregi strati on Eligibility Decision (IRED)1. The assessment identified the need for the
following information related to potential ecological risks: 1) a monitoring program to identify
and evaluate potentially vulnerable water bodies in corn, sorghum, and sugarcane use areas; and
2) further information on potential amphibian gonadal developmental responses to atrazine. On
October 31, 2003, EPA issued an addendum that updated the IRED issued on January 31, 2003.
This addendum described new scientific developments pertaining to monitoring of watersheds
and potential effects of atrazine on endocrine-mediated pathways of amphibian gonadal
development.
As discussed in the October 2003 IRED, an evaluation of the submitted studies regarding the
potential effects of atrazine on amphibian gonadal development was conducted and presented in
the form of a white paper for external peer review to a FIFRA Scientific Advisory Panel (SAP)
in June 20032. In the white paper dated May 29, 2003, seventeen studies consisting of both open
literature and registrant-submitted laboratory and field studies involving both native and non-
native species of frogs were summarized. It was concluded that none of the studies fully
accounted for environmental and animal husbandry factors capable of influencing endpoints that
the studies were attempting to measure. It was also concluded that the current lines-of-evidence
did not show that atrazine produced consistent effects across a range of exposure concentrations
and amphibian species tested.
Based on this assessment (U.S. EPA, 2003a), it was concluded that there was sufficient evidence
to formulate a hypothesis that atrazine exposure may impact gonadal development in
1 The 2003 Interim Reregistration Eligibility Decision for atrazine is available via the internet at
http://www.epa.gov/oppsrrdl/REDs/0001 .pdf
2 The Agency's May 2003 White Paper on Potential Developmental Effects of Atrazine on Amphibians is available
via the internet at http://www.epa.gov/oscpmont/sap/meetings/2003/iune/finaliune2002telconfreport.pdf.
23
-------
amphibians, but there were insufficient data to confirm or refute the hypothesis
(http://www.epa.gov/oscpmont/sap/meetings/2003/iune/finaliune2002telconfreport.pdf).
Because of the inconsistency and lack of reproducibility across studies and an absence of a dose-
response relationship in the currently available data, it was determined that the data did not alter
the conclusions reached in the January 2003 IRED regarding uncertainties related to atrazine's
potential effects on amphibians. The SAP supported EPA in seeking additional data to reduce
uncertainties regarding potential risk to amphibians. Subsequent data collection has followed the
multi-tiered process outlined in the white paper to the SAP (U.S. EPA, 2003d). In addition to
addressing uncertainty regarding the potential use of atrazine to cause these effects, these studies
are expected to characterize the nature of any potential dose-response relationship. A data call-in
for the first tier of amphibian studies was issued in 2005 and studies are on-going; however, as of
this writing, the results are not available.
2.4. Characterization of Waters Inhabited by the Six Assessed Species
Because all six species assessed are aquatic, the conceptual model of exposure is based on the
nature of surface water body types within the watershed. Surface water within the watershed
consists of surface streams ranging from headwater streams (first and second order by the
Strahler classification; Allan 1995) to mid-size reaches and rivers to main stem navigable rivers
such as the Potomac and Susquehanna rivers. In addition, the watershed is defined by the main
stem of the Chesapeake Bay, as well as the estuarine mouths of main rivers and a multitude of
minor stream/estuary inlets, which rim the Bay but are not connected to any major river systems.
For the purposes of this assessment, the surface water network of the Chesapeake Bay watershed
was divided into six broad classifications for comparison with monitoring data and modeled
estimated exposure concentrations (EECs). These broad classes include headwater streams, mid
range streams, major rivers, estuarine inlets of rivers (river mouths), minor estuarine inlets, and
the main stem (or open water portion of the bay) of the Chesapeake Bay. Representative
examples of this classification scheme are provided in Figures 2-2 through 2-8. These
classifications are for characterization purposes only (e.g. comparing exposure estimates with
regions) and do not define distinct regions within the watershed. In other words, the
classification is qualitative in nature and does not define "bright lines" between regions but is
based on a comparative analysis of the stream network. These classifications are used more fully
in comparison with monitoring data collected specifically for the Chesapeake Bay watershed and
this analysis is described in Section 3.4.
The principal significance of this type of scheme is to allow for a simplified comparison of
species location information with monitoring data and modeled EECs. This type of
generalization is particularly significant for modeled exposure estimates. Surface water
modeling was conducted using the Pesticide Root Zone Mode and Exposure Analysis Modeling
System (PRZM/EXAMS, described in more detail below) using existing scenarios linked to the
standard water body for ecological assessments. The standard water body is a static water body
and was developed to represent high-end estimates of atrazine that might be found in
ecologically sensitive environments (i.e., headwater streams) near agricultural fields.
24
-------
Characterization of the water bodies of the Chesapeake Bay watershed is used to compare
PRZM/EXAMS exposure estimates to monitoring data from the assessed species location. In the
Chesapeake Bay, exposure estimates generated using the static water body most closely
represent short-term exposures in low-order streams, such as those found on the Eastern Shore
where agriculture is dominant, and the minor type of estuary, where relevant land use abuts the
water body. Exposure concentrations (particularly longer-term concentrations) in other water
body types in the watershed (mid-size and main-stem rivers, estuarine mouths of rivers, and the
main stem of the Bay) are not expected to be as well represented by the standard water body,
thus, EECs from modeling likely over-estimate exposure in these settings. Analysis of the
impact of flow on modeled predictions of peak and long-term exposures (see Section 3) suggests
that model predictions are over-estimating longer-term exposures in water bodies with moderate
to high flow, or with larger volumes, while providing a reasonable approximation of exposure in
slowly flowing water bodies. Overall, the classification of water body types and the qualitative
comparison of modeled EECs with the classes of water body types is useful for characterizing
where exceedances may, or may not, be likely to occur.
A summary of the assessed species expected to be located in the six classes of water bodies
assessed is in Table 2.4 below. Attempts to better characterize potential exposures in these
settings are described in Section 3.3.
Table 2.4. Summary of Location of The Six Assessed Listed Species Within the Chesapeake Bay
Water Body
Classification
S
pedes"
Dwarf
Wedge-
musselb
Shortnose
Sturgeon0
Kemp's
Ridley
Turtled
Leatherback
Turtlede
Loggerhead
Turtle"
Green
Turtled
Headwater streams
X
Mid-range streams
X
Major rivers
X
X
X
X
X
River mouths
X
X
X
X
X
Minor estuarine inlets
X
X
X
X
Main body of the
Chesapeake Bay
X
X
X
X
X
a See Appendix D for additional information on species habitat.
b Dwarf wedgemussels are confined to small geographic locations and are found in streams with low to moderate
flow.
0 Juvenile sturgeons are found in freshwater rivers; adult sturgeons are found mostly at river mouths, but spawn in
freshwater rivers.
d All of the turtles are transitory within the Bay and located in various habitats within the Bay.
e Leatherback turtles are highly pelagic and are typically found in open waters.
25
-------
Example of
Headwater Streams
Headwater Streams in the
Chesapeake Bay Watershed
Legend
• Atrazine Detections (ng/l)
id Atlantic Streams
¦ Miles
0 0.5 1
Figure 2-2. Conceptual Model of an Exposure Scenario Representing Headwater
Streams in the Chesapeake Bay Watershed
26
-------
44118
•a?9
Mid-Level River Reaches in the
Chesapeake Bay Watershed
Example of Mid-Level Streams
Legend
Atrazine Detections (ng/l)
Mid Atlantic Streams
Figure 2-3. Conceptual Model of an Exposure Scenario Representing Mid-Level River
Reach in the Chesapeake Bay Watershed
27
-------
Main Stem of Major River in the
Chesapeake Bay Watershed
Legend
Atrazine Detections (ng/l)
Mid Atlantic Streams
Example of Main Stem
of a Major River
Figure 2-4. Conceptual Model of an Exposure Scenario Representing Major River Reach
in the Chesapeake Bay Watershed
28
-------
Estuarine Mouth of Major River in the
Chesapeake Bay Votershed
\
Example of
Estuarine Mouth
Legend
* Atrazine Detections (ngfl)
Mid Atlantic Streams
Figure 2-5. Conceptual Model of an Exposure Scenario Representing a Mouth of Major
River in the Chesapeake Bay Watershed
29
-------
Minor Estuarine Inlet in the
Chesapeake Bay V\tatershed
Legend
• Atrazine Detections (ng/l)
Mid Atlantic Streams
Example of Minor Inlet
Figure 2-6. Conceptual Model of an Exposure Scenario Representing a Minor Estuarine
Inlet in the Chesapeake Bay Watershed
30
-------
Main Body of the Chesapeake Bay
Bay
¦ Miles
10 20 40
Figure 2-7. Conceptual Model of an Exposure Scenario Representing the Main Stem of
the Chesapeake Bay
Legend
* Atrazine Detections (ngfl)
Mid Atlantic Streams
Example of Main Stem of
w
31
-------
2.5. Action Area
It is recognized that the overall action area for the national registration of atrazine uses is likely
to encompass considerable portions of the United States based on the large array of both
agricultural and non-agricultural uses. However, the scope of this assessment limits
consideration of the overall action area to those portions that may be applicable to the portion of
the United States that includes the following federally listed species as they occur within the
watershed of the Chesapeake Bay: shortnose sturgeon, dwarf wedgemussel, and loggerhead,
green, leatherback, and Kemp's ridley sea turtles. Deriving the geographical extent of this
portion of the action area is the product of consideration of the types of effects atrazine may be
expected to have on the environment, the exposure levels to atrazine that are associated with
those effects, and the best available information concerning the use of atrazine and its fate and
transport within the Chesapeake Bay Watershed.
Modeled concentrations of atrazine for labeled uses expected to occur within the Chesapeake
Bay watershed exceed established ecological risk levels of concern for aquatic plants and for
some food items of the assessed species suggesting adverse effects on components of the
environment is possible. The results of the screening-level assessment suggest that effects on
components of the environment are possible anywhere in the Chesapeake Bay watershed up to
and including the shallow water fringe of the Bay itself. Estimated atrazine concentrations are
deemed most appropriate for headwater streams and minor inlets surrounding the Bay as
described in Section 2.4; however, the potential for exceedances in other water body types
cannot be precluded. An exception to this is the open waters of the Chesapeake Bay itself.
Model predictions are not representative of the open waters of the Chesapeake Bay due to the
large volume of water present, the influence of tidal fluxes, and differences in water chemistry
relative to the EXAMS water body (freshwater), which are not accounted for in PRZM/EXAMS.
Exposure in the open waters of the Chesapeake Bay is best represented by monitoring data which
suggest that exposure levels in the open water are below levels of concern. Thus, the open water
portion of the Bay is not included in the action area. It is likely that exposure concentrations
predicted with modeling are not uniform throughout the watershed and portions of the action
area may be below levels of concern. However, these areas cannot be definitively drawn on a
map; therefore, the entire area described above includes all land draining to the Bay. Since the
action area is defined as an area where effects may occur, and lack of LOC exceedance indicates
a "no effect" conclusion, the LOC was used to define the action area. In addition, the action area
was limited to the Cheseapeake Bay watershed by the terms of the settlement for the Natural
Resources Defense Council, Civ. No: 03-CV-02444 RDB (filedMarch 28, 2006). Therefore,
species populations that occur outside of the Chesapeake Bay watershed were not included in
this assessment. More detail on the definition of the action area follows.
The named species being assessed as part of this endangered species assessment for the
Chesapeake Bay are generally known to inhabit the main stem of the Bay as well as its main
tributaries and headwaters of the main tributaries. Because this assessment is for multiple
species, the action area is defined using the aggregated greatest extent of the various named
species in conjunction with analysis of land use data, the registered use pattern for atrazine, and
available use information for atrazine. In general, because of the varied use pattern that includes
32
-------
non-agricultural uses on residential, turf, forestry, and rights-of-way sites, the action area is
initially defined as the entire Chesapeake Bay Watershed. Use of atrazine anywhere within the
Chesapeake Bay watershed may lead to exposure within the Bay and its associated tributaries.
An evaluation of usage information was conducted to determine whether any or all of the area
defined by the Chesapeake Bay Watershed should be included in the action area. Current labels
were reviewed to determine which atrazine uses may be present within the defined area. A more
detailed review of the local use information was completed. These data suggest that limited
agricultural uses are present within the defined area and that non-agricultural uses cannot be
precluded. Finally, local land cover data available from the Chesapeake Bay Program
(http://www.chesapeakebav.net/index.cfm) were analyzed to refine the understanding of
potential atrazine use in the areas immediately surrounding the Bay. The overall conclusion of
this analysis is that certain agricultural uses in the Chesapeake Bay watershed can likely be
excluded, and some non-agricultural uses of atrazine are unlikely to occur; however, no areas are
excluded from the final action area based on usage and land cover data.
Finally, environmental fate properties of atrazine were evaluated to determine which routes of
transport are likely to have an impact on the named species. Review of the environmental fate
data as well as physico-chemical properties suggest that transport via overland flow and spray
drift are likely to be the dominant routes of exposure (U.S. EPA, 2003a). Long-range
atmospheric transport of pesticides could potentially contribute to concentrations in the aquatic
habitat used by the listed species in the Bay. Given the physico-chemical profile for atrazine and
the fact that atrazine has been detected in both air and rainfall samples, the potential for long
range transport from outside the area defined by the Chesapeake Bay Watershed cannot be
precluded, but is not expected to result in exposure concentrations that approach those predicted
by modeling using the agricultural and residential scenarios (see Section 3.2).
Transportation of atrazine away from the site of application by both spray drift and volatilization
is well documented. Spray drift is a localized route of transport off of the application site in the
exposure assessments. Currently, quantitative models to address the longer-range transport of
pesticides from application sites are not available. The environmental fate profile of atrazine,
coupled with the available monitoring data, suggest that long-range transport of volatilized
atrazine is a possible route of exposure to non-target organisms. Therefore, the full extent of the
action area could be influenced by this route of exposure. However, given the extent of atrazine
use within the Chesapeake Bay watershed, the magnitude of documented exposures in rainfall at
or below available surface water monitoring data (as well as modeled estimates for surface
water), the extent of the action area is defined by the transport processes of runoff and spray drift
for the purposes of this assessment.
Based on this analysis, the action area for atrazine as it relates to the named species in this
assessment is defined by the full extent of the Chesapeake Bay Watershed, with the exception of
the open waters of the Bay because monitoring data suggest that exposure levels in the open
water are below levels of concern. Figure 2-8 presents the action area graphically. Note that
although the Chesapeake Bay is depicted in Figure 2-8, the open waters of the main trunk of the
Chesapeake Bay are not included in the action area.
33
-------
Chesapeake Bay V\fetershed
Vtermont
Massachusetts
Connecticut
Pennsylvania
New Jersey.
Maryland
V\fest Virginia
North Carolina
Figure 2-8. Action Area Defined by Chesapeake Bay Watershed for the Atrazine
Endangered Species Assessment
As Figure 2-8 indicates, the Chesapeake Bay Watershed encompasses a vast drainage area. The
watershed is defined by a diverse mixture of land covers, soils, and surface stream types, as well
as a varied climate. Because of the limited geographic extent of the species being assessed
relative to the overall watershed extent, the assessment focuses on those areas in closest
proximity to the Chesapeake Bay. The underlying assumption is that use of atrazine in the area
immediately surrounding the Bay is likely to be the most significant contributor to loading of
atrazine to the Bay. This was considered a reasonable assumption considering that the areas of
34
-------
highest atrazine use in the Chesapeake Bay watershed are in the Eastern Shore and southeast
Pennsylvania. Therefore, modeling was conducted for these areas and compared to local
monitoring data. Scenarios modeled for these high use areas are expected to be representative of
the highest exposures in the entire Chesapeake Bay watershed. It is possible that isolated areas
outside of the area immediately surrounding the Bay could have higher exposures, however, it is
expected that these locations are few and the impact of these exposures would be diluted prior to
reaching the Bay. The available monitoring data tend to support this in that the highest atrazine
concentrations detected in regional USGS NAWQA data are from sites closest to the Bay while
sites further removed from the Bay tend to have lower concentrations.
2.6. Stressor Source and Distribution
2.6.1. Environmental Fate and Transport Summary
The following fate and transport description for atrazine was summarized based on information
presented in the 2003 IRED (U.S. EPA, 2003a). In general, atrazine is expected to be mobile
and persistent in the environment. The main route of dissipation is microbial degradation under
aerobic conditions. Because of its persistence and mobility, atrazine is expected to reach surface
and ground water. This is confirmed by the widespread detections of atrazine in surface water
and ground water. Atrazine is persistent in soil, with a half-life (time until 50% of the parent
atrazine remains) exceeding 1 year under some conditions (Armstrong et al., 1967). Atrazine
can contaminate nearby non-target plants, soil and surface water via spray drift during
application or via runoff after application. Atrazine is applied directly to target plants during
foliar application, but pre-plant and pre-emergent applications are generally far more prevalent.
The resistance of atrazine to abiotic hydrolysis (stable at pH 5, 7, and 9) and to direct aqueous
photolysis (stable under sunlight at pH 7), and its moderate susceptibility to degradation in soil
(aerobic laboratory half-lives of 3-4 months) indicates that atrazine is unlikely to undergo rapid
degradation on foliage. Likewise, a relatively low Henry's Law constant (2.6 X 10"9 atm-
m3/mol) indicates that atrazine will probably not undergo rapid volatilization from foliage.
However, its relatively low octanol/water partition coefficient (Log Kow = 2 .7), and its relatively
low soil/water partitioning (Freundlich Kads values < 3 and often < 1) may somewhat offset the
low Henry's Law constant value thereby possibly resulting in some volatilization from foliage.
In addition, its relatively low adsorption characteristics indicate that atrazine may undergo
substantial washoff from foliage.
In terrestrial field dissipation studies performed in Georgia, California, and Minnesota, atrazine
dissipated with half lives of 13, 58, and 261 days, respectively. The inconsistency in these
reported half-lives could be attributed to the temperature variation between the studies in which
atrazine was seen to be more persistent in colder climate. Long-term field dissipation studies
also indicated that atrazine could persist over a year in such climatic conditions. A forestry field
dissipation study in Oregon (aerial application of 4 lb ai/A) estimated an 87-day half-life for
atrazine on exposed soil, a 13-day half-life in foliage, and a 66-day half-life on leaf litter.
Atrazine is applied during pre-planting and/or pre-emergence applications or directly to turf.
Atrazine is transported indirectly to soil due to incomplete interception during foliar application,
35
-------
and due to washoff subsequent to foliar application. The available laboratory and field data are
reported above. For aquatic environments reported half-lives were much longer. In an
anaerobic aquatic study, atrazine overall (total system), water, and sediment half-lives were
given as 608, 578, and 330 days, respectively.
A number of degradates of atrazine were detected in laboratory and field environmental fate
studies. Deethyl-atrazine (DEA) and deisopropyl-atrazine (DIA) were detected in all studies,
and hydroxy-atrazine (HA) and diaminochloro-atrazine (DACT) were detected in all but one of
the listed studies. Deethylhydroxy-atrazine (DEHA) and deisopropylhydroxy-atrazine (DIHA)
were also detected in one of the aerobic studies.
All of the chloro-triazine and hydroxy-triazine degradates detected in the laboratory metabolism
studies were present at less than the 10% of applied that is used to classify degradates as "major
degradates", however, several of these degradates were detected at percentages greater than 10%
in soil and aqueous photolysis studies. Insufficient data were available to estimate half-lives for
these degradates from the available data. The dealkylated degradates are more mobile than
parent atrazine, while HA is less mobile than atrazine and the dealkylated degradates.
2.6.2. Use Characterization
2.6.2.1. National Use Information
Atrazine has the second largest poundage of any herbicide in the U.S. and is widely used to
control broadleaf and many other weeds, primarily in corn, sorghum and sugarcane (U.S. EPA,
2003a). As a selective herbicide, atrazine is applied pre-emergence and post-emergence. Figure
2.1 presents the national distribution of use of atrazine (Kaul et al., 2005).3
3 Kaul et al. from U.S. EPA, Office of Pesticide Programs, Biological and Economic Analysis Division (BEAD)
36
-------
National Distributuion of Atrazine Use (lbs)
Legend
Lbs Atrazine
I 10 - 27729
I 127730 - 82603
¦182604- 165432
H 165433 -441435
¦ 441436- 1090674
Figure 2-9. National Distribution of Atrazine Use
Atrazine is used on a variety of terrestrial food crops, non-food crops, forests,
residential/industrial uses, golf course turf, recreational areas and rights-of-way. Atrazine yields
season-long weed control in corn, sorghum and certain other crops. The major atrazine uses
include: corn (83 percent of total ai produced per year - primarily applied pre-emergence),
sorghum (11 percent of total ai produced), sugarcane (4 percent of total ai produced) and others
(2 percent ai produced). Atrazine formulations include dry flowable, flowable liquid, liquid,
water dispersible granule, wettable powder and coated fertilizer granule. The maximum
registered use rate for atrazine is 4 lbs ai/acre; and 4 lbs ai/acre is the maximum, single
application rate for the following uses: sugarcane, forest trees (softwoods, conifers), forest
plantings, guava, macadamia nuts, ornamental sod (turf farms), ornamental and/or shade trees,
and Christmas trees.
2.6.2.2. Use Information in the Chesapeake Bay Watershed
An analysis of available usage and land cover information was performed to determine which
atrazine uses are likely to be present in the action area. The evaluation also is intended to place
priority on those uses likely to be in closest proximity to the Chesapeake Bay. The analysis
37
-------
indicates that of all registered uses; the agricultural uses are likely to result in the highest
exposures to the listed species. This is due to the preponderance of potential agricultural use
sites (corn and sorghum) in the immediate vicinity of Chesapeake Bay.
Critical to the development of appropriate modeling scenarios and to Office of Pesticide
Program's evaluation of the appropriate model inputs is an assessment of usage information.
The Biological and Economic Analysis Division (BEAD) provided an analysis of both national
and local use information (BEAD: Kaul, et al, 2005, Zinn, et al, 2006, Kaul, et al, 2005a, Kaul, et
al, 2006b). State level usage data for Maryland, Virginia, and Pennsylvania obtained from
USDA-NASS4 and Doane (www.doane.com; the full dataset is not provided due to its
proprietary nature), which were averaged together over the years 2000 to 2004 to calculate
average annual usage statistics by state and crop for atrazine, including pounds of active
ingredient applied, percent of crop treated, number of applications per acre, application rate per
acre, and base acres treated. State level data from 1998 to 2004 were averaged together and
extrapolated down to the county level based on apportioned to county level crop acreage from
the 2002 USDA Agriculture of Census (AgCensus) data. In general, this information suggests
that atrazine use on corn and Sorghum was approximately 500,000 lbs per year in Maryland,
600,000 lbs per year in Virginia, and 1,500,000 lbs per year in Pennsylvania on corn and
sorghum. Other agricultural commodities on which atrazine is used (macadamia nut, guava, and
sugarcane) are not present in the action area. Only information on agricultural uses were
available.
In addition, general use information that indicates where the main uses of atrazine on agricultural
crops are located is in Figure 2-10. A more complete summary of the use information used in
this assessment may be found in Section 3.2.
4 United States Depart of Agriculture (USDA), National Agricultural Statistics Service (NASS) Chemical Use
Reports provide summary pesticide usage statistics for select agricultural use sites by chemical, crop and state. See
http://www.usda.gOv/nass/pubs/estindxl.htm#agchem.
38
-------
Delaware
jrginia;
5^ v;
Legend
r Atrazine Use (lbs)
5001 - 20000
Atrazine Use in the Vicinity
of the Chesapeake Bay
igure 2-10. Atrazine Use (from BEAD; Kaul et al, 2005) in the Immediate Vicinity of t
Chesapeake Bay
le
39
-------
2.7. Mechanism of Herbicidal Action
Atrazine inhibits photosynthesis by stopping electron flow in Photosystem II. Triazine
herbicides associate with a protein complex of the photosystem II in chloroplast photosynthetic
membranes (Schulz et al., 1990). The result is an inhibition in the transfer of electrons that in
turn inhibits the formation and release of oxygen.
2.8. Assessment Endpoints and Measures of Ecological Effect
Assessment endpoints are defined as "explicit expressions of the actual environmental value that
is to be protected."5 Selection of the assessment endpoints is based on valued entities (i.e.,
shortnose sturgeon, dwarf wedgemussel, Kemp's ridley sea turtle, loggerhead sea turtle,
leatherback sea turtle, and green sea turtle), the ecosystems potentially at risk (i.e., Chesapeake
Bay and tributaries), the migration pathways of atrazine (i.e., runoff and spray drift), and the
routes by which ecological receptors are exposed to atrazine-related contamination (i.e., direct
contact and dietary exposure).
Assessment endpoints for the six species included in this assessment are direct toxic effects on
the survival, reproduction, and growth of the species, as well as indirect effects, such as
reduction of food supply and/or modification of their habitat. Each assessment endpoint requires
one or more "measures of ecological effect," which are defined as changes in the attributes of an
assessment endpoint itself or changes in a surrogate entity or attribute in response to exposure to
a pesticide. Specific measures of ecological effect are evaluated based on acute and chronic
toxicity information from the best available data. The reptile effects database is limited;
therefore, birds are used as surrogate species for reptiles as outlined in U.S. EPA (2004). Also,
effects data in sturgeon or dwarf wedgemussels are not available. Therefore, available toxicity
data in surrogate freshwater and estuarine/marine fish species are used to assess potential direct
effects to the shortnose sturgeon, and surrogate oyster and other invertebrate species are used to
assess potential direct effects to the dwarf wedgemussel. Additional ecological effects data from
the open literature, including effects data in reptiles and freshwater and saltwater microcosm and
mesocosm data were also considered.
Measures of effect from microcosm and mesocosm data provide an expanded view of potential
indirect effects of atrazine on aquatic organisms, their populations and communities in the
laboratory, in simulated field situations, and in actual field situations. With respect to the
microcosm and mesocosm data, threshold concentrations were determined from realistic and
complex time variable atrazine exposure profiles (chemographs) for modeled aquatic community
structure changes. Methods were developed to estimate ecological community responses for
monitoring data sets of interest based on their relationship to micro- and mesocosm study results,
and thus to determine whether a certain exposure profile within a particular use site and/or action
area may have exceeded community-level threshold concentrations. Ecological modeling with
the Comprehensive Aquatic Systems Model (CASM) (Bartell et al., 2000; Bartell et al., 1999;
and DeAngelis et al., 1989) was used to integrate direct and indirect effects of atrazine to
indicate changes to aquatic community structure and function.
5 From U.S. EPA (1992). Framework for Ecological Risk Assessment. EPA/630/R-92/001.
40
-------
A complete discussion of all the toxicity data available for this risk assessment, including use of
the CASM model and associated threshold concentrations, and the resulting measures of
ecological effect selected for each taxonomic group of concern are included in Section 4 of this
document. A summary of the assessment endpoints and measures of ecological effect selected to
characterize potential risks to the assessed species associated with exposure to atrazine is
provided in Table 2.5.
41
-------
Table 2.5. Summary of Assessment Endpoints and Measures of Ecological Effect
to Six Listed Species
Species
Assessment Endpoint
Measures of Ecological Effect"
Loggerhead,
Kemp's
ridley, green,
and
leatherback
sea turtles
1. Survival, growth, and reproduction via
direct effects
la. Avian acute oral gavage (LD50) and subacute dietary
(LC50) toxicity studies
lb. Avian reproduction NOAEC
2. Survival, growth, and reproduction via
indirect effects on food supply
2a. Acute toxicity studies in most sensitive surrogate potential
food items; LC50 and EC50
2b. Most sensitive chronic and reproduction toxicity studies in
potential food items; NOAEC
2c. Collective sensitivity distribution of toxicity values of
surrogate food items with effects data
2d. Microcosm/mesocosm threshold concentrations showing
aquatic community-level effects
3. Survival, growth, and reproduction via
indirect effects on habitat and/or primary
productivity (i.e., aquatic plant community)
3a. Most sensitive aquatic plant EC50
3b. Microcosm/mesocosm threshold concentrations showing
aquatic primary productivity community-level effects
4. Survival, growth, and reproduction via
indirect effects on terrestrial vegetation
(riparian habitat) required to maintain
acceptable water quality.
4a. Monocot and dicot seedling emergence EC25
4b. Monocot and dicot vegetative vigor EC25
Shortnose
Sturgeon
1. Survival, growth, and reproduction via
direct effects
la. Most sensitive fish acute LC50
lb. Most sensitive fish chronic NOAEC
2. Survival, growth, and reproduction via
indirect effects on food supply
2a. Most sensitive aquatic invertebrate acute EC50
2b. Most sensitive aquatic invertebrate chronic NOAEC
2c. Collective sensitivity distribution of toxicity values of all
surrogate food items with effects data
3. Survival, growth, and reproduction via
indirect effects on habitat and/or primary
productivity (i.e., aquatic plant community)
3a. Most sensitive aquatic vascular plant EC50
3b. Non-vascular plant (algae) acute EC50 and NOAEC
3c. Microcosm/mesocosm threshold concentrations showing
aquatic primary productivity community-level effects
4. Survival, growth, and reproduction of
shortnose sturgeon individuals via indirect
effects on terrestrial vegetation (riparian
habitat) required to maintain acceptable water
quality and spawning habitat
4a. Monocot and dicot seedling emergence EC25
4b. Monocot and dicot vegetative vigor EC25
Dwarf
Wedgemussel
1. Survival, growth, and reproduction via
direct effects
la. Eastern Oyster EC50
2. Survival, growth, and reproduction of
individuals via indirect effects on food source
or host (i.e., fish)
2a. Aquatic plant, invertebrate, and fish EC50 or LC50 and
NOAEC
3. Survival, growth, and reproduction via
indirect effects on habitat and/or primary
productivity (i.e., aquatic plant community)
3a. Most sensitive vascular aquatic plant EC50
3b. Non-vascular plant (algae) EC50
3c. Microcosm/mesocosm threshold concentrations showing
aquatic primary productivity community-level effects
4. Survival, growth, and reproduction Dwarf
Wedgemussel via indirect effects on
terrestrial vegetation (riparian habitat)
required to maintain acceptable water quality
and habitat.
4a. Monocot and dicot seedling emergence EC25
4b. Monocot and dicot vegetative vigor EC25
a Other data used in the characterization of potential risks to the assessed species are described in Section 4.
42
-------
2.9. Conceptual Model
2.9.1. Risk Hypotheses
Risk hypotheses are specific assumptions about potential adverse effects (i.e., changes in
assessment endpoints) and may be based on theory and logic, empirical data, mathematical
models, or probability models (U.S. EPA, 1998). For this assessment, the risk is stressor-linked,
where the stressor is the release of atrazine to the environment. The following risk hypotheses
are presumed for this endangered species assessment:
• Atrazine in surface water and/or runoff/drift from treated areas into the Chesapeake Bay
and its source waters may directly affect one or more of the assessed species by causing
mortality or adversely affecting growth or reproduction;
• Atrazine in surface water and/or run off/drift from treated areas may indirectly affect one
or more of the assessed species by reducing or changing the composition of food supply in the
Chesapeake Bay and its source waters;
• Atrazine in surface water and/or run off/drift from treated areas may indirectly affect one
or more of the assessed species by reducing or changing the composition of the plant community
in the Chesapeake Bay and its source waters, thus affecting primary productivity and/or cover;
and
• Atrazine in or runoff/drift from treated areas may indirectly affect one or more of the
assessed species by reducing or changing the composition of the terrestrial plant community (i.e.,
riparian habitat) required to maintain acceptable water quality and stream characteristics in the
Chesapeake Bay and its source waters. Runoff or drift into the terrestrial riparian buffer could
damage or destroy the riparian vegetation, which provides important ecosystem services such as
temperature regulation, energy input, and stream bank stabilization.
2.9.2. Diagram
The conceptual model is a graphic representation of the structure of the risk assessment. It
specifies the stressor (atrazine for all assessed species, and two degradates (DEA and DACT)
only for effects to turtles), release mechanisms, abiotic receiving media, biological receptor
types, and effects endpoints of potential concern. The conceptual model for the atrazine
endangered species assessment for the Chesapeake Bay is shown in Figure 2-11. Exposure
routes shown in dashed lines are not quantitatively considered because the resulting exposures
are expected to be sufficiently low as not to cause adverse effects to the species considered in
this assessment.
43
-------
Stressor
Source
Atrazine applied to agricultural fields,
residential lawns, golf courses, and
rights-of-way
^ * r
Groundwater i
Runoff
Spray drift
] Vapor phase and
j long range
transport
*
Receptor
Attribute
Change
Riparian Zone
Terrestrial plants
Habitat of assessed species
Aquatic plants; Aquatic
invertebrates; Aquatic vertebrates
Food chain
Decrease in
abundance;
Shift in prey base;
Individual Assessed
Species
Reduced survival;
Reduced growth;
Reduced
reproduction
Habitat integrity
Decreased water
quality;
Reduced cover;
Stream destabilization
Figure 2-11. Conceptual Model
The conceptual model provides an overview of the expected exposure routes within the action
area previously described in Section 2.5. In addition to the species included in this assessment,
other aquatic receptors that may be potentially exposed to atrazine include freshwater and
marine/estuarine fish, invertebrates, and plants. For turtles, the major exposure routes are
expected to be ingestion of contaminated water and food items. For the sturgeon and mussel, the
major routes of exposure are considered to be via the respiratory surface (gills) or the
integument. Direct uptake and adsorption are the major routes of exposure for aquatic plants.
The source and mechanism of release of atrazine into surface water are ground and aerial
application via foliar spray and coated fertilizer granules to agricultural (e.g., corn and sorghum)
and non-agricultural areas (i.e., golf courses, residential lawns, rights-of-way, etc). Surface
water runoff from the areas of atrazine application is assumed to follow topography, resulting in
direct runoff to the Chesapeake Bay and its source waters. Spray drift and runoff of atrazine may
also affect the foliage and seedlings of terrestrial plants that comprise the riparian habitat
surrounding the Chesapeake Bay and its source waters. Additional release mechanisms include
spray drift and atmospheric transport via volatilization, which may potentially transport site-
related contaminants to the surrounding air. Atmospheric transport is not considered as a
significant route of exposure for this assessment because the magnitude of documented
44
-------
exposures in rainfall are at or below available surface water and monitoring data, as well as
modeled estimates of exposure. In addition, modeling tools are not available to predict the
potential impact of long range atmospheric transport of atrazine.
Direct effects to freshwater or marine/estuarine organisms other than the six endangered species
included in this assessment may occur from exposure to atrazine. Effects to these species may
indirectly affect the sturgeon, mussel, and turtle species via reduction in food or habitat
availability or quality.
In addition to aquatic receptors, terrestrial plants may also be exposed to spray drift and runoff
from atrazine. Detrimental changes in the riparian vegetation adjacent to spawning areas of the
sturgeon or the habitat of the dwarf wedgemussel may adversely affect the assessed species
Indirect effects from riparian habitat alteration may include effects on water temperature, stream
bank stability, and sediment loading. Additional information on riparian habitat is included in
Section 5 (Risk Characterization).
Although the highest fish bioconcentration factor (BCF) of 8.5 (U.S. EPA, 2003c) suggests that
bioconcentration is not a primary concern for atrazine, the principle exposure route for turtles is
expected to be ingestion of contaminated food items. For this reason, bioconcentration is
considered in this assessment as it relates to dietary exposure of contaminated food items by
turtles as described in Section 3.
3.0 Exposure Assessment
An assessment of the potential for the listed species in the Chesapeake Bay watershed to be
exposed to atrazine has been conducted. This exposure assessment represents a modification of
the standard approach outlined in the Overview Document (U.S. EPA, 2004). The
PRZM/EXAMS model has been used to provide estimates of exposure in the standard water
body. Existing and new PRZM scenarios representing both agricultural and non-agricultural use
sites were utilized. A modified approach for assessing residential uses included modeling a
pervious (Vi acre lot) and impervious surface and weighting the output based on local data on the
percentage of impervious surfaces in the region. Finally, non-standard durations of exposure to
match available community-level ecotoxicity threshold concentrations were calculated. The
highest overall exposures were predicted to occur from the agricultural uses of atrazine (corn,
sorghum, and fallow/idle land) that are also likely to be in closest proximity to the species
locations. Modeling indicates that peak exposure estimates from application of atrazine to
forestry is similar to agricultural uses. However, model uncertainties and available information
on atrazine use in forestry operations suggests these are over-estimates and should not be used
for risk estimation. In general, the exposure assessment yields modeled peak exposure estimates
that are consistent with local and national monitoring data, while the modeled annual average
concentrations are two to ten times higher (depending on use site) than those seen in monitoring.
The intermediate duration exposures (14-day, 21-day, 30-day, 60-day, and 90-day averages)
cannot be estimated from the monitoring data due to insufficient sample frequency. However,
additional modeling exercises that simulate flowing water bodies suggest that modeled longer-
term exposures predicted using PRZM/EXAMS are likely overestimated for the environments
inhabited by the assessed species. Taken together, the PRZM/EXAMS modeling and existing
45
-------
monitoring data suggest that longer-term exposures (days to weeks) are expected to be in the
lower |ig/L range. The general approach used for the exposure modeling and discussion of
assumptions in the residential exposure modeling are outlined in Appendices C-l and C-2.
3.1. Conceptual Model of Exposure
The general conceptual model of expected exposure in this assessment is that the highest
exposures will occur in the headwater streams of the tributaries surrounding the Bay. Exposure
models for deriving EECs within an estuarine water body such as the Chesapeake Bay are not
available. However, available monitoring data obtained on May 3, 2006 from the U.S. EPA
Chesapeake Bay Program (http://www.chesapeakebav.net/index.cfm) was used to estimate
exposure to the assessed species within the main stem of the Chesapeake Bay. It is expected
that, given the likelihood that significant amounts of atrazine are being used in watersheds of
southern Pennsylvania and the Eastern Shore of Maryland and Virginia which drain to the Bay,
the available monitoring data are insufficient to predict all possible exposure in these areas.
Therefore, the best available monitoring data from multiple sources together with modeling
estimates were used to characterize potential exposures to the assessed species.
Two general types of estimates were used to characterize potential exposures to the six assessed
listed species: (1) modeling described in Sections 3.2 and 3.3; and (2) monitoring data from
several sources described in Section 3.4. Initial screening-level exposure estimates, derived
using PRZM/EXAMS and the standard water body scenario, were used in risk estimation for risk
quotient calculation. For reasons discussed in Section 3.3, EECs based on the standard water
body pond scenario are likely to be representative of short-term exposure concentrations in
headwater streams and minor estuarine inlets, but may overestimate exposure in larger water
bodies and/or flowing systems. Therefore, if the standard water body EECs resulted in LOC
exceedance, exposure was further characterized using additional modeling exercises and
available monitoring data. Table 3.1 below summarizes the data used to further characterize
exposures when screening-level EECs exceed the LOC for each of the six assessed species. All
methods and data are described in detail in below.
46
-------
Table 3.1. Types of Waters Inhabited by the Assessed Species and Data Used in Exposure
Refinement
Water Body
Data Source for Characterization of Standard
Water Body EECs (if necessary)
Assessed Species
Located In Water
Body
Headwater streams;
mid-range streams
Peak EEC: PRZM/EXAMS standard water body. No
refinement because peak EEC is considered representative of
these waters.
Longer-term EECsa:
(1) Modified PRZM/EXAMS scenarios including incorporation
of site specific flow data into reservoir scenario and variable
volume water model
(2) All available monitoring data.
Dwarf wedgemussel
Major rivers
Peak EEC: All available monitoring data in major rivers.
Longer-term EECsa:
(1) Modified PRZM/EXAMS scenarios including incorporation
of flow data.
(2) All available monitoring data.
Shortnose sturgeon, All
four sea turtle species
assessed
River mouths
(estuarine inlets);
main body of the
Chesapeake Bay
Peak and longer-term EEC: Chesapeake Bay monitoring
data.
Shortnose sturgeon, All
four sea turtle species
assessed
Minor estuarine
inlets
Peak: PRZM/EXAMS standard water body.
Longer-term averages":
(1) Qualitative analysis.
(2) Chesapeake Bay monitoring data.
All four sea turtle species
assessed
" Longer term exposure averages include durations of approximately several days and longer.
3.2. Use of Modeling to Characterize Potential Exposures to Atrazine in the Chesapeake
Bay Watershed
3.2.1. Modeling Approach
The analysis of both available monitoring data and usage information indicates that the exposure
assessment cannot rely exclusively on monitoring data. Although of high quality and generally
located where higher concentrations are expected to occur (Eastern Shore corn belt), the timing,
frequency of sampling, and location of the sample stations are unlikely to capture peak exposure
to atrazine in the high use areas and are unlikely to have sufficient sample frequency to
accurately estimate longer-term exposures. Therefore, an approach was implemented which
blends the standard assessment approach using standard PRZM/EXAMS scenarios for corn,
sorghum and turf with the non-agricultural scenarios (residential, impervious, rights-of- way, and
fallow/idle land) recently developed for use in the Barton Springs endangered species assessment
(U.S. EPA, 2006a). Available usage data (BEAD: Kaul et al, 2005, Kaul et al, 2006a, Kaul et al,
2006b) suggests that the heaviest usage of atrazine is likely to be on corn in the Eastern Shore;
therefore, all selected modeling scenarios were run using the weather data from the Wilmington,
Delaware meteorological station that is closest to the high use area.
47
-------
A total of seven scenarios were utilized for the Chesapeake Bay endangered species assessment.
Of these, three were developed as part of an endangered species assessment of atrazine for the
Barton Springs salamander (U.S. EPA, 2006a). Two of the Barton Springs scenarios were used
in tandem (residential and rights-of-way) with an impervious scenario (described below) while a
third (fallow/idle land) was used by itself as a standard PRZM/EXAMS scenario. The remaining
four scenarios (corn, sorghum, forestry, and turf) were taken from existing scenarios developed
for other regions of the United States and modeled using weather data from the Eastern Shore.
No additional scenarios were developed for this assessment. To address the potential use of
atrazine on the labelled use sites, all of the scenarios have been modeled; however, the results
were characterized to place emphasis on those actually expected to be present. Although not
specifically developed for the Eastern Shore, using the Pennsylvania (corn and turf), Kansas
(sorghum), Oregon Christmas tree (forestry), and non-agricultural scenarios described below
(impervious, residential, rights-of-way, and fallow/idle land) is expected to provide reasonable
high-end estimates of exposure. In addition, the Oregon Christmas tree scenario (developed for
the OP cumulative assessment) was used as a surrogate for forestry use in this area. Further
description and copies of the existing PRZM scenarios may be found at the following website.
http://www.epa.gov/oppefedl/models/water/przmenvironmentdisclaim.htm
One outcome of the 2003 IRED process was a modification to all existing atrazine labels that
stipulated setback distances around intermittent/perennial streams and lakes/reservoirs. The
label changes specify setback distances of 66 feet and 200 feet for atrazine applications
surrounding intermittent/perennial streams and lakes/reservoirs, respectively. These distances
were incorporated into this assessment and, the standard spray drift assumptions were modified
accordingly using AgDrift to estimate the impact of a setback distance of 66 feet on the fraction
of drift reaching a surface water body. The revised spray drift percentages were 0.6% for ground
applications and 6.5% for aerial applications and were incorporated into PRZM/EXAMS
modeling.
Models to estimate the effect of setbacks on load reduction for runoff are not currently available.
It is well documented that vegetated setbacks can result in a substantial reduction in pesticide
load to surface water (USDA, NRCS, 2000). Specifically for atrazine, data reported in the
USD A study indicate that well vegetated setbacks have been documented to reduce atrazine
loading to surface water by as little as 11% and as much as 100% of total runoff without a buffer.
It is expected that the presence of a well vegetated setback between the site of application of
atrazine and receiving water bodies could result in reduction in loading. Therefore, the aquatic
EECs presented in this assessment are likely to over-estimate exposure in areas with well-
vegetated setbacks. While the extent of load reduction can not be accurately predicted through
each relevant stream reach in the action area data from USD A (USD A, 2000) suggests
reductions could range from 11 to 100%.
3.2.1.1. Modeling Agricultural Uses
The non-agricultural scenarios were used within the standard framework of PRZM/EXAMS
modeling using the standard graphical user interface (GUI) shell, PE4v01.pl which may be found
at;
48
-------
http://www.epa.gOv/oppefedl/models/water/index.htm#przmexamsshell
EEC were calculated for the following exposure durations; single day, 14-day, 21-day, 30-day,
60-day, and 90-day. Durations of exposure for 14, 30 and 90 days were post-processed manually
using Microsoft Excel to provide standard one in ten year return frequency exposures.
A complete discussion of the standard modeling approach including further details on
PRZM/EXAMS may be found at the following website:
http://www.epa.gov/oppefedl/models/water/index.htm
3.2.1.2. Modeling Non-Agricultural Uses (Residential and Rights-of-Ways)
A modified approach for assessing residential uses that includes modeling of a pervious (Vi acre
lot) and impervious surface scenario was developed. Model output was weighted based on local
data on the percentage of impervious surfaces in the action area region.
The residential scenario was used in tandem with the impervious scenario. It is likely that some
overspray does reach the impervious surfaces in the residential setting. In order to account for
this, the impervious surface was modeled using three separate assumptions. For the purposes of
risk assessment it was assumed that 1% of the application rate could reach the impervious
surfaces surrounding each residential lot. This amount of overspray is not based on empirical
data (no publicized studies on this occurrence were found in the open literature); however, the
assumption is consistent with the standard assumption of 1% spray drift with ground applications
in ecological risk assessments. It should be remembered that this scenario represents general
impervious surfaces within a watershed not part of the Vi acre lot and includes roads, parking
lots, and buildings where overspray from residential lots is expected to be minimal. The Vi acre
lot, by comparison, was developed with a curve number reflective of the fact that the lot is
covered with both pervious surfaces (grass and landscaped gardens) and impervious surfaces
(driveways, sidewalks, and buildings). To test the assumption and address the uncertainty with
the lack of data for overspray, two alternate scenarios were modeled in order to characterize the
effect the 1% assumption. Modeling was completed for the impervious surface with 0% and
10% over spray to provide a lower bound and an upper bound. The results of these alternate
modeling exercises are discussed more fully in Section 3.3 of this assessment.
Two additional assumptions are critical to modeling the residential use. First, the scenario
assumes that the Vi acre lot is typical for this use pattern. In order to justify the assumption of ]A
acre lot as a typical exposure scenario, publicly available data from the United States Census
(Census) 2003 American Housing Survey (AHS) was reviewed on July 10, 2006 and is available
at the following website.
http://www.census.gov/hhes/www/housing/ahs
Data for all suburban homes available nationally was considered. It is assumed that most
pesticide applications, particularly herbicide applications, will occur in suburban settings. In
order to test the assumption of the Vi acre lot as the best representation, the AHS data for
49
-------
suburban homes that list total number of houses by lot size and by square footage of house (see
Table 1C-3 at the AHS website above) was evaluated. With a total of 45,552,000 total units
reported nationally for all suburban areas, 12,368,000 units (the largest class at 27%) were on
lots between 1/8 acre and Vi acre, while 9,339,000 units (the second largest class at 21%) were
on lots between Vi acre and '/2 acre. Overall, the median lot size was 0.37 acre. This analysis
suggests that the Vi acre lot is a reasonable approximation of suburban pesticide use.
The second critical assumption is that 50% of a Vi acre lot will be treated with atrazine. This
assumption was based partially on data from the AHS website and partially from professional
judgment about typical features and the percentage of a typical lot those features might require.
For example, the AHS survey data reports that of a total of 43,328,000 reported single detached
homes in suburban areas, 10,124,000 (the largest group at 23%) were between 1,500 and 2,000
square feet, while 7,255,000 (the third largest group at 17%) were between 2,000 and 2,500
square feet, and 9,513,000 (the second largest group at 22%) were between 1,000 and 1,500
square feet. From these data, it was assumed that a typical home is 2,000 square feet with a
1,000 square foot footprint. The lower sized houses less than 1,500 square feet are more likely to
represent single floor structures; thus, the 1,000 square foot estimate for a house footprint is
reasonable.
In addition to the footprint of the typical house, it was assumed that a typical house would have a
driveway of approximately 25 by 30 feet or 750 square feet and roughly 250 square feet of
sidewalk. A typical suburban home was also assumed to have roughly 300 square feet of deck
space and 900 square feet of garage. Finally, it was assumed that a substantial portion of the
typical home would be planted in landscaping with an estimate of 2,000 square feet. All of the
previous estimates are based on professional judgment and are not derived from the AHS data.
All of these areas are assumed to not be treated with a turf herbicide, resulting in a total area not
treated with atrazine of 5,200 square feet. Taking a total Vi acre lot size of 10,890 square feet
and subtracting the untreated square footage yields a total remaining area of 5,690, or roughly
50% of the total lot that could be potentially treated.
Currently, two categories of formulations are registered for atrazine use on residential sites.
These are granular and liquid formulations (wettable powder dry flowables). The formulations
have been modeled separately because application rates are different (2 lbs/acre for granular and
1 lb/acre for liquid), and the standard assumption for modeling granular formulations is different
from liquid formulations. Granular formulations are typically modeled as soil applied (in PRZM
the application method, or CAM, must be set to 8 for soil application with a minimized
incorporation depth of 1 cm) and 0% spray drift compared with a foliar application (application
method (CAM) set to 2 for foliar application with a 4-cm depth of incorporation) and standard
spray drift assumption of 1% for ground applications.
For the residential scenarios, it was assumed a percentage of the watershed was represented by
the Vi acre lot and some percentage was represented by impervious surfaces. To account for this
effect in the modeling of the residential scenario, the relative contribution of the impervious and
residential scenarios for different portions of the region surrounding the Chesapeake Bay was
evaluated. Land cover data (http://www.chesapeakebav.net/data/index.cfm) suggests that in the
northern Bay, near Baltimore, Maryland, impervious surface area near the Bay can approach
50
-------
70% of the total area. Alternatively, on the Eastern Shore, the percentage of impervious surface
rarely exceeds 30% and is generally less than 10%. In the southern Bay, near the Tidewater
region of Virginia, the percentage of impervious is roughly 50% of total. Figure 3-1 presents the
analysis of impervious coverage in the area surrounding the Chesapeake Bay relative to available
atrazine use data. For this screening level assessment, it was assumed that 30% of the watershed
is impervious.
51
-------
Chesapeake Bay V\tatershed & Impervious Suraces
Delav^re
Legend
Percent Impervious
i 'ir
_| 0-10
yj 111-20
21 - 3D
Miles
Figure 3-1. Percentage of Impervious Surfaces in the Chesapeake Bay and its Immediate
Tributaries
For the rights-of-way scenario, it was assumed that rights-of-way consist of 50% impervious and
50% pervious cover. In addition, it was assumed that no single watershed will be completely
covered by a rights-of-way use. This assumption seems reasonable given that rights-of-way
(roads, rail and utility lines) are typically long linear features that traverse a watershed. For the
screening level assessment, it was assumed that no more than 10% of the watershed is covered in
rights-of-way.
An analysis was completed for the Chesapeake Bay endangered species assessment for atrazine
to assess the amount of rights-of-way likely to be present in the action area. In this analysis,
52
-------
spatial data specific to the Chesapeake Bay watershed compiled by the Chesapeake Bay Program
(CBP) for roads and railways and obtained on July 12, 2006 that may be found at;
(http://www.chesapeakebav.net/data/data desc.cfm?DB=CBP GIS)
and internal Agency data for pipelines was obtained (spatial data for utility easements was
unavailable). The road, rail, and pipeline land cover data were added to a GIS map of the action
area (Figure 3-2) and a comparison of the density of the total network of potential use sites was
made. Each land cover feature in the GIS map is presented as a line with no width associated.
EFED applied a buffer using the Arc Toolbox within Arc Map in order to account for the
potential width of the each linear feature. This assignment of area to each feature was done in
order to compare the total area of each feature type (e.g. railways) with the total area of the
action area.
For each feature, an assumption was made about the typical width of the feature (e.g. width of
the road surface plus shoulders) plus the right of way area adjacent to the feature that could
potentially be treated. In each case, a conservative assumption for the width of the feature zone
plus the potentially treated area surrounding each was made. These assumed widths were based
on professional judgment but skewed the total feature estimate to the largest feature in the class.
For example, it was assumed that a national highway would yield that largest width which was
then applied to all primary and secondary highways within the action area. This approach is also
assumed to be conservative because it is unlikely that all features within the available data will
be treated with atrazine because many of the areas are likely to be maintained using mechanical
methods (e.g. mowing) or not treated at all. Using the CBP data different road types were able to
be distinguished using the USGS' classification scheme for digital line graph (DLG) data. In
these data set distinguishing between highways (primary and secondary roads), state and county
highways, state and local streets, unimproved trails, as well as minor features such as
interchanges and traffic circles was possible. Different buffer widths were applied to each
category of roadway depending upon the general use of the feature. For example, it is generally
assumed that primary and secondary roads are wider than state/county roads, which in turn are
generally wider than local streets. Based on this approach the following assumptions were made
for the width of each feature.
• Primary/Secondary Roads - 200 feet
• Class 3 Roads (State/County) - 100 feet
• Class 4 Roads (Streets in built up areas) - 50 feet
• Unimproved Trails - 25 feet
• Rail - 200 feet
• Pipeline - 100 feet
• Utility Line - 200 feet
Given these assumptions the percentage of rights-of-way land cover types plus associated buffers
for roads, railways, and pipelines within the action area for the Chesapeake Bay is 0.2 % of the
total area for rail, 3.6 % for all primary and secondary roads (interstates, national and state
highways), and 0.5% for pipelines. Including all classes of roads in the data set yields a road
density of 9.5%. This is believed to be an over estimation because it includes a high number of
53
-------
Right-of-Way Density in
Chesapeake Bay Action Area
^¦;\i
Legend
Pipeline in CB Action Area
CBPJnterchanges
CBP_Misceiianeous_Roads
CBP_Primary_Routes
CBP_Secondary_Roads
CBP_Toii_Roads
CB P_Traffi c_C i rc I e s
Railway in CB Action Area
&
Figure 3-2 Density of Road, Railways, and Pipelines as Surrogate for Right of Way Density
in Chesapeake Bay Watershed (Action Area)
54
-------
roads in urban and suburban areas unlikely to be treated with pesticides for right of way control.
Locally, it appears that higher percentages occur near more urbanized areas, however, it was
assumed that less right of way application of pesticides occur in urbanized areas. Additional
roads may be present in the action area not captured by the available spatial data and the analysis
above does not include utilities for which no spatial data are available. Therefore, the 10 %
assumption of rights-of-way in the action area used in this assessment is an over-estimation but
given the uncertainties is reasonable while still being conservative and protective. The impact of
this assumption on overall EECs is addressed in Section 3.3.
Use of atrazine on commercial forestry operations cannot be precluded as a potential use;
however, the available information suggests that atrazine is rarely used on commercial forestry
operations in the Chesapeake Bay watershed (Powers, 2006; VA DOF, 2004; Muir, 2006;
USD A, 2004; Wagner et al., 2004; Pannill, 2006). However, forestry is a predominant land
cover in some areas of the Chesapeake Bay watershed; therefore, this potential use has been
addressed using the Oregon Christmas tree scenario. This scenario was developed specifically
for the OP cumulative assessment recently completed by the Agency (USEPA, 2006b) and
represents a vulnerable site based on OP use information intended to represent a commercial
nursery operation. Information on the OP cumulative and scenarios used in modeling may be
found at:
http://www.epa.gov/pesticides/cumulative/2006-op/index.htm
The Oregon Christmas tree scenario is expected to approximate commercial forestry operations
where herbicides are typically applied during the seedling emergence and juvenile growth stages
to prevent competition with newly planted trees. The scenario was not modified to represent
local conditions but was modeled using local weather data. Several factors suggest that
modeling of forestry uses of atrazine are likely to result in an over-estimation of exposure. As
previously mentioned, atrazine use in forestry operations in the Chesapeake Bay watershed is
considered to be rare. Secondly, modeled estimates represent a one in ten year return frequency
using 30 years of modeled output; however, if atrazine were used at all, it would likely be
applied for only one or two years during early growth stages. Taken together, the best available
data suggest that the modeled exposures for atrazine forestry use are likely to over-estimate
exposure. In addition, the highest EECs were from the sorghum scenario and these were used for
risk estimation while the EECs for forestry are discussed qualitatively in the risk description.
3.2.2. Model Inputs
3.2.2.1. Label Application Rates and Intervals
Labels may be categorized into two types: labels for technical products and labels for
formulated, or end use, products. Technical products contain atrazine of high purity and are
used only to make formulated products. Formulated products can be applied in specific areas
to control weeds while technical products are not used directly in the environment but rather
to make formulated products. The formulated product labels limit atrazine's potential use to
only those sites where the labels specify.
55
-------
In the January and October IREDs, EPA stipulated numerous changes to the use of atrazine
including label restrictions and other mitigation measures designed to reduce risk to human
health and the environment. Specifically pertinent to this assessment, the Agency entered into a
Memorandum of Agreement (MO A) with the atrazine registrants. In the MO A, the Agency
stipulated certain label changes must be implemented on all atrazine labels prior to the 2005
growing season including cancellation of some uses, reduction in application rates, and
requirements for harmonization across labels including setbacks from waterways. Specifically,
the label changes stipulate no use of atrazine within 50 feet of sinkholes, within 66 feet of
intermittent and perennial streams, and within 200 feet of lakes and reservoirs. It is expected that
a setback distance will result in a reduction in loading due to runoff across the setback zone;
however, current models are not capable of estimating these reductions quantitatively. A
qualitative discussion of the potential impact of these setbacks on estimated environmental
concentrations of atrazine for the assessed species is discussed further in Section 3.2. Table 3.2
provides a summary of label application rates for atrazine uses evaluated in this assessment.
Although currently registered uses of atrazine are numerous, only residential uses, turf, corn,
sorghum, fallow/idle land, and rights-of-ways were modeled because these uses are expected
to predominate in the Chesapeake Bay watershed. Atrazine use in forestry was also
modeled; however, for reasons previously described, atrazine use on forestry is expected to
be minimal.
Atrazine is formulated as liquid, wettable powder, dry flowable, and granular formulations.
Application equipment for the agricultural uses includes ground application (the most
common application method), aerial application, band treatment, incorporated treatment,
various sprayers (low-volume, hand held, directed), and spreaders for granular applications.
Risks from ground boom and aerial applications are quantified in this assessment because
they are expected to result in the highest off-target levels of atrazine due to generally higher
spray drift levels. Due to the high mobility of atrazine, runoff associated with large rain
events is expected to be responsible for the greatest off-target movement of atrazine. Smaller
runoff events resulting from over irrigation would result in lower levels of off-target
movement.
56
-------
Table 3.2 Label Application Information for the Chesapeake Bay Endangered Species
Assessment
Scenario
Maximum
Application
Rate
(lbs/acre)
Maximum
Number of
Applications
Date of First
Application
Formulation
Method
of
Application
Interval
Between
Applications
Residential
2.0
2
April 1
Granular
Ground
30 days
Residential
1.0
2
April 1
Liquid
Ground
30 days
Right-of-
Way
1.0
1
June 1
Liquid
Ground
NA
Fallow/idle
land
2.25
1
November 1
Liquid
Ground
NA
Turf
2.0
2
April 1
Granular
Ground
30 days
Turf
1.0
2
April 1
Liquid
Ground
30 days
Corn
2
l1
April 1
Liquid
Ground and
Aerial
NA
Sorghum
2
l1
April 1
Liquid
Ground and
Aerial
NA
Forestry
4.0
1
June 1
Liquid
Ground and
Aerial
NA
Fallow/idle
land
2.25
1
November 1
Liquid
Ground and
Aerial
NA
1 Actual labeled maximum rates are 2.0 lb/acre for a single application with no more than 2.5 lbs/acre per year. The
rate and number of applications reported in this table are an approximation of the label maximum given the current
limitation in PRZM/EXAMS graphical user interface PE4v01.pl. Currently, PE4v01.pl allows multiple applications
but the rate cannot be varied from one application to the next.
57
-------
3.2.2.2.
Typical Use Rates and Application Intervals
Application rates, number of applications, and application intervals were estimated at the state
level for Maryland, Pennsylvania, and Virginia (BEAD: Kaul et al, 2006a, Kaul et al, 2006b,
Zinn et al, 2006). The information from BEAD was developed from a combination of USDA-
NASS6, and data obtained from Doane (www.doane.com; the full dataset is not provided due to
its proprietary nature). Data from both sources were averaged together over the years 2000 to
2004 to calculate average annual usage statistics by state and crop for atrazine, including pounds
of active ingredient applied, percent of crop treated, number of applications per acre, application
rate per acre, and base acres treated. Application rates are provided at the state level for only
crops grown in the immediate vicinity of the Chesapeake Bay including corn, fallow/idle land (as
a surrogate for rangeland), and sorghum on which atrazine is registered. No other labeled
agricultural uses (sugarcane, guava, and macadamia nuts) are present in the action area.
For atrazine use aggregated for the three states identified above, typical application rates for corn
and sorghum range from 1.0 to 1.2 lbs/acre, while typical rates on fallow/idle land range from
1.0 to 2.0 lbs/acre. Typically the 90th percentile of reported application rates is used as an upper
bound on actual use (U.S. EPA, 2000); however, no data on the 90th percentile is available.
Typical application rates and number of intervals should be evaluated with caution because these
values represent an average , which implies that atrazine is actually applied at rates higher than
those reported as typical a significant percentage of the time.
BEAD also provided additional estimates on the typical number of applications for atrazine in
the Chesapeake Bay area. This information indicates that the typical number of applications for
corn and sorghum is roughly half of the label maximum, while the typical rate on rangeland is
approximately equivalent to the labeled directions. Typical application rates used to characterize
exposure estimates are in Table 3.3.
To refine the risk assessment for the atrazine endangered species assessment, the minimum and
typical application intervals when more than one application is made per year on a site were
evaluated. Intervals were estimated by first determining the registered herbicide/site
combinations within the Chesapeake Bay watershed. The sites chosen were those with the
average number of applications greater than one (BEAD: Kaul, Grube, and Kiely, 2005). If the
average number of applications equals one, it was assumed that only one application is made,
and, therefore, that the typical interval is not needed. Only sites with greater than one
application of a pesticide are discussed below.
For corn, most growers apply atrazine only once per season. However, approximately 12 percent
of growers apply atrazine more than once, following a pre-emergence application with a post-
emergence application (Assessment of Potential Mitigation Measures for Atrazine, 2003).
According to atrazine label information for corn, the minimum application interval is either 14
days or not specified on the label (BEAD: Kaul and Carter, 2005). BEAD contacted experts in
6 United States Depart of Agriculture (USDA), National Agricultural Statistics Service (NASS) Chemical Use
Reports provide summary pesticide usage statistics for select agricultural use sites by chemical, crop and state. See
http://www.usda.gOv/nass/pubs/estindxl.htm#agchem.
58
-------
the States and this interval was described as the absolute minimum interval. Usually application
intervals would be longer.
In Maryland, atrazine intervals are likely to be at least 14 days at a minimum. Whether atrazine
is applied as an early pre-plant application followed by an at-plant application, or atrazine is
applied post-emergence after a pre-emergence treatment, the recommended interval is 14 days or
more (Glenn, 2006). For Virginia, the BEAD estimate for the number of atrazine applications on
corn may be high, with the number approximately 1.1. A typical interval for a post-emergence
treatment after an earlier treatment is approximately 35 days (ranging between 30 to 40 days).
In Pennsylvania, atrazine is usually applied either pre-emergence or post-emergence. For those
situations with both applications of atrazine, a pre-emergence application would be followed by a
lower amount of atrazine as a post-emergence application, with the interval likely to be at least
21 days. Typically, the interval is 28 days.
For sorghum, atrazine may be applied at various timings. "Atrazine is effective at many
application timings including: winter weed control, and pre-plant for control of weeds prior to
planting through post-plant as long as weeds are no more than one and one-half inches and
sorghum is six to 12 inches tall" (Assessment of Potential Mitigation Measures for Atrazine,
2003). According to atrazine label information, for sorghum, the minimum application interval
is either 21 days or not specified on the label (BEAD: Kaul and Carter, 2005). The interval
between applications is likely to be similar to that for corn for Maryland (Glenn, 2006). In
Virginia, atrazine is only applied to sorghum once (Hagood, 2006).
For fallow use, according the Aatrex® 4L label and some other atrazine labels, only one
application of atrazine may be made in fallow period (CDMS search). In addition, the Atrazine
Interim Reregi strati on Decision (IRED) states that only one application per year may be made
for chemical fallow applications (U.S. EPA, 2003a).
Typical application rates and number of intervals must be evaluated with caution in that these
represent an average that implies that a percentage of the time atrazine is actually being applied
at rates higher than those reported as typical. Table 3.3 summarizes the typical application rates
and number of applications relative to those used in this assessment.
59
-------
Table 3.3. Comparison of Modeled Application Rates and Number of Applications with
Typical Use Data Used in the Triazine Cumulative Risk Assessment
Scenario
Maximum
Application
Rate
(lbs/acre)
Maximum
Number of
Applications
Typical
Application
Rate1
(lbs/acre)
Typical Number of
Applications1
Corn
2.0
1
1.0-1.2
1 - 1.3
Sorghum
2.0
1
1.0-1.2
1.0-1.6
Rangeland2
1.0
1
1.0-2.0
0.6-0.73
Reported as range of values from states of Delaware, Maryland, Pennsylvania, and Virginia as prepared for the
triazine cumulative risk assessment
2
Rangeland compared with reported use rates on Fallow; rangeland is no longer a labeled use (U.S. EPA, 2006c)
3
Rates reported as less than 1.0 considered equivalent to 1.0
The appropriate PRZM input parameters were selected from the environmental fate data
submitted by the registrant and in accordance with model parameter selection guidelines
(Guidance for Selecting Input Parameters in Modeling the Environmental Fate and Transport of
Pesticides, Version 2.3, February 28, 2002). These parameters are consistent with those used in
both the 2003 IRED and cumulative triazine risk assessment. The date of first application was
developed reviewing several sources of information including data provided by BEAD, Crop
Profiles maintained by the USD A, and conversations with local experts. More detail on the crop
profiles and the previous assessments may be found at:
http://pestdata.ncsu.edu/cropprofiles/cropprofiles.cfm
http://www.epa.gov/oppsrrdl/REDs/atrazine ired.pdf
http://www.epa.gov/pesticides/cumulative/common mech groups.htm#chlpro
A summary of the model inputs used in this assessment are provided in Table 3.4.
60
-------
Table 3.4. Summary of PRZM/EZAMS Environmental Fate Data Used For Aquatic
Exposure Inputs For Atrazine Endangered Species Assessment for the Chesapeake Bay
Fate Property Value MRID (or source)
Molecular Weight
215.7 g/mol
Product Chemistry
Henry's constant
2.58 xlO"9 atm-m3/mol @
20° C
Product Chemistry
Vapor Pressure
3 x 10"7 mm Hg @ 20° C
Product Chemistry
Solubility in Water
33 mg/1
Product Chemistry
Photolysis in Water
335 days
MRID 42089904
MRID 40431301
Aerobic Soil Metabolism Half-
152 days
lives
MRID 40629303
MRID 42089906
Hydrolysis
stable
MRID 40431319
Aerobic Aquatic Metabolism
304 days
2x aerobic soil
(water column)
metabolism rate constant
Anaerobic Aquatic Metabolism
(benthic)
608 days
MRID 40431323
MRID 40431324
MRID 41257901
Koc
88.78 ml/g
MRID 41257902
MRID 41257904
MRID 41257905
MRID 41257906
Application Efficiency
95 percent for aerial
99 percent for ground
default value
Spray Drift Fraction
6.5 % for aerial
0.6 % for ground
AgDrift value based on
setback distance of 66 ft
3.2.3. Model Results
Model estimated surface water concentrations are summarized in Table 3.5. These EECs
represent the screening level EEC which are used in risk estimation. Discussion and additional
characterization of these EEC is in Sections 3.3.
61
-------
Table 3.5. Summary of PRZM/EXAMS Output for all Scenarios Modeled for the Atrazine Endangered
Species Assessment for the Chesapeake Bay Watershed using the Standard Water Body
90th Percentile
Use Site
Application
Rate
(lbs/acre)
Number of
Applications
(interval)
First
Application
Date
Peak
EEC
fag/L)
14-
day
EEC
21-
day
EEC
30-
day
EEC
60-
day
EEC
90-
day
EEC
(hs/L)
(hs/L)
(hs/L)
(hs/L)
(hs/L)
Residential
Granular1
2.0
z
(30 days)
April 1
11.5
11.4
11.3
11.3
11.1
10.8
Residential
- Liquid1
1.0
2
(30 days)
April 1
7.7
7.6
7.6
7.5
7.3
7.2
Right-of-
Way 1
1
1
June 1
2.9
2.9
2.8
2.8
2.8
2.7
Corn
2
l2
April 1
47.2
46.9
46.8
46.5
45.6
44.4
Sorghum
2
l2
April 1
55.4
54.8
54.6
54.3
53.7
52.5
Fallow/idle
land
2.25
1
November 1
45.1
45.0
45.0
45.0
45.0
44.9
Forestry3
4.0
1
June 1
50.5
49.7
49.7
49.2
47.5
45.9
Turf-
Granular
2.0
2
(30 days)
April 1
9.9
9.9
9.9
9.9
9.8
9.8
Turf-
Liquid
1.0
2
(30 days)
April 1
5.3
5.2
5.2
5.1
5.0
4.9
1 Assumes 1% overspray of atrazine to the impervious surfaces. Alternate assumptions of 0% and 10% overspray to
impervious surfaces are tested in Section 3.3.
2 Actual labeled maximum rates are 2.0 lb/acre for a single application with no more than 2.5 lbs/acre per year. The
rate and number of applications reported in this table are an approximation of the label maximum given the current
limitation in PRZM/EXAMS graphical user interface PE4v01.pl. Currently, PE4v01.pl allows multiple applications
but the rate cannot be varied from one application to the next.
3 - Forestry EEC not recommended for risk estimation due to uncertainty in actual use pattern and overestimation of
application frequency
62
-------
3.3. Additional Modeling Exercises Used to Characterize Potential Exposures
3.3.1. Residential Uses
As noted previously, it was assumed that that a reasonable exposure scenario for the residential
use would be that 30% of the residential scenario is impervious surface. To evaluate this
assumption, and to get a perspective of how atrazine EECs might vary throughout the
Chesapeake Bay Watershed, the impact of alternate percentages of impervious surface coverage
on overall EECs were evaluated. Table 3.6 below presents the results of this analysis. The
analysis indicates that the overall EEC decreases as the percentage of impervious surface
increases. This is likely due to the overall increase in runoff volume which dilutes the total
pesticide mass loading.
Table 3
Expc
Use Site
.6. Summary
•sures from th
Percent
Impervious
of the Impacl
e PRZM/KW
Application
Rate
(lbs/acre)
of Variations in Pervious to Impervious Ratio on Predicted
lMS Residential Scenario (granular) with 1% Overspray1
90th Percentile
Number of u n 3Q 6Q 9Q
Applications Peak
r wr day day day day day
(interval) E£C E£C E£C E£C E£C
(Mg ) (US/L) (ug/L) (ug/L) (ug/L) (ug/L)
Residential
in Eastern
Shore
30%
2.0
2
(30 days)
11.5
11.4
11.3
11.3
11.1
10.8
Residential
in
Northern
Bay
70%
2.0
2
(30 days)
9.0
8.9
8.9
8.9
8.8
8.7
Residential
in
Southern
Bay
50%
2.0
2
(30 days)
10.3
10.2
10.2
10.2
10.0
9.9
1 In this case overspray represents an amount of granules landing on impervious (non-target) surfaces
To evaluate the assumption of 1% overspray, alternatives varying percentage of overspray that
could occur on the impervious surface were modeled. For the residential and rights-of-way
scenarios, it was assumed 1% overspray onto the impervious scenario. An alternative modeling
exercise was conducted to evaluate the significance of overspray. To account for potential
overspray, the impervious scenario (assuming 30% of watershed is impervious and 50% of the ]A
acre lot is treated as above) with a percentage of the application rate as being applied to the non-
target surface was modeled. It was assumed that no more than 10% of the intended application
rate would end up on the impervious surface. Given that the impervious scenario is intended to
represent non-target surfaces such as roads, parking lots and buildings, it seems reasonable to
assume that 10% overspray is an over-estimation of what would likely occur. To model the
overspray, the binding coefficient was set to zero and the aerobic soil metabolism half life set to
stable in lieu of actual data. Thus, it is assumed that non-binding would occur on these surfaces
63
-------
and that limited degradation would occur. The total application rate was then multiplied by the
percentage overspray. For the residential scenario this yielded an application rate on the
impervious surface of 0.2 lbs/acre. In addition, the same analysis using an assumption of 0%
over spray was modeled.
Comparison of the resulting EEC indicates that with 10% overspray the overall EECs for the
residential use pattern are increased by nearly a factor of three, while assuming 0% overspray
only slightly decreases the EEC compared to 1% overspray. This is not unexpected given the
increased runoff, lack of binding, and lack of degradation being assumed. Without actual data
for these processes, it is impossible to determine whether these exposures reflect reality;
although, it is expected that these assumptions are likely to be conservative (some binding and
degradation could occur). The analysis does suggest that overspray onto impervious surfaces
can, and possibly is a significant issue when the percentage of overspray is high. The
comparison is presented in Table 3.7.
Table 3.7. Comparison of Residential Scenario with an Assumption of No Overspray on Impervious
Surface to the Alternate Assumptions of 10% and 1% Overstay to Impervious Surfaces1
90th Percentile of 30 Years of Output
Use Site
Application
Rate
(lbs/acre)
Number of
Applications
(interval)
First
Application
Date
Peak
EEC
(Jig/L)
14-
day
EEC
(US/L)
21-
day
EEC
(US/L)
30-
day
EEC
(US/L)
60- 90-
day day
EEC EEC
(Ug/L) (jig/L)
Residential
with 1%
Overspray
to
2.0
2
(30 days)
April 1
11.5
11.4
11.3
11.3
11.1 10.8
Impervious
Residential
with no
Overspray
to
2.0
2
(30 days)
April 1
9.3
9.2
9.2
9.1
8.8 8.6
Impervious
Residential
with 10%
Overspray
to
2.0
2
(30 days)
April 1
34.5
34.3
34.2
34.1
33.9 33.6
Impervious
1 assumes 30% impervious surface in watershed
For this assessment it was assumed that 50% of the Vi acre lot is treated. To test the significance
of this assumption, the exposure scenario was re-run using different assumptions of 75% and
10% treatment of the ]A acre lot. Modeling an increasing percentage of the ]A acre lot that is
treated to 75% of the total area increases the EECs by roughly 50%, while decreasing the
64
-------
percentage treated to 10% of the total area decreases the EECs by a factor of three. The results
of both analyses are presented in Table 3.8.
Table 3.8.
Comparison of Residential Scenario with an Assumption of 50% of the lA acre lot treated to
the Residential Scenario (granular) with an Assumption of 75% and 10% of the % acre lot treated
90th Percentile of 30 Years of Output
Use Site
Application
Rate
(lbs/acre)
Number of
Applications
(interval)
First
Application
Date
Peak
EEC
(Jig/L)
14-
day
EEC
(Hg/L)
21-
day
EEC
(Hg/L)
30-
day
EEC
(Hg/L)
60-
day
EEC
(Hg/L)
90-
day
EEC
(Hg/L)
Residential
with 50%
of lot
2.0
2
(30 days)
April 1
11.5
11.4
11.3
11.3
11.1
10.8
treated
Residential
with 75%
of lot
2.0
2
(30 days)
April 1
16.1
16.0
15.9
15.8
15.5
15.1
treated
Residential
with 10%
of lot
2.0
2
(30 days)
April 1
4.0
3.9
3.9
3.9
3.9
3.8
treated
In the initial screening level assessment, it is assumed that the ratio of pervious to impervious
surface (70/30) accounts for the difference in exposure and runoff. This ratio is best
characterized as a conservative estimate of runoff potential. In fact, it is expected that the
differential will be more highly skewed for runoff from the impervious scenario than is reflected
in the ratio.
To test this differential and the potential effect it has on runoff and ultimately exposure, both
scenarios were modeled using local weather data. The analysis indicates that, when run with the
Wilmington, Delaware weather station data, the impervious surface scenario yields greater than
five times more runoff than does the Vi acre lot scenario. In areas where impervious cover
approaches 100% of the total, the impervious scenario best represents the amount of runoff.
Alternatively, in more rural areas where impervious cover is less than 5% of the total, the
pervious scenario bests represent the runoff amount. The screening level EECs presented in
Table 3.5 through Table 3.8 were generated by weighting the EECs based on the percentage of
impervious and not the differential in runoff potential. To test the impact of the differential in
runoff potential, an analysis of weighting the runoff yields from the two scenarios relative to the
percentage of cover in the Northern Bay, Eastern Shore, and Southern Bay was conducted. The
analysis indicates that where impervious cover is present, the straight weighting based on
percentage of impervious under-estimates potential runoff and loading. This suggests that for the
no-overspray scenario above, the effect is to over-estimate exposure. Alternatively, if the
assumptions of runoff differential were applied to the overspray evaluation, the EEC would
65
-------
increase. Table 3.9 presents a summary of the impact of these varying impervious percentages
on total runoff amounts.
Table 3.9. Percentage of Runoff Resulting from Impervious Surfaces
within Different Parts of the Chesapeake Bay Watershed1
Residential Residential
Location in Residential Runoff 0/ T . Runoff 0/ f Total
Chesapeake Max % Max % Runoff Weighted „ . . ml)c^|()us Weighted T °°. Weighted
,, , . .. , Residential Runon trom , ... Impervious „ ....
Bay Impervious Pervious from by % ofTotal PR7M (cm) by % nfTotal Runoff
Watershed PRZM (cm) Pervious ot Total FRZM (cm) Impervious ot Total (cm)
(cm) (cm)
North Bay 0.7 0.3 325.6 97.68 0.06 2126 1488.2 0.94 1585.88
Eastern Shore 0.3 0.7 325.6 227.92 0.26 2126 637.8 0.74 865.72
South Bay 0.5 0.5 325.6 162.8 0.13 2126 1063 0.87 1225.8
Runoff reported as centimeter represents the depth of water running off of the entire watershed. For example, a
runoff depth of 1 cm from a 10 hectare watershed equals 1,000,000,000 cm3, or 1,000,000 liters
The impact of this effect on the residential scenario was evaluated by adjusting the assumed
percentage of pervious to impervious from the 70 to 30 ratio used in the screening level
assessment (Table 3.5.) and replacing it with the ratios in Table 3.9 for different portions of the
Bay using the same assumptions of 50% treated lot and 1% overspray. The results of this
analysis are presented in Table 3.10 and indicate that the estimates above are likely over-
predicting exposure when no overspray is modeled.
66
-------
Table 3.10. Summary of the Impact of Runoff Differential on the Predicted Exposures from the
PRZM/EXAMS Residential Scenario (granular) Using the Analysis from Table 3.91
90th Percentile
Use Site
Ratio of
Pervious to
Impervious
Application
Rate
(lbs/acre)
Number of
Applications
(interval)
Peak
EEC
fag/L)
14-
day
EEC
21-
day
EEC
30-
day
EEC
60-
day
EEC
90-
day
EEC
(Hg/L)
(Hg/L)
(Hg/L)
(Hg/L)
(Hg/L)
Residential
70/30
2.0
2
(30 days)
11.5
11.4
11.3
11.3
11.1
10.8
Residential
in
Northern
06/94
2.0
2
(30 days)
8.0
7.9
7.9
7.8
7.8
7.7
Bay
Residential
in Eastern
Shore
26/74
2.0
z
(30 days)
8.9
8.8
8.8
8.8
8.7
8.6
Residential
in
Southern
Bay
13/87
2.0
2
(30 days)
8.3
8.2
8.2
8.1
8.0
8.0
1 Assuming 1% overspray to the impervious scenario
3.3.1.1. Summary of Non-Agricultural EECs
The above analysis suggests that EECs in the standard water body scenarios are likely in the low
|ig/L range (<10 |ig/L). However, if substantial overspray occurs, resulting in higher atrazine
levels on impervious surfaces, then EECs could be higher. However, all EECs for the residential
uses were lower than those estimated for agricultural uses.
3.3.2. Additional Characterization of Agricultural Use EECs
3.3.2.1. Modeling Using the Standard Water Body and Typical Use Rates Within the
Chesapeake Bay Watershed
In addition to the analysis of modeling data discussed above, alternative modeling of the corn
and sorghum scenarios were conducted using the typical application rates information because
these were the two scenarios yielding the highest exposures in the assessment. The rates and
number of applications are similar for both uses with a typical application rate of 1 lb/acre with
number of applications between 1 and 1.5 (1.5 applications represent an average of multiple
applications applied at lower than maximum rates). To simplify this part of the assessment the
refined application rate was modeled at 1 lb/acre with one application. Comparison of typical
applications rates (essentially equivalent to the average of all available reported data) with
monitoring data and modeling with labeled maximum rates should only be used for
67
-------
characterization because by its nature a typical, or average, rate implies that roughly 50% of the
applications are occurring above this value. Given the site-specific nature of an endangered
species assessment it is impossible to rule out that some percentage of actual applications are
occurring in proximity to listed species. However, overall the results of this analysis are not
unexpected and indicate that use at the typical application rates results in a reduction on EEC
across the board by a factor of two. The results of this analysis are summarized in Table 3.11.
Table 3.11. Summary of PRZM/EXAMS Output for all Scenarios Modeled for the
Atrazine Endangered Species Assessment for the Chesapeake Bay Watershed Using a
Standard Water Body
90th Percentile
Application Number of First jj. 21- 30- 60- 90-
Use Site Rate Applications Application Peak ^ (|ay (|ay (|ay (|av
(lbs/acre) (interval) Date ,EE5s EEC EEC EEC EEC EEC
°18'L) (U2/L) (U2/L) (ujj/L) (ujj/L) (ujj/L)
Corn
2
1
April 1
47.2
46.9
46.8
46.5
45.6
44.4
Corn
1
1
April 1
23.6
23.5
23.4
23.3
22.8
22.2
Sorghum
2
1
April 1
55.4
54.8
54.6
54.3
53.7
52.5
Sorghum
1
1
April 1
27.6
27.4
27.3
27.1
26.9
26.2
3.3.2.2. Characterization of Potential Exposures in Flowing Waters
The standard assessment for aquatic organisms relies on estimates of exposure derived from
PRZM/EXAMS using the standard water body. The standard water body is a 1 hectare water
body that is 2 meters deep with a total volume of 20,000,000 liters and is modeled without flow.
The standard water body was developed to provide an approximation of high-end exposures
expected in water bodies, lakes, and perennial/intermittent streams adjacent to treated
agricultural fields. Typically, this has been interpreted as a stream with little or low flow. For
non-persistent pesticides, the standard water body provides a reasonably high-end estimate of
exposure in headwater streams and other low flow water bodies for both acute and longer-term
exposures. For more persistent compounds, the non-flowing nature of the standard water body
still provides a reasonable high-end estimate of peak exposure for many streams found in
agricultural areas; however, it appears to over-estimate exposure for longer-term time periods in
all but the most static water bodies.
The hydrologic landscape of the Chesapeake Bay watershed was characterized by generalizing
the stream network into six general classifications (Section 2.4). In the case of the modeled
concentrations (presented in Table 3.5) that were derived with a non-flowing standard water
body, it is expected that the peak exposures are generally representative of the headwater streams
in areas of low topography such as the Eastern Shore and parts of the Coastal Plain province, as
68
-------
well as minor inlets surrounding the Bay. It is also expected that the standard water body
scenarios is over-estimating exposure in water bodies with flowing water, including the lower
and main channels of the Bay tributaries, the mouths of the tributaries, and the Bay itself.
To characterize the potential impact of flowing water on the longer-term exposures (14-day, 21-
day, 30-day, 60-day, 90-day, and annual average), additional modeling and analysis of available
monitoring data was conducted. Alternate approaches to modeling with the standard water body
were conducted to provide a general sense of the relative reduction in long term exposure which
might be occurring in water bodies where flow is higher than small headwater streams in low
topographic regions (interior of the Eastern Shore).
The sorghum scenario was selected and modeled for the Chesapeake Bay watershed using the
same input parameters presented in Table 3.4 with assumptions for flow (described below). This
scenario provided the highest EEC of any scenario modeled. In fact, usage data from BEAD
suggests (Kaul et al, 2005) that relatively little sorghum is grown in the watershed when
compared to corn. However, given that sorghum is present in the watershed and the fact that the
predicted EEC for corn was similar the selection is reasonable and representative of both corn
and sorghum.
The standard EXAMS standard water body for ecological risk assessment was used as the
receiving body for runoff from a 10 hectare field and is a static water body. The standard water
body is intended to represent a water body or an ecologically sensitive stream adjacent to an
agricultural field. Typically, this is conceptualized as a headwater stream, however; there are
examples of higher order streams with very low flow rates (e.g. small tidal inlets, oxbow lakes
only occasionally fed by stream flow, etc.). In order to test the effect of flow on these predicted
concentrations, the standard water body was modeled as above but allowed the model to route
runoff water from the 10 hectare field through the 1 hectare water body. The results of all the
modeling are presented in Table 3.12.
Further analysis was conducted by pairing PRZM output from the sorghum scenario with the
variable volume water model (VVWM) that was developed for the Probabilistic Risk
Assessment (PRA) process. The VVWM was developed based on the recommendation of the
Scientific Advisory Panel (SAP) to account for the influence of input and output (flow) on model
predictions. The VVWM was used to evaluate the impact of varying volume on the overall
EECs. In general, the WWM yielded EECs below the EXAMS water body, but still above the
annual averages from the available monitoring data (see discussion below). Two alternate model
runs with the VVWM were conducted. The first was done using standard assumptions and
environmental fate parameters generally consistent with the non-flowing standard water body
discussed previously. The assumptions in this model run included a 2 meter depth water body
which can drop to 0.02 meter and rise to 3 meters before flow occurs. The second assumption
was designed to represent a larger volume water body that maximizes flow into the water body.
This was accomplished by increasing the overall maximum depth of the water body to 10 meters.
The net effect of this change is to reduce the original estimates with the VVWM by roughly
50%. The results are summarized in Table 3.12. Documentation and rationale for the
assumptions used in the VVWM may be found at:
69
-------
http://www.epa.gOv/scipoly/sap/2004/index.htm#march
To further characterize the impact of larger water bodies with flow, the sorghum scenario was
run using the Index Reservoir as the receiving water body. The Index Reservoir represents a 5.3
hectare water body draining a 172 hectare watershed and is used for drinking water assessments
for human health risk assessment. In the case of the Index Reservoir, the standard approach is to
take the total runoff from the 172 hectare watershed calculated by PRZM through the Index
Reservoir in EXAMS and route that volume of water as flow through the reservoir while
assuming no change in reservoir volume. The predicted EECs and flow rates from these
alternate approaches that assume flow are slightly below the original non-flowing EEC and are
summarized in Table 3.12. More information on the Index Reservoir may be found at
http://www.epa.gov/oppfeadl/trac/science/reservoir.pdf
The modeled output relative to actual flowing streams was evaluated to provide context to these
estimates. The USGS collected flow rates from 734 streams, creeks and rivers from across
Virginia representing the range of physiographic provinces in Virginia that are typical of stream
types found in the Chesapeake Bay watershed. The flow data was subsequently separated into
regions representing the Eastern Shore, Coastal Plain, Piedmont, and Mountain regions of
Virginia. The modeled flow rates from PRZM were then compared with the regional dataset of
flow developed by the USGS for the Virginia Department of Environmental Quality (VADEQ)
and obtained on May 30, 2006 which can be found at:
http://va.water.usgs.gov/vadeq data/number scroll.htm
As shown in Table 3.12., the 7Q10 (7 day average with a return frequency of 10 years that is
indicative of base-flow values) and Q50 (50th percentile of reported values) values indicate that
flow varies dramatically from the low topography Eastern Shore to the mountainous regions of
western Virginia. Neither flow estimate is a perfect representation of flow conditions as
modeled, but is intended to provide a range of possible flow rates. These flow values range by a
factor of two orders of magnitude across the state. Comparison with the modeled flow rates
suggests that the PRZM modeling is yielding significantly lower flow rates than the Virginia data
particularly when comparing the Q50 data.
To test the influence of these flow data on modeled EECs, a final analysis with the Index
Reservoir was conducted that consisted of modifying the GUI (PE4v01.pl) for running
PRZM/EXAMS. The modification consisted of altering the stream flow (STFLO) parameter in
PRZM responsible for reporting flow through the receiving water body by using the VADEQ
data as opposed to a runoff volume as described previously. Three alternate Index Reservoir
scenarios were modeled using the 7Q10 flow rate for the Eastern Shore, the 7Q10 flow rate for
the Coastal Plain, and the Q50 value for the Coastal Plain (no Q50 value was reported for the
Eastern Shore). This was intended to provide a bracket on possible flow rates and modeled
EECs within the regions most representative of the tributaries (streams, creeks and rivers) in the
immediate vicinity of the Chesapeake Bay (Eastern Shore and Coastal Plain). The results of this
analysis are presented in Table 3.12 and indicate that using 7Q10 values yield EECs comparable
to the standard water body modeling, while using the Q50 values yield long-term EEC
70
-------
appreciably below those predicted using the static water body. This is consistent with the
expectation that modeling is likely to be conservative relative to actual long-term average
concentrations in flowing water.
Table 3.12. Comparison of Alternative PRZM Modeling with EEC Generated Using a
Static Water Body
Flow
(ffVsec)
Peak
96
hour
EEC
(HS/L)
21 day
60 day
90 day
Yearly
Scenario
EEC
EEC
EEC
EEC
EEC
(Jig/L)
(Jig/L)
(Jig/L)
(Jig/L)
(Jig/L)
CB sorghum with
static water body1
0
55.4
55.6
54.9
53.7
52.9
42.9
CB sorghum with
flow thru standard
0.022
30.2
29.9
28.6
26.7
25.3
16.5
water body
CB sorghum with
VVWM with 3 meter
0.023
22.4
22.1
21.7
21.3
21.3
13.9
depth
CB sorghum with
VVWM with 10
0.020
14.2
14.2
14.2
13.9
13.9
10.9
meter depth
CB sorghum with
Index Reservoir2
0.380
49.9
49.1
46.6
41.1
36.8
17.9
CB sorghum (IR)
with 7Q10 flow from
0.138
53.0
52.5
50.8
47.1
44.6
30.2
VA Eastern Shore
CB sorghum (IR)
with 7Q10 flow from
1.730
44.2
41.6
35.3
23.1
17.3
4.6
VA Coastal Plain
CB sorghum (IR)
with Q50 flow from
105.3
32.7
7.3
1.7
0.6
0.4
0.1
VA Coastal Plain
Flow Data From
VADEQ Data3
Q50
7Q10
Eastern Shore
No Data
0.138
Coastal Plain
105.3
1.730
EEC generated using PE4v01 .pi in this table are slightly different from those presented in Table 3.5 due to different duration of
exposure and slight differences in the manual estimation technique used in Table 3.5.
2 Sorghum IR scenario EEC reported using percent cropped area (PCA) of 87%
3 VADEQ flow data reported as 7Q10 values
71
-------
3.3.3. Specific Characterization for Headwater Streams at Locations of the Dwarf Wedge-
mussel
One final piece of characterization of the PRZM/EXAMS modeling was performed relative to
the headwater streams of known locations of the dwarf wedgemussel. The dwarf wedgemussel
is unique with respect to habitat compared with the other assessed organisms in that it is
immobile and resides in headwater and low-order streams with low to moderate flow. Therefore,
the dwarf wedgemussel is located for most of its life cycle in types of waters estimated to have
the highest atrazine concentrations using PRZM/EXAMS. However, because the standard water
body likely overestimates long-term concentrations in flowing waters, potential effects of flow
on atrazine concentrations were evaluated for known locations of the dwarf wedgemussel to
allow for characterization of longer-term EECs in the types of waters listed for this mussel.
Flow rates of streams known to be inhabited by the dwarf wedgemussel are in Table 3.13. Data
were available for only three of the known locations of the dwarf wedgemussel. Therefore, the
stream flow information for the three locations with data were used as surrogates for all locations
for the purpose of estimating flow conditions where the dwarf wedgemussel is located.
Table 3.13. Estimated Flow Rate for Water Bodies Known to be Inhabited by the Dwarf
Wedgemussel.
Location of Dwarf Wedge
Estimated Flow Rate
Basis for Conclusion
Mussel
April to June (cfs)a
Aquia Creek
23
USGS data for Aquia Creek
Stafford county, VA
South Anna River
250
USGS data for South Anna River
Louisa county, VA
Po River
40
USGS data for Po River
Spotsylvania county, VA
a Flow rates represent median daily rate over the time frame of available data (30 to 70 years of data). Data obtained
from http://waterdata.usgs.gov/nwis/rt
As a test against the exposure estimates provided in Section 3.2, the daily information from the
USGS flow rates from Aquia Creek were compared. The list of water body's names and reaches
within which the mussel resides was obtained. Available flow data from the USGS for Aquia
Creek near Garrisonville, Virginia (station 01660400) and South Anna River near Ashland,
Virginia (station 01672500) were obtained, which are two of the water bodies where the mussel
is found within the Chesapeake Bay watershed.
Flow data was provided as a mean daily value for each day of the year based on measurements
recorded between 1972 and 2004 for Aquia Creek and between 1931 and 2004 for South Anna
River (http://waterdata.usgs.gov/nwis/rt). The data are reported for each day as percentiles at the
5th%, 1Q,h 0/oi 20th0/o? 25tho/o> 50th0/o? 75tiy0i 80"% 90th%, and 95,h% of all recorded data. These
data was compared with estimated flow rates from the modeling discussed above and estimated
an appropriate flow rate for additional modeling specific to the dwarf wedgemussel. The
average yearly and four quarter seasonal flow rates for January to March, April to June, July to
September, and October to December were calculated. In general, the highest flow rates were
72
-------
found to be the first quarterly rates (January to March), followed by the second quarterly rates
(April to June). The lowest rates were found during the summer months represented by the third
quarter. In general, the flow rates for the Aquia Creek site are between the 7Q10 and Q50 values
presented in Table 3.14 for the coastal plain, while the South Anna River flow rates are higher
than those previously modeled for coastal plain. This exercise is intended to provide specific
characterization as to the representative nature of the EECs modeled using the static water body
relative to the flowing water system where the mussels are known to be located.
Annual average rates at the 50th percentile of all years of flow for both data sets were selected for
additional flow modeling. It is believed that this is a reasonable value to represent an annual
average and is slightly more conservative (defined by lower flow) than the second quarterly
value that represents the window when most atrazine is expected to be applied as a pre-emergent
herbicide. Table 3.14 summarizes the data for Aquia Creek, and Table 3.15 summarizes the flow
data for South Anna River.
73
-------
Table 3.14. Aquia Creek Flow Information (ft3/sec)2
5th % 10th % 20th % 25th % 50th % 75th % 80th %
90th % 95th %
Annual average of
18.891
daily mean values
4.09
6.43
9.48
11.01
34.89
41.93
81.08
195.29
Average of first quarter
seasonal daily mean values
(January to March)
7.95
13.41
18.42
20.71
33.35
56.47
66.80
120.95
252.46
Average of second quarter
seasonal daily mean values
(April to June)
6.24
8.52
12.61
14.67
23.14
38.29
45.56
86.00
202.01
Average of third quarter
seasonal daily mean values
(July to September)
0.29
0.54
1.10
1.44
4.40
13.00
16.71
43.61
129.54
Average of fourth quarter
seasonal daily mean values
(October to December)
2.00
3.34
5.93
7.36
14.86
32.06
38.96
74.26
198.48
Highlighted column represents the flow data used in the refined modeling for characterization
2 Data source: http://waterdata.usgs.gov/nwis/rt
74
-------
Table 3.15. South Anna River Flow Information (ffVsec)2
5th % 10th % 20th % 25th % 50th % 75th % 80th % 90th % 95th %
Annual average of
daily mean values 72.09 90.51 123.40 137.91 215.741 372.66 442.44 791.56 1411.49
Average of first quarter
seasonal daily mean values
(January to March) 130.04 165.34 230.68 256.53 385.11 650.10 773.46 1311.54 1981.67
Average of second quarter
seasonal daily mean values
(April to June) 100.30 118.55 150.40 163.32 245.52 400.82 463.23 782.62 1324.74
Average of third quarter
seasonal daily mean values
(July to September) 20.54 26.73 38.41 45.01 81.47 153.63 186.91 394.70 1076.04
Average of fourth quarter
seasonal daily mean values
(October to December) 39.05 52.53 75.59 88.36 153.02 289.39 349.97 682.93 1274.96
Highlighted column represents the flow data used in the refined modeling for characterization
2Data source: http://waterdata.usgs.gov/nwis/rt
75
-------
Modeling with the median (50th percentile) of the annual average flow rates from both sites
yields considerably lower EECs, particularly longer-term EECs, than those predicted using the
static water body.
As evidenced by the summary in Table 3.14., the daily flow within Aquia creek varies
significantly from season to season. Previous analysis suggests that flow through the receiving
water body will have a dramatic impact on the longer-term averages of exposure. Analysis
shows that using the Index Reservoir as a receiving water body and modifying the flow rate to
represent different conditions drops the long-term averages below levels of concern (See Section
5). This analysis is intended to characterize the influence of flow on EECs. A closer look at the
daily flow data indicates that periods of much lower flow are possible. At the lower percentiles,
there are periods of time within Aquia Creek when flow rates are consistently below 2 cfs for
extended periods of time. For example, at the 5th percentile, daily flow rates drop below 2 cfs on
June 17 and do not increase above 2 cfs until November 22. By way of comparison, the daily
flow rates at the 50th percentile only drop below 2 cfs between September 3 and September 11.
This analysis indicates that it is possible that there can be long periods of low flow within Aquia
Creek (and other waters inhabited by the dwarf wedgemussel). However, this is believed to be
an unlikely occurrence. First, although it is expected that low flow conditions will generally
occur during the summer and early autumn in this area, it is unlikely that the duration will be as
long as suggested above. This is supported by the fact that the minimum flow year (the year
from the daily distribution with the lowest flow) varies from day to day, suggesting that
continuous low flow conditions in any year are unlikely. Second, even if the low flow conditions
described above did occur continuously in a given year, the return frequency at the 5th percentile
would be 1 in 20 years, suggesting a relatively unlikely occurrence. Third, low flow conditions
generally represent conditions when runoff is unlikely to occur, and analysis suggests runoff is
the dominant mechanism of atrazine exposure. Finally, the time period for low flow conditions
do not coincide with the main time of application of atrazine. The dominant use of atrazine is on
corn and sorghum as a pre-plant and pre-emergent application and is generally applied in the
early- to mid-spring.
In general, EECs predicted using the Index Reservoir and the flow rate from the Aquia Creek
data (18.9 ft3/s) are roughly two times lower for peak exposures and one to two orders of
magnitude lower (depending on duration) for the longer term averages when compared to the
EECs generated using the static water body. Using the flow rate (215.7 ft3/s) for South Anna
River yields an EEC roughly two times lower for the peak and two to three orders of magnitude
lower (depending on duration) than the longer-term averages than the static water body. The
results of this alternate modeling compared to the original EECs predicted using the static water
body are presented in Table 3.16.
76
-------
Table 3.16. Summary of Alternative PRZM Modeling Using the Index Reservoir and Site
Specific Flow data from Aquia Creek and South Anna River Compared with PRZM EEC
Modeling Using a Static Water Body
Scenario
Flow
(ffVsec)
Peak
96
hour
21 day 60 day 90 day Yearly
CB sorghum with
static water body1
0
55.4
55.6
54.9 53.7 52.9 42.9
CB sorghum IR with
Aquia Creek Flow
Data
CB sorghum IR with
South Anna River
Flow Data
18.9
215.7
33.1
32.6
21.6
6.1
7.5 2.7 1.8 0.5
1.3 0.45 0.30 0.07
EEC generated using PE4v01 .pi in this table are slightly different from those presented in Table 3.5. due to different duration of
exposure and slight differences in the manual estimation technique used in Table 3.5.
This analysis suggests that, in streams with flowing water, the predicted EECs using the static
water body are over-estimating exposure for longer duration periods. The modeled values using
flow rates from two of the streams that this species inhabit suggests that the peak exposure EEC
is roughly two times less than that for the static water body and that the longer term exposures
are several orders of magnitude below the static water body EECs. Also, a single sample was
analyzed for atrazine in Aquia Creek once (station id = 01660350) on August 23, 1994 with a
detection of 0.006 |ig/L (Figure 3-2) as part of the Chesapeake Bay monitoring program
(discussed in Section 3.4.). The results of the alternative modeling coupled with the fact that the
only atrazine detection within any of the water bodies where the dwarf wedgemussel resides is
well below both the peak and longer-term EECs predicted both from the static water body and
the alternative flowing water body suggests that the EECs presented above are more
representative of the types of exposures to which the dwarf wedgemussel is likely to encounter.
77
-------
Location of Aquia Creek NAWQASite
Relative to Chesapeake Bay and it's Principal Tributaries
Legend
• CBP Atrazine Detections
O NAWQ SW Sites
¦ Miles
0 5 10
20
figure 3-3. Location of NAWQA Surface Water Sites on Aquia Creek Relative to
Chesapeake Bay Program (CBP) Surface Water Sites.
(Source: http://water.usgs.gov/nawqa and http://www.chesapeakebay.net/data/index.cfm, respectively)
78
-------
3.3.4. Summary of Alternative Modeling Exercises
The modeling data suggest that peak EECs estimated using the standard water body
remained relatively consistent across the modeled scenarios, and that peak atrazine
concentrations calculated using the standard water body are high-end approximations.
However, alternative modeling exercises, which incorporate flow rates representative of
locations of the assessed species, suggests that longer-term atrazine EECs (days to
weeks) in flowing water bodies are likely overestimated by the standard water body, and
the inclusion of flow on estimated atrazine concentrations increases with increasing
duration of exposure.
3.4. Monitoring Data
Unlike many pesticides, atrazine has a fairly robust data set of surface water monitoring
from a variety of sources. Included in this assessment are atrazine data from the USGS
National Water Quality Assessment Program (NAWQA) (http://water.usgs.gov/nawqa/),
Chesapeake Bay Program (http://www.chesapeakebay.net/data/index.cfm), and
Heidelberg College (http://wql-data.heidelberg.edu/). In each case, the data was
characterized in terms of general statistics (number of samples, frequency of detection,
maximum concentration, and mean from all detections). In addition, several sample sites
were evaluated from each data set for more detailed analysis including calculation of
annual maximum and annual time weighted mean concentrations by site by year. For all
data described below, each site characterized by maximum and annual average (or time
weighted mean) concentration represent sites with multiple samples all of which were
evaluated and include in selecting maximum and annual average concentrations. The
sample sites chosen for this additional analysis were selected by choosing those locations
from the national and local data with the highest detected concentrations of atrazine.
Finally, an interpolation of single year's worth of data from one sample site in the
Heidelberg College data to estimate 14-day, 30-day, 60-day, and 90-day averages was
conducted.
NAWQA groundwater data was evaluated to determine the importance of groundwater
on potential loadings to the Chesapeake Bay. Groundwater data from Maryland,
Pennsylvania, and Virginia was downloaded from the USGS NAWQA data warehouse
(http://water.usgs.gov/nawqa/) on May 11, 2006. The three states were selected because
of proximity to the Chesapeake Bay. Delaware was excluded from this analysis because
most of the state lies outside the boundaries of the watershed.
A total of 725 well samples were analyzed for atrazine in groundwater between 1993 and
2004. Of these samples, a total of 373 had positive detections of atrazine with 25 of
those estimated at below the limit of quantitation (LOQ). The frequency of detection for
all detections was 51%. The maximum concentration detected was an estimated value of
4.2 |ig/L (above the LOQ) in an urban setting in Virginia Beach, Virginia while the
highest non-estimated value was 2.2 |ig/L in an urban setting in Lebanon, Pennsylvania.
Of all detections, only 8 samples had detections greater than 1.0 |ig/L. Overall, the data
79
-------
suggest that atrazine recharge to the waters of the Chesapeake Bay watershed are possible
but that the detection frequency, travel times, and magnitude of exposures are such that
they are likely to be dwarfed by the surface runoff route.
3.4.1. National USGS NAWQA Data
An analysis of the entire USGS NAWQA data set for atrazine was conducted. A data
download from the USGS data warehouse (http://water.usgs.gov/nawqa) on May 11,
2006 provided the data set used in this analysis. Overall, a total of 20,812 samples were
analyzed for atrazine and of these, 16,742 had positive detections (including
concentrations estimated above or below the limit of quantitation) yielding a frequency of
detection of roughly 80%. The maximum detection from all samples was 201 |ig/L from
the Bogue Chitto Creek in Alabama in 1999. Overall, the average concentration detected
was 0.26 |ig/L when considering only detections and 0.21 |ig/L when considering all
detections and non-detections (using the detection limit as the value for estimation).
Using the top ten sites with the highest atrazine concentration more refined analysis of
the detections was conducted. All values from the national data set were ranked and the
top ten sites were selected based on maximum concentration. Each location was
analyzed separately by year and the annual maximum and annual time weighted mean
concentrations were calculated. The minimum criterion for calculating time weighted
means for each sampling station was at least 4 samples in a single year. The equation
used for calculating the time weighted annual mean is as follows:
[(( To+i-To) + ((T0+2-T0+i )/2))*C to+i)] + (((Ti+i-Tj-i )/2)*C0 + [((Tend-Tend.i) + ((Tend.i-
Te„d-2)/2)*CTend-l)]/365
where: Ci = Concentration of pesticide at sampling time (Ti)
Ti = Julian time of sample with concentration Ci
T0 = Julian time at start of year = 0
Tend = Julian time at end of year = 365
Generally, the maximum values from this analysis are similar to, or above (by as much as
two to three times) the model predictions from PRZM/EXAMS from the Chesapeake Bay
watershed, while the annual time weighted mean (TWM) concentrations are roughly an
order of magnitude below the static water body model predictions for annual average and
are roughly two to three times below the flow influenced model predictions described
above.
The modeling and national NAWQA monitoring data are not directly comparable
because the monitoring data are generally from high atrazine use areas in the Midwest
and South vulnerable to runoff while the modeling was conducted exclusively for the
action area of the Chesapeake Bay. In the Chesapeake Bay watershed the atrazine use
intensity is similar to the areas in the Midwest and South but the runoff vulnerability is
lower as identified by Williams et al (2004). The Ecological Exposure in Flowing Water
Bodies (Williams, et al, 2004) utilized the WARP model to identify highly vulnerable
80
-------
watersheds for sampling and determined that the top 20% watersheds were
predominantly in the Midwest and South while the watersheds in the immediate vicinity
of the Chesapeake Bay are between the 40% and 50%.
Given the fact that the watersheds surrounding the Chesapeake Bay are less vulnerable to
atrazine runoff a comparison with monitoring data from more vulnerable areas was
conducted to provide context to the modeled exposures. Modeled concentrations that
exceed monitoring data in highly runoff vulnerable atrazine use areas would suggest that
the modeling is either overly conservative or the monitoring is not representative.
Conversely, modeled concentrations that are less than the monitoring data from the
highly runoff vulnerable atrazine use area suggest that modeling is not conservative. In
the case of atrazine, the modeling tends to under predict the highest single day
concentrations and over predicts the annual average concentration from the national
NAWQA data. This is not unexpected given that the majority of the high atrazine
detections are from the 1990s. Also, because runoff vulnerability is much lower in the
area surrounding the Chesapeake Bay. The analysis suggests that modeling in the action
area for atrazine is providing a reasonable estimate of short term exposure but is over-
estimating longer term exposure.
No comparison has been made between these data and model predictions for the
intermediate durations exposures (14-day, 30-day, etc.) because the NAWQA data
generally do not have the frequency needed to conduct a meaningful interpolation
between data points. Table 3.17 presents a summary of the annual time weighted mean
concentrations, and Table 3.18 presents a summary of the annual maximum
concentrations.
81
-------
Table 3.17. Annualized Time Weighted Mean (TWM) Concentration (jig/L) for the Top Ten NAWQA Surface Water Sites
Bogue
Chitto
Creek,
Year near
Memphis,
TN
(02444490)
1991
1992
0.98
1.32
1993
0.77
3.80
1.43
1994
0.87
2.56
1995
2.28
0.74
1996
1.30
4.32
2.18
1997
5.36
3.45
5.55
1.03
2.82
1998
0.82
1.79
2.94
1.21
1.88
1999
9.62
0.28
2.50
0.68
2000
6.49
0.56
1.26
0.15
2001
1.20
0.83
0.78
0.22
1.28
2002
2.88
0.51
2.22
1.26
0.80
2003
2.14
4.46
0.70
7.83
2.23
1.42
2004
1.77
68.781
0.67
1.24
3.31
1.93
1 TWM concentration likely biased due to fact that first sample on May 8 is the peak sample from this year, and interpolation method likely resulted in an inflated time
weighted mean value.
Ranked by Maximum Concentration Detected
Station Name (ID)
Tributary
to S Fork
Dry
Creek,
near
Schuyler,
NE
(06799750)
Sugar Creek, New
Palestine, IN
(394340085524601)
Kessinger
Ditch,
near
Monroe
City, IN
(03360895)
LaMoine
River @
Colmar,
IL
(05584500)
Sugar
Creek @
Milford,
IL
(05525500)
Tensas
River @
Tendal,
LA
(07369500)
Maple
Creek
near
Nickerson,
NE
(06800000)
Auglaize
River near
Ft
Jennings,
OH
(04186500)
-------
Table 3.18. Maximum Concentration (jig/L) for the Top Ten NAWQA Surface Water Sites Ranked by Maximum
Concentration Detected
Year
Station Name (ID)
Bogue
Chitto
Creek,
near
Memphis,
TN
(02444490)
Tributary
to S Fork
Dry Creek,
near
Schuyler,
NE
(06799750)
Sugar Creek, New
Palestine, IN
(394340085524601)
Kessinger
Ditch,
near
Monroe
City, IN
(03360895)
LaMoine
River @
Colmar,
IL
(05584500)
Sugar
Creek @
Milford,
IL
(05525500)
Tensas
River @
Tendal,
LA
(07369500)
Maple
Creek
near
Nickerson,
NE
(06800000)
Auglaize
River near
Ft
Jennings,
OH
(04186500)
1991
1992
14
25
1993
8.5
120
11.2
1994
11
24
1995
27
2.6
1996
14.2
30
18
1997
129
108
92.3
10.3
85.2
1998
7.88
27.7
19.3
30
9.96
1999
201
2.39
13.9
10.7
2000
136
3.84
23
0.87
2001
4.5
14.4
6.96
1.21
10.4
2002
24.8
4.01
21.3
16.4
2.58
2003
18.8
21.3
10.5
108
34.8
13.4
2004
14.6
191
28.3
10.9
91.9
18.7
-------
3.4.2. USGS Watershed Regression of Pesticides (WARP) Data
The NAWQA data were then compared against the percentiles used to develop the USGS WARP
(provided by Charlie Crawford of USGS on June 2, 2006 via email). Comparison against
WARP percentiles was conducted because the WARP model has been to be a valuable tool for
site selection and assessing overall vulnerability. More information on the WARP model may be
found at:
http://pubs.usgs.gov/wri/wri034047/wrir034Q47.pdf
The WARP data were developed using a subset of the national data described above (all WARP
data are included in the national data analysis described above). The USGS National Stream
Water Quality Accounting Network (NASQAN) data was also included in the WARP dataset;
however, it is not included in this assessment as it represents major rivers. Data collected
between 1992 and 1999 from a total of 113 sample sites were used to create the model. Sample
sites were selected based on the robustness of the data available at a given site. The model yields
predicted daily exposures at various percentiles of occurrence. National NAWQA data and the
model predictions against the mean and 95th percentile values from the data used were compared.
The maximum 95th percentile value from the WARP data was 20.2 |ig/L compared to a
maximum of 201 |ig/L from all data. The maximum mean value used in the WARP model
development data was 3.82 |ig/L which is consistent with the annual TWM values discussed
above.
3.4.3. Regional USGS NAWQA Data
The PRZM/EXAMS EECs were compared to data from surface water sites specific to the
Chesapeake Bay watershed (defined by the Lower Susquehanna River Study Unit and the
Potomac River Study Unit). The same technique as applied to the national data (maximum and
TWM) was applied to these two study units to provide a more regionally specific snapshot of the
available NAWQA data. Generally, these data are well below the national data for maximum
exposures with a peak concentration of 25 |ig/L compared to 201 |ig/L nationwide, while the
average concentration from all data are comparable with an average for all detections of 0.28
|ig/L and an average for all data (detects and non-detects) of 0.27 |ig/L. The results of the
refined analysis indicate that, as expected, the overall exposures in the Chesapeake Bay
watershed, typified by the peak and annual TWM are generally less than those seen in the
national and WARP data. A summary of the results for the Potomac River Study Unit is
presented in Table 3.19, and the Lower Susquehanna River Study Unit results are provided in
Table 3.20.
-------
Table 3.19. Annual Time Weighted Mean and Maximum Concentration from the Top
Three USGS NAWQA Surface Water Sites Located in the Potomac Study Unit
Station Name (ID)
MUDDY CREEK AT MONOCACY RIVER AT MORGAN CREEK NEAR
MOUNT CLINTON, VA BRIDGEPORT, MD KENNEDYVILLE, MD
(01621050) (01639000) (01493500)
Year TWM Max TWM Max TWM Max
1993
0.31
18.60
1994
0.13
0.16
0.38 6.90
1995
0.11
0.14
0.97 8.00
1996
0.24
1.66
1.97 4.30
1997
0.26
2.14
1998
0.18
1.96
1999
1.59
25.00
2000
0.22
1.55
2001
0.28
2.87
2002
0.44
2.73
0.36 4.08
2003
0.16
0.74
0.46 6.53
2004
0.89 7.95
85
-------
Table 3.20. Annual Time Weighted Mean and Maximum Concentration from the Top
Three USGS NAWQA Surface Water Sites Located in the Lower Susquehanna River
Study Unit
Station Name (ID)
EASTMAHANTANGO SUSQUEHANNA MILL CREEK AT ESHELMAN
KLINGERSTOWN, PA HARRISBURG, PA MILL PA
(01555400) (01570500) (01576540)
Year TWM Max TWM Max TWM Max
1993
0.16
0.90
0.13 0.78
1994
0.31
3.20
0.21 1.50
1995
0.04 0.81
1996
1997
0.12
0.39
1998
0.33
3.37
1999
0.14
0.58
2000
0.37
3.33
3.4.4. Chesapeake Bay Program Data
A similar analysis of the limited monitoring data obtained on May 3, 2006 from the Chesapeake
Bay Program (CBP) office website was also conducted. The data may be found at the following
website.
http://www.chesapeakebav.net/data/index.htm
The data consists of 686 samples analyzed for atrazine between the years of 1978 (one sample)
through 1999 with the bulk of the samples collected in the 1990s. A total of 74 stations were
present within the data set and all data were analyzed as part of this assessment. An analysis of
the distribution of the site locations relative to the characterization of the watershed (described in
Section 2.4) was completed and is summarized in Table 3.21. The analysis confirms that the
bulk of the monitoring data, including the highest detections described below, were collected
from streams and rivers interior of the bay proper. The analysis also confirms that the general
trend is decreasing concentrations as water moves from headwater streams adjacent to treated
fields to the open waters of the bay where concentrations were well below 1 ug/1.
86
-------
Table 3.21 Summary of Chesapeake Bay Program (CBP) Sample Sites Relative to Watershed
Characterization Described in Section 2.4
Number Number Average Maximum
Watershed of of Concentration Concentration
(PPb) (ppb)
Region
CBP Sample Locations
Samples
Detections
Open Bay
L0002598
L0002599,
L0002600
L0002601,
4
4
L0002602
Bay Inlet
L0001829
L0002632,
20
20
L0002638
Estuarine
L0001822
L0001824,
Mouth of
L0001825
L0001826,
Rivers
L0001827
L0002624
L0002630
L0001828,
L0002229,
L0002631
72
71
Main Stem
L0001084
L0001085,
River
L0001817
L0001818,
L0001819
L0001820,
367
364
L0001821
L0001823,
L0002598
L0002625,
L0002626
L0002627
River
L0001083
L0002463,
Tributary
L0002470
L0002475
L0002473,
L0002476,
L0002479
L0002478,
105
70
L0002481
L0002510,
L0002628
L0002633,
L0002634
L0002635,
L0002636
Headwater
L0002450
L0002451,
Streams
L0002452
L0002454
L0002453,
L0002455,
L00002456, L00002457,
L0002458
L0002459,
L0002460
L0002461,
L0002462
L0002463,
L0002464
L0002465,
98
46
L0002466
L0002467,
L0002468
L0002469,
L0002471
L0002474,
L0002477
L0002478,
L0002487
L0002488,
L0002489
L0002511,
L0002637
0.04
0.18
0.04
0.05
0.43
0.09
0.10
3.06
0.13
1.00
0.74
30.00
87
-------
The peak concentration detected in any sample was 30 |ig/L (station L0002488), which appears
to be located in a tributary on the Eastern Shore. In fact, the general trend in these data are that
the main detections of atrazine are in the tributaries while significantly lower concentrations
have been found in the Bay itself. This pattern suggests that an emphasis on predicting
exposures in the tributaries is the most conservative approach for assessing both direct and
indirect effects to the named species. Five of the top sample locations were selected based on the
highest detected concentration of atrazine, to calculate annual TWM and maximum
concentration. Of these five sites, only two were deemed to have a sufficient number of samples
(minimum needed is four annually) to analyze for TWM. The data, which are summarized
below, indicate that for the location with the peak concentration (30 |ig/L), the annual TWM
concentration is similar to other monitoring data analyzed previously. The results are
summarized in Table 3.22.
Analysis of the data indicate that the maximum atrazine level found was 30 |ig/L, while the 99th,
95th, 90th, 75th, and 50th percentile values were 2.5 |ig/L, 0.5 |ig/L, 0.28 |ig/L, 0.1 |ig/L, and 0.05
|ig/L respectively. A summary of the total data are presented in Figure 3-4 while the general
locations of the sampling stations are presented in Figure 3-5.
Table 3.22. Annual Time Weighted Mean and Annual Maximum Concentrations
from Selected Sample Locations from the Surface Water Monitoring Data from
the Chesapeake Bay Program
Station ID
L0001818
L0002488
Year
TWM Max
TWM Max
1991
2.03 30.00
1992
0.01 0.06
1993
1994
1995
1.93 3.06
1996
0.25 1.29
Overall, there is only limited monitoring data from the Chesapeake Bay itself. The data from the
Chesapeake Bay Program indicate that of the 686 samples analyzed for atrazine, only a handful
are actually from the Bay, with most of the samples collected from tributaries and rivers feeding
into the Bay. Of the samples in the main stem of the Bay, most of the detections for atrazine are
well below 1 |ig/L. The location of the maximum atrazine detection of 30 |ig/L is located in a
headwater stream near the edge of the Chesapeake Bay watershed (Figure 3-6).
88
-------
Atrazine Concentration in Chesapeake Bay and its Tributaries
35
30
25
'20
S 15
o
o
10
oo
00
oo
00
o
oo
oo
00
oo
00
oo
00
CO
oo
oo
00
00
oo
00
LO
00
oo
00
CD
00
oo
00
oo
00
00
00
oo
00
CD
oo
oo
00
o
CD
oo
00
oo
00
oo
00
CO
CD
oo
00
G>
oo
00
LO
CD
oo
00
CD
CD
oo
00
oo
00
00
CD
oo
00
Date
Atrazine Concentration (ppb)
Figure 3-4. Range of Atrazine Concentrations Detected in the Chesapeake Bay and its
Immediate Tributaries.
89
-------
Location of Chesapeake Bay Program (CBP)
Sample Locations
(distinguishing detections and non-detections of atrazine)
~ [-•
/hw.
' • Y. t ¦ "*•
"iAi
y ^« x
- I. v. ^ 7
fo
0" >
I
4
*v# _ «4 v 9
..V •• 1 "«
~~ ~ i- .' *
153012
Legend
• CBP_Atrazine_Monitoring_Data
All CBP Water Quality Stations
Miles
3-5. Location of Surface Water Monitoring Sites in the Chesapeake Bay and Its
Immediate Tributaries.
Figure
90
-------
30000
Location of Maximum Atrazine Detection (ng/l)
in the Chesapeake Bay Program (CBP) Monitoring Data
Legend
• Atrazine Detections (ng/l)
Mid Atlantic Streams
O Site of Maximum Atrazine Detection (ng/l)
» All CBP Water Quality Stations
~] Chesapeake Bay Watershed Boundary
Figure 3-6. Location of Maximum Atrazine Detection (ng/l) in the Chesapeake Bay
Watershed in Data from the Chesapeake Bay Program.
91
-------
3.4.5. Heidelberg College Data
Data from Heidelberg College, consisting of two intensively sampled watersheds (Maumee and
Sandusky) in Ohio were also analyzed. Like the national NAWQA data, these data are outside
of the action area but is included in this analysis to provide context to the modeled exposures.
More information on the water quality monitoring program at Heidelberg College may be found
at the following website:
http://wql-data.heidelberg.edu/
The Heidelberg data are collected more frequently than other data included in this assessment.
The study design was specifically established to capture peak and longer term trends in pesticide
exposures. Data were collected between 1983 and 1999 and consist of an average of roughly
100 samples per year with several days of multiple sampling.
For the Sandusky watershed, a total of 1,597 samples were collected with 1,444 detections of
atrazine (90.4% frequency of detection). The maximum concentration detected in the Sandusky
watershed was 52.2 |ig/L, and the overall average concentration was 4.5 |ig/L. For the Maumee
watershed, a total of 1,437 samples were collected with 1,305 detections of atrazine (90.8%
frequency of detection). The maximum concentration detected in the Maumee watershed was
38.7 |ig/L with an overall average concentration of 3.7 |ig/L.
This analysis was furthered refined by deriving the annual TWM and maximum concentrations
by sampled watershed by year. The results of this analysis are presented in Table 3.23. The
results show a consistent pattern with that seen in other data collected from high atrazine use
areas with general TWM concentrations between 1 and 3 |ig/L.
Table 3.23. Annual Time Weighted Mean and Maximum Concentrations (jig/L) for
Atrazine in Two Ohio Watersheds from the Heidelberg College Data
Sandusky
Maumee
Year
TWM
Max
TWM
Max
1983
1.34
7.97
0.98
5.42
1984
1.08
8.73
1.27
11.71
1985
1.83
19.46
1.00
6.21
1986
3.32
24.61
1.64
10.01
1987
1.76
16.45
1.80
9.92
92
-------
Table 3.23. Annual Time Weighted Mean and Maximum Concentrations (jig/L) for
Atrazine in Two Ohio Watersheds from the Heidelberg College Data
Sandusky Maumee
Year TWM Max TWM Max
1988
0.41
1.53
0.43
2.15
1989
1.30
15.71
1.07
8.49
1990
1.96
19.31
1.69
14.78
1991
1.49
20.59
2.044
21.45
1992
0.39
40.53
0.51
7.35
1993
1.27
26.34
1.21
22.66
1994
0.86
10.10
0.82
4.02
1995
1.39
15.46
1.30
14.06
1996
1.56
23.40
1.19
16.19
19971
2.16
53.21
2.09
38.74
1998
1.49
40.03
1.41
27.62
1999
1.57
17.11
1.88
19.37
1 Sample year 1997 from Sandusky selected for data infilling by interpolation in order to calculate CASM duration exposure
values
Unlike other data sets included in this assessment an effort at interpolation between data points
was completed in order to estimate 14-day, 30-day, 60-day, and 90-day average concentrations.
A final analysis of the data was completed by selecting one year worth of data from the
Heidelberg data. 1997 was selected because it was one of the more recent data sets and because
the maximum and TWM concentrations were higher than most other year's data. To process
these data it was necessary to "fill in the gaps". A total of 126 samples were collected during
1997 with 50 days with multiple samples yielding a time series of roughly 75 days. A step-wise
approach was used to estimate daily concentrations between sampling dates that consisted of
simply extending an analytical result from the date of analysis to the next date. For example, on
January 6, 1997, atrazine was detected at a concentration of 0.475 |ig/L with the next sample
date on January 20, 1997 with a concentration of 0 |ig/L. In the step-wise interpolation all dates
between January 6 and January 20 were assigned the concentration of 0.475 |ig/L. Also, because
93
-------
January 6 was the first sample date of the year, all previous days were also assigned a value of
0.475 |ig/L. This process was repeated throughout the year to fill in the time series and yield 365
days worth of data. In addition, where multiple samples were analyzed on any given day, the
highest of the values on that day was assigned. There is significant uncertainty with this type of
interpolation because there is no information to suggest whether the interpolated value represents
actual exposure. For example, where a significant gap in time exists between two samples, it is
unlikely that a continuous concentration exists. More likely is that there are upward and
downward fluctuations in exposure, with a greater likelihood that higher exposures are missed
between sample times with larger gaps in data points.
Table 3.24 presents the results of this analysis. The analysis suggests that at least for the
Sandusky watershed in 1997 the estimated longer-term exposures are less than the modeled
estimates for the sorghum scenario by a factor of two to three times.
Table 3.24. Magnitude and Duration Estimates from the 1997 Data from
Sandusky Watershed1
14 day
21 day
30 day
60 day
90 day
Maximum
28.26
21.11
18.30
12.38
8.89
90th
Percentile
7.55
7.08
7.82
10.23
8.22
Stepwise interpolation was used between samples
3.4.6. U.S. EPA ORD Great Lakes Program - Lake Michigan Mass Balance Project
The U.S. EPA's Office of Research and Development (ORD) and the Great Lakes National
Program Office (GLNPO), working with state and local partners, has been collecting and
analyzing atrazine data for Lake Michigan in the Lake Michigan Mass Balance (LMMB) study.
In addition, ORD has developed and implemented a model to predict future trends in atrazine
occurrence in the Lake. The LMMB project and data are not directly comparable to the action
area because of its location in the upper Midwest. However, it is included in this assessment
because of the fact that it represents a large water body comparable in size to the Chesapeake
Bay
The LMMB study includes analytical results for atrazine occurrence in the atmosphere (vapor
phase, dry deposition, and wet deposition), surface water tributaries to the Lake, and within Lake
Michigan itself. The LMMB modeling framework includes computational transport, mass
balance, and bioaccumulation and has 3 levels of spatial resolution (whole Lake, 10 surface
segments and 41 water segments, and a high- resolution model consisting of 2,318 surface
segments and 44,042 water segments).
In the Lake Michigan basin, atrazine is usually applied to cornfields in the spring to control
broadleaf and some grassy weeds, and approximately 850,000 kg is applied annually in the Lake
Michigan basin. In the atmospheric component, study results indicate that the predominant
94
-------
atmospheric source of atrazine is precipitation. Atrazine was only detected in 3.7% of vapor
phase samples, and while the detection frequency was higher for particulate samples (dry
deposition), the mean concentrations during spring (peak atrazine use season) were found at
concentrations up to 370 pg/m3. In precipitation, atrazine was detected at concentrations as high
as 2,800 ng/1 (2.8 |ig/L).
In the tributaries feeding Lake Michigan, atrazine was detected in 99% of samples, and
concentrations ranging from 0.064 to 2.7 |ig/L were strongly influenced by geography with
higher concentrations in the south near atrazine use sites and lower in the north. In the open
water portion of the study, the Lake monitoring data showed fluctuating values of atrazine
between 0.03 to 0.06 |ig/L. Seasonal loadings (spring to early summer) tend to be focused on the
southeast and northwest shores. Modeling predicts that, with no increase or decrease in loadings,
concentrations in the Lake will increase slightly and level off thereafter.
In general, the monitoring data and modeling from the LMMB study found that overall loadings
are expected to be similar to those seen in the CBP data in the open Bay with concentrations
generally in the sub-|ig/L range in Lake Michigan and Chesapeake Bay. Similarly, significantly
higher concentrations of atrazine were found in both settings in the tributaries feeding both Lake
Michigan and the Chesapeake Bay. More recent work has been done to develop a model to
predict long-term exposures to atrazine throughout the entire Lake and indicate that long term
atrazine concentrations tend to be seasonal and higher near shore than in the central portions of
the lake.
More details on the LMMB study and atrazine can be found at
http://www.epa.gove/glnpo/lmmb/results/atra datarpt.html
3.4.7. Summary of Open Literature Sources of Monitoring Data for Atrazine
Atrazine is likely to be persistent in ground water and in surface waters with relatively long
hydrologic residence times (such as in some reservoirs) where advective transport (flow) is
limited. The reasons for atrazine's persistence are its resistance to abiotic hydrolysis and direct
aqueous photolysis, its only moderate susceptibility to biodegradation, and its limited
volatilization potential as indicated by a relatively low Henry's Law constant. Atrazine has been
observed to remain at elevated concentrations longer in some reservoirs than in flowing surface
water or in other reservoirs with presumably much shorter hydrologic residence times in which
advective transport (flow) greatly limits its persistence.
A number of open literature studies have been cited in the 2003 IRED (U.S. EPA, 2003a) which
document the occurrence of atrazine and its degradates in both surface water and groundwater.
These data support the general conclusion of the analysis above that higher exposures tend to
occur in the most vulnerable areas in the Midwest and South and that the most vulnerable water
bodies tend to be headwater streams and water bodies with little or no flow.
The analysis in the IRED also documents the occurrence of atrazine in the atmosphere. The data
indicate that atrazine can enter the atmosphere via volatilization and spray drift. The data also
95
-------
suggest that atrazine is frequently found in rain samples and tends to be seasonal probably
related to application timing. Finally, the data suggest that although frequently detected, the
concentrations being detected are less than those seen in the monitoring data and modeling
conducted as part of this assessment and support the contention that runoff and spray drift are the
principal routes of exposure. In general, these detections are located in areas of high atrazine use
such as the Midwestern US It is expected that lower amounts will be present in the action area
due to lower relative use. More details on these data can be found in the 2003 IRED (U.S. EPA,
2003a).
3.5 Summary of Modeling vs. Monitoring Data
Overall, comparison of the monitoring data with the modeling indicates that, in general, the peak
concentrations are reasonably well predicted by modeling with PRZM/EXAMS for all scenarios
and iterations of the modeling but that the longer-term average concentrations are over-estimated
for flowing water bodies. For this analysis, only the peak and annual average (approximated by
averaging across the sample range from the monitoring data) from the monitoring data were
comparable to the model output, with the exception of the analysis from the Heidelberg data.
The Heidelberg analysis, although highly uncertain due to the nature of the interpolation
necessary, suggests that in a highly vulnerable watershed, the longer-term exposures will be less
than predicted in streams and rivers with even moderate flow rates.
A number of uncertainties should be considered when comparing the modeled EECs from the
static water body with various habitat types and monitoring data. Specifically, the modeled
water body represents static water; however, in reality, many water bodies have some amount of
flow. For the Chesapeake Bay watershed, it is expected that no-flow, and low-flow water bodies
are representative of the headwater streams adjacent to agricultural fields. In addition, water
bodies in the Chesapeake Bay watershed increase in flow rate, volume, salinity, and the
influence of tidal fluxes and increasing watershed size will result in some dilution due to the
influx of non-impacted water. None of these factors are accounted for in the modeled estimates
presented in Table 3.5 used for risk estimation. In general, it is expected that modeled atrazine
concentrations in the static water body will over-estimate exposure in settings where flow is
greater than those modeled and where the volume of the water body is greater than that modeled
(20,000,000 liters). It is uncertain what impact differences in water chemistry and tidal
influences would have on modeled exposures.
Overall, the uncertainties inherent in the exposure assessment tend to result in over estimation of
exposures. This is apparent when comparing modeling results with monitoring data. In general,
the monitoring data should be considered a lower bound on exposure while modeling represents
an upper bound. Factors influencing the over-estimation of exposure include the assumption of
no degradation, dilution, or mixing in the subsurface transport from edge of field to springs. The
modeling exercises presented is in actuality assuming the assessed water bodies and application
sites are adjacent. In reality, there are likely to be processes at work which cannot be accounted
for in the modeling which will reduce the predicted exposures. In addition, the impact of
setbacks on runoff estimates have not been quantified while acknowledging that these buffers,
especially well-vegetated buffers, are likely to result in considerable reduction in runoff loading
of atrazine.
96
-------
3.6. Oral Exposure to Sea Turtles
3.6.1. Dietary Exposure from Contaminated Food Items
Dietary exposure to the four species of sea turtles considered in this assessment was estimated
using the highest reported bioconcentration factor of atrazine of 8.5 L/kg (U.S. EPA, 2003c) and
the peak EEC from PRZM/EXAMS reported in Section 3.2. Atrazine concentration in sea turtle
food items was estimated using the following calculation:
BCF of 8.5 L/kg x EEC of 0.055 mg/L = 0.47 mg/kg = 0.47 ppm
The dietary concentration of 0.47 ppm was compared with the avian LC50 (ppm) and NOAEC
(ppm) for derivation of dietary based risk quotients.
Daily doses (mg/kg-bw) of atrazine were estimated for sea turtles by assuming that a turtle
consumes approximately 100% of its weight daily (see below for explanation of this assumption)
using the following equation:
Dietary concentration (0.47 mg/kg = 0.47 ppm) / 100% bw consumed = 0.47 mg/kg-bw
The assumption that sea turtles consume 100% of their body weight daily was based on a report
by Lutcavage and Lutz (1986), who reported that hatchling leatherbacks consume their weight in
food daily. Duron (1978) estimated that adult leatherbacks would need to consume
approximately 200 lbs jellyfish daily to satisfy their energy requirements, resulting in
consumption of considerably less than 100% of body weight daily, given a small adult
leatherback turtle of 260 kg (approximately 570 lbs). Therefore, the assumption that turtles
consume 100% of their body weight daily would result in a conservative estimation of exposure.
Food consumption data were not located for other sea turtle species; therefore, it was assumed
that other sea turtle species also consume no more than 100% of their body weight daily.
3.6.2. Potential Exposure to Sea Turtles from Water Intake
Exposure from water flow-through was estimated based on water turnover rates reported by
Wallace et al (2005) in leatherback turtles and Ortiz et al. (2000) in Kemp's ridley turtles using
the following equation:
Water turnover rate (mL / kg-bw) x (1 L / 1000 mL) x EEC (|ig/L) = Dose (|ig/kg-bw)
Water influx data were not located for green turtles and loggerhead turtles. Therefore, there is
additional uncertainty in the EECs for these turtle species. Results from this analysis are in
Table 3.24.
97
-------
Table 3.24. Water Intake Exposure Estimations for Sea Turtles
Species
Water Influx
Atrazine EEC
Atrazine Dose in
(mL/kg-bw)c
(Ug/L)d
Turtle (fig/kg-bw)
Leatherback3
233
55
13 (0.013 mg/kg-bw)
Kemp's Ridleyb
123
55
6.8 (0.0068 mg/kg-bw
a Maximum of 5 values, which ranged from 106 to 233
b Average of 4 values; range not reported
c Water influx data reported from Wallace et al (2005) and Ortiz el al. (2000)
Atrazine EECs are from Section 3.2 and were derived using PRZM/EXAMS.
d Peak EEC from PRZM/EXAMS using the standard water body scenario
These data will be compared with acute avian LD50 values (mg/kg-bw) for direct effects risk
estimation in turtles.
4.0 Effects Assessment
This ecological risk assessment evaluates the potential for atrazine to affect six species:
shortnose sturgeon, dwarf wedgemussel, and four sea turtle species. Assessment endpoints
include direct toxic effects on the survival, reproduction, and growth of the species as well as
indirect effects such as reduction of the food supply and/or habitat modification. Direct effects
include reduced survival and reproductive impairment from both direct acute (short-term) and
direct chronic (long-term) exposures to atrazine. These assessment endpoints, while measured at
the individual level, provide insight about risks at higher levels of biological organization (e.g.,
populations) as described in U.S. EPA (2004).
With respect to atrazine degradates, including hydroxyatrazine (HA), deethylatrazine (DEA),
deisopropylatrazine (DIA), and diaminochloroatrazine (DACT), it is assumed that each of the
degradates are less toxic than the parent compound. As shown in Table 4.2, comparison of
available toxicity information for HA, DIA, and DACT indicates lesser aquatic toxicity than the
parent for freshwater fish, invertebrates, and aquatic plants.
Table 4.1 Comparison of Acute Freshwater Toxicity Values for Atrazine and
Degradates
Substance
Fish LCso
Daphnid ECS0
Aquatic Plant ECS0
Tested
Qig/L)
Qig/L)
Qig/L)
Atrazine
5,300
3,500
1
HA
>3,000 (no effects at
saturation)
>4,100 (no effects at
saturation)
>10,000
DACT
>100,000
>100,000
No data
DIA
17,000
126,000
2,500
DEA
No data
No data
1,000
Although degradate toxicity data are not available for terrestrial plants, lesser or equivalent
toxicity is assumed, given the available ecotoxicological information for other taxonomic groups
including aquatic plants and the likelihood that the atrazine degradates are expected to lose
efficacy as an herbicide.
98
-------
Therefore, given the lesser toxicity of the degradates, as compared to the parent, concentrations
of the atrazine degradates are not assessed, and the focus of this assessment is limited to parent
atrazine. The available information also indicates that aquatic organisms are more sensitive to
the technical grade (TGAI) than the formulated products of atrazine; therefore, the focus of this
assessment is on the TGAI. A detailed summary of the available ecotoxicity information for all
atrazine degradates and formulated products is presented in Appendix A.
In addition to registrant-submitted and open literature toxicity information, the community-level
endpoints were also used in the evaluation of the potential for atrazine to induce indirect effects
to the assessed species via impacts to aquatic plant community structure and function (See
Section 4.6). Other sources of information, including use of the acute probit dose response
relationship to establish the probability of an individual effect and reviews of the Ecological
Incident Information System (EIIS), are conducted to further refine the characterization of
potential ecological effects associated with exposure to atrazine. A summary of the available
aquatic and terrestrial plant ecotoxicity information, the community-level endpoints, use of the
probit dose response relationship, and the incident information for atrazine are provided in
Sections 4.1 through 4.8.
As previously discussed in the problem formulation, the available toxicity data show that other
pesticides may combine with atrazine to produce synergistic, additive, and/or antagonistic toxic
interactions. The results of available toxicity data for mixtures of atrazine with other pesticides
are presented in Section A.6 of Appendix A. Synergistic effects with atrazine have been
demonstrated for a number of organophosphate insecticides including diazanon, chlorpyrifos,
and methyl parathion, as well as herbicides including alachlor. If chemicals that show
synergistic effects with atrazine are present in the environment in combination with atrazine, the
toxicity of the atrazine mixture may be increased relative to the toxicity of each individual
chemical, offset by other environmental factors, or even reduced by the presence of antagonistic
contaminants if they are also present in the mixture. The variety of chemical interactions
presented in the available data set suggest that the toxic effect of atrazine, in combination with
other pesticides used in the environment, can be a function of many factors including but not
necessarily limited to (1) the exposed species, (2) the co-contaminants in the mixture, (3) the
ratio of atrazine and co-contaminant concentrations, (4) differences in the pattern and duration of
exposure among contaminants, and (5) the differential effects of other physical/chemical
characteristics of the receiving waters (e.g. organic matter present in sediment and suspended
water). Quantitatively predicting the combined effects of all these variables on mixture toxicity
to any given taxa with confidence is beyond the capabilities of the available data. However, a
qualitative discussion of implications of the available pesticide mixture effects data involving
atrazine on the confidence of risk assessment conclusions for the assessed species is addressed as
part of the uncertainty analysis for this effects determination.
4.1. Toxicity Data Used to Evaluate Assessment Endpoints
Toxicity endpoints are established based on data generated from guideline studies submitted by
the registrant and from open literature studies that meet the criteria for inclusion into the
99
-------
ECOTOX database maintained by EPA/ORD. Open literature data presented in this assessment
were obtained from the 2003 atrazine IRED as well as new information obtained from the
ECOTOX database on February 16, 2006. The February 2006 ECOTOX search included all
open literature data for atrazine (i.e., pre- and post-IRED). In order to be included in the
ECOTOX database, papers must meet the following minimum criteria:
(1) the toxic effects are related to single chemical exposure;
(2) the toxic effects are on an aquatic or terrestrial plant or animal species;
(3) there is a biological effect on live, whole organisms;
(4) a concurrent environmental chemical concentration/dose or application rate is
reported; and
(5) there is an explicit duration of exposure.
Data that pass the ECOTOX screen are evaluated along with the registrant-submitted data, and
may be incorporated qualitatively or quantitatively into this endangered species assessment. In
general, effects data in the open literature that are more conservative than the registrant-
submitted data are considered. Based on the results of the 2003 IRED for atrazine, potential
adverse effects on sensitive aquatic plants and non-target aquatic organisms including their
populations and communities, are likely to be greatest when atrazine concentrations in water
equal or exceed approximately 10 to 20 [j.g/L on a recurrent basis or over a prolonged period of
time. Given the large amount of microcosm/mesocosm and field study data for atrazine, only
effects data that are less than or more conservative than the 10 [j,g/L aquatic-community effect
level identified in the 2003 atrazine IRED were considered. The degree to which open literature
data are quantitatively or qualitatively characterized is dependent on whether the information is
relevant to the assessment endpoints (i.e., maintenance of survival, reproduction, and growth)
identified in the problem formulation. For example, endpoints such as behavior modifications
are likely to be qualitatively evaluated, because it is not possible to quantitatively link these
endpoints with reduction in species survival, reproduction, and/or growth (e.g., the magnitude of
effect on the behavioral endpoint needed to result in effects on survival, growth, or reproduction
is not known).
Citations of all open literature not considered as part of this assessment because it was either
rejected by the ECOTOX screen or accepted by ECOTOX but not used (e.g., the endpoint is less
sensitive and/or not appropriate for use in this assessment) are included in Appendix J.
Appendix J also includes a rationale for rejection of those studies that did not pass the ECOTOX
screen and those that were not evaluated as part of this ESA.
The most sensitive endpoint for each taxa evaluated was used for risk quotient calculation (U.S.
EPA, 2004). For this assessment, the toxicity data were used to assess endpoints listed in Table
4.1 for the six species considered in this analysis. A description of all effects data considered for
this assessment is in Appendix A. Currently, no studies have been conducted on sea turtles,
sturgeon, or freshwater mussels. Therefore, surrogate species were used as outlined in U.S. EPA
(2004) for characterization of atrazine toxicity to the assessed species and toxicity to other
animals on which the assessed species rely for sustenance. Avian studies were used for
surrogates for reptiles; the most sensitive fish and bivalve species tested were used to assess
100
-------
potential direct effects to the shortnose sturgeon and dwarf wedgemussel, respectively. In
addition, studies located in the open literature were considered for use in the characterization of
potential toxicity of atrazine to each of the species assessed as the data allow. A summary of the
toxicity data used for this assessment is in Table 4.1.
Table 4.2. Summary of Toxicity Data Used to Evaluate the Assessment Endpoints for the Six
Assessed Listed Species.
Toxicity Data"
Assessment Endpoint
Species Assessed
Freshwater and saltwater
invertebrates EC50, LC50, and
NOAEC
Survival, growth, and reproduction via
direct effects
Dwarf wedgemussel
Survival, growth, and reproduction via
indirect effects on food supply
Shortnose sturgeon, dwarf
wedgemussel, sea turtles (all four
species assessed)
Freshwater and saltwater fish
LC50 and NOAEC
Survival, growth, and reproduction via
direct effects
Shortnose sturgeon
Survival, growth, and reproduction via
effects on food supply
Sea turtles (ambient exposure, all four
species assessed)
Survival, growth, and reproduction via
effects on host fish needed to complete life
cycle
Dwarf wedgemussel
Acute avian LD50, LC50, and
reproduction NOAEC
Survival, growth, and reproduction via
direct effects
Sea turtles (oral exposure, all four
species assessed)
Freshwater and saltwater
aquatic plant EC50
Survival, growth, and reproduction via
indirect effects on habitat and/or primary
productivity
Shortnose sturgeon, dwarf
wedgemussel, sea turtles (all four
species assessed)
Survival, growth, and reproduction via
indirect effects on food supply
Green turtle, dwarf wedgemussel
Terrestrial plant EC25
Survival, growth, and reproduction via
indirect effects on terrestrial vegetation
(riparian habitat) required to maintain
acceptable water quality and spawning
habitat
All species assessed
a Most sensitive single species was initially used in risk estimation for indirect effects; however, dietary requirements and
behavior of the assessed species were used as the data allow to refine potential risks if use of the most sensitive food item species
resulted in LOC exceedances.
4.2. Toxicity Classification Scheme
Toxicity to fish and aquatic invertebrates is categorized using the following system as outlined in
U.S. EPA (2004):
LCso (ppm)
Toxicity Category
<0.1
Very highly toxic
>0.1-1
Highly toxic
>1-10
Moderately toxic
> 10 - 100
Slightly toxic
> 100
Practically nontoxic
101
-------
The following classification system (U.S. EPA, 2004) was used to characterize
toxicity of atrazine to birds (surrogate for sea turtles):
LC5o (ppm)
LD5o (mg/kg-bw)
Toxicity Category
<50
<10
Very highly toxic
50 - 500
10-50
Highly toxic
501 - 1000
51- 500
Moderately toxic
1001 -5000
501 -2000
Slightly toxic
>5000
>2000
Practically nontoxic
Toxicity categories are currently not defined for plants.
4.3. Laboratory Effects Data
This assessment considered both EPA guideline studies and studies located in the open literature.
A summary of registrant-submitted and open literature data used in risk estimation is provided in
this section. Additional information is in Appendix A.
4.3.1. Toxicity to Fish
4.3.1.1. Acute Exposure (Mortality) Studies
Fish toxicity studies were used to assess potential direct effects to the shortnose sturgeon and
potential indirect effects to the dwarf wedgemussel and sea turtles. Dwarf wedgemussels depend
on fish to complete their life cycle, and each of the four turtle species may consume fish as part
of their diet during all or part of their life cycle (see Appendix D).
Atrazine toxicity has been evaluated in numerous fish species, and the results of these studies
demonstrate a wide range of sensitivities to atrazine. LC50 values range from 2000 to 60,000
|ig/L (2 mg/L to 60 mg/L, see Appendix A for additional details on these studies). Therefore,
atrazine is classified as moderately toxic to fish on an acute basis.
Atrazine has been tested in both saltwater and freshwater species. The most sensitive species
was used to calculate risk quotients regardless of the salinity environment because this
assessment includes the Chesapeake Bay and its tributaries, which encompass both freshwater
and saltwater environments. However, species habitat would be considered if LOCs are
exceeded based on RQs derived using the most sensitive LC50. Therefore, the lowest LC50,
2,000 |ig/L reported in the estuarine fish sheepshead minnows (MRID 45208303) was used for
risk quotient calculations (Table 4.2).
102
-------
4.3.1.2.
Chronic Exposure Studies
The most sensitive chronic studies in fish used in this assessment indicate that atrazine is
associated with reduced juvenile survival in the estuarine fish, sheepshead minnows (LOAEC =
3400 |ig/L) and with reduced growth (7% reduction in length and 16% reduction in weight
relative to controls) in freshwater brook trout (LOAEC = 120 |ig/L). No effects (NOAEC) were
observed in these species at 1900 |ig/L and 65 |ig/L, respectively. The most sensitive NOAEC
of 65 |ig/L (MRID 00024377) was used to calculate risk quotients.
Table 4.3. Summary of Fish Toxicity Studies
Jsed In Risk Quotient Calculations.
Reference
(MRID)
Species Tested
Study
Type/Endpoints
Toxicity Value
Comment
Hall etal. 1994
(MRID 45208303)
Sheepshead Minnow
96-hour acute /
mortality
LC50: 2000 jig/L
Probit slope: 4.4a
(95% CI: 2.8-5.9)
None
Maceke^a/. 1976
(MRID 00024377)
Brook trout
44-Week life-
cycle /growth
and reproduction
NOAEC: 65 jig/L
LOAEC: 120 jig/L
NOAEC based on reduced
size and weight
Ward &
Ballantine 1985
(MRID 45202920)
Sheepshead Minnow
Early life stage /
growth and
reproduction
NOAEC: 1900 jig/L
LOAEC: 3400 jig/L
89% reduction in juvenile
survival was observed at
the LOAEC of 3400 jig/L.
a A reliable probit slope could not be estimated for the most sensitive study; therefore, a slope of 4.4 was used from
a different study in sheepshead minnows of equivalent duration (MRID 43344901). This analysis is consistent with
methods described in U.S. EPA (2004a) and results in a more conservative estimation of the probability of an
individual effect than the default slope recommended in U.S. EPA (2004a) of 4.5.
4.3.1.3. Sublethal Effects and Additional Open Literature Information In Freshwater
Fish
In addition to submitted studies, data were located in the open literature that report sublethal
effect levels to freshwater fish that are less than the selected measures of effect summarized in
Table 4.1. Although these studies report potentially sensitive endpoints, effects on survival,
growth, or reproduction were not observed in the four available life-cycyle studies at
concentrations that induced the reported sublethal effects described below and in Appendix A.
Therefore, these sublethal endpoints were not used for risk estimation purposes. In the life-cycle
study design, fish are exposed to atrazine from one stage of the life cycle to at least the same
stage of the next generation (e.g. egg to egg). Therefore, exposure occurs during the most
sensitive life stages and during the entire reproduction cycle.
Reported sublethal effects in adult largemouth bass show increased plasma vitellogenin levels in
both female and male fish at 50 [j,g/L and decreased plasma testosterone levels in male fish at
atrazine concentrations greater than 35 [j,g/L (Wieser and Gross, 2002 [MRID 456223-04]).
Vitellogenin (Vtg) is an egg yolk precursor protein expressed normally in female fish and
dormant in male fish. The presence of Vtg in male fish is used as a molecular marker of
exposure to estrogenic chemicals. It should be noted, however, that there is a high degree of
variability with the Vtg effects in these studies, which confounds the ability to resolve the effects
of atrazine on plasma steroids and vitellogenesis.
103
-------
Effects of atrazine on freshwater fish behavior, including a preference for the dark part of the
aquarium following one week of exposure (Steinberg et al., 1995 [MRID 452049-10]) and a
reduction in grouping behavior following 24-hours of exposure (Saglio and Trijase, 1998 [MRID
452029-14]), have been observed at atrazine concentrations of 5 [ig/L. In addition, alterations in
rainbow trout kidney histology have also been observed at atrazine concentrations of 5 [j,g/L and
higher (Fischer-Scherl et al., 1991 [MRID 452029-07]).
In salmon, potentially sensitive endpoints that have been reported included effects on gill
physiology and endocrine-mediated olfactory functions. Data from Waring and Moore (2004;
ECOTOX #72625) suggest that salmon smolt gill physiology, represented by changes in Na-K-
ATPase activity, was altered at 2 [j,g/L atrazine and higher. Survival was evaluated after transfer
to full salinity sea water (33 °/00). Atrazine exposure for 5 to 7 days in freshwater followed by
transfer to full salinity sea water resulted in higher mortality at atrazine concentrations of 14
ug/L (14% mortality) and higher in one study and at 1 ug/L (15% mortality) and higher in a
separate experiment presented in the publication (no controls died; statistical significance was
not indicated). As noted in Appendix D, observational and experimental evidence suggests that
shortnose sturgeon prefer habitats with less than 5°/00 for all life history stages during summer
months (U.S. EPA, 2003b). Based on distributional evidence, older juvenile and adult shortnose
sturgeon are limited to oligohaline and low mesohaline regions of estuaries (<15700). The
salinity used in by Waring and Moore (2004) simulated full strength seawater (33 °/00).
Therefore the relevance of findings from this study to the shortnose sturgeon is questionable.
Moore and Lower (2001; ECOTOX #67727) reported that endocrine-mediated functions of male
salmon parr were affected at 1 [j,g/L atrazine. The reproductive priming effect of the female
pheromone prostaglandin F2a on the levels of expressible milt in males was reduced relative to
controls after exposure to atrazine at 0.5 (J,g/L. Although the hypothesis was not tested, the study
authors suggest that exposure of smolts to atrazine during the freshwater stage may potentially
affect olfactory imprinting to the natal river and subsequent homing of adults. However, no
quantitative relationship is established between reduced olfactory response of male epithelial
tissue to the female priming hormone in the laboratory and reduction in salmon reproduction
(i.e., the ability of male salmon to detect, respond to, and mate with ovulating females). A
negative control was not included as part of the study design; therefore, potential solvent effect
cannot be evaluated. Furthermore, the study did not determine whether the decreased response
of olfactory epithelium to specific chemical stimuli would result in similar responses in intact
fish.
Although these studies raise questions about the effects of atrazine on plasma steroid levels,
behavior modifications, gill physiology, and endocrine-mediated functions in freshwater and
anadromous fish, the data do not allow for a derivation of a quantitative link between these
sublethal effects and the selected assessment endpoints for the assessed species (i.e., survival,
growth, and reproduction of individuals). Also, effects on survival, growth, or reproduction were
not observed in the four available life-cycle studies at concentrations that induced these reported
sublethal effects. Therefore, potential sublethal effects to fish are considered qualitatively in
104
-------
Section 5.2, but are not used as part of the quantitative risk characterization. Further detail on
sublethal effects to fish is provided in Sections A.2.4a and A.2.4b of Appendix A.
4.3.2. Toxicity to Aquatic Invertebrates
Aquatic invertebrate toxicity studies were used to assess potential direct effects to the dwarf
wedgemussel and potential indirect effects to the shortnose sturgeon, all four sea turtles, and the
dwarf wedgemussel as outlined in Table 4.3.
4.3.2.1. Acute Toxicity Studies
Atrazine is classified as very highly toxic to slightly toxic to aquatic invertebrates on an acute
exposure basis with LC50 and EC50 values ranging from 88 |ig/L to >33,000 |ig/L. A chemical
is considered very highly toxic if the LC50 is less than 100 |ig/L and slightly toxic if the LC50 is
between 10,000 and 100,000 |ig/L. The acute toxicity data in invertebrates indicate a wide range
of sensitivity across species. Furthermore, considerable variability in sensitivity was observed
across studies conducted using the same species (Figure 4-1). The most sensitive (lowest) LC50
value for a given species was used for risk estimation. Therefore, this risk assessment may
overestimate or underestimate toxicity to some taxa under some environmental conditions.
Data Used for Direct Effects Assessment
The dwarf wedgemussel is the only listed aquatic invertebrate included in this assessment for
direct effects. The Eastern oyster was used as a surrogate species for the dwarf wedgemussel.
The acute toxicity data demonstrated that the shell deposition EC50 value in Eastern oyster
(Crassostrea virginica) was >1,700 |ig/L (MRID 46648201); no treatment related effects were
observed in this study at any concentration. A second study in the Eastern oyster also produced
no effects at the highest concentration tested of 1000 |ig/L (MRID 46648201). In addition, the
Pacific oyster was tested with a wettable powder formulated product. That study (MRID
45227722) produced an EC50 >100 |ig/L.
Because none of the studies produced definitive EC50 values (no clear treatment-related effects
at any concentration tested), an EC50 of >1700 |ig/L was used for risk estimation for direct
effects to the dwarf wedgemussel.
One additional acute study in freshwater mussels was located in the open literature. The results
of the study by Johnson et al. (1993) suggest that 48-hour exposures at atrazine concentrations
up to 60 mg/L (60,000 |ig/L) do not affect the survival of juvenile and mature freshwater
mussels, Anodonta imbecilis. This study was not considered suitable for use in RQ calculations;
however, it was considered to be of good quality and useful in risk characterization discussion.
The study in the freshwater mussel, A. imbecilis, do not suggest that use of Eastern Oysters in
risk estimation resulted in an underestimation of potential risk of direct effects to dwarf
wedgemussels.
105
-------
Data Used for Indirect Effects Assessment
Aquatic invertebrate toxicity data were also used to evaluate potential indirect effects to each of
the six listed species because each assessed species depends on aquatic invertebrates for
sustenance. For the indirect effects assessment, the most sensitive aquatic invertebrate species
was initially used for risk estimation, which is consistent with U.S. EPA (2004). The most
sensitive organism tested was the marine copepod. The lowest LC50 in this species was 88
|ig/L; however, a wide range of LC50 values have been reported in copepods from studies that
tested technical grade atrazine (LC50 values of 88, 94, 140, 500, 4300, and 7900 |ig/L have been
reported, see Appendix A). Reasons for the disparity across the reported acute toxicity values in
the copepod are unknown. However, similar variability has been observed in other species that
have been tested by multiple laboratories. For example, studies conducted in the midge
produced LC50s that spanned 2 orders of magnitude (values ranged from 720 to >33,000 |ig/L).
Other than the copepod, all reported acute toxicity values for the other 12 aquatic invertebrate
species tested are 720 |ig/L and higher.
The distribution of available toxicity data are summarized in Figure 4-1 below. These studies are
described in greater detail in Appendix A.
Genus Level Geometric Means of Reported Asute LC50 and EC50 Values in
Aquatic Invertebrates for Atrazine
LC/EC50
(ug/L)
17000
16000
15000
14000
13000
12000
11000
10000
9000
8000
7000
6000
5000
4000
3000
2000
1000
0
£
73
*
E
£
CO
Species
Figure 4-1. Range of Aquatic Invertebrate Acute Toxicity Values Reported for Atrazine
The columns in the above graph represent geometric means of the acute toxicity values (genus
levels). The error bars represent the range of reported values. Error bars higher than the
maximum value of 17,000 |ig/L were reported for two species. These values are 33,000 |ig/L for
106
-------
the midge and 30,000 |ig/L for the waterflea. Values in parentheses represent the number of
studies included in the analysis. No effects were observed in the mud crab and eastern oyster
studies. See Appendix A for a description of the studies used in the generation of Figure 4-1.
4.3.2.2. Chronic Exposure Studies
The most sensitive chronic endpoint for freshwater invertebrates was based on a 30-day flow-
through study on the scud, which showed a 25% reduction in the development of Fi to the
seventh instar at atrazine concentrations of 140 |ig/L; the corresponding NOAEC was 60 |ig/L
(MRID 00024377).
The most sensitive chronic bioassay in saltwater species was a 28-day study in mysid shrimp
(Americamysis bahia) that reported a NOAEC of 80 |ig/L; a 37% reduction in juvenile survival
occurred at the LOAEC of 190 |ig/L. Additional details on this study (MRID 45202920) and
other chronic bioassays are described in Appendix A.
An uncertainty in the chronic bioassay data is that chronic toxicity data suitable for risk quotient
derivation are not available on the most acutely sensitive marine invertebrate (copepod). The
potential impact of this uncertainty in risk estimation is described in Section 6. However, the
absence of a chronic NOAEC in copepods is not expected to change conclusions of this risk
assessment.
Also, a chronic study in bivalves was not available. However, the direct effects assessment to
the dwarf wedgemussel was considered protective because the acute data in the Eastern oyster
and in a freshwater mussel (A. imbecilis ) demonstrated low sensitivity to atrazine relative to
other aquatic invertebrates tested. Therefore, use of the most sensitive aquatic invertebrate
(scud, MRID 00024377) chronic NOAEC in risk estimation for direct effects to the dwarf
wedgemussel is considered protective. Acute and chronic studies used to calculate risk quotients
for aquatic invertebrates are summarized in Table 4.3.
107
-------
Table 4.4. Acute and Chronic Aquatic Invertebrate Toxicity Values Used
in Initial Risk Estimation of Atrazine
Reference
(MRU))
Species
Tested
Species
Assessed
Study Type /
Endpoints
Toxicity
Value
(hs/l)
Comment
Thursby el al.,
1990 (MRID
45202918)
Saltwater
invertebrate,
Copepod
(Acartia tonsa)
Shortnose
sturgeon and sea
turtles / indirect
effects from
reduction in
animal food
supply
Acute toxicity /
mortality
LC50: 88
Probit Slope:
0.95
Data used as initial screen
to assess indirect effects
to listed species from
reduction of animal food
supply.
Ward &
Ballantine,
1985
(MRID
45202920)
Saltwater
invertebrate,
Mysid shrimp
(Americamysis
bahia)
Sea turtles /
indirect effects
from reduction in
animal food
supply
Chronic
exposure /
growth and
survival
NOAEC: 80
37% Reduction in
survival occurred at the
LOAEC of 190 (ig/L.
Data used as initial screen
to assess indirect effects
to listed species from
reduction of animal food
supply.
Macek et al.
1976
(MRID
00024377)
Freshwater
invertebrate,
Scud
Dwarf
wedgemussel /
direct chronic
effects
Shortnose
sturgeon /
indirect effects
from reduction in
animal food
supply
Chronic
exposure / 25 %
red. in
development of
Fi to seventh
instar.
NOAEC: 60
Chronic bivalve data were
not available; therefore,
this study, as the most
sensitive aquatic
invertebrate chronic
study, was used to
characterize potential
chronic toxicity of
atrazine to the dwarf
wedgemussel.
Johnson 1986
(MRID
45087413)
Freshwater
invertebrate,
daphnid
Dwarf
wedgemussel /
indirect effects
from reduction in
food supply
Acute exposure /
immobilization
EC50: 3500
Probit slope:
Sufficient data
not available3
Raw data not included.
Caferalla,
2005b (MRID
46648201);
Mayer 1986
(MRID 40228-
01)
Bivalve,
Eastern oyster
Dwarf
wedgemussel /
direct acute
effects
Acute exposure /
shell deposition
EC50: >1000
and >1700
Probit slope:
None (no
effects
occurred)
Endpoint chosen to assess
potential direct effects to
the dwarf wedgemussel
was 1700 |ig/L for risk
quotient calculations
because no treatment
related effects occurred in
either study.
a Slope information on the toxicity study that was used to derive the RQ for freshwater invertebrates is not
available. Therefore, the probability of an individual effect was calculated using a probit slope of 4.4, which is the
only technical grade atrazine value reported in the available freshwater invertebrate acute studies; 95% confidence
intervals could not be calculated based on the available data (Table A-18). Use of a probit slope of 4.4 would result
in a more conservative estimation of the probability of an individual effect than the default slope recommended in
U.S. EPA (2004a) of 4.5.
108
-------
4.3.3. Toxicity to Sea Turtles
Available toxicity data in turtles or other reptiles are limited (summarized in Table 4.4), and no
data in sea turtles were located. Therefore, birds were used as a surrogate species for the
characterization of atrazine effects to turtles, in accordance with the Overview Document (U.S.
EPA, 2004). Birds are considered a conservative surrogate species for the evaluation of potential
risks to sea turtles the following reasons:
• Reptiles are poikilotherms (body temperature varies with environmental temperature)
while birds are homeotherms (temperature is regulated, constant, and largely independent
of environmental temperatures). As a consequence, the caloric requirements of reptiles
are markedly lower than birds. Therefore, on a daily dietary intake basis, birds consume
more food than reptiles. This can be seen when comparing the caloric requirements for
free living iguanid lizards to Passeriformes (song birds) (U.S. EPA, 1993):
iguanid FMR (kcal/day)= 0.0535 (bw g)A0.799
passerine FMR (kcal/day) = 2.123 (bw g)A0.749
With relatively comparable slopes to the allometric functions, one can see that, given a
comparable body weight, the free living metabolic rate of birds can be 40 times higher
than reptiles, though the requirement differences narrow with high body weights.
Because the existing risk assessment process is driven by the dietary route of exposure, a
finding of safety for birds, with their much higher feeding rates and therefore higher
dietary exposure, is reasoned to be protective of reptiles. For this not to be the case, a
reptile would have to be 40 times more sensitive than birds for the differences in dietary
uptake to be negated. The existing reptile toxicity data (Table 4.4), although limited in its
utility, do not suggest that reptiles are more sensitive than birds to atrazine. In addition,
conservative assumptions were made to estimate exposure to sea turtles (Section 3).
For these reasons, the assessment based on toxicity studies in birds as a surrogate species
is considered protective of sea turtles. Toxicity values used to calculate risk quotients for
sea turtles are summarized in Table 4.4 below.
The available data in birds suggest that atrazine is slightly toxic to avian species on an acute oral
exposure basis. The lowest reported LD50 is 940 mg/kg-bw. Signs of intoxication in mallards
first appeared 1 hour after treatment and persisted up to 11 days (U.S. EPA, 2003a). In
pheasants, remission of signs of intoxication occurred by 5 days after treatment. Signs of
intoxication included weakness, hyper-excitability, ataxia, and tremors; weight loss occurred in
mallards.
One degradate (desethyl atrazine, DEA) has been shown to be roughly as toxic as atrazine to
birds on an acute oral basis. Other degradates evaluated, including deisopropyl atrazine (DIA)
109
-------
and hydroxyatrazine (HA) are considerably less toxic than atrazine to birds on an acute oral basis
(Appendix A). However, DACT, which has been shown to be of equivalent toxicity compared
with atrazine in mammals, has not been tested in birds.
Because all subacute avian LC50 values are greater than 5,000 ppm, atrazine is categorized as
practically non-toxic to avian species on a subacute dietary basis. In the subacute dietary study
in mallard ducks, 30% mortality was observed at the highest test concentration of 5,000 ppm
(MRID 00022923). The time to death was Day 3 for the one Japanese quail and Day 5 for three
mallard ducks (U.S. EPA, 2003).
Reproduction studies in birds have reported reproductive effects at atrazine concentrations as low
as 675 ppm. In bobwhite quail, the following endpoints were affected at 675 ppm atrazine: egg
production, embryo viability, hatchling and 14-day weight, and number of defective eggs (MRID
42547102). Bobwhite and mallard tests show similar toxic effects on reduced egg production
and embryo viability/hatchability with LOAEC and NOAEC values of 675 and 225 ppm,
respectively, for both species. Although the bobwhite test showed a 7 to 18% reduction in 14-
day body weight in the 75 ppm treatment group relative to the control group, the reproductive
endpoints were considered to be more biologically significant, given the use of the avian data as
a surrogate for sea turtles in the Chesapeake Bay. However, use of 75 ppm instead of 225 ppm
would not impact conclusion in this assessment as discussed in Section 5.
Several studies in turtles were located in the open literature. These studies, which are described
in Appendix A, suggest that atrazine does not permeate the outer egg shell of reptiles including
turtles and alligators after direct application to the egg (MRIDs 45545303 and 45545302) or
cause significant alteration in gonadal development and aromatase activity in the snapping turtle
or alligator under the conditions of the available studies (De Solla et al., 2005; Crain et al,
1999). Although these data do not allow for derivation of risk quotients, they suggest that
reptiles are not more sensitive than birds to potential atrazine effects.
110
-------
Table 4.5. Summary of Available Acute Oral, Subacute Dietary, and Reproduction Toxicity Studies in
Birds, and Available Studies in Reptiles.
Test material/
Reference
(MRID)
Species Tested
Study
Type/Endpoints
Toxicity Value
Comment
Technical grade
atrazine
Fink 1976
(MRID 00024721)
Northern bobwhite
quail
(Colinus virginianus)
Acute oral gavage
toxicity / mortality
LD50: 940 mg/kg-
bw
Slope = 3.8
(95% CI: 2.0 - 5.7)
The range of acute oral gavage
LD50s in birds is 940 mg/kg-
bw to 4200 mg/kg-bw
(Appendix A).
Degradate Desethyl
Atrazine (DEA)
Stafford, 2005c
(MRID 46500009)
Northern bobwhite
quail
0Colinus virginianus)
Acute oral gavage
toxicity / mortality
LD50: 768 mg/kg-
bw
slope = 6.2
(95% CI: 3.2-9.3)
These data suggest that the
degradate DEA is
approximately as toxic to birds
on an acute oral basis as
atrazine.
Technical grade
atrazine
Hill etal. 1975
(MRID 00022923)
Mallard duck
(Anas platyrhynchos)
Subacute dietary /
mortality
LC50: > 5,000
(30 % mortality
at 5,000 ppm)
All submitted subacute dietary
studies in birds report LC50s
that are higher than 5,000 ppm.
Technical grade
atrazine
Pedersen &
DuCharme 1992
(MRID 42547102)
Northern bobwhite
(Colinus virginianus)
Dietary Exposure /
Reproduction effects
NOAEC: 225 ppm
LOAEC : 675 ppm
At the LOAEC, egg production,
embryo viability, and hatchling
weight were affected.
Technical grade
atrazine
Pedersen &
DuCharme
1992
(MRID 42547101)
Mallard duck
(Anas platyrhynchos)
Dietary Exposure /
Reproduction effects
NOAEC: 225 ppm
LOAEC : 675 ppm
At the LOAEC, egg production,
egg hatchability, and food
consumption were affected.
Technical grade
atrazine
De Solla et al., 2005;
Ecotox Reference No.
82032
Snapping turtles
4-Month exposure
study in developing
embryos / gonad
development
NOAEC: 13.2 lbs
a.i./Acre (8.1 ppm
soil), highest rate
tested
No treatment-related effects
were observed at the highest
concentration tested.
Technical grade
atrazine
Gross, 2001
(MRIDs 45545303
and 45545302)
Red-eared slider turtle
and American Alligator
10-Day egg exposure
/ endocrine effects
NOAEC: 500 ng/L,
highest
concentration tested
No treatment-related effects
were observed at the highest
concentration tested.
Ill
-------
4.4. Terrestrial Plant Toxicity
Terrestrial plant toxicity data are used to evaluate the potential for atrazine to affect the riparian
zone. Riparian zone effects could impact habitat and stream water quality as discussed in detail
in Section 5.2.
Plant toxicity data from both registrant-submitted studies and studies in the scientific literature
were reviewed for this assessment. Registrant-submitted studies are conducted under conditions
and with species defined in EPA toxicity test guidelines. Sub-lethal endpoints such as plant
growth, dry weight, and biomass are evaluated for both monocots and dicots, and evaluate
effects at both seedling emergence and vegetative life stages. A guideline study generally
evaluates toxicity to ten crop species. A drawback to these tests is that they are conducted on
herbaceous agricultural crop species only, and extrapolation of effects to other species, such as
woody shrubs and trees and wild herbaceous species, contributes uncertainty to risk conclusions.
However, atrazine is labeled for use in forestry production; therefore effects to these types of
trees are not anticipated at concentration anticipated in the environment. In addition, preliminary
data (discussed below) suggests that sensitive woody plant species exist; however, damage to
most woody species at labeled application rates is not expected.
Commercial crop species have been selectively bred, and may be more or less resistant to
particular stressors than wild herbs and forbs. The direction of this uncertainty for specific plants
and stressors, including atrazine, is largely unknown. Homogenous test plant seed lots also lack
the genetic variation that occurs in natural populations, so the range of effects seen from tests is
likely to be smaller than would be expected from wild populations.
Based on the results of the submitted terrestrial plant toxicity tests, it appears that emerged
seedlings are more sensitive to atrazine via soil/root uptake exposure than emerged plants via
foliar routes of exposure. However, all tested plants, with the exception of corn in the seedling
emergence and vegetative vigor tests and ryegrass in the vegetative vigor test, exhibited adverse
effects following exposure to atrazine.
For Tier II seedling emergence, the most sensitive dicot is the carrot and the most sensitive
monocots are oats. EC25 values, on an equivalent application rate basis, for oats and carrots,
which are based on a reduction in dry weight, are 0.003 and 0.004 lb ai/A, respectively; NOAEC
values for both species are 0.0025 lb ai/A. Table 4.5 summarizes the Tier II terrestrial plant
seedling emergence toxicity data.
For Tier II vegetative vigor studies, the most sensitive dicot is cucumber and the most sensitive
monocot is onion. In general, dicots appear to be more sensitive than monocots via foliar routes
of exposure with all tested monocot species showing a significant reduction in dry weight at
EC25 values ranging from 0.008 to 0.72 lb ai/A. In contrast, two of the four tested monocots
showed no effects from atrazine (corn and ryegrass), while EC25 values for oats and onion were
0.61 and 2.4 lb ai/A, respectively. Table 4.6 summarizes the terrestrial plant vegetative vigor
toxicity data used to derive risk quotients in this assessment.
112
-------
Table 4.5. Nontarget Terrestrial Plant Seedling Emergence Toxicity (Tier II).
Surrogate Species % ai EC25 / NOAEC (lbs ai/A) Endpoint Affected MRID No. Study
Author/Year Classification
Monocot - Corn
{Zea mays)
97.7
>4.0/>4.0
No effect
420414-03
Chetram 1989
Acceptable
Monocot - Oat
(Avena sativa)
97.7
0.004/0.0025
red. in dry weight
420414-03
Chetram 1989
Acceptable
Monocot - Onion
(Allium cepa)
97.7
0.009/0.005
red. in dry weight
420414-03
Chetram 1989
Acceptable
Monocot - Ryegrass
(.Lolium perenne)
97.7
0.004/0.005
red. in dry weight
420414-03
Chetram 1989
Acceptable
Dicot - Root Crop - Carrot
(.Daucus carota)
97.7
0.003 / 0.0025
red. in dry weight
420414-03
Chetram 1989
Acceptable
Dicot - Soybean
(4.0/>4.0
No effect
420414-03
Chetram 1989
Acceptable
Monocot - Oat
(Avena sativa)
97.7
2.4 / 2.0
red. in dry weight
420414-03
Chetram 1989
Acceptable
Monocot - Onion
(Allium cepa)
97.7
0.61 / 0.5
red. in dry weight
420414-03
Chetram 1989
Acceptable
Monocot - Ryegrass
(Lolium perenne)
97.7
>4.0/>4.0
No effect
420414-03
Chetram 1989
Acceptable
Dicot - Root Crop - Carrot
(Daucus carota)
97.7
1.7 / 2.0
red. in plant height
420414-03
Chetram 1989
Acceptable
Dicot - Soybean
(Glycine max)
97.7
0.026/0.02
red. in dry weight
420414-03
Chetram 1989
Acceptable
Dicot - Lettuce
(Lactuca sativa)
97.7
0.33 / 0.25
red. in dry weight
420414-03
Chetram 1989
Acceptable
Dicot - Cabbage
(Brassica oleracea alba)
97.7
0.014/0.005
red. in dry weight
420414-03
Chetram 1989
Acceptable
Dicot - Tomato
(Lycopersicon esculentum)
97.7
0.72 / 0.5
red. in plant height
420414-03
Chetram 1989
Acceptable
Dicot - Cucumber
(Cucumis sativus)
97.7
0.008/ 0.005
red. in dry weight
420414-03
Chetram 1989
Acceptable
113
-------
In addition, a report on the toxicity of atrazine to woody plants (Wall et al., 2006; MRID
4687040001) was reviewed by the Agency. A total of 35 species were tested at application rates
ranging from 1.5 to 4.0 lbs ai/A. Twenty-eight species exhibited either no or negligible
phytotoxicity. Seven of 35 species exhibited >10% phytotoxicity. However, further
examination of the data indicate that atrazine application was clearly associated with severe
phytotoxicity in only one species (Shrubby Althea). These data suggest that, although sensitive
woody plants exist, atrazine exposure to most woody plant species at application rates of 1.5 to
4.0 lbs ai/A is not expected to cause adverse effects. A summary of the available woody plant
data is provided in Table A-39b of Appendix A.
4.5. Aquatic Plant Toxicity Data
Aquatic plant toxicity studies were used to evaluate whether atrazine may affect primary
productivity in the Chesapeake Bay and its source waters or direct food source for the dwarf
wedgemussel and green turtles, both of which use plants as a primary component of their diets.
Two types of studies were used to evaluate the potential of atrazine to affect primary
productivity. The most sensitive EC50 from available laboratory studies was initially used to
derive risk quotients to determine whether atrazine may affect aquatic plants. Threshold
concentrations predictive of potential community level effects to aquatic plants were also used to
further characterize indirect effects to the assessed species. Laboratory data are described in
Section 4.5.1., field studies are described in Section 4.5.2., and community-level threshold
concentrations are described in Section 4.5.3.
Recovery from the effects of atrazine and the development of resistance to the effects of atrazine
in some vascular and non-vascular aquatic plants has been reported and may add uncertainty to
these findings. However, reports of recovery are often based on differing interpretations. Thus,
before recovery can be considered as an uncertainty, an agreed upon interpretation is needed. For
the purposes of this assessment, recovery is defined as a return to pre-exposure levels for the
affected community, not for a replacement community of more tolerant species. Further research
is needed to quantify the impact that recovery and resistance would have on aquatic plants.
4.5.1. Laboratory Data
Numerous aquatic plant toxicity studies have been submitted. A summary of these studies is
presented below. See Appendix A for a more comprehensive description of these data. The Tier
II results for freshwater aquatic plants indicate that atrazine causes a 41 to 98% reduction in
chlorophyll production of freshwater algae; the corresponding EC50 value for four different
species of freshwater algae is 1 |ig/L, based on data from a 7-day acute study (MRID 00023544).
Vascular plants are less sensitive to atrazine than their freshwater non-vascular plants with an
EC50 value of 37 |ig/L, based on reduction in duckweed growth (MRID 43074804).
In marine species, the marine algae Isochrysis galbana is the most sensitive nonvascular aquatic
plant (EC50 = 22 |ig/L; MRID 41065204), and the most sensitive vascular aquatic plant is Sago
pondweed (EC50 = 7.5 |ig/L; MRID 45088231). EC50s for sea grasses, which are important
114
-------
forage material for green turtles, range from approximately 70 |ig/L (MRID 45227729) to 30,000
|ig/L (MRID 45205101) in laboratory studies.
4.5.2. Field Data
Microcosm and mesocosm studies with atrazine provide measurements of primary productivity
that incorporate the aggregate responses of multiple species in aquatic plant communities.
Because plant species vary widely in their sensitivity to atrazine, the overall response of the plant
community may be different from the responses of the individual species measured in laboratory
toxicity tests. Mesocosm and microcosm studies allow observation of population and
community recovery from atrazine effects and of indirect effects on higher trophic levels. In
addition, mesocosm and microcosm studies, especially those conducted in outdoor systems,
incorporate partitioning, degradation, and dissipation, factors that are not usually accounted for
in laboratory toxicity studies, but that may influence the magnitude of ecological effect.
Atrazine has been the subject of many mesocosm and microcosm studies in ponds, streams,
lakes, and wetlands. The duration of these studies have ranged from a few weeks to several
years in duration at exposure concentrations from 0.1 |ig/L to 10,000 |ig/L. Most of the studies
have focused on atrazine effects on phytoplankton, periphyton, and macrophytes; however, some
have also included measurements on animals.
Based on the results of the 2003 IRED for atrazine, potential adverse effects on sensitive aquatic
plants and non-target aquatic organisms including their populations and communities are likely
to be greatest when atrazine concentrations in water equal or exceed approximately 10 to 20
|ig/L on a recurrent basis or over a prolonged period of time. A summary of all the freshwater
aquatic microcosm, mesocosm, and field studies that were summarized as part of 2003 IRED is
included in Appendix A. In addition, a number of estuarine/marine field studies are available,
which are also discussed in Appendix A (Section A.3.7). Given the large amount of microcosm
and mesocosm and field study data for atrazine, only effects data less than or more conservative
than the 10 |ig/L aquatic community effect level identified in the 2003 IRED were considered as
part of the open literature search. Based on the selection criteria for review of new open
literature, all of the available studies show effects levels to freshwater and estuarine/marine fish
and invertebrates at concentrations greater than 10 |ig/L.
4.6. Community-Level Endpoints: Threshold Concentrations
In this ESA, direct and indirect effects to the assessed listed species are evaluated in accordance
with the screening4evel methodology described in the Agency's Overview Document (U.S.
EPA, 2004). If aquatic plant RQs exceed the Agency's non-listed species LOC (because the
assessed species do not have an obligate relationship with any one particular plant species, but
rather rely on multiple plant species), based on available EC50 data for vascular and non-vascular
plants, risks to individual aquatic plants are assumed.
It should be noted, however, that the indirect effects and components of the critical habitat
impact analyses in this assessment are unique, in that the best available information for atrazine-
115
-------
related effects on aquatic communities is significantly more extensive than for other pesticides.
Hence, atrazine effects determinations can utilize more refined data than is generally available to
the Agency. Specifically, a robust set of microcosm and mesocosm data and aquatic ecosystem
models are available for atrazine that allowed EPA to refine the indirect effects and critical
habitat impact analysis associated with potential aquatic community-level effects (via aquatic
plant community structural change and subsequent habitat modification) to the listed species.
Use of such information is consistent with the guidance provided in the Overview Document
(U.S. EPA, 2004), which specifies that "the assessment process may, on a case-by-case basis,
incorporate additional methods, models, and lines of evidence that EPA finds technically
appropriate for risk management objectives" (Section V, page 31 of EPA, 2004). This
information, which represents the best scientific data available, is described in further detail
below and in Appendix B. This information is also considered a refinement of the 10-20 |ig/L
range reported in the 2003 IRED (U.S. EPA, 2003a).
The Agency has selected an atrazine level of concern (LOC) in the 2003 IRED (U.S. EPA, 2003a
and b) that is consistent with the approach described in the Office of Water's (OW) draft atrazine
aquatic life criteria (U.S. EPA, 2003c). Through these previous analyses (U.S. EPA, 2003a, b,
and c), which reflect the current best available information, predicted or monitored aqueous
atrazine concentrations can be interpreted to determine if a water body is likely to be
significantly affected via indirect effects to the aquatic community. Potential impacts of atrazine
to plant community structure and function that are likely to result in indirect effects to the rest of
the aquatic community, including the listed species, are evaluated as described below.
As described further in Appendix B, responses in microcosms and mesocosms exposed to
atrazine were evaluated to differentiate no or slight, recoverable effects from significant,
generally non-recoverable effects (U.S. EPA, 2003e). Because effects varied with exposure
duration and magnitude, there was a need for methods to predict relative differences in effects
for different types of exposures. The Comprehensive Aquatic Systems Model (CASM) (Bartell
et al., 2000; Bartell et al., 1999; DeAngelis et al., 1989) was selected as an appropriate tool to
predict these relative effects, and was configured to provide a simulation for the entire growing
season of a 2nd and 3rd order Midwestern stream as a function of atrazine exposure. CASM
simulations conducted for the concentration/duration exposure profiles of the micro- and
mesocosm data showed that CASM seasonal output, represented as an aquatic plant community
similarity index, correlated with the micro- and mesocosm effect scores, and that a 5% change in
this index reasonably discriminated micro- and mesocosm responses with slight versus
significant effects. The CASM-based index was assumed to be applicable to more diverse
exposure conditions beyond those present in the micro- and mesocosm studies.
To avoid having to routinely run the CASM model, simulations were conducted for a variety of
actual and synthetic atrazine chemographs to determine 14-, 30-, 60-, and 90-day average
concentrations that discriminated among exposures that were unlikely to exceed the CASM-
based index (i.e., 5% change in the index). It should be noted that the average 14-, 30-, 60-, and
90-day concentrations were originally intended to be used as screening values to trigger a CASM
run (which is used as a tool to identify the 5% index change LOC), rather than actual thresholds
to be used as an LOC (U.S. EPA, 2003e). The following threshold concentrations for atrazine
were identified (U.S. EPA, 2003e):
116
-------
• 14-day average = 38 [^g/L
• 30-day average = 27 [^g/L
• 60-day average =18 [^g/L
• 90-day average =12 [^g/L
Effects of atrazine on aquatic plant communities that have the potential to subsequently pose
indirect effects to the listed species and their designated critical habitat are best addressed using
the robust set of micro- and mesocosm studies available for atrazine and the associated risk
estimation techniques (U.S. EPA, 2003a, b, c, and e). The 14-, 30-, 60-, and 90-day threshold
concentrations developed by EPA (2003e) are used to evaluate potential indirect effects to
aquatic communities for the purposes of this ESA. Use of these threshold concentrations is
considered appropriate because: (1) the CASM-based index meets the goals of the defined
assessment endpoints for this assessment; (2) the threshold concentrations provide a reasonable
surrogate for the CASM index; and (3) the additional conservatism built into the threshold
concentration, relative to the CASM-based index, is appropriate for an endangered species risk
assessment (i.e., the threshold concentrations were set to be conservative, producing a low level
(1%) of false negatives relative to false positives). Therefore, these threshold concentrations are
used to identify potential indirect effects (via aquatic plant community structural change) to the
listed species and their designated critical habitat. If modeled atrazine EECs exceed the 14-, 30-,
60- and 90-day threshold concentrations following refinements of potential atrazine
concentrations with available monitoring data, the CASM model could be employed to further
characterize the potential for indirect effects. A step-wise data evaluation scheme incorporating
the use of the threshold concentrations is provided in Figure 4.2. Further information on
threshold concentrations is provided in Appendix B.
117
-------
I Action Area
Exposure
Profile
Data
No effect" W-
Aquatic
Yes 1
Derive EECs for
various averaging
periods from
modeling data
90-day
rolling
averages
60-day
rolling
averages
30-day
rolling
averages
No
\
'May affect, but
not likely to
.adversely affect"
60-day
AVG
>18 ug/L?
30-day
AVG
> 27 ug/L?
90-day
AVG
> 12 ug/L 2
14-day
rolling
averages
14-day
AVG
> 38 ug/L2
Refine EECs based on site-specific information and/or monitoring data.
Do refined EECs exceed the threshold concentrations above?
"May affect, but
not likely to
adversely affect"
No
Likely to
adversely affect'
Figure 4.2 Use of Threshold Concentrations in Endangered Species Assessment
4.7.
Use of Probit Slope Response Relationship to Provide Information on the
Endangered Species Levels of Concern
118
-------
The Agency uses the probit dose-response relationship as a tool for providing additional
information on the potential for acute direct effects to individual listed species and aquatic
animals that may indirectly affect the listed species of concern (U.S. EPA, 2004). As part of the
risk characterization, an interpretation of acute RQ for listed species is discussed. This
interpretation is presented in terms of the chance of an individual event (i.e., mortality or
immobilization) should exposure at the EEC actually occur for a species with sensitivity to
atrazine on par with the acute toxicity endpoint selected for RQ calculation. To accomplish this
interpretation, the Agency uses the slope of the dose response relationship available from the
toxicity study used to establish the acute toxicity measures of effect for each taxonomic group
that is relevant to this assessment (i.e., freshwater fish used as a surrogate for aquatic-phase
amphibians and freshwater invertebrates). The individual effects probability associated with the
acute RQ is based on the mean estimate of the slope and an assumption of a probit dose response
relationship. In addition to a single effects probability estimate based on the mean, upper and
lower estimates of the effects probability are also provided to account for variance in the slope, if
available. The upper and lower bounds of the effects probability are based on available
information on the 95% confidence interval of the slope. A statement regarding the confidence
in the estimated event probabilities is also included. Studies with good probit fit characteristics
(i.e., statistically appropriate for the data set) are associated with a high degree of confidence.
Conversely, a low degree of confidence is associated with data from studies that do not
statistically support a probit dose response relationship. In addition, confidence in the data set
may be reduced by high variance in the slope (i.e., large 95% confidence intervals), despite good
probit fit characteristics.
Individual effect probabilities are calculated based on an Excel spreadsheet tool IECV1.1
(Individual Effect Chance Model Version 1.1) developed by the U.S. EPA, OPP, Environmental
Fate and Effects Division (June 22, 2004). The model allows for such calculations by entering
the mean slope estimate (and the 95% confidence bounds of that estimate) as the slope parameter
for the spreadsheet. In addition, the acute RQ is entered as the desired threshold.
4.8. Incident Database Review
A number of incidents have been reported in which atrazine has been associated with some type
of environmental effect. Incidents are maintained and catalogued by EFED in the Ecological
Incident Information System (EIIS). Each incident is assigned a level of certainty from 0
(unrelated) to 4 (highly probable) that atrazine was a causal factor in the incident. As of the
writing of this assessment, 358 incidents are in EIIS for atrazine spanning the years 1970 to
2005. Most (309/358, 86%) of the incidents involved damage to terrestrial plants, and most of
the terrestrial plant incidences involved damage to crops treated directly with atrazine. Of the
remaining 49 incidents, 47 involved aquatic animals and 2 involved birds. Because the species
included in this effects determination are aquatic species, incidents involving aquatic animals
assigned a certainty index of 2 (possible) or higher (N=33) were re-evaluated. Results are
summarized below, and additional details are provided in Appendix E. The 33 aquatic incidents
were divided into three categories:
119
-------
1. Aquatic incidents in which atrazine concentrations were confirmed to be sufficient to
either cause or contribute to the incident, including directly via toxic effects to aquatic
organisms or indirectly via effects to aquatic plants, resulting in depleted oxygen levels;
2. Aquatic incidents in which insufficient information is available to conclude whether
atrazine may have been a contributing factor - these may include incidents where there
was a correlation between atrazine use and a fish kill, but the presence of atrazine in the
affected water body was not confirmed; and
3. Aquatic incidents in which causes other than atrazine exposure are more plausible (e.g.,
presence of substance other than atrazine confirmed at toxic levels).
The presence of atrazine at levels thought to be sufficient to cause either direct or indirect effects
was confirmed in 3 (9%) of the 33 aquatic incidents evaluated. Atrazine use was also correlated
with 11 (33%) additional aquatic incidents where its presence in the affected water was not
confirmed, but the timing of atrazine application was correlated with the incident. Therefore, a
definitive causal relationship between atrazine use and the incident could not be established. The
remaining 19 incidents (58%) were likely caused by some factor other than atrazine. Other
causes primarily included the presence of other pesticides at levels known to be toxic to affected
animals. Although atrazine use was likely associated with some of the reported incidents for
aquatic animals, they are of limited utility to this assessment for the following reasons:
• No incidents in which atrazine is likely to have been a contributing factor have been
reported after 1998. A number of label changes, including cancellation of certain uses,
reduction in application rates, and harmonization across labels to require setbacks for
applications near waterbodies, have occurred since that time. For example, several
incidents occurred in ponds that are adjacent to treated fields. The current labels require
a 66-foot buffer between application sites and water bodies.
• The habitat of the assessed species is not consistent with environments in which incidents
have been reported. For example, no incidents in streams or rivers were reported.
Although the reported incidents suggest that high levels of atrazine may result in impacts to
aquatic life in small ponds that are in close proximity to treated fields, the incidents are of limited
utility to the current assessment. However, the lack of recently reported incidents in flowing
waters does not indicate that effects have not occurred. Further information on the atrazine
incidents and a summary of uncertainties associated with all reported incidents are provided in
Appendix E.
120
-------
5.0
Risk Characterization
Risk was estimated by calculating the ratio of the EEC to the appropriate toxicity endpoint as
outlined in U.S. EPA (2004). The resulting value is the risk quotient (RQ), which is then
compared to pre-established levels of concern (LOC) for each category evaluated (Appendix F).
The highest EEC and most sensitive acute and chronic toxicity endpoints from laboratory studies
were used to determine the screening level RQ. However, exceedance of one or more LOC does
not necessarily result in a "likely to adversely effect" determination. In cases where RQs exceed
one or more of the established LOCs, additional factors including biological and ecological
factors of the assessed species and additional characterization of potential exposures were used
to characterize the potential for atrazine to affect the assessed species. RQs were initially
calculated for the use that resulted in the highest EEC (sorghum); other uses were evaluated if
RQs based on the highest EEC result in a "likely to adversely affect" determination.
Potential direct effects to the six listed species from use of atrazine in the action area are
evaluated in Section 5.1. Potential indirect effects to the assessed species from direct effects to
animals and plants are evaluated in Section 5.2. The risk characterization approach used in this
assessment to evaluate direct and indirect effects to listed species is endorsed by the Services
(USFWS/NMFS, 2004b).
As previously discussed in the effects assessment, the toxicity of the atrazine degradates has
been shown to be less than the parent compound based on the available toxicity data for
freshwater fish, invertebrates, and aquatic plants; therefore, the focus of the risk characterization
is parent atrazine (i.e., RQ values were not derived for the degradates).
5.1. Direct Effects Assessment
5.1.1. Risk Estimation
5.1.1.1. Shortnose Sturgeon
RQs used to estimate direct effects to the shortnose sturgeon are in Table 5.1. These RQs are
further characterized in Section 5.1.2.
121
-------
Table 5.1. Summary of Aquatic RQs to Assess Potential Direct Effects
to the Shortnose Sturgeon.
Effect
Surrogate
Species
Toxicity
Value
(ug/L)
EEC
(Mg/L)
RQ
Probability
of Individual
Effectab
LOC
Exceedance
Direct Acute
Effects to
Shortnose Sturgeon
Sheepshead
Minnow
LC50: 2000
Peak: 55
0.028
1 in2xlOn
None
Direct Chronic
Effects to
Shortnose Sturgeon
Brook Trout
NOAEC: 65
60-Day: 54
0.83
Not
calculated for
chronic
endpoints
None
a No slope was available for the most sensitive study; therefore, a slope of 4.4 (95% CI of 2.8 - 5.9) was used from
a different acute study in sheepshead minnows of equivalent duration (MRID 43344901).
b Based on the 95% CI on the slope, the probability of an individual effects would be from 1 in 146,000 to 1 in
4xl019.
5.1.1.2. Dwarf Wedgemussel
RQs used to estimate direct effects to the dwarf wedgemussel are in Table 5.2 below. RQs are
further interpreted in the risk description, Section 5.1.2.
Table 5.2. Summary of RC
|s Used to Estimate Direct Effects to the Dwarf Wedgemussel.
Effect
Surrogate
Species
Toxicity Value
fig/L
EEC
fig/L
RQ
Probability of
individual Effect
LOC
Exceedance
Direct Acute
Effects to Dwarf
Wedgemussel
Eastern oyster
EC50: >1700
Hg/La
Peak:55
<0.032
<1 in 4xl010c
None
Direct Chronic
Effects to Dwarf
Wedgemussel
Scud
NOAEC: 60
Hg/Lb
21-Day:
55
0.92
Not calculated for
chronic endpoints
None
a Two studies in the Eastern Oyster were located. No treatment-related effects were observed in either study.
Therefore, the study that tested the highest concentration (1700 |ig/L) was used to estimate risk.
b A chronic study in mussels was not located for this assessment; therefore, the most sensitive chronic invertebrate
NOAEC was used. It is uncertain if use of scud as a surrogate species results in an under or overestimation of risk.
However, toxicity studies in invertebrates that are closer in taxonomy to mussels (e.g., snail and leech; Section
5.1.2.) suggest that use of the scud as a surrogate species is protective.
0 The probability of an individual effect was calculated using a probit dose response slope of 4.4; this is the only
slope for technical grade atrazine reported in available ecotoxicity data for freshwater invertebrates (MRID
45202917, scud). Use of a slope of 4.4 results in a more conservative estimation of the probability of an individual
effect than the default slope recommended in U.S. EPA (2004a) of 4.5.
5.1.1.3. Sea Turtles
RQs used to estimate potential direct effects to sea turtles are provided in Table 5.3 below.
These RQs are further characterized in Section 5.1.2.
122
-------
Table 5.3. Summary of RQs Used to Assess Potential Direct Effects to Sea Turtles.3
Effect
Surrogate
Species
Toxicity Value
EEC
RQ
Probability of
Individual Effect
LOC
Exceedance
Direct
Acute
Effects to
Sea Turtles
Bobwhite
quail and
mallard duck
Dietary LC50:
>5000 ppm
0.47
ppm
<0.01
Not calculated;
sufficient dose-
response not
available.
None
LD50: 940 mg/kg-
bw
Probit slope: 3.8
(95% CI: 2.0-5.7)
0.48
mg/kg-
bwb
<0.01
<1 in 1,000,000
(95% CI: <1 in
1,000,000)
Direct
Chronic
Effects to
Sea Turtles
Bobwhite
quail and
mallard duck
Dietary NOAEC:
225 ppm
0.47
ppm
<0.01
Not estimated for
chronic endpoints
None
a Sea turtles include green, loggerhead, leatherback, and Kemp's ridley sea turtles.
b The dose-based EEC represents addition of the dietary EEC (0.46 mg/kg-bw) + the water flow through EEC (0.01
mg/kg-bw) as presented in Section 3.
5.1.2. Risk Description, Direct Effects
5.1.2.1. Shortnose Sturgeon
RQs were derived using standard laboratory studies and PRZM/EXAMS estimated standard
water body EECs. No acute or chronic concern levels were exceeded for direct effects to fish.
The highest acute RQ for fish was 0.028. At this RQ, the estimated probability of an individual
effect (i.e., mortality) would be 1 in 2xlOn. This analysis is based on an assumption of a probit
dose response relationship with an estimated slope of 4.4 for sheepshead minnows (MRID
43344901). The acute LC50 for sheepshead minnows was 2000 |ig/L. It is recognized that
extrapolation of very low probability events is associated with considerable uncertainty in the
resulting estimates. In order to explore the possible bounds to such estimates, the upper and
lower 95% confidence limits of 2.8 to 5.9 were used to calculate upper and lower estimates of
the effects probability associated with the acute RQ. Probability of an individual effect based on
the upper and lower confidence intervals are 1 in 146,000 to 1 in 4xl019. Based on the lack of
acute and chronic LOC exceedance and the low probability of an individual mortality, atrazine is
not likely to cause direct adverse effects to the shortnose sturgeon.
The highest chronic RQ was 0.83 based on a 60-day EEC of 54 |ig/L and a NOAEC of 65 |ig/L
in brook trout. Although an RQ of 0.83 approaches the chronic LOC of 1.0, the exposure value
used in the RQ calculation is expected to produce a conservative measure of exposure for
habitats of the shortnose sturgeon (major rivers, river mouths). The EECs used to derive chronic
RQs were estimated using PRZM/EXAMS EECs, which is based on a standard water body
scenario. Additional modeling and the available monitoring data presented in Section 3
collectively suggest that long-term EECs used to derive RQs for locations of the shortnose
sturgeon (major river systems and river mouths) are expected to be considerably lower than 54
Hg/L.
123
-------
As discussed in Section 4, several open literature studies raise questions about sublethal effects
of atrazine on plasma steroid levels, behavior modifications, gill physiology, and endocrine-
mediated functions in freshwater fish and anadromous fish. Consideration of the sublethal data
indicates that effects associated with alteration of gill physiology and endocrine-mediated
olfactory functions may occur in salmon at atrazine concentrations lower than the lowest
NOAEC reported from submitted life-cycle studies (Waring and Moore, 2004; Moore and
Lower, 2001). However, there are a number of factors that limit the utility of these studies for
this assessment, which are addressed in detail in Sections A.2.4 of Appendix A. For example,
Moore and Lower (2001) measured olfaction responses in exposed epithelial tissue (after
removal of skin and cartilage) and not intact fish to atrazine, and potential solvent effects could
not be reconciled (i.e., no negative (solvent free dilution water) control was tested).
Furthermore, no quantitative relationship is established between reduced olfactory response
(measured as electrophysiological response) of male epithelial tissue to the female priming
hormone in the laboratory and reduction in salmon reproduction (i.e., the ability of male salmon
to recognize and mate with ovulating females). Also, Waring and Moore (2004) evaluated
survival of salmon in full salinity seawater after atrazine exposure in freshwater. However, the
relevance of direct transfer from freshwater to full-salinity seawater to the assessed species is
questionable given the habitats of the assessed species (Appendix D). Other sublethal effects
observed in fish studies have included behavioral modifications, alterations of plasma steroid
levels, and changes in kidney histology at atrazine concentrations ranging from 5 to 35 |ig/L (see
Section 4). However, a number of uncertainties were also identified with each of the studies,
which are discussed in Section A.2.4 of Appendix A.
In summary, it is not possible to quantitatively link the sublethal effects to the selected
assessment endpoints for the assessed listed species (i.e., survival, growth, and reproduction of
individuals). Also, effects to reproduction, growth, and survival were not observed in the four
submitted fish life-cycle studies at levels that produced the reported sublethal effects (Appendix
A). In addition, there are a number of factors in the design of these studies, which are addressed
in detail in Sections A.2.4a and A.2.4b of Appendix A, that preclude quantitative use of the data
in risk assessment.
Based on the lack of LOC exceedance for acute and chronic effects to the most sensitive fish
species tested in acute and life-cycle studies and PRZM/EXAMS standard water body, the best
available information suggests that atrazine use in the Chesapeake Bay watershed will have "no
effect" on the shortnose sturgeon via direct effects.
5.1.2.2. Dwarf Wedgemussel
No acute or chronic LOCs were exceeded for the dwarf wedgemussel. The acute RQ is based on
an EC50 of >1700 |ig/L. No effects were observed in this study resulting in an acute RQ of
<0.032. The probability of individual effect could not be calculated based on the dose-response
from this study because no effects were observed in the acute toxicity study used in RQ
calculation. Therefore, the probability of an individual effect was calculated using a probit dose
response slope of 4.4; this is the only slope for technical grade atrazine reported in available
ecotoxicity data for freshwater invertebrates (MRID 45202917, scud). Use of a probit slope of
4.4 results in a more conservative estimation of the probability of an individual effect than the
124
-------
default slope recommended in U.S. EPA (2004a) of 4.5. Based on a probit slope of 4.4, the
probability of an individual mortality at an RQ of <0.032 is Approximately 1 in 4xl010. In
addition, data located in the open literature (Johnson et al., 1993) suggest use of the saltwater
Eastern Oyster as a surrogate for the dwarf wedgemussel (a freshwater mussel) in risk estimation
was protective because the EC50 in the only freshwater mussel tested (A. imbecilis) of >60,000
|ig/L is considerably higher than the EC50 used in RQ derivation of 17,000 ug/L in Eastern
Oysters. Based on the lack of acute and chronic LOC exceedance, a "no effect" determination is
made for potential direct effects to dwarf wedgemussels.
Chronic toxicity data in mussels were not located for use in this assessment; therefore, the most
sensitive chronic invertebrate NOAEC was used (60 |ig/L in the scud, MRID 00024377) to
derive RQs. It is uncertain if use of the scud as a surrogate species results in an under or
overestimation of risk. However, acute toxicity studies in invertebrates that are closer in
taxonomy than the scud (phylum arthropoda) to the dwarf wedgemussel (i.e., snail and leech;
phylum mollusca) suggest that use of the scud as a surrogate species is likely protective of acute
effects to the dwarf wedgemussel. Acute LC50 values in both the snail and leech are >16,000
|ig/L (although effects occurred in the leech study after approximately 30 days, see Appendix A),
compared with scud LC50s, which range from 5700 |ig/L (MRID 00024377) to 15,000 |ig/L
(MRID 45202917). The slightly lower LC50s in the scud, compared with the snail and leech,
suggest that use of a chronic NOAEC in the scud is unlikely to result in an underestimation of
risk to the dwarf wedgemussel.
In addition, as previously discussed (Section 3), EECs used to derive RQs likely overestimate
potential long-term exposures to the dwarf wedgemussel. Incorporation of location-specific
factors into modeling including representative flow rate from water bodies where the mussels are
expected to occur (Section 3.3.) together with monitoring data from the Chesapeake Bay
tributaries (Section 3.4) suggest that longer term EECs (days to weeks) are expected to be
considerably lower than the modeled values using the standard water body and are likely in the
low |ig/L range.
Based on the lack of LOC exceedances for acute effects to the Eastern oyster, chronic effects to
the most sensitive invertebrate species tested, and EECs derived from the PRZM/EXAMS
standard water body, the best available information suggests that atrazine use in the Chesapeake
Bay watershed will have "no effect" on the dwarf wedgemussel via direct effects.
5.1.2.3. Sea Turtles
No acute or chronic LOCs are exceeded. All acute RQs are less than 0.01. Although there is
uncertainty associated with the RQs, the methods used to derive surrogate effects endpoints for
reptiles and derivation of EECs for sea turtles is considered protective (conservative). Key
uncertainties are as follows:
1. Dietary exposure to turtles was estimated using the highest BCF of 8.5, which was
reported in fathead minnows and the peak PRZM/EXAMS EEC of 55 |ig/L from the
standard water body. Use of a fish BCF to estimate concentrations in turtle dietary items
such as aquatic invertebrates may result in an over or under estimation of atrazine intake
125
-------
by sea turtles. Bioconcentration of neutral organic chemicals such as atrazine is typically
influenced by lipid content of the bioaccumulating organism. Therefore, food items with
lipid content higher than the fish species used in the BCF study may accumulate more
atrazine, resulting in an underestimation of exposure. Conversely, organisms with lipid
contents lower than the fish species used in the BCF study may accumulate less atrazine,
resulting in an underestimation of exposure. However, given the low magnitude of RQs,
variations in lipid content of dietary items relative to fathead minnows are not expected to
alter the conclusions of this assessment. In addition, the resulting dietary EEC is
considered conservative because dietary EECs for food items were derived based on an
assumption of continuous exposure to atrazine at 55 |ig/L. As discussed in Section 3,
short-term peak atrazine concentrations are expected to be similar to the 55 |ig/L
estimate; however, longer term atrazine concentrations at locations within the Bay where
turtles are expected to feed are estimated to be considerably lower than 55 |ig/L.
2. Water intake data used to estimate EECs were available for only two of the four turtle
species, and food intake levels were only available for one of the four species assessed.
An assumption was made in this assessment that the food and water intake values are
representative of all four turtle species. Given the low magnitude of the RQs and the
conservative nature of the EECs used to derive RQs, differences in food or water intake
levels across the four turtle species are not to likely impact conclusions of this
assessment.
3. Ecotoxicity data from birds were used as surrogates for turtles. Use of birds as a
surrogate species for reptiles is considered protective, as discussed in Section 4.
4. One degradate, DEA, was found to be approximately as toxic as atrazine to birds on an
acute oral basis. DACT has been shown to be of equivalent acute toxicity compared with
atrazine in mammals; however, DACT has not been tested in birds. However, both of
these degradates are expected to form a maximum of 18% of atrazine in the environment
(See Section 2). Even if both degradates of concern were found at concentrations equal
to atrazine (55 |ig/L atrazine, 55 |ig/L DEA, and 55 |ig/L DACT) and assuming
equivalent toxicity for all three degradates, acute and chronic RQs would remain lower
than 0.01, which is well below the level of concern (EEC = 1.4 ppm; LC50 = 940 ppm;
NOAEC = 225 ppm). The assumption that all three compounds are present at
concentrations equal to the peak atrazine concentration is conservative given that the
degradates are expected to have similar physicochemical properties and have been shown
to form no more than 18% of atrazine in available degradation studies. Therefore, this
analysis suggests that quantification of exposure to degradates would have negligible
impact on this assessment.
126
-------
5. A NOAEC of 225 ppm was used for the chronic RQ calculation. However, reduced body
weight gain was observed at 75 ppm (the lowest concentration tested; MRID 42547102).
As discussed in Section 4, the reproduction NOAEC of 225 ppm was considered more
biologically relevant for this assessment. However, use of 75 ppm as the NOAEC in RQ
calculations would not change the risk conclusions, and chronic RQs would remain <
0.01, well below the LOC.
Based on an assumption of a probit dose response relationship with an estimated slope of 3.8,
95% confidence intervals of 2.0 to 5.7 (MRID 00024721), and RQs presented in Table 5.3,
probability of an individual effect based on the slope and the 95% confidence intervals would be
<1,000,000.
Based on the lack of LOC exceedances and the conservative assumptions of exposure to sea
turtles, the best available information suggests that atrazine use in the Chesapeake Bay watershed
will have "no effect" to any of the four sea turtle species assessed via direct effects.
5.1.3. Summary of Direct Effects Conclusions.
Table 5.4. Summary of Direct Effects Determinations to the Six Assessed Listed Species.
Species
Direct Effects Conclusion
Basis for Conclusion
Dwarf wedgemussel
"No effect"
No acute or chronic LOCs
are exceeded
Shortnose sturgeon
"No effect"
No acute or chronic LOCs
are exceeded
Four sea turtle species
"No effect"
No acute or chronic LOCs
are exceeded
5.2. Indirect Effects
Pesticides have the potential to exert indirect effects upon the listed organisms by inducing
changes in structural or functional characteristics of affected communities (U.S. EPA, 2004).
For example, perturbation of forage or prey availability and alteration the extent and nature of
nesting habitat are examples of indirect effects.
In conducting a screen for indirect effects, the direct effects LOCs for each taxonomic group are
used to make inferences concerning the potential for indirect effects upon listed species that rely
upon non-endangered organisms in these taxonomic groups as resources critical to their life
cycle (U.S. EPA, 2004). If no direct effect RQs exceed any LOCs for a taxonomic group
(presented in Section 5.1), then the concern for indirect effects to the assessed species that rely
on the taxonomic group is presumed to be lower than LOCs. If direct effects LOCs are
exceeded, then further analysis on the potential for indirect effects to occur depends on the taxa
for which LOCs were exceeded as described below.
127
-------
When LOCs are exceeded for animals that may be food items of the six assessed animals, there
is a potential for atrazine to indirectly affect the assessed animals by reducing available food
supply. In such cases, the dose-response relationship from the toxicity study used for calculating
the RQ of the surrogate prey item is evaluated to estimate the probability of acute effects
associated with an exposure equivalent to the EEC. The greater the probability that exposures
will produce effects on a taxa, the greater the concern for potential indirect effects for listed
species dependant upon that taxa. Indirect effects RQs were initially calculated using
PRZM/EXAMS EEC and the most sensitive laboratory studies for the broad taxonomic groups
outlined in Section 4. Therefore, when direct effects LOCs were exceeded for food items,
additional analysis was conducted to allow for a determination of potential effects to dietary
items more relevant to the assessed species, and potential exposures were further characterized
for waters more reflective of habitats of the assessed species.
When LOCs are exceeded for plants, then the potential exists for indirect effects to occur from
reduction in food source or habitat alteration. The initial plant LOCs were interpreted using the
following (U.S. EPA, 2004):
• plant RQ < endangered species LOC: a no effect determination to listed species that rely
either on a specific plant species (plant species obligate) or multiple plant species (plant
dependant) for some important aspect of their life cycle are not expected;
• plant RQ > endangered species LOC and < non-endangered species LOC: a no effect
determination is made for listed species that rely on multiple plant species to
successfully complete their life cycle (plant dependent species);
• plant RQ > non-endangered species LOC: potential for adverse effects to listed species
that rely either on a specific plant species (plant species obligate) or multiple plant
species (plant dependant) for some important aspect of their life cycle.
If aquatic plant LOCs are exceeded, further evaluation is conducted to determine whether
effects to plants are likely to result in adverse effects to the assessed species. Further
evaluation included analyses of the geographical and temporal nature of the exposure and
characterization of the biological and ecological requirements of potentially impacted listed
species.
A summary of the methods used to evaluate the potential for atrazine to adversely affect the
six assessed species via indirect effects from potential adverse effects to animals and plants
is discussed below. Methods are consistent with those presented in U.S. EPA (2004).
• Potential indirect effects on the assessed species from direct effects on animal food items
were evaluated by considering the diet of the assessed species, the potential magnitude
of effect to dietary species, and the potential number of species affected relative to the
number of species that serve as dietary items.
• Potential indirect effects on the assessed species from effects on habitat and/or primary
productivity were assessed using RQs based on standard water body EECs and the most
sensitive aquatic plant EC50 as a screen. If aquatic plant RQs exceed any LOC,
potential community level effects were evaluated using community-level effect threshold
128
-------
concentrations, as described in Section 4.6; if standard water body EECs do not exceed
the threshold concentrations, a "may effect, but not likely to adversely affect"
determination is made. However, if EECs based on the standard water body exceed the
threshold concentrations, further characterization of the EECs is performed using
additional modeling and monitoring information for locations where the species is
expected to occur.
• The potential for atrazine to affect the assessed species indirectly by affecting riparian
zones in the Chesapeake Bay watershed and areas adjacent to the habitat or spawning
areas of the assessed species is evaluated using submitted terrestrial plant toxicity data
and preliminary studies on woody plants. If terrestrial plant RQs exceed the LOC for
direct effects to non-endangered species based on EECs derived using Terrplant (Version
1.2.1) and submitted guideline terrestrial plant toxicity data, a conclusion that atrazine
may affect the assessed species is made. Further analysis of the potential for atrazine to
affect the assessed species via reduction in riparian habitat includes an evaluation of the
magnitude of the potential effect to riparian habitat, type of riparian area most vulnerable
to atrazine, and relevance of sensitive riparian zones to water quality in the Chesapeake
Bay watershed.
5.2.1. Summary of Biological and Ecological Information Used to Evaluate Potential
Indirect Effects of Atrazine
Location and dietary information on the shortnose sturgeon, dwarf wedgemussel, and the four
sea turtles used to perform the indirect effects assessment are summarized below. These data are
summaries of information provided in Section 2.2 and Appendix D. Additional information can
be obtained from those sections.
5.2.1.1. Shortnose Sturgeon
Data for the shortnose sturgeon were primarily obtained from NMFS (1998); U.S. EPA (2003b);
and Gilbert, 1989).
Diet: Shortnose sturgeon are non-selective continuous benthic omnivores. Dietary items consist
of insect larvae, worms, and mollusks; however, the dietary preferences appear to change with
age. Insect larvae (e.g. Hexagenia, Chaobrus, and Chironomus), and small crustaceans (e.g.
Gammarus, Asellus, and Cyathura) are the predominate food items for juveniles (NMFS, 1998).
Adults feed primarily on small mollusks. In freshwater, these mollusks include Physa,
Helisoma, Corbicula, Amnicola, Valvata, Pisidium, and small Elliptio (NMFS, 1998). In saline
areas, molluscan prey include small Mya and Macoma. Recent data show that adult sturgeon
feed on gammarid amphipods and zebra mussels (NMFS, 1998). Juveniles are located in
freshwater systems from hatching until adulthood (approximately 3 to 7 years); therefore, only
toxicity data in freshwater organisms were used to assess potential effects to juvenile sturgeon
prey items. Both marine and freshwater toxicity data were used to assess potential effects to
adult food items because they may reside in freshwater or low salinity areas of the Chesapeake
Bay.
129
-------
Habitat: Shortnose sturgeon are found primarily in the main stems of larger rivers and river
mouths and the Chesapeake Bay. They reproduce in deeper freshwater rivers with a swift
current and remain in this environment until maturity, when they migrate to mouths of rivers
with slow to no current.
Reproduction: Shortnose sturgeon depend on free-flowing rivers and seasonal floods to provide
suitable spawning habitat. For shortnose sturgeon, spawning grounds have been found to consist
mainly of gravel or rubble substrate in regions of fast flow. Flowing water provides oxygen,
allows for the dispersal of eggs, and assists in excluding predators. Seasonal floods scour
substrates free of sand and silt, which might suffocate eggs (U.S. EPA, 2003b).
Shortnose sturgeon spawn in upper, freshwater sections of rivers and feed and overwinter
in both fresh and saline habitats. In populations that have free access to the total length of a river
(absent of dams), spawning areas are located at the farthest accessible upstream reach of the
river, often just below the fall line (U.S. EPA, 2003b). Tributaries of the Chesapeake Bay that
appear to have suitable spawning habitat for the shortnose sturgeon include the Potomac,
Rappahannock, James, York, Susquehanna, Gunpowder and Patuxent rivers (U.S. EPA, 2003b).
Other scientists believe that very little, if any, suitable spawning habitat remains for shortnose
sturgeon, due to past sedimentation in tidal freshwater spawning reaches (U.S. EPA, 2003b).
"able 5.5. Summary of Shortnose Sturgeon Dietary Items
Life Stage
Location
Dietary items
Examples
Surrogate Species
with Toxicity
Data
Range of
Toxicity Values
(ug/L)
Juveniles
Freshwater
Insect larvae
Hexagenia,
Chironomus,
Chironomus
Rivers
Chaobrus, and
Chironomus
stonefly
LC50: 720 -
>33,000
Stonefly: 6700
Small
Gammarus,
Scud, waterflea
Scud LC50:
crustaceans
Asellus, and
Cyathura
4700 to 15,000
Waterflea EC50:
3500 to >30,000
Adults
Low
Freshwater
Physa,
Snail (Ancylus
Acute Snail and
salinity
Mollusks
Helisoma,
fluviatilis), leech
Leech: LC50:
areas
Corbicula,
>16,000
including
Amnicola,
river
Valvata,
mouths
Pisidium, and
small Elliptic*.
Saltwater
Small Mya
Eastern Oyster
EC50: >1700
Mollusks
(a soft shelled
clam) and
Macoma
(a clam)
130
-------
5.2.1.2. Dwarf Wedgemussel
Primary data sources on the dwarf wedgemussel include USFWS (1993) and U.S. EPA (2003b).
Diet: Little specific information is available on the food items of the dwarf wedgemussel. In
general, mussels are filter-feeders that feed on detritus (dead organic matter), zooplankton, and
phytoplankton. Phytoplankton is typically considered to be the most important food item for
sustenance of mussels. Waterfleas and algae were used as surrogate food species for freshwater
invertebrates and zooplankton, respectively.
Habitat: The dwarf wedgemussel is a freshwater mussel that lives on muddy sand, sand, and
gravel bottoms in creeks and rivers of varying sizes. Its habitat is also characterized by slow to
moderate current with little silt deposition.
Several locations within the Chesapeake Bay watershed are known for the dwarf wedgemussels.
These locations were described in Section 2 (Table 2.3) and include waters in the Potomac River
drainage (Mcintosh Run, Nanjemoy Creek, and Aquia Creek), the York River drainage (South
Anna and Po Rivers), the Tuckahoe Creek Drainage (Norwich Creek, Long Marsh Ditch,
Mason's Branch), the Southeast Creek watershed (Browns branch, Granny Finley, Corsica River
tributary, and Southeast Creek tributary), and the Rappahannock River drainage (Rappahannock
River and Carter Run). All locations are small, flowing streams that are consistent with the
description of headwater streams or mid-level reaches presented in Section 2.4.
5.2.1.3. Sea turtles
A summary of the information on sea turtle's diet and habitat is presented in this section. These
data are a summary of information presented in Section 2.2 and Appendix D and were obtained
from a number of sources including the recovery plans for the four sea turtle species and
information from the Virginia Institute of Marine Science (VIMS), the Chesapeake Bay
Program, and Fish and Wildlife Services. Specific references are presented in Section 2.2 and in
Appendix D.
Diet: The diet of each sea turtle is described in detail in Appendix D and summarized in Table
5.6. Loggerhead and Kemp's ridley sea turtles eat food items including crustaceans, plants,
mollusks, other invertebrates, and fish. Leatherback turtles consume mainly jelly fish, but also
ingest other invertebrates. Green turtles, however, eat invertebrates as hatchlings, but feed
almost exclusively on aquatic plants (i.e., sea grasses) as adults.
131
-------
Table 5.6 Summary of Dietary Items of the Assessed Sea Turtles
Species
Food Item
(taxa)
Food Item
(common name)
Surrogate Species with Toxicity Data
Loggerhead and
Kemp's ridley
Chelicerate
Horseshoe crab
Mud crab
Brown shrimp
Grass shrimp
Pink shrimp
Crustacean
Blue crab
Hermit crab
Mantis shrimp
Plant
Eelgrass
Widgeon grass
Aquatic plant EC50; Community level effects
thresholds
Invertebrate
Jellyfish
Other invertebrates
including sea urchins
No adequate surrogate species; Most sensitive
species (copepod) was used for initial screen.a
Mollusk
Oysters and clams
Other bivalves
Eastern oyster
Fish
Various Species
Atlantic menhaden
Spot
Atlantic croaker
Bluefish
Striped bass
Oyster toadfish
Sheepshead minnow
Brook trout
Leatherback
Invertebrates
Pink comb (jellyfish)
Sea walnut (jellyfish)
Other jellyfish
Other invertebrates
No adequate surrogate species; Most sensitive
species (copepod) was used for initial screen.3
Green, Adults
Plants
Algae
Sea grass
Aquatic plant EC50; community level effects
thresholds
Green, Juveniles
Invertebrates
Variety of invertebrates
Most sensitive species (copepod) was used for
initial screen.3
" Distribution of toxicity values in aquatic invertebrates was used to characterize potential risks if LOCs were
exceeded based on use of the copepod LC50 in RQ calculations.
Habitat: A summary of the expected habitats of the four sea turtles included in this assessment
is in Section 2. Generally, these turtles are expected to be found in the main portion of the Bay,
but may also be found in main stems of major rivers, river mouths, and estuarine inlets. A more
thorough description of the expected locations of these turtles is in Appendix D.
5.2.2. Evaluation of the Potential for Atrazine to Induce Indirect Effects on the Shortnose
Sturgeon, Dwarf Wedgemussel, and Sea Turtles from Reduction in Animal Food
Items
5.2.2.1. Potential Effects to Aquatic Invertebrate Food Items
Potential effects to the six assessed species from reduction in food (aquatic animals) availability
are presented below. As shown in Table 5.7, aquatic animals consumed by the six species
132
-------
assessed include fish, a variety of aquatic invertebrates, and zooplankton. However, an indirect
effects assessment was not performed from direct acute or chronic effects to fish or chronic
effects to aquatic invertebrates, because the direct effects assessment presented in Section 5.1
was conducted using all available toxicity data for these endpoints, and a concern for direct
effects was not identified. Therefore, indirect effects from potential direct effects to these
surrogate species are also presumably lower than LOCs. Potential indirect effects to the six
assessed listed species from potential effects to aquatic plants are evaluated in Section 5.2.4.
Table 5.7. Summary of Animal Prey Items of the Six Assessed Species
Species
Life
Stage
Location
Animal Dietary Items
Surrogate Species with
Available Toxicity Data3
Shortnose
sturgeon
Juveniles
Freshwater rivers
Insect larvae
Chironomus, stonefly
Small crustaceans
Scud, waterflea
Adults
River mouths, main stem of
the Chesapeake Bay, and
large rivers
Freshwater mollusks
Snail, leech
Saltwater mollusks
Eastern oyster
Dwarf
wedgemussel
Allb
Freshwater headwater
streams and mid-level
reaches
Freshwater zooplankton
Daphnia
Sea turtles
All
Main stem of the Chesapeake
Bay, major rivers, river
mouths, estuarine inlets
Crustaceans, mollusks,
other invertebrates, and
fish
Brown shrimp, pink
shrimp, grass shrimp, mud
crabs, oysters, fish
a Copepods, the most sensitive aquatic invertebrate tested, were not considered an appropriate surrogate food item
for any of the assessed species. However, the lowest copepod LC50 was used for risk estimation because copepods
were the most sensitive invertebrate species tested as outlined in U.S. EPA (2004). Toxicity data on more
appropriate dietary invertebrate species were used to further characterize potential risks to the assessed species if
RQs based on the lowest copepod LC50 exceed acute LOCs.
b Glochidial stage receives sustenance from fish. No direct effects RQs exceeded LOCs for fish; therefore, indirect
effects to glochidial stage mussels from potential effects to fish are also presumably lower than LOCs.
RQs initially used to screen whether atrazine may indirectly affect the six listed species
considered in this assessment via reduction in available prey were based on acute ecotoxicity
data from the most sensitive invertebrates tested and the PRZM/EXAMS estimated peak EEC of
55 |ig/L in the standard water body. Results of this analysis are presented in Table 5.8.
Table 5.8. Summary of RQs Used to Estimated Indirect Effect to Shortnose Sturgeon and Sea Turtles via
Potential Direct Effects on Dietary Items
Surrogate
Food Item
Assessed
Species for
Indirect
Effects
Toxicity
Value
(Mg/L)
EEC
Oig/L)
RQ
Probability of
individual Effect"
Risk Interpretation
Aquatic
invertebrates,
Copepod
Shortnose
sturgeon;
Sea turtles
LC50: 88
55
0.62
1 in 2
Availability of aquatic
invertebrate food items may
be affected by atrazine use.
a Probability based on the assumption of a probit dose-response relationship and an estimated slope of 0.95 (MRID
45202918)
133
-------
Table 5.8 indicates that atrazine may affect the shortnose sturgeon and sea turtles via potential
direct effects on sensitive aquatic invertebrate food items. However, this analysis was based
only on the most sensitive aquatic invertebrate species in laboratory studies and did not consider
specific dietary characteristics of the assessed species or the expected location of the species
within the Bay and its source waters. Therefore, additional characterization of the potential for
atrazine to affect aquatic invertebrate food items of the six assessed species is presented below.
The potential for atrazine to elicit indirect effects via effects on food items is dependent on
several factors including: (1) the potential magnitude of effect on invertebrate populations; and
(2) the number of prey species potentially affected relative to the number of prey species needed
for sustenance. Together, these data provide a basis to evaluate whether a sufficient number of
individuals within a prey species and the number of prey species may be reduced such that an
adverse effect to the shortnose sturgeon, dwarf wedgemussel, or sea turtles is likely, unlikely, or
unable to be determined. Therefore, the sensitivity of all aquatic invertebrates to atrazine and the
types of organisms consumed by the assessed listed species were considered.
Table 5.9 below presents RQs for surrogate food items that are representative of dietary items of
the assessed species. The sensitivity of all aquatic invertebrates tested to atrazine is represented
in Figure 5-1. This analysis considers only acute risk to aquatic invertebrate food items. Acute
and chronic RQs for fish and chronic RQs for invertebrates were lower than LOCs for direct
effects; therefore, potential indirect effects to listed species from direct effects on these endpoints
were presumably lower than LOCs. Although marine copepods were the most sensitive aquatic
invertebrate tested, more suitable surrogate food items for the assessed species were used in the
refined analyses (Table 5.9).
134
-------
Table 5.9. Summary of RQs Used to Assess Potential Risk to Animal Food Items of the
Shortnose Sturgeon, Dwarf Wedgemussel, and Four Sea Turtles.
Surrogate
Food
Item
Species
Listed Species
That Receive
Sustenance
From Food
Item Or
Similar Food
Item
Acute
Toxicity
Value Range
(No. of
Studies)
RQ Range
Based on an
EEC of 55
Qig/L)
Probability of
Individual
Effect3
Risk Interpretation
Midge
Juvenile
shortnose
sturgeon
720 - >33,000
(3)
<0.01 -
0.076
<1 in
l,000,000a
Based on LOC exceedance,
atrazine may affect food
items that are as sensitive as
the midge; however, the
magnitude of potential effects
on food availability (<1 in
1,000,000) is not likely
sufficient to induce indirect
effects to juvenile shortnose
sturgeon.
Brown
shrimp
Sea turtles
1000 (1)
0.055
<1 in
l,000,000c
Based on LOC exceedance,
atrazine may affect food
items that are as sensitive as
the brown shrimp; however,
the magnitude of potential
effects on food availability
(<1 in 1,000,000) would not
likely be sufficient to induce
indirect effects to sea turtles.
Mysid
shrimp
Juvenile
shortnose
sturgeon
1000 - 5400
(2)
0.01-0.055
<1 in
l,000,000c
Based on LOC exceedance,
atrazine may affect food
items that are as sensitive as
the mysid shrimp; however,
the low probability of an
individual effect suggests that
the magnitude of potential
effects would not likely be
sufficient to induce indirect
effects to predators.
Waterflea
Dwarf
wedgemussel;
juvenile
shortnose
sturgeon
3500 -
>30,000 (5)
0.02
<1 in
l,000,000a
Based on low probability of
individual effects and lack of
LOC exceedance, atrazine is
not likely to affect food items
that are as sensitive as scud
or waterfleas to the extent
that indirect effects on
predators are expected
Scud
Juvenile
shortnose
sturgeon
5700 - 15,000
(3)
0.01
<1 in
1,000,000
135
-------
Table 5.9. Summary of RQs Used to Assess Potential Risk to Animal Food Items of the
Shortnose Sturgeon, Dwarf Wedgemussel, and Four
Sea Turtles.
Surrogate
Food
Item
Species
Listed Species
That Receive
Sustenance
From Food
Item Or
Similar Food
Item
Acute
Toxicity
Value Range
(No. of
Studies)
RQ Range
Based on an
EEC of 55
Qig/L)
Probability of
Individual
Effect3
Risk Interpretation
Adult
Not estimated
Eastern
shortnose
>1000 - >1700
<0.03 -
based on lack
oyster
sturgeon; sea
turtles
(2)
<0.055 b
of dose-
response
Adult
shortnose
sturgeon
Not estimated
Leech
>16000 (2)
<0.01
based on lack
of acute dose-
response
Based on lack of LOC
Not estimated
exceedance and/or lack of
Mud crab
Sea turtles
>1000(1)
<0.055b
based on lack
of dose-
response
effects in the study at highest
concentrations tested, atrazine
is not likely to affect food
Snail
Adult
shortnose
sturgeon
>16,000 (1)
<0.01
Not estimated
based on lack
of dose-
response
items that are as sensitive as
these organisms to the extent
that indirect effects to
predators would be expected
Grass
shrimp
Sea turtles
9000 (1)
<0.01
<1 in
l,000,000c
Pink
shrimp
Sea turtles
6900 (1)
<0.01
<1 in
l,000,000c
Juvenile
<1 in
l,000,000c
Stonefly
shortnose
sturgeon
6700 (1)
<0.01
a The probability of an individual effect was calculated using a probit dose response slope of 4.4 (MRID 45202917,
scud); this is the only slope for technical grade atrazine reported in available ecotoxicity data for freshwater
invertebrates. 95% Confidence intervals could not be calculated based on the available data (Table A-18).
b No effects were observed in the mud crab or the eastern oyster studies, and the EECs used in risk estimation were
considered conservative; therefore, these slight LOC exceedances, should be interpreted with caution.
c Dose-response based on probit slope of 4.5 from study in mysid shrimp (MRID 43344902).
136
-------
Genus Leusl Geometric Means of Reported Axite LC50 and EC50 Values in
Aquatic Invertebrates for Atrazine
17000
16000
15000
14000
13000
12000
11000
LC/EC5010000
(ug/L) 9000
8000
7000
6000
5000
4000
3000
2000
1000
0
<5\
_c
CO
n
o
-------
for technical grade atrazine reported in available ecotoxicity data for freshwater invertebrates and
results in a more conservative analysis than use of the default slope of 4.5 recommended by U.S.
EPA (2004a).
Based on the lack of LOC exceedance for the surrogate freshwater zooplankton species used in
this assessment, the best available data suggests that atrazine use in the Chesapeake Bay
watershed is expected to have "no effect" to the dwarf wedgemussel via reduction in
zooplankton as food supply.
Shortnose Sturgeon
Shortnose sturgeon are omnivores and continuously feed on benthic and epibenthic invertebrates
including mollusks, crustaceans, and oligochaete worms (U.S. EPA 2003b). Table 5.9 indicates
that the endangered species LOC was exceeded for several surrogate food species for juveniles
(midge, RQ = 0.076; mysid shrimp, RQ = 0.055).
The probability of an individual mortality to food items as sensitive as the midge was estimated
using a probit dose response slope of 4.4 (MRID 45202917); this is the only slope for technical
grade atrazine reported in available ecotoxicity data for freshwater invertebrates. Based on an
assumption of a probit dose response relationship with a mean estimated slope of 4.4, the
probability of an individual effect to this food item species at an RQ of 0.076 would be
approximately 1 in 2,000,000.
In addition, the endangered species LOC is exceeded for mysid shrimp (RQ = 0.055). Based on
the assumption of a probit dose-response relationship and a mean estimated slope of 4.5 in mysid
shrimp (MRID 43344902), the probability of an individual effect at an RQ of 0.055 is 1 in lxlO8
(represented by <1 in 1,000,000 in Table 5.9). No LOCs were exceeded for the other juvenile
shortnose sturgeon dietary organisms or for any surrogate adult shortnose sturgeon dietary
organisms tested including mollusks (snail, leech, oyster), small crustaceans (daphnia), and
stoneflies (Table 5.9).
The EECs used to estimate potential effects to invertebrate food items were based on
PRZM/EXAMS estimated values using the standard water body, which are considered to be
representative of short-term atrazine concentrations in headwater streams and small estuarine
inlets. However, juvenile sturgeon are located in relatively deep river channels with high flow
rates, and adult sturgeon are located in river mouths and in the main stem of the Chesapeake
Bay. Based on additional modeling exercises and monitoring data presented in Section 3.4,
EECs in these locales are likely to be lower than those estimated using the standard water body
scenario; therefore, the results of this indirect effects analysis are considered conservative.
Based on the non-selective nature of feeding behavior, the conservative nature of the EECs used
to derive RQs, and the low estimated magnitude (<1 in 1,000,000) of anticipated effects on the
availability of food items as sensitive as those with LOC exceedances, the availability of dietary
items of juvenile or adult shortnose sturgeon is not likely to be affected to an extent that would
constitute a "take", as defined in Section 2.1. Therefore, the data suggest that the significance of
potential effects to the shortnose sturgeon from reduction in food is negligible, and that atrazine
138
-------
is "not likely to adversely affect" the shortnose sturgeon via effects on aquatic animals as food
supply.
Sea Turtles
Loggerhead, leatherback, Kemp's ridley, and juvenile green turtles consume a variety of
crustaceans, mollusks, jellyfish, other invertebrates, and fish. The endangered species level of
concern was exceeded for one surrogate food item, brown shrimp (RQ = 0.055). Sufficient dose-
response data were not available to allow for an evaluation of the probability of an individual
effect to the brown shrimp. However, based on the lowest reported slope for surrogate saltwater
aquatic invertebrate food items of 4.5 (mysid shrimp, MRID 43344902) and an assumption of
probit dose-response relationship, the probability of an individual effect at an RQ of 0.055 would
be <1 in 1,000,000 for brown shrimp. LOCs were not exceeded for other surrogate species
tested including grass shrimp, pink shrimp, Eastern oysters, fish, or mud crabs. Additional
formulated product data suggest that effects to fiddler crabs would also not be expected at the
EECs derived in this assessment. The LC50 in fiddler crabs was >198,000 |ig/L (MRID
00024395; described in Appendix A).
Each of the four sea turtle species assessed are located and feed in a variety of locations within
the Chesapeake Bay, including main stem of major rivers, river mouths, the main stem of the
Bay, and estuarine inlets. The EECs derived using the PRZM/EXAMS water body scenario of
55 |ig/L may be representative of short-term peak exposures in headwater streams and estuarine
inlets, but likely overestimates exposure and risk to invertebrates found in other areas of
Chesapeake Bay where turtles are expected to be located (main stem of the Bay, major rivers,
and river mouths).
Therefore, the above analysis based on a conservative exposure estimate suggests that a low
number of dietary species relative to the number of species that serve as food to sea turtles are
may be impacted by atrazine use in the Chesapeake Bay watershed. The magnitude of potential
effects to the single dietary species (brown shrimp) with an LOC exceedance (RQ = 0.055) is
low (estimated probability of an individual effect was <1 in 1,000,000) such that indirect effects
to predators would be insignificant. Consequently, potential effects to the assessed sea turtle
species from potential effects to aquatic animals as food items are anticipated to be negligible.
Based on insignificant magnitude of potential effects to sea turtles from potential reduction in
food supply organisms in the Chesapeake Bay, atrazine use in the Chesapeake Bay watershed is
not likely to adversely affect sea turtles.
139
-------
5.2.3. Summary of Effects Determinations: Indirect Effects from Direct Effects to Aquatic
Animals
Table 5.10. Summary of Indirect Effects Determinations to the Six Assessed Listed
Species Resulting from Direct Effects to Aquatic Animals
Species
Inirect Effects Conclusion
Basis for Conclusion
Dwarf wedgemussel
"No effect"
No acute or chronic LOCs are
exceeded for surrogate fish or
freshwater zooplankton species.
Shortnose sturgeon
"May affect, but not likely to
adversely affect"
Some food items of these species
may be affected; however, the
significance of such effects to the
shortnose sturgeon and to the four
sea turtles are considered negligible
based on the low anticipated
magnitude of such an effect on the
available food supply and the
conservative exposure assumptions
used in this analysis.
Four sea turtle species
"May affect, but not likely to
adversely affect"
5.2.4. Evaluation of Potential Indirect Effects to the Six Listed Species from Potential
Effects to Aquatic Plants
Potential indirect effects to the six assessed listed species from effects on habitat and/or primary
productivity were assessed using RQs based on the most sensitive aquatic plant study and
PRZM/EXAMS estimated EECs from the standard water body scenario as a screen. If aquatic
plant RQs exceed the LOC of 1.0, potential community level effects were evaluated using
threshold concentrations for community level effects, as described in Section 4.
RQs used to estimate potential indirect effects to shortnose sturgeon, dwarf wedgemussel, and
sea turtles based on primary productivity effects and/or decrease in available plants as food are in
Table 5.11. Aquatic plants serve as important food sources for the dwarf wedgemussel and adult
green turtles.
140
-------
Table 5.11 Summary of RQs Used to Estimate Indirect Effects to Shortnose
Sturgeon, Dwarf Wedgemussel, and Sea Turtles via Direct Effects on Aquatic
Plants.
Surrogate Species
Toxicity Value
(Hg/L)
mrii)
EEC
fag/L)
RQ
Risk Interpretation
Freshwater Plants
Atrazine may affect
the assessed listed
species via effects on
aquatic plants; further
analysis of potential
risks is necessary.
Non-vascular plants,
green algae
EC50: 1 (ig/L
00023544
55 ng/L
55
Vascular plants,
Cyanophyceae
Anabaena cylindrica
EC50: 37 (ig/L
43074804
55 ng/L
1.5
Marine Plants
Marine algae,
Isochrysis galbana
EC50: 22 ng/L
41065204
55 ng/L
2.5
Sago pondweed
EC50: 7.5 ng/L
45088231
55 ng/L
7.3
Based on the results in Table 5.11, atrazine may indirectly affect each of the six assessed listed
species via effects on aquatic plants. However, this analysis was based on the most sensitive
aquatic plant species tested and a PRZM/EXAMS EEC generated based on the standard water
body scenario. No known obligate relationship exists between any single plant species and the
assessed listed species; therefore, additional analyses were performed to determine whether
potential effects to individual plant species would likely result in community level effects. As
previously discussed in Section 4, threshold concentrations were determined from realistic and
complex time variable atrazine exposure profiles (chemographs) for modeled aquatic community
structure changes (see Section 4.6 and Appendix B). If the following threshold concentrations
are exceeded based on the EECs presented in Section 3 for the standard water body, then EECs
based on the expected location of the assessed species within the Chesapeake Bay are further
characterized. Exceedance of the following threshold concentrations indicates that changes in
the aquatic plant community structure are possible.
• 14-day average = 38 |ig/L
• 30-day average = 27 |ig/L
• 60-day average =18 |ig/L
• 90-day average =12 |ig/L
The only uses that resulted in exceedance of the preceding thresholds are corn, sorghum, and
fallow/idle lands; 14-, 30-, 60-, and 90-day average concentrations for these uses are in Table
5.12.
141
-------
Table 5.12. Summary of PRZM/EXAMS Output for Corn, Sorghum, and Fallow/Idle
Land and Comparison of Estimated Atrazine Concentrations to Community Level
Effects Thresholds
Use Site
14-Day
EEC
(Jig/L)
14-Day
Effect
Threshold
30-Day
EEC
(Jig/L)
30-Day
Effect
Threshold
60-Day
EEC
(Jig/L)
60-Day
Effect
Threshold
90-Day
EEC
(Jig/L)
90-Day
Effect
Threshold
Corn
46.9
38
46.5
27
45.6
18
44.4
12
Sorghum
54.8
54.3
53.7
52.5
Fallow/
idle land
45.0
45.0
45.0
44.9
The above uses result in EECs that exceed the 14-, 30-, 60-, and 90-day thresholds for
community level effects; however, the EECs from this analysis were estimated using
PRZM/EXAMS and the standard water body scenario. As previously discussed in Section 3.3,
the standard water body may not accurately represent EECs in expected locations of the assessed
species. Therefore, additional information on the location of the assessed species was used to
further characterize potential exposures relative to those presented for the standard water body
scenario. This analysis was presented in detail in Section 3.2 and is summarized below for each
of the assessed species.
5.2.4.1. Additional Characterization of EECs in Flowing Streams and Rivers
The six species considered in this assessment are located in headwaters with low to moderate
flow (mussel); in larger rivers (shortnose sturgeon and sea turtles); at river mouths (shortnose
sturgeon and sea turtles); in the main stem of the Chesapeake Bay (sea turtles); and in estuarine
inlets (sea turtles). Outside of short-term concentrations in small estuarine inlets and headwater
streams, none of the locations where the assessed species are located are likely well represented
by the standard water body, which was used to derive EECs used in RQ calculations. The
primary reason that long-term concentrations estimated using the standard water body are not
representative of these environments is that creeks and rivers are flowing water bodies, inlets and
the Bay are subject to extensive mixing. In contrast, the standard water body is a static water
body.
As described in detail in Section 3.2, a number of additional modeling exercises were performed
to allow for characterization of potential effects of flow rate on the EECs. This analysis, together
with available monitoring data, was used to further characterize potential exposures to the habitat
of the listed species.
First, the variable volume water model (VVWM) was used to account for the influence of input
and output (flow) on model predictions. Two alternate model runs were conducted using the
VVWM. The first was done using standard assumptions and environmental fate parameters
142
-------
generally consistent with the non-flowing standard water body. The second assumption was
designed to represent a larger volume water body that maximizes flow into the water body.
Second, the impact of flow was characterized using the Index Reservoir as the receiving water
body and various flow rates. Flow assumptions considered representative of the headwater
streams and mid-level reaches where the dwarf wedgemussel is located were evaluated. The
EECs in larger rivers would be expected to be lower than those estimated for headwater streams
due to higher flow rate and greater dilution potential.
In addition to the modeling exercises, existing monitoring data were used to characterize atrazine
concentrations in the Chesapeake Bay and its tributaries. A detailed description of these data are
in Section 3.4. Table 5.13 below provides a brief summary of the key results from the additional
modeling and monitoring data used to characterize potential exposures.
Table 5.13. Characterization of Exposures Based on Additional Modeling and
Available Monitoring Data
Analysis
Results
Modeling using WWM
EECs are below the 14- and 30-day community level effects
thresholds, but above the 60- and 90-day community-level
threshold concentrations.
Modeling using the Index
Reservoir and various flow rates
EECs decreased as flow rate increased. Modeling with flow rates
representative of dwarf wedgemussel locations results in EECs
that are lower than all community-level threshold concentrations.
Monitoring data, Chesapeake Bay
and its tributaries
The maximum atrazine level was 30 |ig/L. while the 99th, 95th,
90th, 75th, and 50th percentile values were 2.5 |ig/L. 0.5 |ig/L. 0.28
Hg/L, 0.1 ng/L, and 0.05 ng/L, respectively.
Monitoring, other representative
water bodies
High peak atrazine concentrations have been observed; however,
longer-term (>14 days) durations (when the data allow for
calculation) are in the low |ig/L range.
Collectively, the alternative modeling exercises and the monitoring data discussed in Section 3
suggest that atrazine concentrations in the headwater streams, rivers, river mouths, and
Chesapeake Bay are expected to be in the low |ig/L range, and lower than the 14-, 30-, 60-, and
90-day threshold concentrations for community-level effects. Therefore, community level
effects to aquatic plants are considered unlikely in the Chesapeake Bay. No obligate relationship
between the assessed species and any single aquatic plant species is known to exist. Therefore, it
is concluded that atrazine is not likely to adversely affect any of the six assessed listed species
via effects to aquatic vegetation.
There is additional concern, however, for green turtles from effects to aquatic plants because
green turtles are primarily herbivores and may be found in minor estuarine inlets where atrazine
concentrations may be higher than concentrations in the Bay, main rivers, and river mouths.
However, due to extensive mixing within these inlets, peak atrazine concentrations that may
occur after a run-off event are expected to be diluted rather quickly, such that any effects to
aquatic vegetation would be anticipated to be temporary. Also, estuarine sea grasses, a forage
item of green turtles, was shown to be less sensitive to atrazine than those used to derive aquatic
plant risk quotients. EC50s for sea grasses range from approximately 70 |ig/L (MRID
45227729) to 30,000 |ig/L (MRID 45205101) in laboratory studies, which result in RQs that are
143
-------
less than LOCs. In addition, recovery of aquatic plants is expected to occur from the short-term
exposures to atrazine within these minor inlets (Appendix A). Also, green turtles are highly
mobile and are transitory at any single location within the Bay, and these minor inlets are not
expected to be important to the green turtle's feeding or reproduction. Therefore, potential
temporary effects to aquatic vegetation are not likely to result in harm or harassment of green
turtles. This observation combined with the transient nature of atrazine within these inlets
caused by extensive water mixing and the presence of a large pool of food within the Bay as it
relates to the small areas potentially impacted (See Appendix G for map of submerged aquatic
vegetation in the Chesapeake Bay), suggests that atrazine is not likely to adversely affect the
green turtle via potential effects to food supply.
Table 5.14. Summary of Indirect Effects Determinations to the Six Assessed Listed
Species Resulting from Effects to Aquatic Plants
Species
Direct Effects Conclusion
Basis for Conclusion
Dwarf wedgemussel
"May affect, but not likely to
adversely affect"
Individual aquatic plant species
within the Chesapeake Bay
watershed may be affected.
However, atrazine concentrations
are not anticipated to exceed
community-level effect threshold
concentrations, and no known
obligate relationship between the
assessed species and any single
aquatic plant species exists.
Therefore, potential effects are
considered to be insignificant in the
context of a take as defined in
Section 2.1.
Shortnose sturgeon
"May affect, but not likely to
adversely affect"
Four sea turtle species
"May affect, but not likely to
adversely affect"
5.2.5. Potential Indirect Effects to the Listed Species via Direct Effects to Terrestrial
Plants
Riparian plants beneficially affect water and stream quality in a number of ways in both adjacent
stream reaches and areas downstream of the riparian zone. A general discussion of riparian
habitat and its relevance to the assessed species is provided below followed by a discussion of
potential risks to the assessed species caused by effects on riparian areas from use of atrazine in
the Chesapeake Bay watershed.
5.2.5.1. Discussion of Riparian Habitat and Its Relevance to the Assessed Species
Riparian vegetation serves several functions in the stream ecosystem including: serving as an
energy source; providing organic matter to the stream; providing shading, which ensures thermal
stability of the stream; and service as a buffer filtering out sediment, nutrients, and contaminants
before they reach the stream. Criteria based largely on professional judgment have been
proposed to assess the health of riparian zones and their ability to support fish habitat that may
be used to assess the health of riparian zones (Fleming et al. 2001). These criteria are in Table
5.15 below. General criteria are identified for the width of vegetated area (i.e. distance from
cropped area to water), structural diversity of vegetation, and canopy shading.
144
-------
Table 5.15. Semi-quantitative Criteria Related to Riparian Vegetation for Assessing the
Health of Riparian Areas for Supporting Aquatic Habitats.1
Criteria
Quality
Excellent
Good
Fair
Poor
Buffer width
>18m
12- 18m
6 - 12m
<6m
Vegetation
diversity
>20 species
15-20 species
5-14 species
<5 species
Structural
diversity
3 height classes
grass/shrub/tree
2 height classes
1 height class
sparse
vegetation
Canopy shading
mixed sun/shade
sparse shade
90% sun
no shade
1 Adapted from
"leming etal. 2001.
Additional discussion of the importance of riparian areas in the Chesapeake Bay and its source
waters can be found at http://www.chesapeakebav.net/riparl.htm. Three attributes of habitat
quality were linked to riparian vegetation for this assessment: water temperature, stream bank
stability, and sediment loading. Each of these attributes are discussed briefly below.
Stream bank Stabilization: Riparian vegetation typically consists of three distinct types of
plants; a groundcover of grasses and forbs, an understory of shrubs and young trees, and an
overstory of mature trees. These plants serve as structural components for streams, with the root
systems helping to maintain stream stability, and the large woody debris from the mature trees
providing in stream cover. Riparian vegetation has been shown to be essential to maintenance of
a stable stream (Rosgen, 1996). Destabilization of the stream can have a severe impact on
aquatic habitat quality. Following a disturbance, the stream may down cut and widen, releasing
sediment from the stream banks and scouring the stream bed. Bed scour can move redds (egg
nests) after spawning, and/or decrease the number of good spawning sites by changing the size of
gravel available. Destabilization of the stream can have a severe effects aquatic habitat quality
by increasing sedimentation within the watershed. Effects of sedimentation are summarized
below.
Sedimentation: Riparian vegetation is also important in moderating the amount of sediment
loading from upland sources. The roots and stems of riparian vegetation can intercept eroding
upland soil (USDA NRCS, 2000) and riparian plant foliage can reduce erosion from within the
riparian zone by covering soil and reducing the impact energy of raindrops onto soil (Bennett,
1939). Sediment can smother benthic plants and animals. Increased turbidity from sediment
loading could also reduce light transmission, potentially affecting aquatic plants (Cloern, 1987,
Weissing and Huisman 1994) that are important for shelter and food. Increased suspended solids
could also affect foraging behavior of sea turtles (US NMFS, 2004).
145
-------
In addition, sediment carries excess nutrients, particularly phosphorus, into Bay waters,
compromising water quality (www.chesapeakebav.net). Fine sediments can physically occlude
interstitial pore spaces preventing fry from emerging and altering the habitat of mussel locations.
In addition, sediment can bury and effectively suffocate mussels. The U.S. Fish and Wildlife
Services reported that as little as one-quarter of an inch of silt covering the substrate resulted in
death of 90% of mussel species evaluated (U.S. FWS, 2006). Increased siltation in the stream
may also affect spawning, by settling on spawning gravel and reducing flow of water and
dissolved oxygen to the eggs and fry (Everest et al. 1987). Reduced oxygen levels can result in
direct mortality. In addition, fine particles settling on the streambed can also disrupt the food
chain by reducing habitat quality for aquatic invertebrates, and adversely affect groundwater-
surface water interchange (Nelson el al. 1991).
Thermal stability. Riparian habitat provides shading, which provides thermal stability.
However, the sensitivity of the assessed species to temperature fluctuation is unknown, although
U.S. EPA (2003b) reported that temperatures >29 degrees C are stressful to the shortnose
sturgeon. Studies on the temperature sensitivity of the dwarf wedgemussel were not located.
5.2.5.2. Terrestrial Plant Exposure Analysis
The potential for atrazine to affect riparian areas in the Chesapeake Bay watershed was initially
evaluated using terrestrial plant RQs (U.S. EPA, 2004). However, exceedance of terrestrial plant
LOCs does not imply that atrazine use would be expected to result in adverse effects to the
assessed species from riparian zone alterations. Discussion and interpretation of LOC
exceedances is in Sections 5.2.5.8 and 5.2.5.9.
Plants in riparian areas may be exposed to atrazine residues carried from application sites via
surface water runoff or spray drift. Atrazine residues can directly expose seedlings breaking
through the soil surface and expose more mature plants through root uptake or by direct
deposition onto foliage. Although both seedlings and more mature plants can be exposed to
atrazine residues on the soil, seedlings are understood to be the more sensitive life stage. Runoff
or drift into the terrestrial riparian buffer could damage or destroy the riparian vegetation, which
provides important ecosystem services previously discussed such as temperature regulation,
energy input, and stream bank stabilization.
Based on the results of the submitted terrestrial plant toxicity tests, it appears that emerged
seedlings are more sensitive to atrazine via soil/root uptake exposure than emerged plants via
foliar routes of exposure. However, all tested plants, with the exception of corn in the seedling
emergence and vegetative vigor tests and ryegrass in the vegetative vigor test, exhibited adverse
effects following exposure to atrazine. Therefore, a variety of herbaceous plants that may inhabit
riparian zones may be sensitive to atrazine exposure. However, most woody plants are not
expected to be sensitive to atrazine at environmentally relevant concentrations (MRID
4687040001), and atrazine is labeled for use in forestry production. Therefore, most woody
plants are not expected to be affected by atrazine at the labeled application rates.
146
-------
Atrazine exposure to riparian vegetation was estimated using TerrPlant (version 1.2.1),
considering use conditions likely to occur in the Chesapeake Bay watershed. The TerrPlant
model evaluates exposure to plants via runoff and spray drift. The runoff loading of TerrPlant is
estimated based on the solubility of the chemical and assumptions about the drainage and
receiving areas. The spray drift component of TerrPlant assumes that 1% and 5% of the
application rate deposits in the receiving area for ground boom and aerial applications,
respectively.
TerrPlant calculates exposure values for terrestrial plants inhabiting two environments: dry
adjacent areas and semi-aquatic areas. The 'dry, adjacent area' is considered to be representative
of a slightly sloped area that receives relatively high runoff and spray drift levels from
upgradient treated fields and was used as a surrogate for riparian areas.
The following input values were used to estimate terrestrial plant exposure to atrazine from all
uses: solubility = 33 ppm; minimum incorporation depth = 0 (from product labels); application
methods: ground boom, aerial, and granular (from product labels). The following agricultural
and non-agricultural scenarios were modeled: ground/aerial application to fallow land at 2.25 lbs
ai/A, granular application to residential lawns at 2 lbs ai/A, and aerial or ground application to
corn or sorghum at 2 lbs a.i./Acre. Although atrazine is also labeled for forestry use on conifers
at an application rate of 4 lb ai/A, this use was not modeled because the best available
information indicates that atrazine is rarely used in forestry in the Chesapeake Bay watershed
(Section 2). However, potential impacts to riparian vegetation resulting from atrazine use on
forestry (should herbicide use patterns on forestry in the Chesapeake Bay watershed change in
the future) are discussed as part of the Risk Description. If forestry uses of atrazine are
considered (at an application rate of 4.0 lb ai/A), the EECs in Table 5.16 and the resulting RQs
would be expected to increase by a factor of approximately two.
Terrestrial plant EECs for non-granular and granular formulations are summarized in Table 5.16.
EECs resulting from spray drift are derived for non-granular applications only.
Table 5.16. Screening-Level Exposure Estimates for Terrestrial Plants to Atrazine
Use/ App. Rate
(lbs/acre)
Application
Method
Total Loading to
Dry Adjacent Areas
(lbs/acre)
Drift EEC
Fallow land / 2.25
Aerial
0.16
0.14
Ground
0.07
0.02
Corn/Sorghum / 2.0
Aerial
0.14
0.10
Ground
0.06
0.02
Residential / 2.0
Granular
0.04
NA
147
-------
5.2.5.3. Risk Quotients for Riparian Vegetation via Runoff Exposure
Comparison of plant EECs to the seedling emergence EC25 values presented in Section 4
indicates that terrestrial plant RQs are above LOCs for non-endangered plants for all species
except corn and soybeans. RQs range from <1 (corn and soybeans) to 53 (carrots). LOCs were
exceeded for both ground and aerial applications and for both granular and spray applications.
Monocots and dicots show similar sensitivity to atrazine; therefore, RQs were similar across both
taxa. Seedling emergence RQs are in Tables 5.17.
Table 5.17. Nontarget Terrestrial Plant Seedling Emergence RQs
Surrogate Species
EC25
EEC
RQ
(lbs ai/A)
Dry adjacent areas
Dry ad jacent areas
Aerial: 0.16
4.0
Ground: 0.07
(Zea mays)
Granular: 0.04
Aerial: 0.16
Aerial: 40
Monocot - Oat
0.004
Ground: 0.07
Ground: 18
(Avena sativa)
Granular: 0.04
Granular: 10
Aerial: 0.16
Aerial: 18
Monocot - Onion
0.009
Ground: 0.07
Ground: 7.8
(Allium cepa)
Granular: 0.04
Granular: 4.4
Aerial: 0.16
Aerial: 40
Monocot - Ryegrass
0.004
Ground: 0.07
Ground: 18
(Lolium perenne)
Granular: 0.04
Granular: 10
Aerial: 0.16
Aerial: 53
Dicot - Root Crop - Carrot
0.003
Ground: 0.07
Ground: 23
(Daucus carota)
Granular: 0.04
Granular: 13
Aerial: 0.16
-------
a. Inhibition of new growth could result in degradation of high quality riparian
habitat over time because as older growth dies from natural or anthropogenic
causes, plant biomass may be prevented from being replenished in the riparian
area.
b. Inhibition of new growth may also slow the recovery of degraded riparian areas
that function poorly due to sparse vegetation because atrazine deposition onto
bare soil would be expected to inhibit the growth of new vegetation.
3. Because LOCs were exceeded for most species tested (8/10) in the seedling emergence
studies, it would be expected that many species of herbaceous plants may be affected by
atrazine exposure.
5.2.5.4. RQs for Riparian Vegetation via Spray Drift Exposure
Vegetative vigor RQs exceeded LOCs for 3 dicot species (soybeans, cabbage, and cucumber) of
10 plants tested. Vegetative vigor RQs were not exceeded for any of the monocot species tested.
The highest vegetative vigor RQ was 14 (cucumbers).
Table 5.18. Nontarget Terrestrial Plant Vegetative Vigor Toxicity1
Surrogate Species
EC25
(lbs ai/A)
Drift EEC
(lbs ai/A)
RQ
Monocot - Com
(Zea mays)
>4.0
Aerial: 0.11
Ground: 0.02
4.0
Aerial: 0.11
Ground: 0.02
-------
Table 5.18. Nontarget Terrestrial Plant Vegetative Vigor Toxicity1
Surrogate Species
EC25
(lbs ai/A)
Drift EEC
(lbs ai/A)
RQ
Dicot - Cucumber
(Cucumis sathnis)
0.008
Aerial: 0.11
Ground: 0.02
Aerial: 14
Ground: 2.5
All Toxicity Values are from Chetram (1989), MRID 42041403.
The vegetative vigor RQs exceeded the LOC of 1.0 for three dicot plants (soybeans, cucumbers,
and cabbage) with a maximum RQ of 14 in cucumbers. This analysis suggests that some dicots
in riparian habitat are expected to be at risk from foliar exposure via spray drift. Therefore,
riparian habitats comprised of herbaceous plants sensitive to atrazine may be adversely affected
by spray drift. RQs were not exceeded for monocots; therefore, drift would not be anticipated to
affect riparian zones comprised primarily of monocot species such as grasses.
Because RQs for terrestrial plants listed in Tables 5.17 and 5.18 are above LOCs, atrazine use is
considered to have the potential to affect aquatic species by impacting plants in riparian areas
potentially resulting in degradation of stream water quality. These potential effects are evaluated
below.
5.2.5.5. Types of Riparian Zones Sensitive to Atrazine Effects
The parameters used to assess riparian quality that are potentially sensitive to atrazine were
outlined in Table 5.15 and include buffer width, vegetation diversity, vegetation cover, structural
diversity, and canopy shading. Buffer width, vegetation cover, and/or canopy shading could be
reduced if atrazine exposure impacted plants in the riparian zone or prevented new growth from
emerging. Plant species diversity and structural diversity may also be affected if only sensitive
plants are impacted (Jobin etal., 1997, Kleijn and Snoeijing, 1997), leaving non-sensitive plants
in place. Atrazine may also affect the long term health of high quality riparian habitats by
affecting seed germination. Thus, if atrazine exposure impacted these riparian parameters, water
quality within the Chesapeake Bay watershed could be affected.
Because woody plants are typically not sensitive to environmentally-relevant atrazine
concentrations (MRID 4687040001), effects on shading and structural diversity (height classes)
of vegetation are not expected. The potential for effects is limited to herbaceous (non-woody)
plants, which are not generally associated with shading or considered to represent vegetation of
higher height classes. The most sensitive riparian quality criteria are expected to be plant
diversity, vegetation cover, and buffer width because the more sensitive plants (young,
herbaceous plants) are expected to important in maintaining these parameters. A reduction in the
quality in any of these parameters may have the potential to reduce water quality and thus
adversely affect the assessed listed species.
The riparian health criteria described in Fleming etal. (2001; Table 5.15) and the characteristics
associated with effective vegetative buffer strips suggest that healthy riparian zones would be
less sensitive to the impacts of atrazine runoff than poor riparian zones. Riparian zones rich in
species diversity and woody species may contain sensitive species; however, they would also be
150
-------
less likely to consist of a high proportion of very sensitive plants. Wider buffers have greater
potential to reduce atrazine residues over a larger area, resulting in lower levels. In addition,
trees and woody plants in a healthy riparian area would act to filter spray drift (Koch et al. 2003)
and push spray drift plumes over the riparian zone (Davis et al. 1994) thus reducing exposure to
herbaceous plants, which tend to be more sensitive to atrazine. Thus, high quality riparian zones
would be expected to be less sensitive to atrazine's effects than riparian zones that are narrow,
low in species diversity, and comprised of young herbaceous plants or unvegetated areas.
Therefore, the available data suggest that riparian zones comprised largely of herbaceous plants
and grasses would likely be most sensitive to atrazine effects. Bare ground riparian areas could
also be adversely affected by prevention of new growth of grass which can be an important
component of riparian vegetation for maintaining water quality.
Effects from atrazine are more likely to occur in reaches abutting sparsely vegetated riparian
zones because these are the areas where sediment loading to surface water and the potential for
significant deposition is expected to be highest. However, high quality habitat for the assessed
species and high quality spawning habitat of the shortnose sturgeon are more likely to occur in
areas that have not previously been affected by sedimentation.
Cropping to the edge of surface water bodies is expected to result in the greatest level of
sedimentation in adjacent water bodies because no riparian vegetation is present to reduce the
amount of sediment reaching the water. However, the lack of riparian vegetation in these areas
precludes atrazine-induced effects to such vegetation. Therefore, the use of atrazine on fields
without riparian vegetation draining into the Chesapeake Bay and its source waters is not
expected to significantly affect erosion from fields and subsequent sediment loading into the
waters.
5.2.5.6. Agricultural Practices and Sedimentation
In row crop agriculture, land and soil management practices have been identified as having a
large effect on erosion (Green et al. 2003, Tebriigge and During 1999). The practices identified
as erosion reducing, some of which employ herbicide use, are consistent with recent U.S.
government policies encouraging soil conservation (Uri and Lewis, 1998). Also, current atrazine
labels mandate a 66 foot set back for streams. These setbacks are expected to result in lower
loading from application sites to riparian areas; however, the reduction cannot be quantified.
In preparing soil for crops, seeding, and controlling pests, a number of different practices may be
employed that have a large effect on erosion levels and, presumably on subsequent sediment
loading to receiving water bodies. Those practices that disturb the soil are correlated to a greater
extent with increased erosion; conversely, management practices that do not disturb the soil
result in lowered erosion levels. For example, the method of tilling is strongly correlated with
erosion levels (Shiptalo and Edwards 1998). No-till and chisel plow practices result in relatively
low disruption of the soil and are associated with significantly reduced erosion levels. These two
methods of tillage are commonly referred to as conservation tillage, based on their ability to
preserve topsoil. Combining conservation tillage methods with the use of "cover crops" (not
removing the crop residue after harvest to reduce the surface area of soil directly exposed the
impact of rain drops) has been shown by numerous researchers to be an effective means of soil
151
-------
conservation, resulting in a significant reduction in erosion under a wide range of conditions (e.g.
Williams et al. 2000, Jacinthe et al. 2004). An integral part of many soil conservation plans
includes the use of herbicides (Mickelson et al. 2001, Kelly et al. 1996). Some soil conservation
scenarios require greater use of herbicides relative to conventional tillage to control weeds that
would be managed as a result of plowing (Kelly et al. 1996). Atrazine may be used as part of the
soil conservation methods that reduce erosion or in more traditional farming methods, which
may increase erosion due to the inherent nature of conventional tillage as opposed to any direct
cause-effect relationship to atrazine use. Thus, atrazine use may be associated with relatively
high or low sediment loadings resulting from upland agricultural erosion.
Another key factor in the evaluation of potential risks from use of atrazine to riparian habitats is
that a number of sediment reducing strategies are currently in place for the Chesapeake Bay and
its source waters. For example, Maryland (Maryland Department of Natural Resources, 2004)
and Virginia (Commonwealth of Virginia, 2005) have sediment reduction strategies for the
Chesapeake Bay Watershed including its tributaries as part of the Chesapeake Bay 2000
agreement (Chesapeake Bay Program, 2000; http://www.chesapeakebav.net/agreement.htm).
These strategies include implementation of BMPs to reduce loading of nutrients, chemicals, and
sediment into the Chesapeake Bay and its source waters. In some of the locations of the assessed
species (for example, in the coastal zone or in critical areas), a sediment control and water
quality (SCWQ) plan must be submitted to the state for agricultural land adjacent to tributaries of
the Chesapeake Bay. These plans can include a number of BMPs such as tillage practices, land
retirement, cover crops, tree planting, and riparian buffers. However, any number or
combination of BMPs may be included in the SCWQ plan, and a quantitative relationship
between the presence of these BMPs in combination with each other and reduction in sediment
or nutrient loading or reductions in pesticide loading to riparian areas has not been established.
It would be anticipated that atrazine use would have negligible impact on sediment loading in
areas where these state adopted BMPs are implemented.
5.2.6. Potential for Atrazine to Affect the Assessed Species via Effects on Riparian
Vegetation in the Chesapeake Bay Watershed
It is difficult to estimate the magnitude of potential impacts of atrazine use on riparian habitat
and the magnitude of potential effects on stream water quality from such impacts as they relate to
survival or reproduction of the assessed species. The level of exposure and any resulting
magnitude of effect on riparian vegetation are expected to be highly variable and dependent on
many factors. The extent of runoff and/or drift into stream corridor areas is affected by the
distance the field is offset from the stream, local geography, weather conditions, and quality of
the riparian buffer itself. The sensitivity of the riparian vegetation is dependent on the
susceptibility of the plant species present to atrazine and composition of the riparian zone (e.g.
vegetation density, species richness, height of vegetation, width of riparian area).
152
-------
Quantification of risk to the assessed species is precluded by the following factors:
A quantitative relationship between factors such as temperature fluctuation and increased
sedimentation and survival or reproduction of the assessed species is not known;
Relationship between distance of soil input into the stream and sediment deposition in
areas critical to survival and reproduction of the assessed species is not known;
Riparian areas are highly variable in their composition and location with respect to
atrazine use; therefore, their sensitivity to potential damage is also variable.
Locations of shortnose sturgeon spawning habitat within the Chesapeake Bay watershed
are not known;
In addition, even if plant community structure was quantifiably correlated with riparian function,
it may not be possible to discern the effects of atrazine on species composition separate from
other agricultural actions or determine if atrazine is a significant factor in altering community
structure. Plant community composition in agricultural field margins is likely to be modified by
many agricultural management practices. Driving on and mowing of field margins and off-target
movement fertilizer and herbicides are all likely to cause changes in plant community structure
of riparian areas adjacent to agricultural fields (Jobin et al. 1997, Kleijn and Snoeijing 1997,
Schippers and Joenje 2002). Although herbicides are commonly identified as a contributing
factor to changes in plant communities adjacent to agricultural fields, some studies identify
fertilizer use as the most important factor affecting plant community structure near agricultural
fields (e.g. Schippers and Joenje 2002) and community structure is expected to be affected by a
number of other factors (de Blois et al. 2002). In addition, urbanization and development are
also critical factors that may affect stream quality. Thus, the effect of atrazine on riparian
community structure would be expected to be one influence complicated by a myriad of other
factors. Although the data do not allow for a quantitative estimation of risk from potential
riparian habitat alteration, a qualitative discussion is presented below.
As previously discussed, the potential for atrazine to affect the six assessed listed species via
effects to riparian vegetation depends on the potential exposure to and extent of sensitive
(herbaceous/grassy) riparian zones and the importance of sensitive riparian zones to water
quality in the Chesapeake Bay. As of 2004, there were approximately 44,507 acres of grassy
riparian buffers in the Chesapeake Bay watershed (Sweeney, 2006), which represents
approximately 0.1% of the total land and 0.4% of the agricultural land in the Chesapeake Bay
watershed (land areas obtained from www.chesapeakebav.net/wspv31). There are a total of
185,500 miles of riparian forest buffers in the Chesapeake Bay
(www.chesapeakebav.net)/wspv31). which corresponds to approximately 1,539,000 acres of
forested riparian areas in the Chesapeake Bay watershed.7 Therefore, the acreage of grassy
riparian areas is approximately 2.8% of the forested riparian land (44,500 / 1.5 million = 0.028).
Given that forested and grassy riparian areas represent only a fraction of the total riparian areas
in the Chesapeake Bay watershed, the area of grassy riparian zones relative to all riparian areas
in the Chesapeake Bay is likely considerably less than 2.8%. In addition, only a fraction of the
7 The Chesapeake Bay Program (http://www.chesapeakebav.net/wspv31/) reports forested riparian buffer miles that
are at least 100 feet in width and those that are less than 100 feet in width. This calculation was performed using a
weighted average forested riparian buffer width. Riparian forest miles with >100 feet in width were assigned a
width of 100 feet; riparian forest miles with a width <100 feet were assigned a width of 50 feet.
153
-------
grassy/riparian areas are expected to be adjacent to cropland labeled for atrazine use, which
would further diminish the extent of potential impacts of atrazine on grassy riparian buffers and
resulting impacts to water quality in the Chesapeake Bay. Most (70%) of the grassy riparian
buffers in the Chesapeake Bay is on Maryland's Eastern Shore (Sweeney, 2006). However, no
potential suitable spawning habitat of shortnose sturgeon has been located in major rivers on
Maryland's Eastern shore (U.S. EPA, 2003b), and Maryland's Eastern shore is not expected to be
a critical feeding area for the sea turtles, which are transient throughout the Bay and tend to
prefer the higher salinities of the Virginia portion of the Bay
(http://www.chesapeakebav.net/info/seaturtle.cfm).
The low acreage of grassy, herbaceous riparian areas in the Chesapeake Bay watershed sensitive
to atrazine exposure suggests that potential impacts of atrazine to these riparian areas and
resulting effects on sedimentation in the Chesapeake Bay as a whole are expected to be minimal.
This does not imply that grassy riparian buffers are ineffective in reducing nutrient or sediment
loading, but rather that the acreage of land currently devoted to grassy riparian buffers is
sufficiently low such that potential impacts of atrazine to sensitive riparian buffers are not
expected to result in a measurable effect to the assessed species that reside in the main stem of
the Chesapeake Bay, major rivers, or river mouths (shortnose sturgeon and sea turtles). For these
reasons, potential impacts of atrazine to sensitive riparian areas are not likely to adversely affect
shortnose sturgeon or sea turtles. This determination is based on insignificance of the effects
because atrazine effects to grassy, herbaceous riparian vegetation in the Chesapeake Bay, major
rivers, or river mouths cannot be meaningfully measured or detected in the context of a level of
effect where "take" of a single short-nosed sturgeon or assessed sea turtle would occur.
However, it is possible that localized areas may exist where sensitive riparian areas are important
with respect to soil retention and sediment loading prevention, particularly in small headwater
streams. The only species included in this assessment located in small headwater streams is the
dwarf wedgemussel. Therefore, additional analyses were conducted to allow for an evaluation of
potential effects to the dwarf wedgemussel via impacts to riparian areas. The potential for
atrazine to affect riparian areas of the dwarf wedgemussel was evaluated by assessing land use in
the local watersheds of the known dwarf wedgemussel populations. If local land use data
suggests a potential for atrazine exposure to affect the riparian areas to an extent that water
quality may be affected, further evaluation of the potential sensitivity of the type of riparian area
(if any) present around the known streams of dwarf wedgemussels is conducted. If land cover is
consistent with atrazine use and riparian areas surrounding the known locations of the dwarf
wedgemussel are expected to be sensitive to atrazine, then a "likely to adversely effect"
determination could be made.
Land use within the watershed of known dwarf wedgemussel locations was evaluated to
determine the possible extent of riparian area potentially exposed to and affected by atrazine. An
example map used for this analysis, the Po River watershed, is in Figure 5-2 below. Figure 5-2
is one example of the analyses performed; similar maps were created for all known dwarf
wedgemussel populations (Appendix I) except for Mcintosh Run and Nanjemoy Creek because
land use has been previously evaluated for these watersheds by U.S. FWS (1997). In addition,
land use maps were not created for watersheds on Maryland's Eastern shore because agriculture
and cropland are clearly a predominant land cover (Appendix I) in these watersheds based on
154
-------
data obtained from the Chesapeake Bay Program (http://www.chesapeakebav.net/wspv31/) and
USD A (http://www.ams.usda.gov/statesummaries/). Therefore, it was assumed that atrazine
exposure to riparian areas of these dwarf wedgemussel populations could be significant. Results
of these analyses are summarized in Table 5.19 and are presented in greater detail in Appendix I.
Land use data utilized to create the maps were obtained from the Regional Earth Science
Applications Center (RESAC) of the University of Maryland and is available on-line at
http://www.geog.umd.edu/resac/outgoing/.
ireenfh
~
Po River Watershed
~
VA Counties
Po River and Streams
Land Cover
croplands
pasture/hay
¦1
"natural" grass
barren
¦n
deciduous forests
evergreen forests
mixed forests
open water
¦I
urban
DWM Spotsylvania County
Po River Range
Figure 5-2. Map of Land Cover Data in the Po River Watershed.
The map in Figure 5-2 illustrates that land cover in the Po River watershed is mainly forested
cover with some pastureland and urban areas. The area of cropland in the watershed is minimal,
particularly areas in close proximity to the Po River. Atrazine use on forestry in Virginia is
negligible (VA DOF, 2004), and pastureland is not a labeled use (U.S. EPA, 2006c). Therefore,
atrazine exposure to riparian areas of the Po River is expected to be minimal, and the significance
of any potential effects to the dwarf wedgemussel resulting in effects to riparian areas is expected
to be low such that a take is not anticipated. For these reasons, atrazine is not likely to adversely
affect dwarf wedgemussels in the Po River via effects to riparian areas. Comparable analyses
were performed for other dwarf wedgemussel populations. Results of these analyses are
summarized in Table 5.19 and are presented in greater detail in Appendix I. DWM = dwarf
wedgemussel.
Cropland within the watersheds was further characterized using data from USD A
(http://www.ams.usda.gov/statesummaries/) at the county level and from the Chesapeake Bay
program (http://www.chesapeakebav.net/wspv31/) at the sub-watershed level. If the
predominant land cover surrounding waters of specific dwarf wedgemussel habitats was found to
be inconsistent with land cover on which atrazine would be expected to be applied, then a
determination that atrazine will "not likely adversely affect" the dwarf wedgemussel population
within the watershed could be made.
155
-------
In the Po River example above (Figure 5-2), land cover within the watershed is mainly forest
cover with some pastureland and urban areas; 77% of land cover forest, open water, wetland, or
barren land (http://www.chesapeakebav.net/wspv31/). The extent of cropland in close proximity
to the Po River is minimal (Figure 5-2), and only a small proportion of cropland (approximately
20%) in Spotsylvania County was harvested for corn from 1987 to 2002 (no sorghum was
harvested in Spotsylvania County from 1987 to 2002;
http://www.ams.usda.gov/statesummaries/). Atrazine use in forestry operations in Virginia is
minimal (VA DOF, 2004). Together, these data suggest that the extent of riparian areas in the
Po River watershed expected to be exposed to and affected by atrazine is minimal, and the
significance of any resulting potential effects to water quality in the Po River and effects to the
dwarf wedgemussel resulting from such effects is expected to be insignificant (as defined in
Section 2.1) such that a take is not anticipated. For these reasons, it was concluded that atrazine
is not likely to adversely affect dwarf wedgemussels in the Po River.
Similar analyses were performed for other dwarf wedgemussel populations. Results of these
analyses are summarized in Table 5.19 and presented in greater detail in Appendix I. As
summarized in Table 5.19, land use data surrounding riparian areas of waters inhabited by the
dwarf wedgemussel in Virginia and two populations in Maryland suggest that riparian area
exposure to atrazine is expected to be minimal. These populations include Aquia Creek, South
Anna River, Po River, Carter Run, Nanjemoy Creek, and Mcintosh Run. These data would also
suggest that the water concentrations of atrazine estimated using PRZM/EXAMS modeling,
which assumes that 87% to 100% of the watershed is cropped, over-estimate potential atrazine
exposures to these populations of the dwarf wedgemussels.
As previously discussed, atrazine use in forestry is considered negligible in the Chesapeake Bay
watershed (Powers, 2006; VA DOF, 2004; Muir, 2006; USD A, 2004; Wagner etal., 2004,
Pannill, 2006). However, even if atrazine was used in forestry operations, increased
sedimentation from potential effects of atrazine on riparian areas may not occur. Intensive forest
management practices, particularly road building, harvesting and mechanical site preparation,
result in the greatest increases in erosion from forest sites. The available studies on the impact of
mechanical versus chemical (i.e., herbicide) site-preparation for forestry demonstrate that use of
mechanical site preparation methods result in 20 to 400% more sediment than observed on paired
sites which are prepared with herbicides (Michael et al., 2000). Therefore, even if atrazine is
used in forestry operations near dwarf wedgemussel locations, its use may or may not be
associated with increased sedimentation.
As discussed above, land use data suggest that exposure to riparian areas of some dwarf
wedgemussel populations is expected to be minimal. However, land use surrounding Long
Marsh Ditch, Mason's Branch, Aquia Creek, and tributaries of the Corsica River and Southeast
Creek is predominantly agriculture (including cropland,
http://www.chesapeakebav.net/wspv31 /). Each of these locations are along Maryland's Eastern
shore. Presence of large acreage of cropland does not necessarily imply that atrazine use would
be expected to affect dwarf wedgemussels via effects to riparian areas. For atrazine to affect
water quality via impacts on riparian vegetation, riparian areas would need to be present and to
be comprised predominantly of grassy or herbaceous vegetation. Therefore, a qualitative
analysis of the riparian areas along waters inhabited by dwarf wedgemussels where cropland is
156
-------
predominant land cover using aerial photography images obtained from Google Earth (Version
4.0, available at http://earth. googl e. com/). These analyses were not conducted for the dwarf
wedgemussel populations where land use data suggests that atrazine exposure to riparian areas is
expected to be minimal. An analysis of the riparian area of Norwich Creek is presented below as
an example. Analyses conducted for riparian areas of other known dwarf wedgemussel
populations in Maryland's Eastern shore are presented in Appendix I and summarized in Table
5.19.
Norwich Creek is part of the Tuckahoe drainage system. Land use in the watershed is
predominantly (73%) cropland (U.S. FWS, 1997). The area of Norwich creek with dwarf
wedgemussel habitat is surrounded by a 50 meter forested riparian zone (U.S. FWS, 1997). In
addition, upstream locations are surrounded by predominantly forested riparian areas on both
sides of the stream bank (Figures 5-3 and 5-4). Atrazine is not expected to affect forested
riparian areas based on the low sensitivity of woody plants to atrazine. Therefore, riparian areas
of Norwich Creek are not expected to be affected by atrazine to an extent that would be
anticipated to have significant impacts on the dwarf wedgemussel.
Figure 5-3. Example of a Riparian Area of Norwich Creek
The area in the photograph was the subject of analyses in U.S. FWS, 1997.
157
-------
ip, 300$ C'-rfef-i
irrijg« c ;qq« DiH»lOVt*
Figure 5-4. Example of Riparian Area Upstream of the Norwich Creek Site (U.S. FWS, 1997). Presented in
Figure 5-3.
158
-------
5.2.6. Summary of Conclusions: Potential of Atrazine to Affect the Six Listed
Species via Impacts on Riparian Habitat.
Conclusions of the potential for atrazine to affect the six assessed listed species from potential
terrestrial plant and riparian habitat effects are shown in Figure 5-5 and Table 5.19 below. The
best available data suggests that sedimentation from agricultural land could have a negative
impact on some dwarf wedgemussel populations; however, based either on land use or the type
of riparian areas surrounding the known habitats of the dwarf wedgemussel (cropped or
forested), atrazine use is expected to have an insignificant adverse impact on dwarf
wedgemussels via effects to riparian vegetation. However, if the composition of riparian areas in
the Chesapeake Bay and its tributaries changes over time or if land use patterns change over time
such that atrazine use increases considerably, then this conclusion would need to be reevaluated.
Table 5.19. Conclusions for the Potential of Atrazine to Affect Specific Dwarf Wedgemussel
Populations
Site
Basis for Effects Determination
Effects Determination
Dwarf Wedgemussel Habitats Expected to Have Minimal Exposure to Atrazine
Aquia Creek,
Stafford
County, VA
The Aquia Creek watershed is 88 square miles with 81% of land cover
forest, open water, wetland, or barren land and 13% agriculture
(http://www.chesapeakebay.net/wspv31/).
An analysis of the land cover surrounding the Aquia Creek watershed
(Figure 1-4, Appendix I) indicates forestland is the predominant land cover
with minimal cropland in close proximity to the creek. Data from USD A
indicate that 1% of the land cover in Stafford County (1500 of the 173,000
acres) was harvested for corn and sorghum (Attachment 1), and forestry is
a rare use in Virginia (VA DOF, 2004). This analysis suggests that the
extent of riparian areas of the Aquia Creek watershed that may be subject
to atrazine exposure is minimal. Therefore, potential effects to riparian
areas and resulting potential effects to dwarf wedgemussels are expected to
constitute an insignificant effect. d
May effect, but not likely
to adversely affect
South Anna
River, Louisa
County, VA
An analysis of the land use surrounding the South Anna River watershed
(Figure 1-3, Appendix I) indicates that land use is predominantly forest and
pastureland, with minimal cropland or residential land cover. Data from
USD A indicate that 1% and 4% of the land cover in Louisa and Hanover
counties, respectively, was harvested for corn or sorghum (Attachment 1 of
Appendix I). In addition, row crops constitute approximately 3% of land
cover in Louisa countyb, and approximately 4% of land cover is
residential.13 Atrazine use in forestry operations is minimal in Virginia (VA
DOF, 2004), and pastureland is not a currently labeled use (U.S. EPA,
2006). This analysis suggests that the extent of riparian areas of the South
Anna River that may be subject to atrazine exposure is minimal.
Therefore, potential effects to riparian areas and resulting potential effects
to dwarf wedgemussels are expected to constitute an insignificant effect in
the South Anna River.d
May effect, but not likely
to adversely affect
159
-------
Table 5.19. Conclusions for the Potential of Atrazine to Affect Specific Dwarf Wedgemussel
Populations
Site
Basis for Effects Determination
Effects Determination
Po River,
Spotsylvania
County, VA
The predominant land cover surrounding the Po River is forest land; 77% of
land cover is forest, open water, wetland, or barren land.3 Approximately 2%
of the land cover in Spotsylvania county (4,300 of the 257,000 acres) was
harvested for commodities labeled for atrazine uses (corn or sorghum; 2002
data, Attachment 1), and minimal cropland surrounds the Po River in
Spotsylvania County (Figure 1-1, of Appendix I). Atrazine use on forestry,
the predominant land cover in the Po River watershed, in Virginia is rare
(VA DOF, 2004), and atrazine is not labeled for use on pastures (U.S. EPA,
2006a). This analysis suggests that the extent of riparian areas of the Po
River that may be subject to atrazine exposure is minimal. Consequently, the
significance of any potential effects to the dwarf wedgemussel resulting in
effects to riparian areas is expected to be low such that a take is not
anticipated.d For these reasons, atrazine is not likely to adversely affect
dwarf wedgemussels in the Po River via effects to riparian areas.
May effect, but not likely to
adversely affect
Carter Run,
Fauquier County,
VA
The Carter Run watershed is approximately 56 square miles (36,000 acres).
The predominant land cover in the watershed is forest (63%) and pasture
land (34%). Cropland constitutes 1.5% of the land cover in the Carter Run
watershed. Figure 1-4 (Appendix I) illustrates land cover data in the
watershed. These data were presented in the Bacteria TMDL for Carter Run,
Fauquier County, Virginia (January, 2005). Atrazine use in forestry
operations is minimal (VA DOF, 2004), and pasture land is not a labeled use
for atrazine (U.S. EPA, 2006). Therefore, this analysis suggests that the
extent of riparian areas of Carter Run that may be subject to atrazine
exposure is minimal. Therefore, potential effects to riparian areas and
resulting potential effects to dwarf wedgemussels are expected to constitute
an insignificant effect.d
May effect, but not likely to
adversely affect
Rappahannock
River,
Spotsylvania and
Stafford County
Figure 1-2 (Appendix I) indicates that land cover near the Rappahannock
River in Spotsylvania and Stafford Counties is predominantly forested with
minimal cropland. Approximately 2% of the land cover in Spotsylvania
County and approximately 1% of the land cover in Stafford County were
harvested for commodities labeled for atrazine use (corn or sorghum, 2002
data; Attachment 1 of Appendix I). Atrazine use in forestry operations is
minimal in Virginia (VA DOF, 2004), and atrazine is not labeled for use on
pastures (U.S. EPA, 2006a). This analysis suggests that the extent of
riparian areas of the Rappahannock River watershed that may be subject to
atrazine exposure is minimal, and the significance of any potential effects to
the dwarf wedgemussel resulting in effects to riparian areas from atrazine is
expected to be low such that a take is not anticipated.d For these reasons,
atrazine is not likely to adversely affect dwarf wedgemussels in the
Rappahannock River via effects to riparian areas.
May effect, but not likely
to adversely affect
160
-------
Table 5.19. Conclusions for the Potential of Atrazine to Affect Specific Dwarf Wedgemussel
Populations
Site
Basis for Effects Determination
Effects Determination
Nanjemoy Creek,
Charles County,
MD
Land cover data for the Nanjemoy Creek watershed was evaluated and
presented in U.S. FWS (1997). The Nanjemoy Creek watershed is a
predominantly forested watershed with 90% forested area within 100 meters
of streams in the watershed. In addition, approximately 13% of land cover in
the Nanjemoy Creek watershed is agriculture,3 and approximately 2% of land
in Charles County was harvested for corn or sorghum (Attachment 1 of
Appendix I). Together, these data suggest that the extent of potential
atrazine exposure to riparian areas of Nanjemoy Creek is minimal and that
the types of riparian areas in the Nanjemoy Creek watershed (forestland,
typically 100 meters or more) are not expected to be sensitive to atrazine
exposure based on the low sensitivity of woody plants to atrazine (Wall et
al., 2006; MRID 4687040001). Therefore, potential effects to riparian areas
and resulting potential effects to dwarf wedgemussels are expected to
constitute an insignificant effect.
May effect, but not likely to
adversely affect
Mcintosh Run,
St. Mary's
County, MD
Land cover data for the Mcintosh Run watershed was evaluated and
presented in U.S. FWS (1997). The Mcintosh Run watershed is largely
forested; 85% of the area within 100 meters of streams in the watershed is
forested and 9% is cropland. In addition, 4% of land cover was harvested for
corn or sorghum0 (Attachment 1, appendix I). Bank vegetation is dominated
by mature- and sapling-aged trees.0 Together, these data suggest that the
extent of potential exposure to riparian areas of Mcintosh Run is minimal
and that the types of riparian areas surrounding Nanjemoy Creek (forestland,
typically 100 meters or more) are not expected to be sensitive to atrazine
based on the low sensitivity of woody plants (Wall et al., 2006; MRID
4687040001). Therefore, potential effects to riparian areas and resulting
potential effects to dwarf wedgemussels are expected to constitute an
insignificant effect.d
May effect, but not likely to
adversely affect
Dwarf wedgemussel Habitats with Predominant Agriculture Landcover in the Watershed
Longmarsh
Ditch; Mason
Branch, MD
Riparian area of Longmarsh Ditch is cropped to the edge of the stream bank
(Figure 1-11, Appendix I); riparian buffer is absent (see Appendix I for
photograph). Riparian area of Mason Branch, however, is primarily forested
with some areas cropped to the streambank where Longmarsh Ditch becomes
Mason Branch (Appendix I). Neither of these types of riparian areas are
expected to be sensitive to atrazine. Based on the predominance of cropland
in the watershed, atrazine exposure to the riparian areas is expected;
however, based on the low anticipated sensitivity of wooded or cropped
riparian areas to atrazine, potential effects to the riparian area are expected to
constitute and insignificant effect to the dwarf wedgemussel.d
May effect, but not likely to
adversely affect
Browns Branch,
Granny Finley,
and tributaries of
Southeast Creek
and Corsica
River, MD
Land cover surrounding these habitats is primarily agriculture; therefore,
atrazine exposure to these riparian areas may occur. However, aerial
photography of these waters indicates the presence of wooded riparian
buffers on both sides of the streambanks (Appendix I). Atrazine is not
expected to detrimentally impact primarily forested riparian areas based on
the low sensitivity of atrazine on woody plants (Wall et al., 2006; MRID
4687040001). Therefore, potential effects to dwarf wedgemussels resulting
from potential impacts to riparian areas at these locations is expected to
constitute an insignificant effect.d
May effect, but not likely to
adversely affect
Norwich Creek,
Talbot County,
MD
Norwich Creek is part of the Tuckahoe drainage system. Land use in the
watershed is predominantly (72%) agriculture (U.S. FWS, 1997). The area
of Norwich creek with dwarf wedgemussel habitat is surrounded by a 50
meter forested riparian zone (U.S. FWS, 1997; Figure I-6a of Appendix I).
In addition, upstream locations are surrounded by predominantly forested
May effect, but not likely to
adversely affect
161
-------
Table 5.19. Conclusions for the Potential of Atrazine to Affect Specific Dwarf Wedgemussel
Populations
Site
Basis for Effects Determination
Effects Determination
riparian areas on both sides of the stream bank (Figure I-6b of Appendix I).
Atrazine is not expected to detrimentally impact primarily forested riparian
areas based on the low sensitivity of atrazine on woody plants (Wall et al.,
2006; MRID 4687040001). Therefore, riparian areas of Norwich Creek are
not expected to be affected by atrazine use to an extent that would be
anticipated to have significantd impacts on the dwarf wedgemussel.
a http://www.chesapeakebav.net/wspv31
b http://fisher.lib.virginia.edu/collections/gis/nlcd/browse countv.html
0 http://www.ams.usda.gov/statesummaries/)
d Significance of Effect: Insignificant effects are those that cannot be meaningfully measured, detected, or evaluated
in the context of a level of effect where take occurs for even a single individual
e U.S. FWS. 1997. Characterization of Endangered Dwarf Wedgemussel (Alasmidonta heterodon) habitats in
Maryland. CBFO-C97-01. January, 1997.
162
-------
Figure 5-5. Summary of the Potential of Atrazine to Affect the Six Assessed Listed Species
via Riparian Habitat Effects.
Water
Temnerature
Streambank
Stability
Sedimentation
Wider and shallower
channels resulting from
eroding streambanks could
adversely modify habitat
Terrestrial plant risk quotients are above LOCs; riparian vegetation may be affected
No effect. Woody
vegetation, which
provides thermal
stability, is not
expected to be
substantially affected
Water temperature
increases in the absence
of shading
Increased suspended sediment
or deposited sediment may
affect all assessed species
either directly or via effects on
aquatic plants and
invertebrates
Riparian health is associated with many water quality parameters. The assessment links
riparian vegetation to the following potential effects
Not likely to adversely affect.
Atrazine is not expected to harm the
roots of large, mature plants
providing stablity to streambanks
and denuded streambanks, which
would be most sensitive to the plant
growth inhibition effects of atrazine.
Effects to vegetation are expected to be limited to a fraction of species. Most young
herbaceous plants are expected to be affected, although most woody species are not
expected to be affected. More species are expected to be sensitive to atrazine at the
seedling stage.
Not likely to adversely affect (Table
5.19). Acreage of riparian habitat
expected to be sensitive to atrazine is
sufficiently low in the Chesapeake Bay
watershed such that potential impacts of
atrazine to sensitive riparian buffers are
not expected to result in a measurable
effect to the assessed species that reside
in the main stem of the Chesapeake Bay
and the Major river systems. In small
headwater streams of the dwarf
wedgemussel, either potential exposure
to riparian areas is expected to be
minimal, or the composition of riparian
areas is such that they are not expected
to be sensitive to atrazine.
6.0 Uncertainties
A number of uncertainties are inherent in ecological risk assessment, which are discussed in
detail in U.S. EPA (2004). Principle uncertainties in this risk assessment are discussed below.
Additional uncertainties were discussed in Sections 2 through 5.
163
-------
6.1. Exposure Assessment Uncertainties
6.1.1. PRZM Modeling Inputs and Predicted Aquatic Concentrations
Overall, the uncertainties inherent in the exposure assessment tend to result in over estimation of
exposures. This is apparent when comparing modeling results with monitoring data. In general,
the monitoring data should be considered a lower bound on exposure, while modeling represents
an upper bound. Factors influencing the over-estimation of exposure include the assumption of
no dilution or flow within the receiving water body. In addition, the impact of setbacks on
runoff estimates have not been quantified while acknowledging that these buffers, especially
well-vegetated buffers, are likely to result in significant reduction in runoff loading of atrazine.
In general, the simplifying assumptions used in this assessment appear from the characterization
above to be reasonable especially in light of the analysis completed and the available monitoring
data. There are also a number of assumptions that tend to result in over-estimation that cannot be
quantified, but can be qualitatively described. For instance, modeling in this assessment for each
use site assumes (with the exception of the right of way scenario) that the entire 10-hectare
watershed is taken up by the respective use pattern. The assessment assumes that all applications
have occurred concurrently on the same day at the exact same application rate. This is unlikely
to occur in reality but is a reasonable assumption in lieu of actual data. In addition, the use of the
standard water body assumes no flow through and thus the longer-term average concentrations
presented above are likely reasonable approximations of headwater streams and water bodies but
are also likely over-estimates of what is expected in lower reaches of the tributaries and within
the Chesapeake Bay itself.
In general, buffer restrictions are an effective means of reducing movement of pesticides via drift
to non-target aquatic resources. Effectiveness of a spray drift buffer can be evaluated
quantitatively using AgDrift, which estimates the percentage of drift that is expected from a
given buffer distance. This buffer-specific drift fraction is then used in PRZM/EXAMS in lieu
of the default spray drift percentages that assume no buffer to quantitatively evaluate the
effectiveness of a given buffer distance on off-site drift loadings. Currently, atrazine labels
specify setback (or buffer) distances between applications and surface water bodies. These
distances were integrated into this assessment using AgDrift to estimate distance specific spray
drift values as substitute for the standard edge-of-field assumptions.
Unlike spray drift, there are currently no models that evaluate the effectiveness of a vegetative
buffer on runoff and loadings. The effectiveness of vegetative buffers is highly dependent on the
condition of the buffer. For example, a well-established, healthy vegetative buffer can be a very
effective means of reducing runoff and erosion from agricultural fields. Alternatively, a buffer
of poor vegetative quality or a buffer that is channelized can be ineffective at reducing loadings.
Until such time that a quantitative method for estimate the effect of vegetative buffers of various
conditions on pesticide loadings, it can only be stated that aquatic exposure predictions are likely
to overestimate exposure where healthy vegetative buffers exist and likely do not overestimate
exposure where poorly developed, channelized, or bare buffers exist.
164
-------
In general, the linked PRZM/EXAMS model produces estimated aquatic concentrations that are
exceeded once within a ten-year period. PRZM is a process or "simulation" model that
calculates what happens to a pesticide in a farmer's field on a day-to-day basis. It considers
factors, such as rainfall and plant transpiration of water, as well as how and when the pesticide is
applied. It has two major components: hydrology and chemical transport. Water movement is
simulated by the use of generalized soil parameters, including field capacity, wilting point, and
saturation water content. The chemical transport component can simulate pesticide application
on the soil or on the plant foliage. Dissolved, adsorbed, and vapor-phase concentrations in the
soil are estimated by simultaneously considering the processes of pesticide uptake by plants,
surface runoff, erosion, decay, volatilization, foliar wash-off, advection, dispersion, and
retardation.
Uncertainties surrounding each of the individual components named above add to the overall
uncertainty of the modeled concentrations. Additionally, model inputs from the environmental
fate degradation studies are chosen to represent the upper confidence bound on the mean, values
that are not expected to be exceeded in the open environment 90 percent of the time. Mobility
input values are chosen to be representative of conditions in the open environment. The natural
variation in soils adds to the uncertainty of modeled values. Factors such as application date,
crop emergence date, and canopy cover can also affect estimated concentrations, adding to the
uncertainty of modeled values. Factors within the ambient environment such as soil
temperatures, sunlight intensity, antecedent soil moisture, and surface water temperatures can
cause actual aquatic concentrations to differ for the modeled values.
Additionally, the rate at which atrazine is applied, the percent of a watershed that is cropped, and
the percent of crops in that watershed that was actually treated with atrazine may be lower than
the default assumption of the maximum allowable application rate being used, the entire crop
being treated, and the default estimate of the area within a watershed planted with agricultural
crops. The geometry of a watershed, and limited meteorological data sets also add to the
uncertainty of estimated aquatic concentrations.
There is significant uncertainty with the quantitative use of the predicted EEC generated from
these alternative scenarios. The standard water body and the Index Reservoir has been
developed and vetted through a public peer review process. Both were developed with a specific
range of exposure settings in mind. For the EXAMS static water body the 1 hectare body is
intended to represent highly vulnerable water bodies, streams, creeks and rivers in headwater
areas adjacent to agricultural fields. The Index Reservoir was developed to represent a small
highly vulnerable drinking water reservoir. Neither water body was intended to represent larger,
faster flowing water bodies. Therefore, the use of these two water bodies to represent flowing
streams, creeks and rivers is intended only to provide a sense of the impact of flow on the
modeled EEC for characterizing what those EEC represent and is not intended to provide the
means to better characterize the exposure and potential risks.
165
-------
6.1.2. Monitoring Data
The monitoring data used in this risk assessment were not collected for the purpose of supporting
an ecological risk assessment; therefore they are not likely to be representative of peak atrazine
concentrations.
6.1.3. Exposure to Degradates
Some degradates of atrazine are common to other triazine herbicides such as simazine or
propazine. Therefore, exposure to degradates could be higher because some of the degradates
are common to other triazines that may co-occur in the action area. However, given the low
magnitude of risk of the degradates, this uncertainty is not expected to impact the conclusions of
this assessment.
6.1.4. Use Characterization
Corn is expected to be the predominant atrazine use in the Chesapeake Bay watershed.
However, atrazine may also be used on conifers and softwoods, and there is considerable
forestland in some areas of the Chesapeake Bay watershed. Exposure to the assessed listed
species from atrazine use in forestry operations was considered negligible because a total of 24
pounds of atrazine was used in the Commonwealth of Virginia by Virginia's forestry community
in 2003 (VA DOF, 2004). USDA (2004) indicates that use of atrazine in coniferous evergreen
operations is also low. In addition, atrazine is not used in the maintenance of State forestland in
Maryland (MD DNR, 2006) and is not used in National Forestland
(http://www.fs.fed.us/foresthealth/publications/pesticide/pur/reports.htm). The Maryland
Department of Natural Resources also indicated that atrazine use in tree farms is uncommon in
Maryland (MD DNR, 2006; Pannel, 2006). In addition, atrazine is applied only during the first
year of tree growth. Based on the extended duration between planting and harvest for trees, only
a small proportion of forestland is expected to be less than 1 year and, thus, treated with atrazine.
For these reasons, potential risk from atrazine use on trees is expected to be less than risks based
on agricultural crops described in this assessment.
Aquatic EECs from forestry (PRZM conifers scenario) were lower than those for sorghum and
were, therefore, not used in RQ calculations. Therefore, the assumption that atrazine use in
forestry is negligible would not impact aquatic EECs or RQs. However, the assumption of
negligible atrazine use in forestry was important in the evaluation of land use surrounding dwarf
wedgemussel populations, which was used to evaluate potential effects to riparian areas of dwarf
wedgemussel locations, particularly in Virginia. If agricultural practices in forestry change such
that atrazine use increases dramatically, then risks to Virginia dwarf wedgemussel populations
would need to be re-evaluated.
166
-------
6.1.5. Long-range Transport of Volatilized Atrazine
The environmental fate and monitoring data suggest that long range transport of volatilized
atrazine is a possible route of exposure for the listed species. However, given the magnitude of
documented atrazine concentrations in rainfall at or below available surface water and
groundwater monitoring data (as well as modeled estimates for surface water), and the lack of
modeling tools to predict the impact of long range transport of atrazine, the extent of the action
area is defined by the transport processes of runoff and spray drift for the purposes of this
assessment.
6.2. Effects Assessment Uncertainties
6.2.1. Age Class and Sensitivity of Effects Thresholds
It is generally recognized that test organism age may have a significant impact on the observed
sensitivity to a toxicant. The acute toxicity data for fish are collected on juvenile fish between
0.1 and 5 grams. Aquatic invertebrate acute testing is performed on recommended immature age
classes (e.g., first instar for daphnids, second instar for amphipods, stoneflies, mayflies, and third
instar for midges).
Testing of juveniles may overestimate toxicity at older age classes for pesticidal active
ingredients, such as atrazine, that act directly (without metabolic transformation) because
younger age classes may not have the enzymatic systems associated with detoxifying
xenobiotics. In so far as the available toxicity data may provide ranges of sensitivity information
with respect to age class, this assessment uses the most sensitive life-stage information as
measures of effect for surrogate aquatic animals. Nonetheless, no data on glochidial stage
mussels were available; therefore, data on juvenile bivalves were used for RQ calculations used
to estimate risk to mussels. It is unknown if glochidial stage mussels are expected to be more,
less, or equivalent in sensitivity to atrazine as the juvenile mussels that were tested in the
available studies.
6.2.2. Use of Surrogate Species Effects Data
Guideline toxicity tests are not available for turtles or freshwater mussels; therefore, surrogate
species were used as outlined in U.S. EPA (2004). Therefore, birds were used as a surrogate for
reptiles and saltwater mussels were used as a surrogate for the freshwater dwarf wedgemussel.
The available open literature information on atrazine toxicity to reptiles was insufficient to allow
for a direct comparison of the surrogate species to the assessed species. Extrapolating the risk
conclusions from the surrogate tested species to the assessed species may either underestimate or
overestimate potential risks. However, as described in Section 4, use of birds as a surrogate for
sea turtles was considered conservative. Efforts are made to select the organisms most likely to
be affected by the type of compound and usage pattern; however, there is an inherent uncertainty
in extrapolating across phyla. LOCs are intentionally set low, and conservative estimates are
made in the screening level risk assessment to account for these uncertainties.
167
-------
Aquatic invertebrates were used as surrogates for jellyfish, which is the primary dietary item of
leatherback turtles. It is unknown if the available toxicity data are representative of the
sensitivity of jellyfish to atrazine. If jellyfish are particularly sensitive to atrazine, then jellyfish
availability could be affected by its use. However, leatherback turtles are highly pelagic and
reside in the main stem of the Chesapeake Bay. At atrazine concentrations expected in the main
stem of the Bay (low |ig/L range), none of the surrogate aquatic invertebrate species tested are
likely to be adversely affected to such an extent that a "take" as defined in Section 2.1 is
expected for turtles resulting from a reduction in food supply. Also, data from the Chesapeake
Bay Monitoring Program suggests that jellyfish populations have remained stable in the main
channel of Chesapeake Bay since 1984, where this highly pelagic species is most likely to be
found (Figure 6-1). The presence of a stable jellyfish population does not necessarily indicate
that atrazine have not impacted jellyfish numbers prior to 1984 or that atrazine may not affect
seasonal fluctuations of jellyfish numbers. However, these data support the conclusion that
atrazine does not appear to be affecting jellyfish numbers in the Chesapeake Bay.
Average Annual Jellyfish Count Reported by the Chesapeake Bay Program
from 1984 to 2002
700
600 *
500
400
Jellyfish No.
300
200 *
^ ~
100 <( . ~ *
~ ~ ~
0 ~ ~ ~ ~
U i 1 1 1 1 1 1 1 1 1
1984 1986 1988 1990 1992 1994 1996 1998 2000 2002 2004
year
~ ~
~
~ ' ~ . '
~ ~ ~
Figure 6-1. Jellyfish Monitoring Data in the Chesapeake Bay
Values represent average number of jellyfish across all monitoring stations that reported jellyfish numbers >0.
Figures representing all data (count and biovolume across all monitoring stations) are in Appendix H). Data
obtained from (http://www.chesapeakebaY.net/bavbio.htm).
6.2.3. Use of the Lowest Aquatic Invertebrate Toxicity Value to Estimate Risk to Potential
Food Items
Several of the aquatic invertebrates (e.g., midge, copepod, daphnid) showed a wide range of
sensitivity within and between species of the same genus (2 orders of magnitude). Therefore,
acute RQs based on the most sensitive toxicity endpoint for aquatic invertebrates may represent
an under- or over-estimation of potential direct risks to freshwater invertebrates and indirect
effects to the assessed species via a reduction in available food.
168
-------
6.2.4. Absence of a chronic study in the most acutely sensitive marine/estuarine
invertebrate
Acute toxicity data suggest that the copepod was the most sensitive species in acute studies.
However, the copepod is not used as a surrogate for direct effects to any of the assessed species.
In addition, the copepod was not a surrogate food item of any of the assessed species. Therefore,
potential impacts of this uncertainty to the conclusions of this assessment are expected to be
minimal. However, one study in open literature was located that allows for some
characterization of potential chronic effects of atrazine on the copepod. Forget-Leray eial.
(2004) reported results from a 96-hour, 10-day, and 30-day exposure study. An acute 96-hour
LC50 of 125 |ig/L was reported for the copepod E. ajfinis nauplii. A 10-day NOAEC of 25 |ig/L
was reported; increased incidences of mortality were observed at 49 |ig/L. In addition, delayed
maturity (time from nauplius to adult before molting) was observed at 25 |ig/L in the 30-day
exposure study. This study is discussed in further detail in Appendix A. Although there are
uncertainties associated with the study that may limit its utility in ecological risk assessment,
including reporting deficiencies and use of an unacceptable solvent, these data suggest that the
copepod may represent the most sensitive species tested in available chronic studies. Potential
impacts of this uncertainty on this risk assessment are discussed in Section 5; however, use of a
NOAEC of 25 |ig/L in place of the NOAEC of 60 |ig/L used in risk estimation would not be
expected to alter the conclusions of this assessment.
6.2.5. Extrapolation of Long-term Environmental Effects from Short-term Laboratory
Tests
The influence of length of exposure and concurrent environmental stressors (e.g., urban
expansion, habitat modification, decreased quantity and quality of water, predators, etc.) to the
assessed species may affect the species response to atrazine. The most probably effect of these
types of uncertainty is that the effect is underestimated. Timing, peak concentration, and
duration of exposure are critical in terms of evaluating effects, and these factors will vary both
temporally and spatially within the action area. Overall, the effect of this variability may result
in either an overestimation or underestimation of risk.
6.2.6. Use of Threshold Concentrations as Community-Level Endpoints
For the purposes of this endangered species assessment, threshold concentrations are used to
predict potential indirect effects (via aquatic plant community structural change) to the assessed
species. The conceptual aquatic ecosystem model used to develop the threshold concentrations
is intended to simulate the ecological production dynamics in a 2nd or 3rd order Midwestern
stream; however, the model has been correlated to the micro- and mesocosm studies, which were
derived from a wide range of experimental studies (i.e., jar studies to large enclosures in lentic
and lotic systems), that represent the best available information for atrazine-related community-
level endpoints.
Although it is not possible to determine how well the responses observed in the micro- and
mesocosm studies reflect the Chesapeake Bay community, available microcosm and mesocosm
169
-------
data and laboratory studies (Appendix A) do not indicate that estuarine systems are more
sensitive than the freshwater systems on which the threshold concentrations were based. In
addition, the available laboratory saltwater plant studies do not suggest increased sensitivity
compared with freshwater aquatic plant species. Given that threshold concentrations were
derived based on the best available information from available community-level data for
atrazine, these values are intended to be protective of the aquatic community. Additional
uncertainties associated with use of the screening thresholds to estimate community-level effects
are discussed in Section B.8 of Appendix B.
6.2.7. Sediment Loading from Riparian Effects
No standard methods are available for assessing the effects from increased sedimentation that is
a potential effect of damaging riparian vegetation.
6.2.8. Exposure to Pesticide Mixtures
This assessment considered only the single active ingredient of atrazine. However, the assessed
species and their environments may be exposed to multiple pesticides simultaneously.
Interactions of other toxic agents with atrazine could result in additive effects, synergistic effects
or antagonistic effects. Conceptually, the combined effect of the mixture is equal to the sum of
the effects of each stressor (1 + 1=2) for additive toxicity. Synergistic effects occur when the
combined effect of the mixture is greater than the sum of each stressor (1 + 1 >2), and
antagonistic effects occur when the combined effect of the mixture is less than the sum of each
stressor (1 + 1 <2).
The available data suggest that pesticide mixtures involving atrazine may produce either
synergistic, additive, or antagonistic effects. Mixtures that have been studied include atrazine
with insecticides such as organophosphates and carbamates or with herbicides including alachlor
and metolachlor. Additive or synergistic effects have been reported in several taxa including
fish, amphibians, invertebrates, and plants.
As previously discussed, evaluation of pesticide mixtures is beyond the scope of this assessment
because of the myriad of factors that cannot be quantified based on the available data. Those
factors include identification of other possible co-contaminants and their concentrations,
differences in the pattern and duration of exposure among contaminants, and the differential
effects of other physical/chemical characteristics of the receiving waters (e.g. organic matter
present in sediment and suspended water). Evaluation of factors that could influence
additivity/synergism is beyond the scope of this assessment and is beyond the capabilities of the
available data to allow for an evaluation. However, it is acknowledged that not considering
mixtures could over- or under-estimate risks depending on the type of interaction and factors
discussed above.
170
-------
6.2.9. Sublethal Effects
The assessment endpoints used in ecological risk assessment include potential effects on
survival, growth, and reproduction of the assessed species and organisms on which the species
depend for survival. A number of studies were located that evaluated potential sublethal effects
to fish from exposure to atrazine. Many of these studies reported toxicity values that were less
sensitive than the submitted studies, and were not considered for use in risk estimation.
However, several fish studies were located in the open literature that reported effects on
endpoints other than survival, growth, or reproduction at concentrations that were considerably
lower than the most sensitive endpoint from submitted studies.
Reported sublethal effects including changes in hormone levels, behavioral effects, kidney
pathology, gill physiology, and potential olfaction effects have been observed at concentrations
lower than 65 |ig/L, the most sensitive fish life-cycle NOAEC (see Appendix A and Section
4.1.2.). These studies were not considered appropriate for risk estimation in place of the life
cycle studies because quantitative relationships between these sublethal effects and the ability of
fish to survive, grow, and reproduce has not been established. The magnitude of the reported
sublethal effect associated with reduced survival or reproduction has not been established;
therefore it is not possible to quantitatively link sublethal effects to the selected assessment
endpoints for this ESA. In addition, in the fish life-cycle studies, no effects were observed to
survival, reproduction, and/or growth at levels associated with the sublethal effects. Also, there
were limitations to the studies that reported sublethal effects that preclude their quantitative use
in risk assessment (see Appendix A and Section 4.2.1). Nonetheless, if future studies establish a
quantitative link between the reported sublethal effects and fish survival, growth, or
reproduction, the conclusions with respect to potential effects to fish may need to be revisited.
Upon evaluation of the available studies, however, the most sensitive NOAEC from the
submitted life-cycle studies was considered to be the most appropriate chronic endpoint for use
in risk assessment. In the life-cycle study design, fish are exposed to atrazine from one stage of
the life cycle to at least the same stage of the next generation (e.g. egg to egg). Therefore,
exposure occurs during the most sensitive life stages and during the entire reproduction cycle.
Four life cycle studies have been submitted in support of atrazine registration. Species tested
include brook trout, bluegill sunfish, and fathead minnows. The most sensitive NOAEC from
these studies was 65 |ig/L.
6.3. Assumptions Associated with the Acute LOCs
The risk characterization section of this endangered species assessment includes an evaluation of
the potential for individual effects. The individual effects probability associated with the acute
RQ is based on the mean estimate of the slope and an assumption of a probit dose response
relationship for the effects study corresponding to the taxonomic group for which the LOCs are
exceeded. These slopes from surrogate species could over- or under-estimate potential risks.
171
-------
7.0 Conclusions
Conclusions of this assessment are summarized in Table 7.1. The best available data suggest
that atrazine will either have no effect or is not likely to adversely affect any of the assessed
species either by direct toxic effects or by indirect effects resulting from effects to aquatic or
terrestrial plants or aquatic animals.
172
-------
Table 7.1. Summary of Effects Determinations For Six Listed Species
Assessment Endpoint
Species
Effects
Determination
Basis for Determination
Direct effects to listed
species (Section 5.1)
All six assessed
species
No Effect
No acute or chronic LOCs for endangered species are exceeded.
Indirect effects to listed
species via reduction of
aauatic animals as food
supply
(Section 5.2.2.)
Shortnose
sturgeon,
loggerhead turtle,
Kemp's ridley
turtle, green turtle,
leatherback turtle
Not likely to
adversely affect
Acute LOCs are exceeded for some animals that are food items
of the assessed species. However, the low magnitude of
potential effects on any one species, the low number of dietary
species potentially affected (indicated by LOC exceedances)
relative to the number potentially consumed by the assessed
species, and the conservative nature of the EECs used to derive
RQs for organisms in flowing water systems suggests that the
potential effects to the food supply of the assessed species
constitutes an insignificant effect.3
Dwarf
wedgemussel
No effect
No acute or chronic LOCs are exceeded.
Indirect effects to listed
species via reduction of
aauatic olants as food
items or primary
productivity
(Section 5.2.4.)
All six assessed
species
Not likely to
adversely affect
No known obligate relationship between the assessed species
and any single aquatic plant species exists, and short-term and
long-term atrazine concentrations were estimated to be lower
than established thresholds for community-level effects to
aquatic vegetation.
Indirect effects to listed
species via direct
effects on riparian areas
required to maintain
acceptable water quality
and spawning habitat
(Section 5.2.5.)
Shortnose sturgeon
and each of the
four assessed sea
turtles
Not likely to
adversely affect
Acreage of riparian habitat expected to be sensitive to atrazine is
sufficiently low in the Chesapeake Bay watershed such that
potential impacts of atrazine to sensitive riparian buffers are not
expected to result in a measurable effect to the assessed species
that reside in the main stem of the Chesapeake Bay and the
Major river systems. Therefore, potential effects to riparian
areas from use of atrazine are expected to constitute an
insignificant effecta.
Dwarf
wedgemussel
Not likely to
adversely affect
Landcover data surrounding watersheds of dwarf wedgemussel
habitats suggest that riparian area exposure to atrazine is
expected to be minimal and/or that the predominant riparian
area adjacent to waters of dwarf wedgemussel habitats is not
expected to be sensitive to atrazine. Therefore, potential effects
to the dwarf wedgemussel from effects to riparian areas are
expected to constitute an insignificant effect.3
a Significance of Effect: Insignificant effects are those that cannot be meaningfully measured, detected, or
evaluated in the context of a level of effect where take occurs for even a single individual
173
-------
8.0 References
Allan, JD. (1995). Stream ecology: structure and function of running waters. Chapman and Hall,
NY.
Armstrong, D. E., C. Chester and R. F. Harris 1967. Atrazine hydrolysis in soil. Soil Sci. Soc.
Amer. Proc. 31:61-66.
BEAD. 2003. Assessment of Potential Mitigation Measures for Atrazine, 2003, Biological and
Economic Analysis Division (BEAD), Office of Pesticide Programs, U.S. Environmental
Protection Agency.
Baker, D. B., K. A. Kieger, R. P. Richards and J. W. Kramer. 1985. Effects of intensive
agricultural land use on regional water quality in northwestern Ohio, p. 201-207. Irr. US EPA.
Perspectives on nonpoint source pollution, proceedings of a national conference, Kansas
City, MO. EPA-440/5-85-001. XI - 75
Bartell, S.M., K.R. Campbell, C.M. Lovelock, S.K. Nair, and J.L. Shaw. 2000. Characterizing
aquatic ecological risk from pesticides using a diquat dibromide case study III.
Ecological Process Models. Environ. Toxicol. Chem. 19(5): 1441-1453.
Bartell, S.M., G. Lefebvre, G. aminski, M. Carreau, and K.R. Campbell. 1999. An ecosystem
model for assessing ecological risks in Quebec rivers, lakes, and reservoirs. Ecol. Model.
124:43-67.
Bennett H.H, (1939). Soil Conservation. New York, New York, 993 pp.
Byles, R. A. (1988) The behavior and ecology of sea turtles Caretta caretta and Lepidochefys
kempi, in Chesapeake Bay. PhD Dissertation. College of William and Mary,
Williamsburg, VA.
Capel, P. D., M. Lin and P. J. Wotzka. 1994. Pesticides in rains in Minnesota 1991-1993: An
interim report. Minnesota Dept. Agr., Agronomy Serv. Div., p. 26. (Additional pages on
1994 data).
CBP (Chesapeake Bay Program). 2006. Data obtained from
http://www.chesapeakebav.net/wspv31/. Accessed in August, 2006.
CDMS search, 2006, https://premier.cdms.net/webapls/
Cloern, J.E.. 1987. Turbidity as a control on phytoplankton biomass and productivity in
estuaries. Continental Shelf Research 7(11-12): 1367-1381.
174
-------
Colby, S.R., E.R. Hill, N.E. Humberg, L.M. Kitchen, R.G. Lym, W.J. McAvoy, R. Prasad.
(1989) The Herbicide Handbook of the Weed Science Society of America. 6th Ed.
Champaign Illinois.
Curran, W., Professor of Weed Science, Pennsylvania State University, Personal Communication
with Nicole Zinn, May 17, 2006.
Davies, P. E., L. S. J. Cook and J. L. Barton. 1994. Triazine herbicide contamination of
Tasmanian streams: Sources, concentrations and effects on biota. Aust. J. Mar.
Freshwater Res. 45:209-226. (MRID 45020003).
Davis B. N. K., M. J. Brown, A. J. Frost, T. J. Yates, R. A. Plant. (1994). The Effects of Hedges
on Spray Deposition and on the Biological Impact of Pesticide Spray Drift.
Ecotoxicology and Environmental Safety. 27(3):281-293.
DeAngelis, D.L., S.M. Bartell, and A.L. Brenkert. 1989. Effects of nutrient recycling and food-
chain length on resilience. Amer. Nat. 134(5):778-805.
de Blois S., G. Domon, A. Bouchard (2002). Factors affecting plant species distribution in
hedgerows of southern Quebec. Biological Conservation 105(3): 355-367.
Dodd, C.K., Jr. (1988) Synopsis of the biological data on the loggerhead sea turtle Caretta
caretta (Linnaeus 1758). Fish and Wildlife Service Biological Report 88(14). 110pp.
Duron, M. 1978. Contribution a letude de la biologie de Dermochelys coriacea (Linne) dans les
pertuis Charentais. These, Doct. Biol. Anim., Univ. Bordeaux. 112p. Cited in NOAA
Fisheries (http://swr.nmfs.noaa.gov/PIR/feis/Appendix%20C.PDF)
Ernst, C.H., Lovich, J.E. and Barbour R.W. (1994) Turtles of the United States and Canada.
Smithsonian Institution Press, Washington, D.C.
Everest, F.H., R.L. Beschta, J.C. Scrivener, K.V. Koski, J.R. Sedell, C.J. Cederholm. (1987).
Fine sediments and salmonid production: a paradox, p. 98-142. In E.O. Salo and T.W.
Cundy [ed.] Proceedings of the Symposium on Streamside Management: Forestry and
Fishery Interactions. University of Washington, Seattle, WA.
Fish and Wildlife Health Program. 2004. Final Report Submitted to National Marine Fisheries
Service. Contract #: 40EMNP200722.
Fleming, W., D. Gait, J. Holechek. (2001). Ten steps to evaluate rangeland riparian health.
Rangelands 23(6):22-27.
Forget-Leray, J., Landriau, I., Minier, C., and Leboulenger, F. (2005). Impact of Endocrine
Toxicants on Survival, Development, and Reproduction of the Estuarine Copepod
Eurytemora affinis (Poppe). Ecotoxicol.Environ.Saf. 60: 288-294. EcoReference No.:
175
-------
80951
Foreman, W.T., Majewski, M.S., Goolsby, D.A., Wiebe, F.W., and Coupe, R.H. 2000.
Pesticides in the atmosphere of the Mississippi River Valley, Part II - air. Sci. Tot.
Environ. 248. 135-146.
Frank, R., H. E. Braun, M. V. Holdrinet, G. J. Sirons and B. D. Ripley. 1982. Agriculture and
water quality in the Canadian Great Lakes Basin. V. Pesticide use in 11 agricultural
watersheds and presence in stream water 1975-1977. J. Environ. Qual. 11:497-505.
Frank, R. and G. J. Sirons. 1979. Atrazine: Its use in corn production and its loss to stream
waters in southern Ontario, 1975-1977. Sci. Total Environ. 12:223-239.
Frank, R., G. J. Sirons, R. L. Thomas and K. McMillan. 1979. Triazine residues in suspended
solids (1974-1976) and water (1977) from the mouths of Canadian streams flowing into
the Great Lakes. J. Great Lakes Res. 5:131-138. XI-79
Giddings, Jeffrey M., Todd A. Anderson, Lenwood W. Hall, Jr., Ronald J. Kendall, R. Peter
Richards, Keith R. Solomon, and W. Martin Williams, "Aquatic Ecological Risk
Assessment of Atrazine - A Tiered Probabilistic Approach, A Report of an Expert
Panel," Novartis Crop Protection, Inc., Greensboro, NC, Section 5.3.1.5, page 147, June
23, 2000.
Gilbert, C.R. 1989. Species profiles: life histories and environmental requirements of coastal
fishes and invertebrates (Mid-Atlantic Bight)--Atlantic and shortnose sturgeons. U.S.
Fish Wildl.Serv. Biol. Rep. 82(11.122). U.S. Army Corps of Engineers TREL-82-4. 28
pp.
Glenn, S., Associate Professor, University of Maryland College Park, Personal communication
with Nicole Zinn, May 15, 2006.
Hagood, S., Professor, Virginia Polytechnic Institute and State University, Personal
communication with Nicole Zinn, May 16, 2006.
Hall, J. K., M. Pawlus and E. R. Higgins. 1972. Losses of atrazine in runoff water and soil
sediment. J. Environ. Qual. 1:172-176.
Hormann, W. D., J. C. Tournayre and H. Egli. 1979. Triazine herbicides residues in central
European streams. Pest. Monit. J. 13:128-131.
Jacinthe P. -A., R. Lai, L.B. Owens, D. L. Hothem (2004). Transport of labile carbon in runoff
as affected by land use and rainfall characteristics. Soil and Tillage Research 77(2): 111-123.
Jobin, B., C. Boutin, J.L. DesGranges. (1997). Effects of agricultural practices on the flora of
hedgerows and woodland edges in southern Quebec. Can J Plant Sci 77:293-299.
176
-------
Kadoum, A. M. and D. E. Mock. 1978. Herbicide and insecticide residues in tailwater pits:
Water and pit bottom soil from irrigated corn and sorghum fields. J. Agric. Food Chem.
26:45-50.
Kaul, M., T. Kiely, and A. Grube, 2005. Triazine pesticides usage data and maps for cumulative
risk assessment, D317992. Unpublished EPA report.
Kaul, M. and J. Carter, 2005, Use Sites for Blanco, Hays, and Travis Counties in Texas for
Atrazine, Simazine, Carbaryl, Metolachlor, Diazinon, and Prometon (D322226).
Kenneth Rygwelski, 2002, personal communication regarding materials submitted for inclusion
in the first draft of the Lake Michigan Lake Wide Management Plan, LaMP 2002
Keinath, J.A., Musick, J.A. and R.A. Byles. (1987) Aspects of the biology of Virginia's sea
turtles: 1979-1986. Virginia J. Sci. 38: 329-336.
Kimmel, T. L. (2004) Sea Turtle Tagging and Health Assessment Study in Maryland's
Chesapeake Bay and Summary of Sea Turtle Strandings in Maryland, 1991 to 2003.
Maryland Department of Natural Resources, Cooperative Oxford Laboratory
Kimmel, Tricia. (2006) Maryland Department of Natural Resources. Personal Communication.
Kimmel, T.L., Driscoll, C. P. and J. R. Brush (2006) A Five Year Summary of Sea Turtle
Tagging Data From an In-Water Study in the Maryland Portion of the Chesapeake Bay:
Possible Evidence of Site Fidelity. Maryland Department of Natural Resources,
Cooperative Oxford Laboratory, Oxford, Maryland, USA
Kleijn, D., G.I. Snoeijing. (1997). Field boundary vegetation and the effects of agrochemical
drift: botanical change caused by low levels of herbicide and fertilizer. Journal of
Applied Ecology 34: 1413-1425.
Koch H., P. Weisser, M. Landfried. (2003). Effect of drift potential on drift exposure in
terrestrial habitats. Nachrichtenbl. Deut. Pflanzenschutzd. 55(9): S. 181-188.
Lutcavage, M., P.L. Lutz. 1986. Metabolic Rate and Food Energy Requirements of the
Leatherback Sea Turtle, Dermochelys coriacea. Copeia 1986 (3): 796-798. Cited in
NOAA Fisheries (http://swr.nmfs.noaa.gov/PIR/feis/Appendix%20C.PDF)
Lutz, P.L. and Musick, J.A., etal. (1997) The Biology of Sea Turtles. CRC Press, LLC.
Washington, D.C.
Majewski, M.S., Foreman, W.T., and Goolsby, D.A. 2000. Pesticides in the atmosphere of the
Mississippi River Valley, Part I - rain. Sci. Tot. Enviorn. 248. 201-212.
Mansfield, Kate. (2006) Virginia Institute of Marine Sciences. Personal Communication.
177
-------
Maryland Department of Natural Resources. 2006. Personal Communication with James M.
McCann, Zoologist on July 25, 2006.
Moody, J. A. and D. A. Goolsby. 1993. Spatial variability of triazine herbicides in the lower
Mississippi River. Environ. Sci. Technol. 27:2120-2126.
Moore, A. and Lower, N. (2001). The Impact of Two Pesticides on Olfactory-Mediated
Endocrine Function in Mature Male Atlantic Salmon (Salmo salar L.) Parr.
Comp.Biochem.Physiol.B 129: 269-276. EcoReference No.: 67727
Muir, D. C. G., J. Y. Yoo and B. E. Baker. 1978. Residues of atrazine and n-deethylated atrazine
in water from five agricultural watersheds in Quebec. Arch. Environ. Contam. Toxicol.
7:221-235.
Muir, M. Department of Natural Resources. 2006. Personal Communication. August 16, 2006.
Musick, J. A. (1979) Loggerhead. Proceedings of the Symposium on Endangered and
Threatened Plants and Animals of Virginia.
Nelson R.L., M.L. McHenry, W.S. Platts. (1991). Mining, Chap 12 in Influences of Forest and
Rangeland Management on Salmonid Fishes and Their Habitats, Meehan, WR, ed.
American Fisheries Society, Bethesda, MD.
NMFS (National Marine Fisheries Service and U.S. Fish and Wildlife Service). 1991. Recovery
Plan for U.S. Population of Atlantic Green Turtle. National Marine Fisheries Service,
Washington, D.C.
NMFS (National Marine Fisheries Service and U.S. Fish and Wildlife Service). 1991. Recovery
Plan for U.S. Population of Loggerhead Turtle. National Marine Fisheries Service,
Washington, D.C.
NMFS (National Marine Fisheries Service) and U.S. Fish and Wildlife Service. 1992. Recovery
plan for the Kemp's ridley sea turtle (Lepidochelys kempii). National Marine Fisheries
Service, St. Petersburg, Florida.
NMFS (National Marine Fisheries Service). 1998. Recovery Plan for the Shortnose Sturgeon
(Acipenser brevirostrum). Prepared by the Shortnose Sturgeon Recovery Team for the
National Marine Fisheries Service, Silver Spring, Maryland. 104 pages.
NMFS. 2004. Endangered species act biological opinion. April 19, 20004.
NOS, CCEHBR, Cooperative Oxford Laboratory, Oxford, Maryland, USA. Poster presented at
the 26th Annual Symposium on Sea Turtle Conservation and Biology in Greece.
178
-------
Ortiz R.M., R.M. Patterson, C.E. Wade, and F.M. Byers. 2000. Effects of actue fresh water
exposure on water flux rates and osmotic responses in Kemp's ridley sea turtles
(Lepidochelys kempi). Comp biochem Physiol. Part A 127: 81-87.
Pannill, 2006. Maryland Department of Natural Resources Forest Service. 2006. Personal
communication on August 29, 2006.
Powers, K. Maryland Department of Natural Resources. 2006. Personal communication on
August 14, 2006.
Richard, J. J., G. A. Junk, M. J. Avery, N. L. Nehring, J. S. Fritz and H. J. Svec. 1975. Residues
in water. Pest. Monit. J. 9:117-123.
Roberts, G. C., G. J. Sirons, R. Frank and H. E. Collins. 1979. Triazine residues in a watershed in
southwestern Ontario, Canada (1973-1975). J. Great Lakes Res. 5:246-255.
Rosgen, D.L. (1996) Applied Fluvial Geomorphology. Wildland Hydrology, Pagosa Springs,
CO.
Schulz, A., F. Wengenmayer and H. M. Goodman. 1990. Genetic engineering of herbicide
resistance in higher plants. Plant Sci. 9:1-15.
Schippers P., W. Joenje (2002). Modelling the effect of fertiliser, mowing, disturbance and width
on the biodiversity of plant communities of field boundaries. Agriculture, Ecosystems &
Environment 93(l-3):351-365.
Seney, E. E. (2003) Historical Diet Analysis of Loggerhead (Caretta caretta) and Kemp's ridley
(Lepidochelys kempi) Sea Turtles in Virginia. A Thesis Presented to the Faculty of the
School of Marine Science, The College of William and Mary in Virginia.
Sweeny, 2006. Personal communication with Jeff Sweeny of the Chesapeake Bay Tributaries
Strategy Program, July 14th, 2006.
Tierney D, Christensen B, Mattan C, and Dando C. 1999. Atrazine voluntary monitoring
program analytical results 1993 - 1998. December 13, 1999. Submitted by Novartis.
U.S. Department of Agriculture (USD A), Natural Resources Conservations Service (NRCS),
2000. Conservation Buffers to Reduce Pesticide Losses. Natural Resources
Conservation Service. Fort Worth, Texas. 21pp.
USDA. 2004. Agricultural Chemical Usage. 2003 Nursery and Floriculture Summary.
September, 2004. Available on-line at
http://usda.mannlib.cornell.edu/usda/current/AgriChemUsNursFlo/AgriChemUsNursFlo-
09-15-2004.pdf
179
-------
U.S. EPA. 1993. Wildlife Exposure Factors Handbook. EPA/600?R-93/187a. Office of
Research and Development, Washington DC.
U.S. EPA. 1998. Framework for Ecological Risk Assessment. EPA/630/R-95/002F
April 1998. Published on May 14, 1998, Federal Register 63(93):26846-26924.
U.S. EPA. 2003a. Interim Reregi strati on Eligibility Decision (IRED) for Atrazine, Case No.
0062, January 2003, EPA/OPP, Web address:
http://www.epa.gov/oppsrrdl/REDs/atrazine_ired.pdf
U.S. EPA. 2003b. Biological Evaluation for the Issuance of Ambient Water Quality Criteria for
Dissolved Oxygen, Water Clarity and Chlorophyll a for the Chesapeake Bay and Its Tidal
Tributaries. April 25, 2003. Region III, Chesapeake Bay Program Office.
U.S. EPA. 2003c. Ambient aquatic life water quality criteria for atrazine - revised draft. EPA-
822-R-03-023. October, 2003.
U.S. EPA. 2003d. White paper on potential developmental effects of atrazine on amphibians.
May 29, 2003. Office of Pesticide Programs, Washington D.C. Available at
http ://www. epa.gov/ scipoly/sap.
U.S. EPA. 2003e. Atrazine MOA Ecological Subgroup: Recommendations for aquatic
community Level of Concern (LOC) and method to apply LOC(s) to monitoring data.
Subgroup members: Juan Gonzalez-Valero (Syngenta), Douglas Urban (OPP/EPA),
Russell Erickson (ORD/EPA), Alan Hosmer (Syngenta). Final Report Issued on October
22, 2003.
U.S. EPA. 2004. Overview of the Ecological Risk Assessment Process in the Office of
Pesticide Programs. Office of Prevention, Pesticides, and Toxic Substances. Office of
Pesticides Programs. Washington, D.C. January 23, 2004.
U.S. EPA. 2006a. Risks of Atrazine Use to Federally Listed Endangered Barton Springs
Salamanders (Eurycea sosorum). Pesticide Effects Determination. Office of Pesticide
Programs, Environmental Fate and Effects Division. August 22, 2006.
U.S. EPA. 2006b. Cumulative Risk Assessment for the Chlorinated Triazines. Office of
Pesticides Programs. EPA-HQ-OPP-2005-0481. Washington, D.C. March 28, 2006.
U.S. EPA. 2006c. Memorandum from Special Review and Reregi strati on Division to
Environmental Fate and Effects Division: Errata Sheet for Label Changes Summary
Table in the January 2003 Atrazine IRED. June 12, 2006.
U.S. Fish and Wildlife Service and National Marine Fisheries Service. 1998. Endangered
Species Consultation Handbook: Procedures for Conducting Consultation and
Conference Activities Under Section 7 of the Endangered Species Act. Final Draft.
March 1998.
180
-------
U.S. FWS. 1993. Dwarf wedgemussel (alasmidonta heterodon) recovery plan. February 08,
1993.
U.S. FWS. 2006. http://www.fws.gov/midwest/mussel/current_threats.html. Accessed July 6,
2006.
USFWS/NMFS. 2004a. 50 CFRPart 402. Joint Counterpart Endangered Species Act Section 7
Consultation Regulations; Final Rule. FR 47732-47762.
USFWS/NMFS. 2004b. Letter from USFWS/NMFS to U.S. EPA Office of Prevention,
Pesticides, and Toxic Substances. January 26, 2004.
(http://www.fws.gov/endangered/consultations/pesticides/evaluation.pdf).
Virginia Department of Forestry. 2004. Operational Forestry Herbicide Use. Virginia Forestry
Herbicide User Group. Chesapeake Bay & Non-Chesapeake Bay Watershed Counties.
Available on-line at http://www.dof.virginia.gov/mgt/herbicide-03-use.shtml.
Virginia Department of Game and Inland Fisheries (VA DGIF). 2006.
DWM_locations_distl783. Vector digital data. Acquired August 01, 2006.
Waldron, A. C. 1974. Pesticide movement from cropland into Lake Eire. US EPA, Athens, GA,
EPA-660/2-74-03.
Wallace P, C.L. Williams, F.V. Paladino, S.J. Morreale, R.T. Lindstrom, and J.R. Spotila. 2005.
Bioenergetics and diving activity of internesting leatherback turtles Dermochelys
coriacea at Parque Nacional Marino Las Baulas, Costa Rico. J. Exp. Biol. 208: 3873 -
3884.
Wauchope, R. D. 1978. The pesticide content of surface water draining from agricultural fields -
a review. J. Environ. Qual. 7:459-472.
Wieser, C. M. and T. Gross. 2002. Determination of potential effects of 20 day exposure of
atrazine on endocrine function in adult largemouth bass (Micropterus salmoides).
Prepared by University of Florida, Wildlife Reproductive Toxicology Laboratory,
Gainesville, FL, Wildlife No. NOVA98.02e; submitted by Syngenta Crop Protection,
Inc., Greensboro, NC. (MRID No. 45622304).
Weissing F.J., J. Huisman. (1994). Growth and Competition in a Light Gradient. Journal of
Theoretical Biology 168(3):323-336.
Williams, W. M., Harbourt, C. M., Matella, M. K., Ball, M. H., and Trask, J. R., 2004. Atrazine
Ecological Flowing Water Chemical Monitoring Study in Vulnerable Watersheds Interim
Report: Watershed Selection Process. Waterborne Environmental, Inc. (WEI). WEI
Report No. WEI 936.32. pp 57.
181
-------
Wolke, R.E., Brooks, D.R. and A George. (1982) Spirorchidiasis in loggerhead sea turtles
(Caretta caretta). Pathology Journal of Wildlife Diseases 18(2): pp. 175-185. Wildlife
Disease Association.
Wu, T. L., D. L. Correl and H. E. H. Remenapp. 1983. Herbicide runoff from experimental
watersheds. J. Environ. Qual. 12:330-336. XI - 92
Wu, T. L. 1981. Atrazine residues in estuarine water and the aerial deposition of atrazine into
Rhode River, Maryland. Water, Air, Soil Poll. 15:173-184.
Zinn, N. and A. Jones, 2006, Refined Usage Data for Atrazine in Alabama, Maryland,
Pennsylvania, and Virginia, D328786.
Zug, G.R., and G.H. Balazs. (1985) Skeletochronological age estimates for Hawaiian green
turtles. Marine Turtle Newsl. (33):9-10.
182
------- |