PRO1
Risks of Atrazine Use to
Federally Listed Endangered
Barton Springs Salamanders
(Eurycea sosorum)
August 22, 2006
-------
Risks of Atrazine Use to Federally Listed
Endangered Barton Springs Salamanders
(Eurycea sosorum)
Pesticide Effects Determination
Environmental Fate and Effects Division
Office of Pesticide Programs
Washington, D.C. 20460
August 22,2006
-------
Table of Contents
1. Executive Summary 6
2. Problem Formulation 8
2.1 Purpose 9
2.2 Scope 10
2.3 Previous Assessments 10
2.4 Stressor Source and Distribution 13
2.4.1 Environmental Fate and Transport Assessment 13
2.4.2 Mechanism of Action 14
2.4.3 Use Characterization 14
2.5 Assessed Species 17
2.6 Action Area 20
2.7 Assessment Endpoints and Measures of Ecological Effect 23
2.8 Conceptual Model 25
2.8.1 Risk Hypotheses 25
2.8.2 Diagram 25
3. Exposure Assessment 28
3.1 Label Application Rates and Intervals 28
3.2 Aquatic Exposure Assessment 29
3.2.1 Background 30
3.2.2 Geology/Hydrogeology 34
3.2.3 Conceptual Model of Exposure 37
3.2.4 Existing Monitoring Data 39
3.2.5 Modeling Approach 43
3.2.5.1 Model Inputs 44
3.2.6 Individual Scenario Results 48
3.2.6.1 Residential 49
3.2.6.2 Turf 54
3.2.6.3 Fallow/Idle Land 54
3.2.6.4 Rights-of-Way 55
3.2.7 Characterization 59
4. Effects Assessment 64
4.1 Evaluation of Aquatic Ecotoxicity Studies 65
4.1.1 Toxicity to Freshwater Fish 67
4.1.1.1 Freshwater Fish: Acute Exposure (Mortality) Studies 67
4.1.1.2 Freshwater Fish: Chronic Exposure (Growth/Reproduction) Studies ..67
4.1.1.3 Freshwater Fish: Sublethal Effects and Additional Open Literature
Information 68
4.1.2 Toxicity to Aquatic-phase Amphibians 69
4.1.2.1 Amphibians: Open Literature Data on Mortality 69
4.1.2.2 Amphibians: Open Literature Data on Sublethal Effects 69
4.1.3 Toxicity to Freshwater Invertebrates 71
-------
4.1.3.1 Freshwater Invertebrates: Acute Exposure Studies 71
4.1.3.2 Freshwater Invertebrates: Chronic Exposure Studies 72
4.1.3.3 Freshwater Invertebrates: Open Literature Data 73
4.1.4 Toxicity to Aquatic Plants 73
4.1.4.1 Aquatic Plants: Laboratory Data 73
4.1.5 Freshwater Field Studies 74
4.2 Community-Level Endpoints: Threshold Concentrations 75
4.3 Use of Probit Slope Response Relationship to Provide Information on the
Endangered Species Levels of Concern 78
4.4 Incident Database Review 78
5. Risk Characterization 79
5.1 Risk Estimation 80
5.1.1 Direct Effects 80
5.1.2 Indirect Effects 81
5.1.2.1 Evaluation of Potential Indirect Effects via Reduction in Food Items
(Freshwater Invertebrates 82
5.1.2.2 Evaluation of Potential Indirect Effects via Reduction in Habitat
and/or Primary Productivity (Freshwater Aquatic Plants) 84
5.2 Risk Description 86
5.2.1 Direct Effects to the Barton Springs Salamander 87
5.2.2 Indirect Effects via Reduction in Food Items (Freshwater Invertebrates) 88
5.2.3 Indirect Effects via Reduction in Habitat and/or Primary Productivity
(Freshwater Aquatic Plants) 90
6. Uncertainties 92
6.1 Exposure Assessment Uncertainties 92
6.1.1 Modeling Assumptions 92
6.1.2 Impact of Vegetative Setbacks on Runoff 92
6.1.3 PRZM Modeling Inputs and Predicted Aquatic Concentrations 93
6.2 Effects Assessment Uncertainties 93
6.2.1 Age class and sensitivity of effects thresholds 93
6.2.2 Use of surrogate species effects data 94
6.2.3 Acute freshwater invertebrate toxicity data for the midge 94
6.2.4 Extrapolation of long-term environmental effects from short-term
laboratory tests 94
6.2.5 Use of threshold concentrations for community-level endpoints 95
6.3 Assumptions Associated with the Acute LOCs 95
7. Summary of Direct and Indirect Effects to the Barton Springs Salamander ... 96
8. References 97
-------
List of Tables
Table 1.1. Effects Determination Summary for the Barton Springs Salamander 8
Table 2.1. Summary of Assessment Endpoints and Measures of Ecological Effect.. 25
Table 3.1. Information for the Barton Springs Salamander Endangered Species
Assessment 29
Table 3.2. Summary of USGS Monitoring Data from the Four Springs Comprising
Barton Springs 40
Table 3.3. Summary of PRZM/EZAMS Environmental Fate Data Used for Aquatic
Exposure Inputs for Atrazine Endangered Species Assessment for the
Barton Springs Salamander 45
Table 3.4. Land Cover Adjustment Factors for the Action Area in the Barton Springs
Segment of the Edwards Aquifer (BSSEA) 47
Table 3.5. Summary of PRZM Output EECs for all Modeled (Edge of Field
Concentrations with Base Flow Incorporated) 58
Table 3.6. Comparison of Maximum, Typical, and 90th Percentile Label Rates and
Number of Applications 59
Table 3.7. Comparison of Residential and Rights-of-Way EECs Assuming Variable
Percentages of Overspray (0, 1, and 10%) 61
Table 3.8. Comparison of Residential EECs (granular) with 1% Over Spray and
Variable Percentages of Impervious Surface (10 and 3 0%) 62
Table 3.9. Comparison of Residential EECs (granular) with 10% Overspray and
Variable Percentages of Impervious Surface (10 and 3 0%) 62
Table 3.10. Comparison of Residential EECs (granular) Assuming Various
Percentages of Treated 1A Acre Lot (10 and 50%) 63
Table 4.1. Aquatic Toxicity Profile for Atrazine 66
Table 4.2. Categories of Acute Toxicity for Aquatic Organisms 67
Table 5.1. Summary of Direct Effect RQs for the Barton Springs Salamander 81
Table 5.2. Summary of RQs Used to Estimate Indirect Effects to the Barton Springs
Salamander via Direct Acute Effects on Dietary Items 83
Table 5.3. Summary of RQs Used to Estimate Indirect Effects to the Barton Springs
Salamander via Direct Chronic Effects on Dietary Items 84
Table 5.4. Summary of RQs Used to Estimate Indirect Effects to the Barton Springs
Salamander via Direct Effects on Aquatic Plants 85
Table 5.5. Summary of RQs Used to Assess Potential Risk to Freshwater Invertebrate
Food Items of the Barton Springs Salamander 89
Table 5.6. Summary of Modeled Scenario Time-Weighted EECs with Threshold
Concentrations for Potential Community-Level Effects 91
Table 7.1. Effects Determination Summary for the Barton Springs Salamander 96
-------
List of Figures
Figure 2.1 National Extent of Atrazine Use (Ibs) 16
Figure 2.2. Barton Springs Complex 19
Figure 2.3. Barton Springs Salamander Action Area 23
Figure 2.4. Conceptual Model for Barton Springs Salamander 26
Figure 3.1. Barton Springs Segment of the Edwards Aquifer with HydroZones 31
Figure 3.2. Hydrogeologic Cross Section of the Barton Springs Segment of the Edwards
Aquifer Showing Dominant Flow Pathways Within Each Hydrozone 33
Figure 3.3. Conceptual Model of Surface and Subsurface Flow Within the Barton Springs
Segment of the Edwards Aquifer Relative to the Barton Springs Salamander34
Figure 3.4. Flow paths within Recharge Zone of the Barton Springs Segment of the
Edwards Aquifer 36
Figure 3.5. Location of Surface Water Sites within the Barton Springs Segment 41
Figure 3.6. Location of Groundwater Sites Within the Barton Springs Segment 42
Figure 3.7. Flow Hydrograph Data for Barton Springs 43
Figure 3.8. Conceptual Model of Paired Residential/Impervious Scenarios 50
Figure 3.9. Percentage of Impervious Surface Coverage in Vicinity of Barton Springs.. 52
Figure 3.10. Representative Time Series Output from Paired Residential/Impervious
PRZM Scenario for Granular Applications 53
Figure 3.11. Conceptual Model of Rights-of-Way Scenario 56
Figure 4.1. Summary of Reported Acute LCso/ECso Values in Freshwater Invertebrates
for Atrazine 72
Figure 4.2. Use of Threshold Concentrations in Endangered Species Assessment 77
Appendices
Appendix A Ecological Effects Data
Appendix B Supporting Information for the Aquatic Community-Level
Threshold Concentrations
Appendix C Supporting Information for the Scenario Development
Appendix D Status and Life History of the Barton Springs Salamander
Appendix E Stepwise Approach to Modeling for the Barton Springs
Salamander Endangered Species Assessment (Using the
Residential Scenario as an Example)
Appendix F Incident Database Information
Appendix G RQ Method and LOCs
-------
1. Executive Summary
The purpose of this assessment is to make an "effects determination" for the Barton
Springs salamander (Eurycea sosorum) by evaluating the potential direct and indirect
effects of the herbicide atrazine on the survival, growth, and reproduction of this
Federally endangered species. 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).
The range of the Barton Springs salamander is restricted to four spring outlets that
comprise the Barton Springs complex, which is located near downtown Austin, Texas.
Subsurface flow from the Barton Springs segment of the Edwards Aquifer and its
contributing zone supply all of the water in the springs that make up the Barton Springs
complex. Therefore, the action area for the Barton Springs salamander is defined by
those areas within the hydro geologic framework of the Barton Springs segment of the
Edwards Aquifer.
Environmental fate and transport models were used to estimate high-end exposure values
expected to occur in the Barton Springs action area as a result of agricultural and non-
agricultural atrazine use in accordance with label directions. Modeled concentrations
provide "edge-of-field" estimates of exposure which are intended to represent atrazine
concentrations transported with runoff water directly to Barton Springs via subsurface
flow through the fractured karst limestone of the Edwards Aquifer. Estimated high-end
exposure values were compared with available monitoring data, although the monitoring
data are unlikely to capture the upper bounds of exposure due to sampling frequency. In
general, the modeled peak exposure estimates are two to ten times higher than
concentrations seen in the monitoring data, while the annual average concentrations
based on modeling are consistent with those seen in monitoring. The highest overall
modeled exposures were predicted to occur from residential uses of atrazine within the
action area.
The assessment endpoints for the Barton Springs salamander include direct toxic effects
on the survival, reproduction, and growth of the salamander itself, as well as indirect
effects, such as reduction of the prey base and/or modification of its habitat. Direct
effects to the Barton Springs salamander are based on toxicity information for freshwater
vertebrates, including fish, which are generally used as a surrogate for amphibians, as
well as available aquatic-phase amphibian data from the open literature. Given that the
salamander's prey items and habitat requirements are dependant on the availability of
freshwater aquatic invertebrates and aquatic plants, respectively, toxicity information for
these taxonomic groups is also discussed. In addition to the registrant-submitted and
open literature toxicity information, indirect effects to Barton Springs salamanders, 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.
Degradates of atrazine include hydroxyatrazine (HA), deethylatrazine (DBA),
deisopropylatrazine (DIA), and diaminochloroatrazine (DACT). Comparison of available
toxicity information for the degradates of atrazine indicates lesser aquatic toxicity than
the parent for freshwater and estuarine/marine fish, invertebrates, and aquatic plants.
Because the degradates are not of greater toxicological concern than atrazine,
concentrations of the atrazine degradates are not assessed further, and the focus of this
assessment is parent atrazine.
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 Barton Springs salamander via direct toxicity to the salamander or indirectly
based on direct effects to their food supply (i.e, freshwater invertebrates) or habitat (i.e.,
aquatic plants). When RQs for a particular type of effect are below LOCs, the potential
for adverse effects to the Barton Spring salamander 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 preliminary conclusion of "may effect". If a
determination is made that use of atrazine within the action area "may affect" the Barton
Spring salamander, additional information is considered to refine the potential for
exposure and effects, and the best available data are 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 Barton Springs salamander.
The best available data suggest that atrazine will either have no effect or is not likely to
adversely affect the Barton Springs salamander by direct toxic effects or by indirect
effects resulting from effects to aquatic invertebrates and plants. A summary of the risk
conclusions and effects determination for the Barton Springs salamander 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 5.2.
-------
Table 1.1. Effects Determination Summary for the Barton Springs Salamander
Assessment Endpoint Effects Determination Basis for Determination
Survival, growth, and
reproduction of Barton
Springs salamander
individuals via direct
effects
No effect
No acute and chronic LOCs are exceeded.
Indirect effects to Barton
Springs salamander via
reduction of prey (i.e.,
freshwater invertebrates)
May affect, but not likely
to adversely affect
Acute LOCs are exceeded based on the most sensitive
ecotoxicity value for the midge; however RQs for other dietary
items (amphipods, leeches, snails) are less than LOCs. Based
on the non-selective nature of feeding behavior in the Barton
Springs salamander and low magnitude of anticipated
individual effects to all evaluated prey species, atrazine is not
likely to indirectly affect the Barton Springs salamander via a
reduction in freshwater invertebrate food items. This finding is
based on insignificance of effects (i.e., effects to freshwater
invertebrates are not likely to result in "take" of a single Barton
Springs salamander) and discountability (i.e., the effect to
freshwater invertebrates is extremely unlikely to occur given
the estimated individual event probability of 1 in 45.5 million).
Indirect effects to Barton
Springs salamander via
reduction of habitat and/or
primary productivity (i.e.,
aquatic plants)
May affect, but not likely
to adversely affect
Although atrazine use may directly affect individual vascular
and non-vascular aquatic plants in Barton Springs, its use
within the action area is not likely to adversely affect the
Barton Springs salamander via indirect community-level
effects to aquatic vegetation. Predicted 14-, 30-, 60-, and 90-
day EECs for all modeled atrazine use scenarios within the
action area are well below the threshold concentrations
representing community-level effects. This finding is based on
insignificance of effects (i.e., community-level effects to
aquatic plants are not likely to result in "take" of a single
Barton Springs salamander).
2.
Problem Formulation
Problem formulation provides a strategic framework for the risk assessment. By
identifying the important components of the problem, 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 CFRPart 402 (USFWS/NMFS, 2004a; FR 69 47732-47762).
The structure of this risk assessment is based on guidance contained in 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
This ecological risk assessment is a component of the settlements for Center for
Biological Diversity and Save Our Springs Alliance v. Leavitt, No. L04CV00126-CKK
(filed January 26, 2004) and Natural Resources Defense Council, Civ. No: 03-CV-02444
RDB (filedMarch 28, 2006). The purpose of this ecological risk assessment is to make
an "effects determination," as directed in Section 7(a) (2) of the ESA, for the Barton
Springs salamander (Eurycea sosorum) by evaluating 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) on the survival, growth, and/or reproduction of this Federally
endangered species. The Barton Springs salamander was federally listed as an
endangered species on May 30, 1997 (62 FR 23377-23392) by the U.S. Fish and Wildlife
Service (USFWS or the Service). No critical habitat has been designated for this species.
In this endangered species assessment, direct and indirect effects to the Barton Springs
salamander are evaluated in accordance with the screening-level methodology described
in the Agency's Overview Document (U.S. EPA, 2004). It should be noted, however,
that the indirect effects analysis in this assessment utilizes 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 associated with potential aquatic community-level effects (via aquatic
plant community structural change and subsequent habitat modification) to the Barton
Springs salamander. Use of such information is consistent with the guidance provided in
the Overview Document, 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", the Agency will reach one of the following three
conclusions regarding the potential for atrazine to affect the Barton Springs salamander:
• "No effect";
• "May affect, but not likely to adversely affect"; or
• "Likely to adversely affect".
If the results of the screening-level assessment show no indirect effects and levels of
concern (LOCs) for the Barton Springs salamander are not exceeded for direct effects, a
"no effect" determination is made, based on atrazine's use within the action area. If,
however, indirect effects are anticipated and/or exposure exceeds the LOCs for direct
effects, the Agency concludes a preliminary "may affect" determination for the Barton
Springs salamander.
If a determination is made that use of atrazine within the action area "may affect" the
Barton Springs salamander, additional information is considered to refine the potential
for exposure at the predicted levels based on the life history characteristics (i.e., habitat
range, feeding preferences, etc.) of the Barton Springs salamander and potential
community-level effects to aquatic plants. Based on the refined information, the Agency
-------
will use the best available information to distinguish those actions that "may affect, but
are not likely to adversely affect" from those actions that are "likely to adversely affect"
the Barton Springs salamander. This information is presented as part of the Risk
Characterization in Section 5.
2.2 Scope
Atrazine is currently registered as a 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. It is one of the most widely used
herbicides in North America (U.S. EPA, 2003a).
The end result of the EPA pesticide registration process is an approved product label.
The label is a legal document that stipulates how and where a given pesticide may be
used. Product labels (also known as end-use labels) describe the formulation type,
acceptable methods of application, approved use sites, and any restrictions on how
applications may be conducted. Thus, the use, or potential use, of atrazine in accordance
with the approved product labels is "the action" being assessed.
This ecological risk assessment is for currently registered uses of atrazine in the action
area associated with the Barton Springs salamander. Further discussion of the action area
for the Barton Springs salamander is provided in Section 2.6.
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 freshwater fish, invertebrates, and aquatic plants. Specifically, the
available degradate toxicity data for HA indicates that it is not toxic to freshwater fish
and invertebrates at the limit of its solubility in water. In addition, available aquatic plant
degradate toxicity data for HA, DEA, DIA, and DACT report non-definitive ECso values
(i.e., 50% effect was not observed at the highest test concentrations) at concentrations
that are 700 to 10,000 times higher than the lowest reported aquatic plant ECso value for
parent atrazine. Given the lesser toxicity of the degradates, as compared to the parent, the
focus of this assessment is parent atrazine. A detailed summary of the available
ecotoxicity information for all of the atrazine degradates is presented in Appendix A.
2.3 Previous Assessments
The Agency completed a refined ecological risk assessment for aquatic impacts of
atrazine use in January 2003 (U.S. EPA, 2003a). This assessment was based on
laboratory ecotoxicological data as well as microcosm and mesocosm field studies found
in publicly available literature, a substantial amount of monitoring data for freshwater
streams, lakes, reservoirs, and estuarine areas, and incident reports of adverse effects on
aquatic and terrestrial organisms associated with the use of atrazine. In the refined
assessment, risk is described in terms of the likelihood that concentrations in water bodies
10
-------
(i.e., lakes/reservoirs, streams, and estuarine areas) equaled or exceeded concentrations
shown to cause adverse effects to aquatic communities and populations of aquatic
organisms. The results of the refined aquatic ecological assessment indicated that
exposure to atrazine is likely to result in adverse community-level and population-level
effects to aquatic communities at concentrations greater than or equal to 10-20 ug/L on a
recurrent basis or over a prolonged period of time.
During this time, the Agency extensively reviewed a probabilistic ecological risk
assessment submitted by the registrant (Giddings et al., 2000). The Agency's review of
Syngenta's probabilistic risk assessment is included in Appendix XVII of the 2003
atrazine IRED. EPA's refined risk assessment incorporates some of the data submitted
by the registrant in its probabilistic risk assessment.
The results of the Agency's ecological assessments for atrazine are fully discussed in the
January 31, 2003, Interim Reregi strati on Eligibility Decision (IRED)1. Because the
Agency had determined that atrazine shares a common mechanism of toxicity with the
structurally-related chlorinated triazines simazine and propazine, a cumulative human
health risk assessment for the triazines was necessary before the Agency could make a
final determination of reregi strati on eligibility. However, the Agency issued the interim
decision in order to identify risk reduction measures that were necessary to support the
continued use of atrazine. The January 2003 IRED requires extensive drinking water
monitoring in Community Water Systems (CWSs) where atrazine levels have exceeded
or are predicted to have the potential to exceed drinking water levels of concern. In
addition, the need for the following information related to potential ecological risks was
established: 1) an ecological monitoring program of potentially vulnerable water bodies
in corn, sorghum, and sugarcane use areas; and 2) further information on potential
amphibian gonadal developmental responses to atrazine.
EPA issued an addendum on October 31, 2003 that updated the IRED issued on January
31, 2003 (U.S. EPA, 2003b). This addendum describes new scientific developments
pertaining to ecological monitoring and mitigation of watersheds and potential effects of
atrazine on endocrine-mediated pathways of amphibian gonadal development.
The January 2003 IRED required atrazine registrants to develop a watershed monitoring
protocol. The resulting protocol identifies 40 indicator watersheds in corn and sorghum
growing areas in which monitoring has been required for a two-year period within each
watershed. The first 20 watersheds were monitored in 2004 and 2005. The second set of
20 watersheds was monitored in 2005, and the second year of sampling for these
watersheds is currently in progress. The goal of the monitoring is to ascertain the extent
to which any of the watersheds have streams with atrazine concentrations that could
cause significant changes in aquatic plant community structure, the most sensitive
endpoint in the aquatic ecosystem. Streams in watersheds exceeding the Agency's levels
of concern will be subject to mitigation consistent with watershed management principles
1 The 2003 Interim Reregi strati on Eligibility Decision for atrazine is available via the
internet at http://www.epa.gov/oppsrrdl/REDs/0001.pdf
11
-------
described by the Agency's Office of Water program requirements
(http://www.epa.gov/owow/tmdl/). These monitoring sites are representative of 1,172
watersheds determined to be among the most vulnerable to atrazine surface water loading
from use on corn and sorghum. Therefore, the results from the 40 watersheds will be
used to determine if further monitoring or remedial efforts are needed in the larger
population of watersheds. EPA has selected an atrazine level of concern (LOG) that is
based on significant aquatic community effects consistent with those described in the
Office of Pesticide Programs (OPP) 2003 ecological risk assessment (U.S. EPA, 2003a
and b) and the Office of Water's (OW) draft atrazine aquatic life criteria (U.S. EPA,
2003c). Further discussion of the aquatic community-level LOG is provided in Section
4.2 and Appendix B of this assessment. Aqueous atrazine concentrations obtained from
monitoring studies can be interpreted with the LOG to determine if a water body is likely
to be significantly affected.
As discussed in the October 2003 IRED, the Agency also conducted an evaluation of the
submitted studies regarding the potential effects of atrazine on amphibian gonadal
development and presented its assessment 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, the Agency summarized seventeen studies consisting of both open
literature and registrant-submitted laboratory and field studies involving both native and
non-native species of frogs. The Agency 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. The Agency 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, the Agency concluded and the SAP concurred that there was
sufficient evidence to formulate a hypothesis that atrazine exposure may impact gonadal
development in amphibians, but there were insufficient data to confirm or refute the
hypothesis (http://www.epa.gov/oscpmont/sap/2003/June/iunemeetingreport.pdf).
Because of the inconsistency and lack of reproducibility across studies and an absence of
a dose-response relationship in the currently available data, the Agency 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 Agency's 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, results are not available.
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/2003/iune/finaliune2002telconfreport.pdf.
12
-------
2.4 Stressor Source and Distribution
2.4.1 Environmental Fate and Transport Assessment
The following fate and transport description for atrazine was summarized based on
information contained 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. 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 only 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-mVmol) 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. It should also be noted that foliar dissipation rates for numerous pesticides
have generally been somewhat greater than otherwise indicated by their physical
chemical and other fate properties.
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 Ib 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 directly to soil during pre-planting and/or pre-emergence applications.
Atrazine is transported indirectly to soil due to incomplete interception during foliar
application, 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.
13
-------
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. Deethylhydoxy-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 the Agency uses to
classify degradates as "major degradates" (U.S. EPA, 2004), 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.
2.4.2 Mechani sm of Acti on
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.
2.4.3 Use Characterization
An analysis of available usage and land cover information, including extensive
discussions with local experts in the fields of agriculture and soil science, was completed
to determine which atrazine uses are likely to be present in the action area. This
evaluation is intended to place priority on those atrazine use areas likely to be in closest
proximity to the salamander's habitat. The analysis indicates that of all registered uses
for atrazine, the non-agricultural uses are likely to result in the highest exposures to the
salamander. This is due to the preponderance of potential residential and other non-
agricultural use sites (i.e., recreational and rights-of-way) in the immediate vicinity of
Barton Springs, and the fact that very little agricultural crops other than fallow uses for
the Conservation Reserve Program (CRP) are actually grown in the action area. Further
details on the analysis used to make this determination are discussed below and included
in Appendix C.
Critical to the development of appropriate modeling scenarios and to the evaluation of the
appropriate model inputs is an assessment of usage information. The Agency's
Biological and Economic Analysis Division (BEAD) provided an analysis of both
national and local use information for atrazine (Kaul et al., 2005, Zinn and Jones, 2006,
Kaul, et al., 2006). State level usage data were used to calculate county level usage
because no reliable county level data are available for Texas. State usage data were
14
-------
obtained from USDA-NASS3 and EPA proprietary data4 sources. 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.
Because no reliable county level usage data are available for Texas, average annual
pounds applied and acres treated by county were calculated by apportioning the estimated
state level usage to counties based on the proportion of total state acres grown of each
crop in each county. The most recently available acreage data were obtained from
USDA's 2002 Census of Agriculture. Estimates of the percent of each crop treated, the
number of applications and the application rate in each county are assumed to be the
same as the state level estimates. Apportioning the usage in this manner may
underestimate or overestimate the actual usage in a particular county.
In this analysis, the Agency gathered information on the agricultural uses of atrazine in
the three counties (Hays, Travis, and Blanco) located within or adjacent to the action area
for atrazine in the context of the Barton Springs salamander. Information was available
on crops for which atrazine is registered, amounts of atrazine used by county, application
rates, methods of application, application timing, and intervals between applications.
Usage information is critical in determining which uses should be modeled, while the
application methods, intervals, and timing are critical model inputs for estimating
atrazine exposure. While the modeling described below relies initially on maximum
label application rates and numbers of applications, the information on typical ranges of
application rates and number of applications is useful for characterization of the modeling
results. In general, for agricultural uses, atrazine is used in limited amounts relative to
national use patterns in Hays, Travis and Blanco counties.
Nationally 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. As a selective herbicide, atrazine is applied pre-emergence and post-
emergence. Figure 2.1 presents the national distribution of atrazine use from data
collected between 1998 and 2004 and used in the cumulative triazine assessment (U.S.
EPA, 2006a; Kaul et al., 2005).
3 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.
4 US EPA proprietary usage databases provide estimates of pesticide usage for select agricultural use sties
by chemical, crop and state.
15
-------
National Distribution of Atrazine Use (Ibs)
Legend
Lbs Atrazine
I 10-27729
H 27730 - 82603
(•82604-165432
M165433-441435
• 441436-1090674
250
.:.. I
Figure 2.1 National Extent of Atrazine Use (Ibs)
Locally, county level estimates of atrazine were derived using state level estimates from
USDA-NASS and EPA proprietary data. 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, in the three county area, approximately 20,000
Ibs of atrazine were used on corn, sorghum, wheat, cotton, and pecans in descending
order of total pounds applied.
Subsequent information based on land cover data (City of Austin, 2003a and b; USGS,
2003) and discussions with local experts (Davis, 2006; Garcia, 2006; Perez, 2006; see
Appendix C for more detail) indicates that most of the agricultural commodities listed
above are actually grown to the east of the action area and thus are not included in this
assessment. The land cover analysis indicates that, of possible agricultural uses for
atrazine, only the fallow/idle land use is likely to be present in the action area. Land
cover data also suggest that many of the currently registered non-agricultural atrazine
uses could not be excluded from the assessment (see Appendix C). However, the non-
agricultural forestry use of atrazine on conifers was not evaluated as part of this
16
-------
assessment because forest land cover data from the U.S. Geological Survey and the U.S.
Forest Service indicate that pine plantations are not present within the action area for the
Barton Springs salamander (http://nationalatlas.gov/atlasftp.html). The usage analysis
suggests that atrazine may be used on outdoor ornamental nurseries, although subsequent
information (Shay, 2006, personal communication; DeLong-Amaya, 2006, personal
communication; City of Austin, 2003a and b) suggests these uses are very limited in
nature and are not assessed (see Appendix C for more detail). Based on this analysis, a
suite of scenarios was developed, including a single agricultural scenario (fallow/idle
land) and four non-agricultural scenarios (residential, impervious, rights-of-way, and turf)
using local land cover, soils, and agronomic and climatic data specific to Travis and Hays
counties in Texas.
Application rates, number of applications, and application intervals were also estimated
(Zinn and Jones, 2006) for the three-county area. The minimum and maximum annual
application rates for atrazine were obtained from EPA data sources. Application rates are
provided at the state level for only crops grown in Blanco, Hays, and Travis counties on
which atrazine is registered. The minimum application rate was reported as the minimum
rate range. The 90th percentile application rate was reported as the highest application
rate at which at least 90% of the averaged total area is treated. Therefore, at least 90% of
the area is treated at this rate or less.
The only typical information available for a use site included in this assessment is for
fallow land (Kaul et al., 2006). This was reported as meadow use; however for this
analysis, it is used as a surrogate for atrazine use on fallow land under the CRP.
Application rates are in units of pounds per acre. The minimum reported application
rates for fallow ranged between 0.25 Ibs/acre and 0.5 Ibs per acre. The typical rate was
reported as 0.8 Ibs/acre for fallow, while the 90th percentile application rate for fallow
was 2 Ibs/acre. For fallow uses, the typical number of applications was 1. Information
on typical intervals for fallow was not available. Overall, atrazine is applied as a pre-
plant or pre-emergent herbicide to most sites in this part of Texas in late winter to mid
spring. No information was available for other agricultural crops and no data were
available for non-agricultural uses.
2.5 Assessed Species
A brief introduction to the Barton Springs salamander, including a summary of habitat,
diet, and reproduction data relevant to this endangered species risk assessment is
provided below. Further information on the status and life history of the Barton Springs
salamander is provided in Appendix D.
The Barton Springs salamander, shown in Figure D.I of Appendix D, is aquatic
throughout its entire life cycle. As members of the Plethodontidae Family (lungless
salamanders), they retain their gills, and become sexually mature and eventually
reproduce in freshwater aquatic ecosystems. The available information indicates that the
Barton Springs salamander is restricted to the immediate vicinity of the four spring
outlets that make up the Barton Springs complex (Figure 2.2), located in Zilker Park near
17
-------
downtown Austin, Texas. As such, this species has one of the smallest ranges of any
vertebrate species in North America (Chippindale, 1993). The Barton Springs segment of
the Edwards Aquifer and its contributing zone supply all of the water in the springs that
make up the Barton Springs complex. Flows of clean spring water are essential to
maintaining well-oxygenated water necessary for salamander respiration and survival.
The subterranean component of the Barton Spring salamander's habitat may provide a
location for reproduction (USFWS, 2005); however, little is known about the
reproductive biology of the Barton Springs salamander in the wild. It appears that
salamanders can reproduce year-round, based on observations of gravid females, eggs,
and larvae throughout the year in Barton Springs (USFWS, 2005).
Based on survey results, Barton Springs salamanders appear to prefer areas near the
spring outflows, with clean, loose substrate for cover, but may also be found in the
aquatic plants, such as moss. In addition to providing cover, moss and other aquatic
plants harbor a variety and abundance of the freshwater invertebrates that salamanders
eat.
18
-------
OJ
txi
Q
c
o
>
i_
01
•o
w
10
Figure 2.2. Barton Springs Complex (from Hauwert et al., 2005)
19
-------
2.6 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
protection of the Barton Springs salamander as they occur within hydrogeologic
framework of Barton Springs. 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 Barton Springs.
Unlike exposure pathways for most aquatic organisms, where stressors are transported
via surface water to the receptor within a defined watershed, the Barton Springs
salamander resides in a unique environment in which the source of the water, hence the
stressor, reaches the salamander via subsurface flow. Thus, the fate and transport of
atrazine is an important factor in defining the action area for the Barton Springs
salamander. The fate profile (see Section 2.4.1) describes why runoff from treated fields,
transported through the fractured limestone of the Edwards Aquifer, is considered the
principal route of exposure for the salamander. Thus, the action area for this assessment
is defined by those areas within the hydrogeologic "watershed" that drain to the springs.
In this case, the area draining to the springs is defined by the subsurface geologic
framework as opposed to surface hydrology. Figure 2.3 depicts the extent of the action
area based on this hydrogeologic framework. More detail on the definition of the action
area follows.
The Barton Springs salamander is known to inhabit only 4 springs (Main Barton Springs,
Eliza Springs, Old Mill Springs, and Upper Barton Springs; see Figure 2.2), located in the
Barton Springs Segment of the Edwards Aquifer (BSSEA), and associated subterranean
areas in the aquifer itself (USFWS, 2005). Barton Springs, located in Zilker Park near
downtown Austin, Texas is an aquifer-fed system consisting of four hydrologically
connected springs: (1) Main Springs (also known as Parthenia Springs or Barton Springs
Pool); (2) Eliza Springs (also known as the Elks Pit); (3) Old Mill Springs (also known as
Sunken Garden or Walsh Springs); and (4) Upper Barton Springs (Pipkin and Freeh,
1993). Collective flow from this group of springs represents the fourth largest spring
system in Texas (Brune, 1981). The springs themselves are fed by the BSSEA, and thus
groundwater input is the primary determinant of water quality for the salamander. Main
Springs supply the water for Barton Springs Pool, and during high groundwater flow
conditions, the surface water flow from Barton Creek may enter the pool if it overtops the
dam at the upper end of the pool. Thus, any pesticide used in the land areas contributing
to the groundwater in the Barton Springs segment of the aquifer or the surface water in
Barton Creek could potentially be transported to these areas.
Flow to the Barton Springs is controlled by the geology and hydrogeology of the BSSEA.
Numerous geological and groundwater studies (Slade et al., 1986, Hauwert et al., 2004)
20
-------
have been conducted that define the extent of the area contributing to the Barton Springs.
The BSSEA represents an approximately 150 square mile portion of the Edwards Aquifer
system in central Texas. Within the BSSEA, both surface water and groundwater flow
are controlled by the subsurface geology principally by the fracture nature of limestone
within portion of the BSSEA. This is particularly relevant for Barton Springs because
surface water flow from Barton Creek into the pool system is diverted via a bypass
channel upstream from the main pool to limit the input of surface water from Barton
Creek. Thus, the dominant source of water to the pool system is via subsurface flow.
Subsurface flow in the BSSEA as it relates to Barton Springs is well defined and includes
the Barton Creek watershed upstream of the springs accounting for potential surface
water inputs into Barton creek. The BSSEA is characterized as a karst system, which
permits relatively rapid transit of groundwater, with velocities along the dominant flow
path of 1-5 miles/day, depending on groundwater flow conditions (USFWS, 2005)
particularly within the fracture portions. Based on dye tracer studies, pesticides applied
within the recharge and contributing zones could potentially be present in the water of the
springs on a time scale of days to weeks (Hauwert et al., 2004).
Four hydrogeologic zones characterize the BSSEA. These are, from west to east, the
Contributing Zone, the Recharge Zone, the Transition Zone, and the Artesian Zone. Of
these zones, the Contributing and Recharge Zones have the greatest and most direct
influence on Barton Springs. There is evidence that the Transition Zone has some limited
input into the Barton Springs, while the Artesian Zone contributes no subsurface flow to
the springs (Slade et al., 1985, Hauwert et al., 2004). A more detailed description of the
geology and hydrogeology of the BSSEA is provided in Section 3.2.2.
In addition, an evaluation of usage information was completed to determine whether any
or all of the area defined by the BSSEA should be included in the action area. Current
labels and local use information were reviewed to determine which atrazine uses could
possibly be present within the defined area. 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 (City of Austin, 2003a and b; USGS, 2003) was
analyzed and interviews with the local agricultural sector (Davis, 2006; Garcia, 2006;
Perez, 2006; see Appendix C for more detail) were conducted to refine the
characterization of potential atrazine use in the areas defined by Hays, Travis, and Blanco
counties. The overall conclusion of this analysis was that while certain agricultural uses
could be excluded, and some non-agricultural uses of atrazine were unlikely, no areas
could be excluded from the final action area based on usage and land cover data.
Finally, the environmental fate properties of atrazine were evaluated to determine which
routes of transport are likely to have an impact on Barton Springs. Review of the
environmental fate data as well as physico-chemical properties suggests that transport via
overland and subsurface flow are likely to be dominant routes. Spray drift and/or long-
range atmospheric transport of pesticides could also potentially contribute to
concentrations in the aquatic habitat used by the salamander. Given the physico-
chemical profile for atrazine and the fact that atrazine has been detected in both air and
21
-------
rainfall samples, the potential for long range transport from outside the area defined by
the BSSEA cannot be precluded, but is not expected to approach concentrations predicted
by modeling (see Section 3.2.5). However, because areas where the atmospheric
component of atrazine loading is considered significant are typically high use areas
(Midwest corn belt), and the area surrounding Barton Springs is not a high use atrazine
area, the expected loadings from atmospheric transport of atrazine are not expected to
approach concentrations predicted by modeling (see Section 3.2.5).
Atrazine has been documented to be transported away from the site of application by both
spray drift and volatilization. The Agency typically addresses spray drift as a localized
route of transport off of the application site in exposure assessments. In the case of the
Barton Springs salamander assessment, spray drift is not considered to be a significant
route of exposure because the source area for atrazine reaching the springs is generally
removed from the spring system where the salamander resides, and the atrazine
exposures that reach the springs do so via subsurface flow. Therefore, there is no direct
pathway between the application site and receptor for drift to occur (no applications of
atrazine are reportedly made within the immediate vicinity of the springs). The Agency
does not currently have quantitative models to address the long range transport of
pesticides from application sites. 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. The full extent of the
action area could hypothetically be influenced by this route of exposure. However, given
the amount of direct use of atrazine within the immediate area surrounding the species
(Kaul, et al., 2006), the magnitude of documented exposures in rain (Majewski et al.,
2000; Majewski and Capel, 1995; Capel et al., 1994) at or below available surface water
and groundwater 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 subsequent
overland and subsurface flow for the purposes of this assessment.
Based on this analysis, the action area for atrazine as it relates to the Barton Springs
salamander is defined by the contributing, recharge, and transition zones within the
BSSEA. Figure 2.3 presents the action area graphically.
22
-------
Barton Springs Salamander Action Area
Blanco
Legend
0 Barton Springs
— Sam pled Creeks
Action Area Zones
Artesian Zone
^| Contributing Zone
I I Recharge Zone
^•Transition Zone
Figure 2.3. Barton Springs Salamander Action Area
2.7 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., Barton Springs salamanders), the ecosystems potentially at risk (i.e, Barton
Springs), the migration pathways of atrazine (i.e., runoff), and the routes by which
ecological receptors are exposed to atrazine-related contamination (i.e., direct contact).
Assessment endpoints for the Barton Springs salamander include direct toxic effects on
the survival, reproduction, and growth of the salamander itself, as well as indirect effects,
such as reduction of the prey base and/or modification of its 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 registrant-submitted
5 From U.S. EPA (1992). Framework for Ecological Risk Assessment. EPA/630/R-92/001.
23
-------
guideline tests that are performed on a limited number of organisms. Given that the
results of the required registrant-submitted amphibian toxicity tests are not available for
this assessment, it is assumed that fish and aquatic-phase amphibian toxicities are similar.
Birds are generally considered as surrogates for terrestrial-phase amphibians; however,
Barton Springs salamanders are neotenic (i.e., retain gills throughout their lives) and are
aquatic-phase amphibians. Therefore, fish are used as a surrogate for
amphibian/salamanders, in accordance with guidance specified in the Agency's Overview
Document (U.S. EPA, 2004). Additional ecological effects data from the open literature,
including effects data on salamanders and aquatic freshwater microcosm and mesocosm
data were also considered.
Measures of effect from microcosm and mesocosm studies 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 on aquatic
community effects were determined from complex time variable atrazine exposure
profiles (chemographs) within these experimental studies. Methods were developed to
estimate ecological community responses for any possible atrazine chemograph based on
the relationships in the micro- and mescocosm study results. This information was used
to determine whether a certain exposure profile within a particular use site and/or action
area may have exceeded a level of concern. 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 calibrate the measured atrazine chemographs in
order to estimate direct and indirect effects of atrazine and to project potential changes in
aquatic community structure and function.
A complete discussion of all the toxicity data available for this risk assessment, including
use of the CASM model and associated aquatic community-lev el 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 Barton Springs salamander risks associated with exposure to atrazine is
provided in Table 2.1.
24
-------
Table 2.1. Summary of Assessment Endpoints and Measures of Ecological Effect
Assessment Endpoint
Measures of Ecological Effect
1. Survival, growth, and reproduction of Barton
Springs salamander individuals via direct effects
la. Rainbow trout acute LC50
Ib. Brook trout chronic NOAEC
Ic. Open literature lab and field NOAEC data for
salamanders
2. Survival, growth, and reproduction of Barton
Springs salamander individuals via indirect effects
on prey (i.e., freshwater invertebrates)
2a. Midge acute EC50
2b. Scud chronic NOAEC
2c. Acute EC/LCso data for freshwater invertebrates
that are potential food items for the Barton Spring
salamander
3. Survival, growth, and reproduction of Barton
Springs salamander individuals via indirect effects
on habitat and/or primary productivity (i.e., aquatic
plant community)
3a. Vascular plant (duckweed) acute EC50
3b. Non-vascular plant (freshwater algae) acute
ECso
3c. Microcosm/mesocosm threshold concentrations
showing aquatic primary productivity community-
level effects
2.8 Conceptual Model
2.8.1 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.
Based on the results of the 2003 atrazine IRED (U.S. EPA, 2003a), the following risk
hypotheses are presumed for this endangered species assessment:
• Atrazine in groundwater, surface water, and/or runoff from treated areas may
directly affect Barton Springs salamanders by causing mortality or adversely affecting
growth or fecundity;
• Atrazine in groundwater, surface water, and/or runoff from treated areas may
indirectly affect Barton Springs salamanders by reducing or changing the composition of
prey populations; and
• Atrazine in groundwater, surface water, and/or runoff from treated areas may
indirectly affect Barton Springs salamanders by reducing or changing the composition of
the plant community in the springs, thus affecting primary productivity and/or cover.
2.8.2 Diagram
The conceptual model is a graphic representation of the structure of the risk assessment.
It specifies the stressor, release mechanisms, abiotic receiving media, biological receptor
types, and effects endpoints of potential concern. The conceptual model for the potential
effects of atrazine on the Barton Springs salamander is shown in Figure 2.4. Exposure
routes shown in dashed lines are not quantitatively considered because these exposures
are expected to be sufficiently low as not to cause direct or indirect effects to the Barton
Springs salamander.
25
-------
Stressor
Source
Atrazine applied to agricultural
fields, residential lawns, golf
courses, and rights-of-way
Receptor
Attribute
Change
Spray drift
^
r
Groundwater/
Surface Water
Atmospheric
transport
Springs
Aquatic plants
Aquatic invertebrates
Aquatic vertebrates
Individual
salamander
Reduced survival
Reduced growth
Reduced reproduction
Food chain
Decrease in abundance
Shift in prey base
Habitat
integrity
Reduced cover
_T
Figure 2.4. Conceptual Model for Barton Springs Salamander
The conceptual model provides an overview of the expected exposure routes for Barton
Springs salamanders within the atrazine action area previously described in Section 2.6.
In addition to freshwater aquatic vertebrates including Barton Springs salamanders, other
aquatic receptors that may be potentially exposed to atrazine include freshwater
invertebrates and aquatic plants. For freshwater vertebrate and invertebrate species, 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. Direct effects to freshwater invertebrates and aquatic plants resulting from
exposure to atrazine may indirectly affect the Barton Springs salamander via reduction in
food and habitat availability. The available data indicate that atrazine is not likely to
bioconcentrate in aquatic food items, with fish bioconcentration factors (BCFs) ranging
from 2 to 8.5 (U.S. EPA, 2003c). Therefore, bioconcentration of atrazine in salamanders
via the diet was not considered as a significant route of exposure.
Individual Barton Springs salamanders with the greatest potential to experience direct
adverse effects from atrazine use are those that occur in surface water and/or groundwater
with the highest concentrations of atrazine. Water passing into, and through Barton
Springs comes from groundwater in the Barton Springs Segment of the Edwards Aquifer.
26
-------
When Barton Creek floods, some of the surface flow enters Barton Springs Pool;
however, during normal flow, the water from Barton Creek enters a bypass channel
upstream from the main pool and does not enter the pool itself.
Based on historical records of pesticide use in Zilker Park and the area surrounding
Barton Springs dating to 1997, atrazine has not been used in this area (personal
communication with Elizabeth McVeety, pesticide applicator at Zilker Park, April 21,
2006). According to the City of Austin Parks and Recreation Department (PARD)
Integrated Pest Management Plan (IPM) (2005), the main concern within the Park is
control of fire ants, and spot treatment of Round-up (glyphosate) is the only herbicide
specified for control of Johnson grass and poison ivy. Although the IPM does not
specifically address atrazine use within Zilker Park, it is currently being revised to
specifically restrict atrazine use within the Park in the future (personal communication
with Elizabeth McVeety, pesticide applicator at Zilker Park, July 24, 2006). Given that
atrazine is not used within the Barton Springs area, it is unlikely that atrazine in runoff
would indirectly affect Barton Springs salamanders by reducing or changing the
composition of riparian zone vegetation and increasing sedimentation of the springs in
the main pool. Increased sedimentation in the main pool is more likely to result from
high groundwater flow conditions, when the surface water flow from Barton Creek
overtops the dam at the upper end of the pool. Therefore, potential indirect effects to
Barton Springs salamanders via reduction or change in the riparian zone vegetation (i.e.,
terrestrial plants) and resulting sedimentation are not considered a significant route of
exposure and are not further addressed in this risk assessment.
The source and mechanism of release of atrazine into surface and groundwater are
ground and aerial application via foliar spray and coated fertilizer granules to agricultural
(i.e., fallow/idle land) and non-agricultural sites (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 Barton Creek and/or runoff to
the recharge area of the Barton Springs Segment of the Edwards Aquifer, where it
becomes groundwater that discharges to the surface water of Barton Springs. Additional
release mechanisms include spray drift and atmospheric transport via volatilization,
which may potentially transport site-related contaminants to the surrounding air.
However, spray drift is not considered to be a significant route of exposure because the
source area for atrazine is generally removed from the spring system where the
salamander resides, and the atrazine exposures that reach the springs do so via subsurface
flow. Atmospheric transport is not considered as a significant route of exposure for this
assessment because the magnitude of documented exposures in rainfall are at or below
available surface water and monitoring data, as well as modeled estimates of exposure
(Majewski et al., 2000; Majewski and Capel, 1995; Capel et al., 1994).
27
-------
3. Exposure Assessment
3.1 Label Application Rates and Intervals
Atrazine labels may be categorized into two types: labels for manufacturing uses
(including technical grade atrazine and its formulated products) and end-use products.
While technical products, which contain atrazine of high purity, are not used directly
in the environment, they are used to make formulated products, which can be applied
in specific areas to control weeds. The formulated product labels legally limit
atrazine's potential use to only those sites that are specified on the labels.
In the January and October 2003 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 (U.S. EPA, 2003a and b). Specifically pertinent to
this assessment, the Agency entered into a Memorandum of Agreement (MOA) with the
atrazine registrants. In the MO A, the Agency stipulated that certain label changes must
be implemented on all manufacturing-use product labels for atrazine and on all end-use
product labels for atrazine prior to the 2005 growing season including cancellation of
certain 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. The modeling discussed below
predicts edge of field concentrations and thus spray drift is not quantitatively included in
the predicted exposures. It is expected that a setback distance will result in a reduction in
loading due to runoff across the setback zone; however, current models do not address
this reduction quantitatively. Therefore, these restrictions are not quantitatively evaluated
in this assessment. A qualitative discussion of the potential impact of these setbacks on
estimated environmental concentrations of atrazine for the Barton Springs salamander is
discussed further in Section 3.2.3. Table 3.1 provides a summary of label application
rates for atrazine uses evaluated in this assessment.
Currently registered non-agricultural uses of atrazine within the Barton Springs action
area include residential areas such as playgrounds and home lawns, turf (golf courses
and recreational fields), and rights-of-way. Agricultural uses within the Barton
Springs action area include fallow/idle land6 including Conservation Reserve
Program (CRP) lands in Texas. According to use data gathered by EPA, there is no
agricultural use of atrazine on corn or sorghum within the Barton Springs action area,
although corn and sorghum represent the greatest use nationally.
Atrazine is formulated as liquid, wettable powder, dry flowable, and granular
formulations. Application equipment for the agricultural uses includes ground
6 Fallow or ideland is defined by the Agency as arable land not under rotation that is set at rest for a period
of time ranging from one to five years before it is cultivated again, or land usually under permanent crops,
meadows or pastures, which is not being used for that purpose for a period of at least one year. Arable land,
which is normally used for the cultivation of temporary crops, but which is temporarily used for grazing, is
also included.
28
-------
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 considered in this assessment because they are expected to
result in the highest off-target levels of atrazine due to generally higher spray drift
levels. Ground boom and aerial modes of application tend to use lower volumes of
application applied in finer sprays than applications coincident with sprayers and
spreaders and thus have a higher potential for off-target movement via spray drift.
Table 3.1. Label Application Information for the Barton Springs Salamander
Endangered Species Assessmentl
Maximum
„ . Application
Scenano „ ,
Rate
(Ibs/acre)
Residential 2.0
Residential 1.0
Rights-of-
Way
Fallow/
Idle land
Turf 2.0
Turf 1.0
Maximum Date of ,„ ,, ... Interval
~T , „ ^. . ^ , .. Method of _ ,
Number of First Formulation . . . Between
Applications Application ^ Applications
2 April 1 Granular Ground 30 days
2 April 1 Liquid Ground 30 days
1 June 1 Liquid Ground NA
i XT 11 T • -j Ground and ,T.
1 November 1 Liquid . . , NA
Aenal
2 April 1 Granular Ground 30 days
2 April 1 Liquid Ground 30 days
1 - Based on 2003 IRED and Label Change Summary Table memorandum dated June 12,2006 (U.S. EPA,
2006b).
3.2 Aquatic Exposure Assessment
The exposure assessment represents an application of the standard approach outlined in
the Overview Document (U.S. EPA, 2004) for the hydrogeologic conditions of the
springs. The Agency's PRZM model was used to provide edge of field estimates of
exposure, which are assumed to be the concentrations of atrazine transported with runoff
water directly to Barton Springs via subsurface flow through the fractured limestone of
the Edwards Aquifer. Actual conditions are likely to result in lower atrazine
concentrations through dilution, mixing, retention, and degradation. Available
monitoring data from the spring systems were also evaluated and compared with model
estimates. While of high quality and targeted to the Barton Springs system, the
monitoring data are not considered to be robust in terms of capturing peak atrazine
concentrations (i.e. the sample frequency is likely to miss the peak concentration).
29
-------
New regionally-specific PRZM scenarios representing both agricultural and non-
agricultural use sites were developed following standard methodology (U.S. EPA, 2005)
to capture the upper bounds of exposure. Residential uses were modeled using pervious
(1/4 acre lot) and impervious surface scenarios and weighting the output based on local
data on the percentage of impervious surfaces in the action area region. Durations of
exposure were used to match available ecotoxicity thresholds. The highest overall
exposures were predicted to occur from the residential uses of atrazine that are in closest
proximity to the spring system. In general, the exposure assessment yields modeled peak
exposure estimates that are two to ten times higher than those seen in monitoring data,
while the annual average concentrations are consistent with those seen in monitoring.
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.
3.2.1 Background
The Barton Springs salamander resides in a geographically limited area defined by a set
of spring fed pools in the outskirts of the city of Austin. These pools represent the total
aerial extent of the salamander, as defined in Sections 2.5 and D.4 of Appendix D. The
pools are a unique system in that they are fed via two sources of water. Surface water has
historically reached the pool system via overland flow through Barton Creek. However,
water from Barton Creek is currently diverted near the inflow to the pool system and
provides only limited input to the pool system during high flow (flood) events. The bulk
of the water reaching the pool system is fed via a series of springs. The springs consist of
the Main Spring, Upper Spring, Old Mill Spring, and Eliza Spring with approximately
80% of the flow originating from the Main Spring. All of the springs are fed via
subsurface flow originating in fractured limestone aquifer of the Edwards Aquifer, which
trends south-southwest away from the pool system. Groundwater from the fractured
limestone (karst) is derived from perennial groundwater flow and via recharge that
originates from both surface streams and infiltration of rainfall. Therefore, the basic
conceptual model of exposure for this assessment focuses on the subsurface pathway
delivering groundwater to the pools via the karst system.
The hydrogeology of the Barton Springs Segment of the Edwards Aquifer (BSSEA)
defines the action area (see Section 2.6) of atrazine use for the Barton Springs
salamander. Several hydrogeologic zones define the BSSEA. From west to east, these
are the Contributing Zone, the Recharge Zone, the Transition Zone, and the Artesian
Zone. The relevance and route of exposure relative to the Barton Springs system is
different for each zone and is defined by the geology of the system. Given the basic
geology and hydrogeology of these zones within the BSSEA, the Contributing Zone and
the Recharge Zone (and to a lesser extent the Transition Zone) are likely to contribute
directly to the Barton Springs pool systems. Therefore, land use patterns within these
zones were considered to determine the potential for atrazine exposure to the Barton
Springs salamander. Figure 3.1 shows the extent of the BSSEA.
30
-------
Barton Spring Segment of the Edwards Aqui
i K^ ^f 'Jri _^---1-'^ p ^ 11 r0 - "
f ^ J j
Figure 3.1. Barton Springs Segment of the Edwards Aquifer with HydroZones
-------
Groundwater flow within the Recharge Zone is dominated by subsurface flow via
fractured limestone. Numerous studies have been conducted which document the nature
of the subsurface geology and the nature and extent of groundwater flow via these
fractures (Slade et al., 1986; Hauwert et al., 2004; Mahler, 2005a). Flow within these
fractures has been documented to travel from the point of origin to outflow at the springs
within hours to days of individual precipitation events, suggesting that atrazine reaching
the Recharge Zone is likely to have the most immediate and significant impact on Barton
Springs.
The Contributing Zone lies due west of the Recharge Zone. In this zone, runoff from
sites treated with atrazine is transported via overland flow to surface water streams and
ponds. Atrazine may then be transported via surface water streams to the Recharge Zone,
where it is available for infiltration into the network of karst fractures that ultimately feed
the Barton Springs system. Unlike stressors originating within the Recharge Zone, some
dilution and degradation is expected during this transport process. "Losing" streams
(defined as a stream where flow is lost to groundwater recharge) within the Recharge
Zone have been reported to provide as much as 85% (Slade et al., 1986) of the annual
recharge to groundwater. Historically, surface water flow through Barton Creek has
contributed to the loading of water, sediment, and contaminants to the Barton Springs
pools. However, in the current configuration of Barton Creek relative to the Barton
Springs pools, the creek has been artificially routed past the pools to ensure that the
springs are providing the bulk of the recharge to the pools. Occasionally, large
precipitation events may result in a bypass of this configuration overflowing of the pool
system. In general, the pools are typically fed by groundwater flow through the karst
fractures of the Recharge Zone that can receive stressors from both direct infiltration and
"loss" from surface water streams.
The Barton Springs system consists of a series of connected pools located within the city
limits of Austin, Texas. The Barton Springs salamander has been found within the
fractures (springs) feeding the pool system and within the pools themselves. Each
receptor location is somewhat unique from the other in how exposures are expected to
interact with the salamander.
Exposures to stressors for salamanders residing within the fracture system are due to a
combination of base flow with occasional runoff derived from pulses of increased flow.
With the increased flow comes the potential for an increase in the magnitude of exposure
that is of short duration depending on the climatic event. Base flow within the spring
systems is fed by loss of volume from surface streams as they traverse the Recharge Zone
of the BSSEA and from groundwater movement out of the Contributing Zone into the
fractured limestone of the Recharge Zone. The short term pulsed increases in runoff-
derived water through the springs are the result of increased loss through surface streams
originating in the Contributing Zone and direct infiltration of precipitation and runoff
from surface areas of the Recharge Zone. Thus, salamanders residing within the fracture
system of the springs are likely to be exposed to longer-term base flow concentrations of
atrazine with occasional shorter duration pulses of higher concentrations correlated with
precipitation derived runoff events transported through the fractures.
32
-------
Salamanders have also been found to reside within the pools themselves. In general, the
organisms residing in the pools will be exposed to the same sources of exposure.
However, it is expected that the magnitude and duration of exposure will be somewhat
different given the tendency of water to move through the pools (except in the most
extreme climatic events) more slowly. This suggests that exposures in the pools will be
generally lower in magnitude than in the springs, but will also tend to have a longer
duration of exposure than in the springs.
Figures 3.2 and 3.3 present the conceptual models of both of these potential exposure
pathways. More details on the geology and hydrogeology may be found in the following
section. Finally, a more complete description of the Barton Springs pool system in which
the salamander resides is provided in Section D.4 of Appendix D.
West
East
Contributing
Zone
Surface
Recharge
Zone
Artesian
Zone
Figure 3.2. Hydrogeologic Cross Section of the Barton Springs Segment of the
Edwards Aquifer Showing Dominant Flow Pathways Within Each Hydrozone
(Taken from Mahler, 2005a)
33
-------
Barton Creek flow
Barton Spring
Figure 3.3. Conceptual Model of Surface and Subsurface Flow Within the Barton
Springs Segment of the Edwards Aquifer Relative to the Barton Springs
Salamander
3.2.2 Geology/Hydrogeology
The Barton Springs pool system lies at the extreme northern end of the BSSEA, which is
a portion of a larger fractured limestone aquifer system known as the Edwards Aquifer.
The Edwards Aquifer and BSSEA are major sources of groundwater used for drinking
water and represent a critical source of water necessary to replenish surface water
resources for both recreational and ecological uses throughout the eastern half of Texas.
The Edwards Aquifer is a karst system of limestone and dolomite of Cretaceous age
(Slade et al., 1986). The aquifer covers roughly 6,000 square kilometers and stretches
from north of Austin to an area southwest of San Antonio. In general, the physical trend
of the Edwards Aquifer (and Barton Springs Segment) is south to north, and the
carbonate rocks within the aquifer dip to the east except where broken by fractures within
the Recharge Zone (Slade et al., 1986). The thickness of the aquifer generally increases
from north to south and is typically 400 to 450 feet thick (Slade et al., 1986). It is a
principal source of groundwater for drinking water in Texas, and where it discharges to
34
-------
the surface, it is critical for providing freshwater for both recreational and ecological
needs.
The Barton Springs Segment extends from the Colorado River south roughly 20 miles
into Hays County and covers 391 square kilometers. The Barton Springs Segment is
separated from the rest of the Edwards Aquifer by a hydrogeologic divide with
groundwater north of the divide flowing north-northeast towards the Colorado River and
south of the divide flowing south-southwest. In general, the BSSEA discharges at a
number of springs along the Colorado River and Barton Creek. Flow through the BSSEA
is typically around 35 cubic feet per second (cfs) during low flow periods, but can reach
above 75 cfs during high flow conditions, while the average flow is reported to range
between 53 cfs (Hauwert et al., 2004) and 56 cfs (Mahler, 2005a). Slade et al. (1986)
also estimated that up to 85% of the recharge reaching the BSSEA was derived from
infiltration from the main creeks crossing the Recharge Zone. The remaining infiltration
was derived from water coming from minor tributaries and from upland areas in the
Contributing Zone and from direct infiltration of precipitation.
Hauwert et al. (2004) conducted dye trace studies of the flow systems in the BSSEA
between 1996 and 2002. In these studies, the authors attempted to discern specific flow
patterns within the Recharge Zone using dye tracing, mapping of the potentiometric table,
water chemistry, local knowledge of geology, and cave mapping. Non-toxic dye
injection into caves, sinkholes, and wells was used to define the route of groundwater
flow, estimate flow velocities, and approximate travel times. The important finding of
this study relative to this assessment is that travel times within the Recharge Zone range
from hours up to one week in close proximity to the springs (defined by Travis County),
while farther south and west, travel times can increase to approximately 4 weeks. Flow
through fractures also may occurr within the Transitional Zone that separates the
Recharge Zone from the eastern artesian portion of the BSSEA. Figure 3.4 presents a
summary of the flow paths defined by this study (Hauwert et al., 2004).
35
-------
Summary of Groundwater
Trace Injections (1996-2002)
AUSTIN
— Barton Springs (See Inset bel
Figure 3.4. Flow paths within Recharge Zone of the Barton Springs Segment of the
Edwards Aquifer (Taken from Mahler, 2005a; originally published in Hauwert et
al.,2004)
36
-------
3.2.3 Conceptual Model of Exposure
Given the understanding of the geology/hydrogeology described above, a combination of
modeling and monitoring data is needed to assess the potential exposures from atrazine to
the Barton Springs salamander. Routes of exposure are dependent on the location of
registered use sites for atrazine within the action area (defined in Section 2.6 as the
Contributing, Recharge, and Transition Zones), the location of those uses, and locations
within the pool system (fractures versus pools) where the salamander resides. For
instance, uses that are predominantly within the Recharge Zone of the BSSEA are likely
to reach the springs via direct transport through the fractures within the karst zone.
Atrazine originating from within the Recharge Zone is analogous to "edge of field"
concentrations because atrazine applied within this area may be transported directly from
the site of application to the subsurface fractures. Thus, atrazine may move directly from
the edge of the treated field (or even from within it) into the fracture system. This route
of exposure is expected to be the most conservative (e.g. represent the highest potential
exposure).
The interconnected nature of the subsurface network can have a significant influence on
mixing, dilution, storage and degradation of flow through karst (Field, 2004). The
simplest, and for purposes of this assessment, most conservative assumption, is a straight
conduit from the source to the springs (defined as a Type I karst network in Field, 2004).
For the BSSEA, it is unclear how much, if any, interconnectedness exists in the Recharge
Zone. Therefore, a conservative assumption is that the system is represented by a straight
conduit network of fractures, and atrazine reaching the springs from source areas has
limited potential for degradation and dilution. The conservativeness of this assumption is
apparent when considering that source areas further removed from Barton Springs have
the potential for some dilution and degradation both at the surface and subsurface.
Atrazine residues derived from application sites that are located predominantly within the
Contributing Zone are expected to travel via overland flow to surface streams to the
Recharge Zone where infiltration from "losing" streams is likely to occur. Thus,
exposures from these sources are expected to be lower relative to the "edge of field"
exposures that originate in the Recharge Zone. Therefore, the "edge of field" approach is
likely to over-estimate exposure originating in these areas (i.e., the Contributing Zone).
Given the limited nature of the available monitoring data both within the spring network
and in the surrounding groundwater and surface water, an analysis of potential use sites
within the action area is needed. Available agricultural statistics, land cover data, usage
information, and soils data were evaluated relative to the hydrogeologic framework
described above. This information was used to determine whether both agricultural and
non-agricultural uses sites are present in the Recharge Zone, the Contributing Zone, or
both.
37
-------
Based on the hydrogeologic configuration, pesticide use is modeled in the Contributing,
Recharge, and Transition Zones. PRZM is used to model edge-of-field runoff
concentrations, assuming that the edge of field runoff concentration is transported
directly to the Barton Springs system via flow through the fracture system. This
exposure pathway represents a simplification of how the species may be exposed to
pesticides used in the BSSEA; however, it provides an upper bound of potential
exposures and is reasonable given that there are no available models for predicting
pesticide fate and transport in karst systems. This approach is considered to be
conservative because the conceptual model of transport does not include degradation or
dilution for use sites in close proximity to the springs.
Analysis of land cover data and usage information suggests that limited agriculture is
present in the Contributing and Recharge Zones of the BSSEA. In order to address the
potential for atrazine exposure from use on these sites, a suite of PRZM modeling
scenarios was developed for the specific agronomic, soil, and climatic data available for
the BSSEA. As noted above, the action area for the development of the Barton Springs
scenarios is comprised of three hydrologic zones of the BSSEA (in order of importance):
1) the Recharge Zone, which consists of a fractured karstic geology; 2) the Contributing
Zone, where surface runoff may flow to the Recharge Zone; and 3) the Transition Zone,
which has a remote potential to contribute to the Recharge Zone
(http://www.edwardsaquifer.net/intro.html). Although the Transition Zone is considered
in this assessment, primary emphasis is given to the Recharge Zone with secondary
emphasis on the Contributing Zone. No scenarios were parameterized based solely on
the Transition Zone. Spatial data containing the hydrozone boundaries were obtained
from the Barton Springs/Edwards Aquifer Conservation district
(ftp://www.bseacd.org/from/HCP Shape Files/). The areas to the east of the Recharge
Zone are not considered relevant to the assessment because groundwater flow to the
Barton Springs system comes either directly from transport through the Recharge Zone,
which occurs generally south to north, or indirectly via the Contributing Zone/Recharge
Zone interaction, where flow is dominantly west to east.
As previously discussed in Section 3.1, label changes, including the establishment of
setback restrictions on application of atrazine around wells and sinkholes, perennial and
intermittent streams, lakes, and reservoirs, were implemented as part of the IRED/MOA.
Specifically, the label changes restrict atrazine use within 50 feet of sinkholes, 66 feet of
intermittent and perennial streams, and 200 feet of lakes and reservoirs. These
restrictions are not quantitatively evaluated in this assessment.
As stated previously, this assessment assumes that the estimated environmental
concentration (EEC) is derived from an edge-of-field exposure; thus, spray drift is not a
factor in the predicted assessment. However, the assessment also conservatively assumes
that the edge of field exposure could occur where surface runoff enters the subsurface
fractures via sinkholes and/or other conduits to the karst fractures adjacent to the site of
application. The net effect of the 50 foot setback would reduce loadings to the subsurface
where the setback zone consists of a healthy vegetated zone. Alternatively, where the
setback zone consists of bare soil, or a poorly maintained vegetative zone, little reduction
38
-------
in runoff would be likely to occur. Current models do not estimate the effect of setbacks
on load reduction for runoff, although it is documented in the literature that well
vegetated setbacks can result in a substantial reduction in pesticide load to surface water
(USDA, NRCS, 2000). Specifically for atrazine, data reported in the USDA study
indicate that well vegetated setbacks 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 50 foot setback between the site of application of atrazine
and sinkholes could result in loading reductions to the subsurface system. Therefore, the
aquatic EECs presented in this assessment are likely to over-estimate exposure in areas
with well-vegetated setbacks. However, given the lack of quantifiable estimates of load
reduction and available data on the effectiveness of vegetated zones surrounding
sinkholes, ranges of potential exposures cannot be estimated. The label changes also
specify setback distances of 66 and 200 feet for atrazine applications surrounding
intermittent/perennial streams and lakes/reservoirs, respectively. Typically, the influence
of setback distances on spray drift loading would be evaluated using AgDrift to estimate
the impact of the setback on the fraction of drift reaching a surface water body (U.S.
EPA, 2004). However, as previously discussed, spray drift was not considered as a
significant route of exposure in this assessment.
Overall, it is expected that well vegetated and maintained setbacks will reduce overall
loading of atrazine from runoff estimates presented in this assessment. However, these
reductions cannot be quantified and are unlikely to be uniform across the action area.
3.2.4 Existing Monitoring Data
USGS provided monitoring data for surface streams, groundwater wells, and the four
springs making up the Barton Springs system (Mahler, 2005a). Specifically, the data
provided long-term trends within all three source types. In addition, recent data from the
USGS targeted single runoff events within the spring systems that included high
frequency sampling to match the hydrograph correlated with the several specific runoff
events.
Four springs were included in the USGS analysis, including Main Spring, Eliza Spring,
Upper Spring, and the Old Mill Spring. All four springs represent the main source of
inflow into the Barton Springs pool system with the Main Spring providing roughly 80%
of overall flow. Sampling and analysis of these springs indicates that the highest
detection of atrazine was 3 ug/L in the Upper Spring, however, most detections of
atrazine were below 1 ug/L. Given the nature of the flow regime within the springs, it is
unlikely that these sampling events have captured the peak exposures.
Evaluation of long-term trends in the monitoring data suggests that atrazine
concentrations have increased since the 1980s; however, given that recent detections
could be related to improved sampling techniques and analytical methods with lower
detection limits, these observations may reflect an increase in monitoring intensity rather
than a trend in atrazine exposures. Long-term trends in the analysis of spring water data
suggest that atrazine occurrence has been sporadic. Atrazine was not detected in spring
39
-------
samples prior to 2000; however, this may be due to changes in detection limits and an
increased attention in recent years on more frequent sampling and selected sampling tied
to specific runoff events (as noted above). More recent sampling suggests that longer-
term concentrations in the four springs tend to be less than 1 ug/L, with most of the
longer-term detections closer to 0.1 ug/L. Consideration of only detections yields an
overall average concentration of 0.45 ug/L (excluding non-detections). Table 3.2
presents a summary of the site-specific monitoring data.
Table 3.2. Summary of USGS Monitoring Data from the Four Springs Comprising
Barton Springs
Spring
Main
Upper
Old
Mill
Eliza
Range
of
Sample
Dates
1978-
2005
2001-
2005
2001-
2005
2000-
2005
#of
Samples
78
44
12
15
#of
Detects
60
44
12
15
Frequency
of Detection
77%
100%
100%
100%
Maximum
Concentration
(jig/L)
0.555
3.190
0.063
0.112
Minimum
Concentration
(Hg/L)
0.042
0.018
0.007
0.007
Average
Concentration
(jig/L)
0.070
0.164
0.015
0.036
Several degradates of atrazine were detected in the samples collected from the springs
including deethyl-atrazine, 2-hydroxyatrazine, chlorodiamino-s-triazine, and
deisopropylatrazine. All four degradates were detected in spring water at concentrations
below those found for atrazine. In general, the concentrations of the degradates were 2 to
3 times less than atrazine during low flow (base flow) periods, while atrazine
concentrations were 10 to 15 times higher than deethyl-atrazine during high flow (storm
derived runoff events).
Analysis of the stream data suggests that trends similar to those seen in the spring data
occurr in surface streams. Similar to the spring detections, most of the atrazine detections
in surface streams have occurred since 2000. The highest detection of atrazine in the
streams is 4.39 ug/L in Slaughter Creek in 2005. Figure 3.5 shows the location of stream
samples within the BSSEA.
40
-------
~^, ,
ak
,5 r
Surface Water Sites in the
Barton Springs Segment of the Edwards Aquifer
Figure 3.5. Location of Surface Water Sites within the Barton Springs Segment
Statewide surface water data for atrazine collected by the USGS NAWQA Program
between 1993 and 2003 shows that atrazine was detected in 792 out of 866 samples
(92% detection frequency). The highest detected concentration of atrazine in all of the
NAWQA samples was 20 ug/L in 1994 and 1995 from two sites in the Trinity River
Study Unit in Ellis and Navarro counties (station ID numbers 321313096415201 and
321017096420099) that are outside of the action area for this assessment.
Analysis of the well data from the USGS collected between 2000 and 2005 suggests that
atrazine was detected in groundwater from five locations within the BSSEA. In general,
with the exception of a single sample analyzed from well # YD-58-34-617 in 2002 at
0.192 ug/L, all atrazine detections were below 0.1 ug/L. More commonly detected were
the principal degradates of atrazine including deethyl-atrazine, 2-hydroxyatrazine,
chlorodiamino-s-triazine, and deisopropylatrazine, none of which are included in this
assessment (rationale for this exclusion is provided in the atrazine 2003 IRED and
Section 2.2 of this assessment). Overall, the low frequency and magnitude of detection
of atrazine suggests that its occurrence in baseflow may be a minimal source of exposure
at Barton Springs. Figure 3.6 presents the location of the groundwater wells within the
BSSEA.
41
-------
fc&kl LynWh Bh&hnson
Groundwater Wells in the
Barton Springs Segment of the Edwards Aquifer
Canyon Lake
Figure 3.6. Location of Groundwater Sites Within the Barton Springs Segment
Overall, the monitoring data provided by the USGS indicate relatively consistent low-
level concentrations of atrazine over time with periodic spikes related to storm-derived
runoff events. Because of the limited nature of the runoff-related sampling, it is not
possible to determine whether these data are representative of overall peak exposures
(Mahler, personal communication, 2005b). Therefore, these data represent a lower bound
on exposures and are considered to be representative of long-term baseflow exposure in
the spring system.
42
-------
A key component of the total load reaching Barton Springs is represented by base flow.
For this assessment, a reasonably conservative estimate of atrazine load arriving via base
flow is represented by the average concentration in all springs of 0.45 ug/L. This
estimation is conservative because it excludes all non-detections of atrazine and includes
concentrations associated with storm events. The USGS has estimated base flow
concentrations of atrazine to be less than 0.1 ug/L (Mahler, 2005a). Information on the
average flow rate through the spring system was evaluated. As noted above, Hauwert et
al. (2004) estimated an average flow rate through the entire system of 53 cfs, while
Mahler (2005a) reported an average of 56 cfs. Hydrograph data for Barton Springs from
the USGS (Figure 3.7) yields an average flow of 62 cfs.
Mean Flow (cfs) for Barton Springs
200 -t
180
o
Date
Figure 3.7. Flow Hydrograph Data for Barton Springs
3.2.5 Modeling Approach
The analysis of available monitoring data and usage information indicates that the
exposure assessment cannot rely exclusively on monitoring data. Although of high
quality and in selected instances targeted to pesticide use and single runoff events, the
unique nature of flow through the BSSEA and the relationship of the flow regime to the
Barton Springs salamander indicates that the exposure assessment should rely on
modeling to augment the available monitoring data.
43
-------
Typically, the Agency conducts modeling using scenarios intended to represent use sites
in areas that are highly vulnerable to either runoff, erosion, or spray drift. Runoff
estimates predicted by the PRZM model are linked to the Exposure Analysis Modeling
System (EXAMS). For ecological risk assessment, the Agency relies on a standard water
body to receive the edge of field runoff estimates. The standard water body is of fixed
geometry and includes processes of degradation and sorption expected to occur in ponds,
canals, and low order streams (e.g. first and second order streams), but with no flow
through the system.
The unique geology/hydrogeology of the BSSEA suggests that, for the use sites being
evaluated, an estimate of exposure in surface water may not be suitable by itself. If the
use site resides exclusively within the Recharge Zone, the principal route of exposure is
expected to be either via "edge of field" runoff into an adjacent fracture or via overland
flow to a stream which subsequently "loses" some of that flow to the fracture system. In
order to provide the most conservative estimate for each scenario modeled, edge of field
concentrations are used to represent sources originating with the Recharge Zone (instead
of EXAMS concentrations) to mimic the pulsed nature of exposures moving through
karst fractures.
Loading of atrazine to the system via base flow is approximated using available
monitoring data. The average concentration from the available USGS monitoring data
from the four springs of 0.45 ug/L is added to the daily edge of field EEC. The volume
associated with the base flow is accounted for by using the long-term average flow rate
from the USGS hydrograph data for Barton Springs of 63 cfs. PRZM reports a daily flow
at the edge of field in units of centimeters representing the depth of water leaving the 10
hectare field. An equivalent volume is calculated by multiplying the edge of field runoff
depth by the total area of the 10 hectare field in centimeters. In order to use the average
flow rate from the hydrograph data, the allocation of 63 cfs was back-calculated to the
entire BSSEA drainage area of 900 square kilometers. This estimation yields an
approximate additional flow through the Recharge Zone of 1,000,000,000 cmVday
(PRZM reports this as the depth of water running off the entire 10 hectare watershed, or 1
cm) which is added to the daily edge of field estimate.
Peak atrazine concentrations, as well as rolling time-weighted averages of 14 days, 21
days, 30 days, 60 days, and 90 days are calculated for comparison with various
ecotoxicity endpoints (including aquatic community-level threshold concentrations) for
atrazine.
3.2.5.1 Model Inputs
EECs from surface water sources were calculated using the Agency's Tier II PRZM
model. PRZM is used to simulate pesticide transport as a result of runoff and erosion
from a standardized watershed. The linkage program shell, PE4v01.pl, which
incorporates the site-specific scenarios developed by the Agency, was used to run PRZM
(U.S. EPA, 2005). However, new, site-specific scenarios were developed for use in this
assessment. Linked site-specific use scenarios and meteorological data are used to
44
-------
estimate exposure for each modeling scenario. Weather and agricultural practices are
simulated over 30 years to estimate the 1 in 10 year exceedence probability at the site.
Further information on these models may be found at:
http://www.epa.gov/oppefedl/models/water/index.htm
The appropriate PRZM input parameters were selected from environmental fate data
submitted by the registrant and in accordance with US EPA-OPP EFED water 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 atrazine IRED (U.S. EPA,
2003a) and the cumulative triazine risk assessment (U.S. EPA, 2006a); no new
environmental fate data were incorporated into this assessment. The date of first
application was identified based on 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 3.3.
Table 3.3. Summary of PRZM/EZAMS Environmental Fate Data Used for Aquatic
Exposure Inputs for Atrazine Endangered Species Assessment for the Barton
Springs Salamander
Fate Property
Molecular Weight
Henry's constant
Vapor Pressure
Solubility in Water
Photolysis in Water
Aerobic Soil Metabolism Half -lives
Hydrolysis
Aerobic Aquatic Metabolism (water
column)
Value
215.7
2.58 xlO'9
3 x 10~7 atm-nrVmol
33 mg/1
335 days
152 days
Stable
304 days
MRID (or source)
MRID 41379803
MRID 41379803
MRID 41379803
MRID 41379803
MRID 42089904
MRID 4043 1301
MRID 40629303
MRID 42089906
MRID 4043 13 19
2x aerobic soil metabolism
rate constant
45
-------
Fate Property Value MRID (or source)
Anaerobic Aquatic Metabolism 6QO 4Q43 ^
(benthic) J
MRID 40431324
MRID 41257901
„ oo-,o „ MRID 41257902
Koc 88.78 ml/g MRID 41257904
MRID 41257905
MRID 41257906
, .. . _„.. . 95% for aerial , ,- ,. ,2
Application Efficiency „„ „/ ^ i default value
^ J 99 % for ground
o T-V -jv T- +• i 5 % for aerial , - ,, , 2
Spray Dnft Fraction , n/ - , default value
v J 1 % for ground
1 - Spray drift not included in final EEC due to edge-of-field estimation approach
2 - Inputs determined in accordance with EFED "Guidance for Chemistry and Management Practice Input Parameters
for Use in Modeling the Environmental Fate and Transport of Pesticides " dated February 28, 2002
Unlike the Agency's standard ecological risk assessment methodology that relies on
EECs derived using an EXAMS standard water body, edge of field concentrations were
predicted for those use sites expected to reside within the action area. It is expected that
infiltration directly into the fractured limestone represents the most direct route of
exposure and is likely to yield the highest EECs. In this instance, PRZM alone was used
to estimate the edge of field concentration. Unlike the typical approach used to estimate
exposure using the EXAMS water body described above, the PRZM output from the *.zts
file (daily time series data) associated with each scenario modeled was used. In this
instance, the *.zts file provides daily estimates of atrazine exposures. The PE4v01.pl
script was modified to provide time series output (TSER in PRZM) as opposed to the
standard cumulative output (TCUM in PRZM) of runoff volume, mass transported with
runoff, infiltration volume, mass transported with infiltration, eroded sediment, and mass
of pesticide transported with sediment. The runoff volume and runoff mass were
extracted and converted to a runoff concentration.
The standard approach for conducting ecological risk assessment assumes that 100% of
the 10-hectare watershed is covered by the relevant use. This approach also assumes that
the standard water body is adjacent to the edge of the field. In this assessment, the
majority of the use sites, with the exception of residential uses, are either sporadically
present within the action area or are predominant further south and west of the spring
systems. Therefore, it is unlikely that edge of field concentrations for these use sites are
equivalent to the standard assumption that the water body is receiving runoff from a small
watershed that is adjacent to a field that is 100% cropped and treated. Although travel
through the fractures is likely to be direct, there is also likely to be non-impacted fracture
flow arriving simultaneously at the springs from the same runoff event. In order to
account for this, an adjustment factor is applied to each of the use sites located outside
the immediate area of the springs. These factors are based on an assessment of recent
land cover data with specific cover types that correlate with the use sites. A summary of
the land cover-based adjustment factors is presented in Table 3.4.
46
-------
Table 3.4. Land Cover Adjustment Factors for the Action Area in the Barton
Springs Segment of the Edwards Aquifer (BSSEA)
Scenario Modeled
Land Cover Adjustment Factor
Residential
70% of action area in vicinity of Barton Springs (%
decreases with distance from spring system)
Impervious
30% of action area in vicinity of Barton Springs (%
decreases with distance from spring system)
Turf
Rights-of-Way
Fallow/Idle land
100% of action area in vicinity of Barton Springs
and 28% of treatable golf course (based on limited
occurrence of golf courses within action area)
10% of action area in vicinity of Barton Springs
(based on density near springs; % decreases with
distance from spring system)
5% of action area in vicinity of Barton Springs
(estimated from land cover of entire action area;
majority of fallow/idle land is located at the
southern and western edge of action area)
The edge of field concentrations are post-processed (see Appendix E for details) in order
to provide durations of exposure. First, daily concentrations were calculated using the
time series data described above for all 30 years of model output. Then, peak, 14-day,
21-day, 30-day, 60-day, and 90-day average concentrations were calculated across the
entire 30 years of data. In order to match the standard PRZM/EXAMS output, the 30
years of daily and rolling averages were separated into individual years and the maximum
value for each of the 30 years was calculated for both peak and rolling averages. Finally,
the 30 years of maximum values were ranked and the 90th percentile from the rankings
was selected as the final EEC for use in risk estimation.
The time series output generated by PRZM provides individual runoff events and does
not capture the influence of base flow in the fracture zone. In order to account for the
influence of base flow, the average of all monitoring data detections (excluded non-
detections) was added to the time series output. For atrazine, the estimated base flow
value of 0.45 ug/L is considered to be conservative because it includes recently analyzed
storm runoff data and excludes all non-detections. In addition, the flow associated with
the base flow exposures was estimated using USGS hydrograph data for Barton Springs
and was also added into the exposure estimate. Addition of base flow exposure to the
estimated exposure was intended to provide an upper bound on base flow and account for
the limitations in the sample frequency that may not completely represent base flow.
47
-------
The 0.45 ug/L estimate of base flow concentration is considered to be a conservative
estimate for several reasons. First, the estimate is based on all detections from the site-
specific USGS monitoring data and includes samples collected during high flow events.
Second, the value excludes all non-detections. These first two points are important
because the USGS estimates that the actual base flow concentration of atrazine from the
non-runoff driven sampling is less than 0.1 ug/L (Mahler, 2005a).
Calculating an annual average concentration from the PRZM generated edge of field
concentrations without the added 0.45 ug/L value yields an annual average concentration
of roughly 0.15 ug/L that is consistent with the USGS estimate. Therefore, addition of
the 0.45 ug/L atrazine concentration is assumed to be conservative and protective. The
analysis indicates that long term averaging of the PRZM edge of field output provides
estimated exposures consistent (within a factor of two to three times) with the estimate of
base flow predicted by the USGS from the site specific monitoring data. Therefore, the
approach taken in this assessment is considered to be reasonable and protective. The
approach is reasonable given that the monitoring data, though of high quality and targeted
to Barton Springs, is not considered to be of sufficient robustness to provide an upper
bound on both peak and longer-term exposures due to the limited sampling frequency in
the study design. This approach is protective because use of the 0.45 ug/L concentration
as a base flow value is greater than any estimate of base flow available based on a limited
monitoring data set.
The calculations described above were completed by a straight comparison of runoff
mass divided by runoff volume to get a daily estimated EEC. However, a second step
was included to evaluate the influence of infiltrating water on the overall exposure. This
additional step was included to mimic the influence of infiltrating water into the fractured
limestone of the Recharge Zone and to account for base flow concentrations associated
with some amount of infiltrating water not accounted for in the runoff edge of field EEC.
The additional volume of water predicted by PRZM to leach out of the bottom of the soil
profile was added back into the edge of field concentration. This additional step accounts
for recharge to the fractured limestone that is likely to occur from infiltration of runoff
water through fractures/sinkholes and direct infiltration throughout the soil profile.
3.2.6 Individual Scenario Results
A total of four scenarios were developed for this assessment, including residential, turf,
rights-of-way, and fallow/idle land. Of these, two scenarios are used in tandem with the
impervious scenario (residential and rights-of-way). Discussions with local experts
(Markwardt, 2006; Ward, 2006; Mason, 2006) suggest that atrazine is never used in the
rights-of-way scenario within the action area. However, the rights-of-way scenario was
modeled, given its potential as a use site in the action area. It should be noted, however,
that the predicted EECs associated with rights-of-way are less relevant to actual
exposures than the remaining scenarios and are presented for a qualitative comparison.
Model inputs were selected for atrazine using the most recent data available from the
atrazine IRED (U.S. EPA, 2003a) and the triazine cumulative assessment (U.S. EPA,
48
-------
2006a). A discussion of each assessed exposure scenario and a summary of the results
for each is provided below. Copies of the model input files along with the stepwise
approach for processing model output are provided in Appendix E.
3.2.6.1 Residential
The residential exposure scenario represents two scenarios modeled in tandem. The first
scenario is intended to reflect runoff and erosion from a typical !/4 acre lot and reflects a
typical urban/suburban use site with homeowner and professional applications. The
residential lot scenario was developed using local soil information and a USDA runoff
curve number developed specifically for Vi acre lots (USDA, 1986). In order to justify
the assumption of !/4 acre lot as a typical exposure scenario, publicly available data was
reviewed from the United States Census (Census). Specifically, data from 2003 from the
American Housing Survey (AHS) available at the following website was reviewed.
http://www.census.gov/hhes/www/housing/ahs
Initially, the data for all suburban homes available nationally was reviewed. It is
assumed that most pesticide applications, particularly for herbicides, occur in suburban
settings. In order to test the assumption of the !/4 acre lot as the best representation, AHS
data for 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 reviewed. 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 located on lots between 1/8 acre and !/4 acre, while 9,339,000
units (the second largest class at 21%) were located on lots between !/4 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 selection of the Vi acre lot was
an assumed to provide an estimate of potential exposure in an urban/suburban scenario
where it is expected that most herbicide use will occur. It is believed that this
representation provides a reasonable estimate of typical uses in an urban/suburban
watershed particularly as it relates to the City of Austin and its rapidly developing
outskirts.
The second scenario was developed to represent general impervious surfaces expected to
be present in an urban/suburban watershed. Examples of representative impervious
surfaces include roads, parking lots, and buildings. These surfaces are distinct from the
impervious surfaces inherent in the !/4 acre lot (driveways, sidewalk, and house). The
impervious surface uses a high-end curve number (98 out of a maximum of 100) to
mimic the runoff expected from these surfaces. Using these in tandem allows for a
weighting of the runoff potential of both surface types within a residential watershed that
is different from the standard agricultural watershed that assumes uniform land cover.
Figure 3.8 presents the conceptual model of the paired impervious/pervious scenario
approach.
49
-------
10 Hectare Field - 50% Impervious/50%
Pervious (1/4 acre lot) Draining to 1
Hectare Water body
Figure 3.8. Conceptual Model of Paired Residential/Impervious Scenarios (green
square represents 1A acre lot while black square represents impervious surface
scenario. Ratio of pervious to impervious surface based on best available land cover
data)
For edge of field EECs, the output is weighted based on the percentage of impervious
surface present in the action area and by the percentage of the 1A acre lot treated. For this
assessment, it is assumed that 30% of the action area in the vicinity of the spring system
is impervious (see Figure 3.9). This assumption is reasonable given the density of
residential development surrounding the springs. Outlying areas are likely to have lower
percentages of impervious surfaces; however, residential areas in close proximity to the
springs are likely to be most representative of the expected edge of field concentrations.
Additional analysis of the impact of alternative assumptions for percent impervious
50
-------
surface, overspray, and percentage of lot treated is included in Section 3.2.7. Because of
the unique nature of the karst system, it is assumed that no direct spray drift will reach the
spring system (personal communication with Elizabeth McVeety, pesticide applicator at
Zilker Park, April 21, 2006). However, it is likely that some overspray may reach
impervious surfaces in the residential setting. In order to account for overspray, the
impervious surface was modeled using three separate assumptions. For the purposes of
risk assessment, it is assumed that 1% of the application rate could reach the impervious
surfaces surrounding each residential lot. This amount of overspray is not based on any
empirical data (studies of this type were not identified); however, the assumption seems
reasonable given the principal drift assumption for ground spray in ecological risk
assessments is 1%. In order to test the assumption of 1% overspray and address the
uncertainty associated with the lack of data for overspray, two alternate scenarios were
modeled. The impervious surface was modeled with 0% overspray and 10% overspray to
provide a lower bound (0% overspray) and an upper bound (10% overspray) on the 1%
assumption. The results of these alternate modeling exercises are discussed more fully in
Section 3.2.7 of this assessment.
In this exercise, it is also assumed that that 50% of the 1A acre lot is treated with atrazine.
This assumption was based on data from the AHS website and 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 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 this data, it is assumed that a typical
home is 2,000 square feet with a 1,000 square foot footprint. 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 considered to be reasonable.
In addition to the footprint of the typical house, it is also assumed that a typical house has
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 landscaped 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 1A
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.
The first assumption may result in an underestimation of exposure, given that more
overspray of impervious surfaces is possible. The impact of this assumption is tested and
characterized in Section 3.2.7. Note that this scenario represents general impervious
surfaces within a watershed not part of the 1A acre lot and includes roads, parking lots,
and buildings among others where overspray from residential lots is expected to be
51
-------
minimal. The 1A 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). In this case, the
assumption that 50% of the lot is treated likely overestimates the amount of landscaped
area treated, but underestimates unintentional overspray of driveways and sidewalks.
The impact of this assumption is also evaluated in Section 3.2.7. Overall, these are
simplifying assumptions that are likely to provide a reasonable high-end estimate of
exposure, given the limitations of the modeling approach.
Percentage of Impervious Surface
in the Austin, Texas Area
Barton Springs
Sampled Creeks
Figure 3.9. Percentage of Impervious Surface Coverage in Vicinity of Barton
Springs
The combined edge of field concentrations are estimated using the *.zts output from
PRZM as described above. In this paired scenario approach, the *.zts output from both
the impervious and residential scenarios are weighted and added together to provide an
52
-------
overall estimate of exposure. Non-detects in the weighted output were converted to 0.45
ug/L in order to capture the potential influence of base flow (described in more detail
above).
Two categories of formulations are currently registered for atrazine use on residential
sites, including granular and liquid formulations (wettable powder dry flowables). Both
formulations were modeled separately because application rates are different (2 Ibs/acre
for granular and 1 Ib/acre for liquid) and the standard assumption for modeling granular
formulations is different from liquid formulations. Granular formulations are typically
modeled as soil applied (CAM is set to 8 with a minimized incorporation depth of 1 cm)
and 0% spray drift, as compared with a foliar application (CAM isset to 2 with a 4-cm
depth of incorporation), which assumes the standard spray drift of 1% for ground
applications. However, because spray drift is not assumed to contribute to the loadings
in the spring and some overspray is expected to impervious surfaces, both residential
scenarios (liquid and granular) were modeled assuming that 1% of the application rate is
applied to the impervious surface.
Figure 3.10 graphically presents the runoff only time series output for the edge of field
concentrations predicted for the granular application of atrazine to the paired
residential/impervious scenario, assuming an overspray of 1% of the application rate to
the impervious surface.
Residential (Granular) Edge of Field Concentration (ppb)
Concentration (ppb)
-»-rOCO-&>.OlCO^JO
DOOOOOOOC
I
J!
lii. L i
it
^
t
JL J
|k
ib
. L
^^
1
,L
|L i
Ik I
Ik.
bb
^k^j
1
.il
lltt 1
k
CO CO CO CO
Date
Edge of Field EEC
Figure 3.10. Representative Time Series Output from Paired Residential/Impervious
PRZM Scenario for Granular Applications
53
-------
3.2.6.2 Turf
The turf scenario was developed consistent with the current PRZM scenarios for turf in
Pennsylvania and Florida (no pre-existing turf scenario for Texas is available). For the
Barton Springs assessment, the turf scenario is intended to represent golf course turf and
recreational fields because residential uses are captured in the residential scenario
(standard turf scenarios are typically used to represent both golf course, recreational, and
residential uses). The standard approach for conducting ecological risk assessments
assumes 100% of the 10-hectare watershed is covered by the relevant use. This approach
also assumes that the receptor (EXAMS standard water body) is adjacent to the edge of
the field. In this assessment, with the exception of the residential use sites described
above, most of the other potential atrazine use sites are not adjacent to the receptor.
Although travel through the fractures is likely to be direct, there is also likely to be non-
impacted fracture flow arriving simultaneously at the springs. In order to account for
this, an adjustment factor was applied to each of the non-residential use sites. For turf, it
was assumed that 100% of the watershed feeding the fractures is represented by golf
course turf, but that the percentage of the golf course expected to be treated in this case is
represented by the golf course adjustment factor of 28%. This seems reasonable given
that the land cover analysis indicates that only a few golf courses are present within the
action area. As with the residential scenario, a base flow concentration of 0.45 ug/L was
added to the overall exposure.
Similar to the residential scenario, two categories of formulations are registered for
atrazine use on turf sites. These are granular and liquid formulations (wettable powder
dry flowables). Both formulations were modeled separately because application rates are
different (2 Ibs/acre for granular and 1 Ibs/acre for liquid) and the standard assumption
for modeling granular formulations is different from liquid formulations. Granular
formulations are typically modeled as soil applied (CAM is set to 8 with a minimized
incorporation depth of 1 cm) and 0% spray drift, as compared with a foliar application
(CAM is set to 2 with a 4-cm depth of incorporation), which assumes the standard 1%
spray drift for ground applications.
3.2.6.3 Fallow/Idle Land
The fallow/idle land scenario represents the only agricultural use present within in the
action area. Generally, this scenario conceptually represents the potential application of
atrazine to fallow lands under the Conservation Reserve Program (CRP) as most of the
available information suggests that no agricultural use of atrazine occurs in the action
area. It is also expected, given the available usage information, that fallow/idle land
represents a minor use in the action area, although there are limitations associated with
this analysis. Regardless, the scenario was included in the assessment to address all
potential atrazine uses.
54
-------
This scenario was developed similar to the standard PRZM scenarios. The bulk of this
use pattern is located south and west of the receptor site; therefore, a land cover
adjustment factor was applied to the modeled output. Although travel through the
fractures is likely to be direct, non-impacted fracture flow is likely to arrive
simultaneously at the springs, especially given that most of this use site is located within
the Contributing Zone. To account for this, an adjustment factor was applied for each of
the non-residential use sites. For the fallow/idle land scenario, this factor represents the
percentage of the land cover within the action area that is in fallow/idle land. As shown
in Table 3.4, the adjustment factor for fallow/idle land is 5%. As with the residential
scenario, a base flow concentration of 0.45 ug/L was added to the overall exposure.
3.2.6.4 Rights-of-Way
The rights-of-way scenario represents a vegetated buffer strip where atrazine could be
applied adjacent to a water body. The vegetative strip as modeled is intended to represent
treated buffers along roadways, railroad lines, and utility rights-of-way. Each of these
sites is conceptualized as a naturally vegetated strip that runs linearly adjacent to a
sensitive water body. Figure 3.11 presents the conceptual model of this scenario relative
to a roadway, although a similar layout would be expected for rail and power rights-of-
way. It is expected that the density of roads, railroads, and utility rights-of-way define
the density of rights-of-way use within the watershed where pesticides may be applied.
Land cover data was used to define this percentage. For this assessment, it is assumed
that the maximum density of treated rights-of-way within the action area is 10%, which is
expected to represent a slight over-estimation.
55
-------
Figure 3.11. Conceptual Model of Rights-of-Way Scenario
56
-------
In addition, the potential impact of impervious surfaces within the treated area was
addressed to estimate its influence on overall runoff and exposure. For the roadway
rights-of-way use, it is assumed that there is an equal amount of impervious and pervious
surface within the treatment area, but that the atrazine treatment is likely to be restricted
to the pervious portion of the rights-of-way. As with the residential scenario, this
assumption accounts for a potential overspray to the impervious surface of 1%.
Conversations with local experts (Markwardt, 2006; Ward, 2006; Mason, 2006) suggest
that the rights-of-way scenario is likely to be conservative, given that atrazine is not
typically used (glyphosate is reportedly the herbicide of choice in Hays and Travis
counties) and treatment zones are typically one to four feet wide. As with the residential
scenario, the impact of 0% and 10% over spray of the impervious surface on rights-of-
way is characterized in Section 3.2.7 of this assessment.
As with the residential scenario, the edge of field EECs for both the runoff only and
runoff plus infiltration scenarios were estimated. Similar to the residential scenario, a
base flow concentration of 0.45 ug/L was added to the overall exposure.
Table 3.5 presents the summary of all relevant time-weighted concentrations for each
scenario modeled at the 90th % of exposure for the edge of field exposure. The EECs
presented in Table 3.5 are used to derive risk quotients, which are presented as part of the
Risk Characterization in Section 5
57
-------
Table 3.5. Summary of PRZM Output EECs for all Modeled Scenarios (Edge of Field Concentrations with Base Flow
Incorporated).
90th Percentile of 30 Years of Output
Use Site
Residential
Granular1
Residential
Liquid1
Right-of-
Way1
Fallow/Idle
land
Turf-
Granular
Turf - Liquid
Application
Rate (Ibs/acre)
2
1
1
2.25
2
1
Number of
Applications
(interval)
2
2
1
1
2
2
First
Application
Date
April 1
April 1
June 1
November 1
April 1
April 1
Peak
(one-day)
EEC
(HS/L)
41.2
26.6
6.2
7.5
22.4
16.2
14-day
EEC
(jig/L)
3.5
2.5
1.1
1.0
2.0
1.7
21-day
EEC
(jig/L)
2.5
1.8
0.9
0.8
1.5
1.3
30-day
EEC
(jig/L)
1.9
1.5
0.8
0.7
1.2
1.0
60-day
EEC
(Hg/L)
1.2
1.0
0.6
0.6
0.8
0.7
90-day
EEC
(jig/L)
1.0
0.8
0.6
0.6
0.7
0.6
Annual
Average
(jig/L)
0.6
0.5
0.5
0.5
0.5
0.5
58
-------
In general, the exposure assessment yields modeled peak exposure estimates that are two
to ten times higher than those seen in monitoring data, while the annual average
concentrations are consistent with 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.
3.2.7 Characterization
Reported use information provides a sense of the actual use on sites similar to those
assessed including fallow/idle land. In this instance, the data for fallow/idle land
suggests that the 90th percentile application rate is similar to the maximum labeled use
rate while the typical use application rate (equivalent to the average of all reported
applications) is roughly half the labeled maximum rate used in this assessment. Table 3.6
summarizes the typical and 90th percentile rates and number of applications relative to
those used in this assessment. If it were assumed that this pattern holds true for all uses
(agricultural and non-agricultural), then modeling with the typical application rates would
yield predicted exposures that are roughly half of those presented in Table 3.5.
Table 3.6. Comparison of Maximum, Typical, and 90th Percentile Label Rates and
Number of Applications
Scenario
Fallow/Idle
land
Maximum
Application
Rate
(Ibs/acre)
2.25
Maximum
Number of
Applications
1
90th
Percentile
Application
Rate
(Ibs/acre)
2
90th Percentile
Number of
Applications
1
Typical
Application
Rate
(Ibs/acre)
0.9
Typical
Number of
Applications
1
In order to account for the variability in overspray, the residential scenario was modeled
assuming two alternate scenarios of 0% and 10% overspray to impervious surfaces. The
alternate assumptions are intended to provide a bound on the 1% assumption. Because
both the residential and rights-of-way scenarios were modeled using the paired
pervious/impervious approach, the alternate scenarios were modeled for both scenarios
(residential was modeled for both granular and liquid formulations). The
conservativeness of these assumptions is unknown, given a lack of data on this
phenomenon. However, given that the impervious scenario is intended to represent non-
target surfaces such as roads, parking lots and buildings, it is seems reasonable to assume
that 10% overspray is an over-estimation of what would likely occur to these off-site
areas, while 0% may be an under-estimation.
In order to model the overspray, the binding coefficient was set to zero and the aerobic
soil metabolism half-life was set to stable in lieu of actual data for the impervious
scenario. It is assumed that non-binding would occur on these surfaces and that limited
degradation would occur. The percentage overspray was then multiplied by the total
application rate to yield an effective application rate for the overspray to impervious
59
-------
surfaces. This analysis yielded an application rate on the impervious surface of 0.2
Ibs/acre (0.23 kg/ha) for 10% overspray and 0.02 Ibs/acre (0.023 kg/ha) for 1%
overspray.
Comparison of the resulting EECs indicates that with 10% overspray, the overall increase
in peak EECs is roughly 30%, while the longer-term EECs are increased by nearly 50%.
For the 1% overspray assumption, there is very little increase in overall EECs for both
peak and average EECs, as compared to the 0% overspray assumption. This is not
unexpected, given the increased runoff, lack of binding, and lack of degradation being
modeled. Without actual data on these processes, it is impossible to determine whether
these exposures reflect reality, especially given that none of the monitoring data indicate
concentrations approaching any of these EECs with or without overspray. The overspray
comparison is presented in Table 3.7.
Other assumptions that can have a significant impact on the overall predicted EECs
include the percentage of impervious surface and the percentage of 1A acre lot that is
treated. In both instances, the relationship between the assumption and the predicted
EEC is linear. The assumed action area impervious surface percentage of 30% in the
vicinity of the Barton Springs decreases dramatically further south and west from the
salamander's habitat. It is apparent from the available data that this value decreases to
less than 10% the further south and west from the springs. The impact of this assumption
was evaluated by readjusting the output to reflect the impact of a 10% impervious cover
assumption on predicted exposures. In general, peak and longer-term average
concentrations are generally doubled as the percentage of impervious decreases. The
comparison of this analysis is presented in Table 3.8. This is likely due to the increase in
treated area contributing more pesticide mass and an increase in the impervious surface,
which yields greater amounts of non-contaminated runoff.
This impact of a decrease in impervious surface will hold only with the assumption of
limited overspray. This assumption was explored by comparing the impact of the change
in percentage of impervious surface on the 10% overspray scenario discussed above. In
this case, peak EECs increase by roughly 50% while the averages are only slightly
increased. The comparison of this analysis is presented in Table 3.9.
Finally, in this assessment it is assumed that 50% of the 1A acre lot is treated. In order to
test the significance of this assumption, the exposure scenario was re-run using a different
assumption of 10% treatment of the 1A acre lot. As expected, peak EECs are reduced by
roughly a factor of five, while the longer term exposures are reduced by a factor of two to
three times. The results of this comparison are presented in Table 3.10.
60
-------
Table 3.7. Comparison of Residential and Rights-of-Way EECs Assuming Variable Percentages of Overspray (0,1, and 10%)
90th Percentile of 30
Use Site
Residential -
1%
Overspray1
Residential -
No
Overspray1
Residential -
10%
Overspray1
Rights-of-
Way - 1%
Overspray
Rights-of-
Way -No
Overspray
Rights-of
Way - 10%
Overspray
Application
Rate
(Ibs/acre)
2.0
2.0
2.0
1.0
1.0
1.0
Number of
Applications
(interval)
2
(30 days)
2
(30 days)
2
(30 days)
1
1
1
First
Application
Date
April 1
April 1
April 1
June 1
June 1
June 1
Peak
EEC
(Hg/L)
41.2
40.0
51.7
6.2
5.9
9.5
Years of Output
14-day 21-day 30-day
EEC EEC EEC
(jig/L) (jig/L) (jig/L)
3.5 2.5 1.9
3.3 2.4 1.8
5.6 4.1 3.4
1.1 0.9 0.8
1.1 0.9 0.8
1.9 1.7 1.5
60-day
EEC
(jig/L)
1.2
1.1
2.3
0.6
0.6
1.1
90-day Annual
EEC Average
(jig/L) (jig/L)
1.0 0.6
0.9 0.6
1.7 0.9
0.6 0.5
0.6 0.5
0.9 0.6
1 - Only the granular application was tested for characterization
61
-------
Table 3.8. Comparison of Residential EECs (granular) with 1% Over Spray and Variable Percentages of Impervious Surface
(10 and 30%)
90th Percentile of 30 Years of Output
Use Site
Residential -
30% Impervious
Residential -
10% Impervious
Applicati
on Rate
(Ibs/acre)
1.0
1.0
Number of
Applications
(interval)
2
(30 days)
2
(30 days)
First „ , „„„ 14-day
. .... Peak EEC rrfs
Application , _. EEC
(us/L) , , „
Date (jig/L)
April 1 41.2 3.5
April 1 62.8 5.0
21-day
EEC
(jig/L)
2.5
3.5
30-day
EEC
(jig/L)
1.9
2.6
60-day
EEC
(Hg/L)
1.2
1.6
90-day Annual
EEC Average
1.0 0.6
1.2 0.6
Table 3.9. Comparison of Residential EECs (granular) with 10% Overspray and Variable Percentages of Impervious Surface
(10 and 30%)
90th Percentile of 30 Years of Output
Use Site
Residential -
30% Impervious
Residential -
10% Impervious
Applicati
on Rate
(Ibs/acre)
1.0
1.0
Number of
Applications
(interval)
2
(30 days)
2
(30 days)
First Peak
Application EEC
Date (jig/L)
April 1 51.7
April 1 67.1
14-day
EEC
(jig/L)
5.6
6.0
21-day
EEC
(jig/L)
4.1
4.2
30-day
EEC
3.4
3.2
60-day
EEC
(jig/L)
2.3
2.0
90-day
EEC
(jig/L)
1.7
1.5
Annual
Average
(jig/L)
0.9
0.7
62
-------
Table 3.10. Comparison of Residential EECs (granular) Assuming Various Percentages of Treated 1A Acre Lot (10 and 50%)
90*hPercentileof30
Use Site
Residential -
50% Treated
Residential -
10% Treated
Application
Rate
(Ibs/acre)
1.0
1.0
Number of
Applications
(interval)
2
(30 days)
2
(30 days)
First
Application
Date
April 1
April 1
Peak
EEC
(jig/L)
41.2
9.5
Years of Output
14-day 21-day
EEC EEC
(jig/L) (jig/L)
3.5 2.5
1.3 1.0
30-day
EEC
(jig/L)
1.9
0.9
60-day
EEC
(Hg/L)
1.2
0.7
90-day
EEC
(jig/L)
1.0
0.6
Annual
Average
(jig/L)
0.6
0.5
63
-------
4. Effects Assessment
This assessment evaluates the potential for atrazine to adversely affect the Barton Springs
salamander. As previously discussed in Section 2.7, assessment endpoints for the Barton
Springs salamander include direct toxic effects on the survival, reproduction, and growth
of the salamander itself, as well as indirect effects, such as reduction of the prey base
and/or modification of its habitat. Direct effects to the Barton Springs salamander are
based on toxicity information for freshwater vertebrates, including fish, which are
generally used as a surrogate for amphibians, as well as available salamander toxicity
data from the open literature. Given that the salamander's prey items and habitat
requirements are dependent on the availability of freshwater aquatic invertebrates and
aquatic plants, toxicity information for various freshwater aquatic invertebrates and
plants is also discussed. Acute (short-term) and chronic (long-term) toxicity information
is characterized based on registrant-submitted studies and a comprehensive review of the
open literature on atrazine. In addition to registrant-submitted and open literature
toxicity information, indirect effects to Barton Springs salamanders, via impacts to
aquatic plant community structure and function are also evaluated based on community-
level threshold concentrations. 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 freshwater 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 4.4, respectively.
With respect to atrazine degradates, including hydroxyatrazine (HA), deethylatrazine
(DEA), deisopropylatrazine (DIA), and diaminochloroatrazine (DACT), it is assumed
that they are of lesser toxicity as compared to the parent compound. Comparison of
available toxicity information for the degradates of atrazine indicates lesser aquatic
toxicity than the parent for freshwater fish, 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,
available aquatic plant degradate toxicity data for HA, DEA, DIA, and DACT report non-
definitive ECso values (i.e., 50% effect was not observed at the highest test
concentrations) at concentrations that are 700 to 10,000 times higher than the lowest
reported aquatic plant ECso value for parent atrazine. 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 of the
atrazine degradates and formulated products is presented in Appendix A.
64
-------
4.1 Evaluation of Aquatic Ecotoxicity Studies
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 ECOTOX database maintained by EPA/Office of Research and
Development (ORD) (U.S. EPA, 2004). Open literature data presented in this assessment
were obtained from the 2003 atrazine IRED (U.S. EPA, 2003a) as well as information
obtained 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 ug/L on a
recurrent basis or over a prolonged period of time (U.S. EPA, 2003a). Given the large
amount of microcosm/mesocosm and field study data for atrazine, only effects data that
are more conservative than the 10 ug/L aquatic-community effect level identified in the
2003 atrazine IRED were considered. In addition, data for taxa that are directly relevant
to the Barton Springs salamander (i.e., aquatic-phase amphibians) were also 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 Barton Springs salamander survival, reproduction, and growth) identified
in Section 2.7. For example, endpoints such as behavior modifications are likely to be
qualitatively evaluated, because quantitative relationships between modifications and
reduction in species survival, reproduction, and/or growth are not available.
As described in Agency's Overview Document (U.S. EPA, 2004), the most sensitive
endpoint for each taxa is evaluated. For this assessment, evaluated taxa include
freshwater fish, freshwater aquatic invertebrates, and freshwater aquatic plants.
Currently, no guideline tests exist for salamanders. Therefore, surrogate species were
used as described in the Overview Document (U.S. EPA, 2004). In addition, aquatic-
phase amphibian ecotoxicity data from the open literature are qualitatively discussed.
Table 4.1 summarizes the most sensitive ecological toxicity endpoints for the Barton
Springs salamander, based on an evaluation of both the submitted studies and the open
literature, as previously discussed. A brief summary of submitted and open literature
65
-------
data considered relevant to this ecological risk assessment for the Barton Springs
salamander is presented below. Additional information is provided in Appendix A. It
should be noted that Appendix A also includes ecotoxicity data for taxonomic groups that
are not relevant to this assessment (i.e., birds, estuarine/marine fish, invertebrates, and
plants) because the Agency is completing endangered species assessments for other
species concurrently with this assessment.
Table 4.1. Aquatic Toxicity Profile for Atrazine
Assessment Endpoint
Acute Direct Toxicity to
Salamander
Chronic Direct Toxicity
to Salamander
Indirect Toxicity to
Salamander via Acute
Toxicity to Freshwater
Invertebrates (i.e. prey
items)
Indirect Toxicity to
Salamander via Chronic
Toxicity to Freshwater
Invertebrates (i.e. prey
items)
Indirect Toxicity to
Salamander via Acute
Toxicity to Non-vascular
aquatic plants
Indirect Toxicity to
Salamander via Acute
Toxicity to Vascular
aquatic plants
Species
Rainbow
trout1
Brook
trout1
Midge
Scud
4 species
of
freshwater
algae
Duckweed
Toxicity Value Used in
Risk Assessment
96-hour LC50 = 5,300
ug/L
Probit slope = 2.72
NOAEC = 65 ug/L
LOAEC = 120 ug/L
48-hour LC50 = 720
ug/L
Probit slope unavailable
NOAEC = 60 jig/L
LOAEC = 120 ug/L
1-week EC50 = 1 ug/L
14-day EC50 = 37 ug/L
Citation
MRID#
(Author &
Date)
000247-16
(Beliles and
Scott, 1965)
000243-77
(Macek et al.,
1976)
000243-77
(Macek et al.,
1976)
000243-77
(Macek et al.,
1976)
000235-44
(Torres &
O'Flaherty,
1976)
430748-04
(Hoberg, 1993)
Comment
Acceptable
Acceptable: 7.2%
reduction in
length; 16%
reduction in
weight
Supplemental:
raw data
unavailable
Acceptable: 25%
reduction in
development of F!
to seventh instar
Supplemental: 41
to 98% reduction
in chlorophyll
production; raw
data unavailable
Supplemental:
50% reduction in
biomass; NOAEC
not determined
1 Used as a surrogate for the Barton Springs salamander.
presented in Section 4.1.2.
Open literature data for the salamander are
Toxicity to aquatic fish and invertebrates is categorized using the system shown in Table
4.2 (U.S. EPA, 2004). Toxicity categories for aquatic plants have not been defined.
66
-------
Table 4.2. Categories of Acute Toxicity for Aquatic Organisms
LC50 (ppm)
<0.1
>0.1-1
>1-10
> 10 - 100
>100
Toxicity Category
Very highly toxic
Highly toxic
Moderately toxic
Slightly toxic
Practically nontoxic
4.1.1 Toxicity to Freshwater Fish
As previously discussed, no guideline tests exist for salamanders; therefore, freshwater
fish are used as surrogate species for amphibians including salamanders (U.S. EPA,
2004). The available open literature information on atrazine toxicity to aquatic-phase
amphibians, which is provided in Section 4.1.2, shows that acute and chronic ecotoxicity
endpoints for amphibians are generally less sensitive than fish. Therefore, endpoints
based on freshwater fish ecotoxicity data are assumed to be protective of potential direct
effects to aquatic-phase salamanders including the Barton Springs salamander. A
summary of acute and chronic freshwater fish data, including sublethal effects, is
provided below.
4.1.1.1 Freshwater Fish: Acute Exposure (Mortality) Studies
Freshwater fish acute toxicity studies were used to assess potential direct effects to the
Barton Springs salamander because direct acute toxicity guideline data on salamanders
are unavailable. Atrazine toxicity has been evaluated in numerous freshwater fish
species, including rainbow trout, brook trout, bluegill sunfish, fathead minnow, tilapia,
zebrafish, goldfish, and carp, and the results of these studies demonstrate a wide range of
sensitivity. The range of acute freshwater fish LCso values for atrazine spans one order of
magnitude, from 5,300 to 60,000 ug/L; therefore, atrazine is categorized as moderately
(>1,000 to 10,000 ug/L) to slightly (>10,000 to 100,000 ug/L) toxic to freshwater fish on
an acute basis. The freshwater fish acute LCso value of 5,300 ug/L is based on a static
96-hour toxicity test using rainbow trout (Oncorhynchus mykiss) (MRID # 000247-16).
No sublethal effects were reported as part of this study. A complete list of all the acute
freshwater fish toxicity data for atrazine is provided in Table A-8 of Appendix A.
4.1.1.2 Freshwater Fish: Chronic Exposure (Growth/Reproduction) Studies
Similar to the acute data, chronic freshwater fish toxicity studies were used to assess
potential direct effects to the Barton Springs salamander because direct chronic toxicity
guideline data for salamanders do not exist. Freshwater fish full life-cycle studies for
atrazine are available and summarized in Table A-12 of Appendix A. Following 44
weeks of exposure to atrazine in a flow-through system, statistically significant
reductions in brook trout mean length (7.2%) and body weight (16%) were observed at a
concentration of 120 ug/L, as compared to the control (MRID # 000243-77). The
67
-------
corresponding NOAEC for this study is 65 ug/L. Although the acute toxicity data for
atrazine show that rainbow trout are the most sensitive freshwater fish, available chronic
rainbow trout toxicity data indicates that it is less sensitive to atrazine, on a chronic
exposure basis, than the brook trout, with respective LOAEC and NOAEC values of
1,100 ug/L and 410 ug/L. Further information on chronic freshwater fish toxicity data
for atrazine is provided in Section A.2.2 of Appendix A.
4.1.1.3 Freshwater Fish: Sublethal Effects and Additional Open Literature
Information
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.
Reported sublethal effects in rainbow trout show increased plasma vitellogenin levels in
both female and male fish and decreased plasma testosterone levels in male fish at
atrazine concentrations of approximately 50 ug/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.
In salmon, endocrine-mediated olfactory functions were affected at 0.5 ug/L atrazine
(Moore and Lower, 2001). The reproductive priming effect of the female pheromone
prostaglandin F2a on the levels of expressible milt in males was reduced after exposure to
atrazine at 0.5 ug/L. Overall, the relationship between reduced olfactory response of
males 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) in the wild is not established. In addition, EPA did not use these data
in development of the aquatic life water quality criteria for atrazine because the test
material was not adequately described or translated. Furthermore, the study did not
determine whether the decreased response of olfactory epithelium to specific chemical
stimuli would likely impair similar responses in intact fish.
Although these studies raise concern about the effects of atrazine on endocrine-mediated
functions in freshwater and anadromous fish, these effects are difficult to quantify
because they are not clearly tied to the assessment endpoints for the Barton Springs
salamander (i.e., survival, growth, and reproduction of individuals). In addition,
differences in habitat and behavior of the tested fish species compared with the Barton
Springs salamander suggest that the results are not readily extrapolated to salamanders.
Furthermore, there is uncertainly associated with extrapolating effects observed in the
laboratory to more variable exposures and conditions in the field. Therefore, potential
sublethal effects on fish are evaluated qualitatively and 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.
68
-------
4.1.2 Toxicity to Aquatic-phase Amphibians
Available toxicity information on potential atrazine-related mortality and sublethal
effects to aquatic-phase amphibians (including salamanders) from the open literature is
summarized below in Sections 4.1.2.1 and 4.1.2.2, respectively. Guideline ecotoxicity
studies for amphibians are not available.
4.1.2.1 Amphibians: Open Literature Data on Mortality
Available acute data for amphibians, including the leopard frog (Ranapipiens), wood
frog (R. sylvaticas), and American toad (Bufo americanus), indicate that they are
relatively insensitive to atrazine with acute LCso values > 20,000 ug/L (Allran and
Karasov, 2001). Acute toxicity data are not available for the salamander.
Chronic mortality data for aquatic-phase amphibians confirm that exposure to atrazine
does not cause direct mortality to frogs and salamanders at concentrations ranging from
approximately 200 to 2,000 ug/L; these concentrations represent the highest tested
atrazine treatment levels within each of the studies. Salamander-specific chronic
mortality data are available for the spotted salamander (Ambystoma maculatum), small-
mouthed salamander (A. texamim), streamside salamander (A. barbouri), and the long-
toed salamander (A. macrodactylum). The available salamander data show no effect to
mortality at the highest treatment concentrations of atrazine in each of the respective
studies, ranging from approximately 200 to 400 ug/L (Boone and James, 2003; Rohr et
al., 2003, Forson and Storfer, 2006). Rohr et al. (2004) reported decreased embryo
survival through Day 16 in streamside salamanders following exposure to 400 ug/L
atrazine (NOAEC = 40 ug/L). However, most embryo mortality was associated with a
white film covering the embryo, suggesting the presence of a fungal pathogen, which
may have decreased survival. According to the study authors, it is unknown whether the
fungi caused or simply followed mortality. In addition, reduced survival was reported in
only one of the two years tested; therefore, there is a high degree of uncertainty
associated with the reported results.
4.1.2.2 Amphibians: Open Literature Data on Sublethal Effects
An evaluation and review of the results of submitted studies regarding potential atrazine
effects data on amphibian gonadal development are presented in the Agency's White
Paper (U.S. EPA, 2003d) and discussed in Section A.2.4c of Appendix A. As previously
discussed in Section 2.3, the Agency has concluded that there is currently insufficient
evidence to confirm or refute the hypothesis that atrazine exposure may impact gonadal
development in amphibians. Therefore, the Agency has requested additional data from
the registrant to reduce uncertainties regarding potential risk to amphibians. In addition
to addressing uncertainty regarding the potential of atrazine to cause these effects, these
studies are expected to characterize the nature of any potential dose-response
69
-------
relationship. The initial results of these amphibian studies are expected to become
available in late 2006 to early 2007; therefore, as of this writing, they are not available for
inclusion in this endangered species risk assessment for the Barton Springs salamander.
Open literature data on sublethal effects of atrazine to aquatic-phase amphibians,
including frogs and salamanders, are summarized below and discussed in greater detail in
Section A.2.4d of Appendix A. The following information includes studies identified as
part of the 2006 open literature search that were not reviewed as part of the Agency's
2003 White Paper (U.S. EPA, 2003d).
Frogs (Anurans)
The reviewed studies were classified as qualitative because they address issues of
concern to the risk assessment, but are not appropriate for quantitative use due to
uncertainties related to a lack of raw data and limitations in the study design. Further
information on the study design and uncertainties associated with each of the reviewed
studies are provided in Section A.2.4d and Table A-16 of Appendix A. In summary, the
microcosm/mesocosm and chronic lab data for frogs indicate that sublethal effects to
amphibians, such as reduced mass and length at metamorphosis may occur at atrazine
exposure concentrations of approximately 200 ug/L and higher under the conditions
tested (Diana et al., 2000; Boone and James, 2003; and Gucciardo, 1999). Decreased
frog weight (and length) at metamorphosis at > 200 ug/L atrazine is hypothesized to
result from atrazine's effect on algal populations, which are a primary source of food for
developing anurans (Diana et al., 2000). Other factors, such as decreasing DO, pH, and
macrophyte biomass following atrazine exposure may also contribute to observed
sublethal effects.
In the lab, plasma testosterone was reduced in male frogs at atrazine concentrations of
259 ug/L; however, an increase in aromatase activity (aromatase increases synthesis of
l?p-estradiol resulting in depletion of testosterone levels) was not observed (Hecker et
al., 2005). Therefore, the mechanism associated with decreased testosterone levels in
adult males is unclear.
The observed effect levels of -200 ug/L are greater than the aquatic community-level
effects of 10-20 ug/L documented in the January 2003 atrazine IRED (U.S. EPA, 2003a).
In addition, uncertainties and associated limitations in the design of the reviewed studies
are similar to the conclusions previously reported (U.S. EPA, 2003d).
Salamanders (Caudates)
The reviewed sublethal studies contain variable results with respect to atrazine exposures
and sublethal effects to aquatic-phase salamanders. Two chronic studies on the
streamside salamander (A. barbouri) and long-toed salamander (A. macrodactylum) show
reduced mass and snout-vent length (SVL) at metamorphosis, in addition to accelerated
metamorphosis, relative to controls, at atrazine concentrations ranging from 184 to 400
ug/L (Rohr et al., 2004; Forson and Storfer, 2006). The NOAEC values for these studies
70
-------
range between 18.4 and 40 ug/L. In another study, the time to metamorphosis was
increased in small-mouthed salamanders (A. texanuni) at the only concentration of
atrazine tested (197 ug/L); however, no effect on the time to metamorphosis was
observed in spotted salamanders at the same concentration of atrazine (Boone and James,
2003).
The interaction of atrazine and one of the iridoviruses (the Ambystoma tigrinum virus
[ATV]) was studied in long-toed salamanders by Forson and Storfer (2006). Larvae
exposed to both atrazine and ATV had lower levels of mortality and ATV infectivity
compared to larvae exposed to virus alone, suggesting that atrazine may compromise
virus efficacy or improve salamander immune competency. Behavioral changes in
locomotion (i.e., increased activity following tapping on tanks) were observed in
streamside salamanders exposed to 400 ug/L; however, this endpoint is not relevant to
the assessment endpoints chosen for this risk assessment. It is unclear how increased
larval salamander activity due to tank tapping in the lab would translate into reduced
fitness in the wild.
All of the reviewed salamander studies from the open literature were classified as
qualitative because they address issues of concern to the risk assessment, but are not
appropriate for quantitative use due to uncertainties related to a lack of raw data and
limitations in the study design. Further information on the study design and uncertainties
associated with each of the reviewed studies is provided in Section A.2.4d and Table A-
17 of Appendix A.
4.1.3 Toxicity to Freshwater Invertebrates
Freshwater aquatic invertebrate toxicity data were used to assess potential indirect effects
of atrazine to the Barton Springs salamander. Direct effects to freshwater invertebrates
resulting from exposure to atrazine may indirectly affect the Barton Springs salamander
via reduction in available food. As discussed in Section D.5.1 of Appendix D, Barton
Springs salamanders feed on a wide range of freshwater aquatic invertebrates including
ostracods, copepods, chironomids, snails, amphipods, mayfly larvae, leeches, and adult
riffle beetles. Based on analysis of the stomach and fecal samples from a limited number
of adult and juvenile Barton Springs salamanders, the most prevalent organisms found
were ostracods, amphipods, and chironomids (USFWS, 2005). However, data on the
relative percentage of each type of aquatic invertebrate in the salamander's diet are not
available.
A summary of acute and chronic freshwater invertebrate data, including published data in
the open literature since completion of the IRED (U.S. EPA, 2003a), is provided below in
Sections 4.1.3.1 through 4.1.3.3.
4.1.3.1 Freshwater Invertebrates: Acute Exposure Studies
Atrazine is classified as highly toxic to slightly toxic to aquatic invertebrates. There is a
wide range of ECso/LCso values for freshwater invertebrates with values ranging from
71
-------
720 to >33,000 ug/L. The freshwater LCso value of 720 ug/L is based on an acute 48-
hour static toxicity test for the midge, Chironomus tentans (MRID # 000243-77). Further
evaluation of the available acute toxicity data for the midge shows high variability with
the LC50 values, ranging from 720 to >33,000 ug/L. With the exception of the midge,
reported acute toxicity values for the other five freshwater invertebrates (including the
water flea, scud, stonefly, leech, and snail) are 3,500 ug/L and higher. All of the
available acute toxicity data for freshwater invertebrates are provided in Section A.2.5
and Table A-18 of Appendix A. The LC5o/EC5o distribution for freshwater invertebrates
is graphically represented in Figure 4.1. The columns represent the lowest reported value
for each species, and the positive y error bar represents the maximum reported value.
Values in parentheses represent the number of studies included in the analyses.
Summary of Reported Acute LC5o/EC5o Values in Freshwater Invertebrates
for Atrazine
ocnnn ^ -
Qnnnn -
ocnnn -
0
m
O • — • onnnn -
Uj .Q ZUUUU
o °~
m ®- icnnn
_l
-i nnnn
5000 -
n
i —
L r
h I
Midge (3) Waterflea (5) Scud (3) Stonefly (1) Leech (1) Snail (1)
Species
Figure 4.1. Summary of Reported Acute LCso/ECso Values in Freshwater
Invertebrates for Atrazine
4.1.3.2 Freshwater Invertebrates: Chronic Exposure Studies
The most sensitive chronic endpoint for freshwater invertebrates is based on a 30-day
flow-through study on the scud (Gammarus fasciatus\ which showed a 25% reduction in
the development of FI to the seventh instar at atrazine concentrations of 140 ug/L; the
corresponding NOAEC is 60 ug/L (MRID # 000243-77). Although the acute toxicity
data for atrazine show that the midge (Chironomus tentans) is the most sensitive
freshwater invertebrate, available chronic midge toxicity data indicate that it is less
sensitive to atrazine, on a chronic exposure basis, than the scud, with respective LOAEC
and NOAEC values of 230 ug/L and 110 ug/L. Additional information on the chronic
toxicity of atrazine to freshwater invertebrates is provided in Section A.2.6 and Table A-
20 of Appendix A.
72
-------
4.1.3.3 Freshwater Invertebrates: Open Literature Data
One additional acute study for an underrepresented taxa of 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 do not affect the survival
of juvenile and mature freshwater mussels, Anodonta imbecilis; therefore, A. imbecilis is
less acutely sensitive to atrazine than other freshwater invertebrates.
4.1.4 Toxicity to Aquatic Plants
Aquatic plant toxicity studies were used as one of the measures of effect to evaluate
whether atrazine may affect primary production. In Barton Springs, primary productivity
is essential for indirectly supporting the growth and abundance of the Barton Springs
salamander. In addition to providing cover, moss and other aquatic plants harbor a
variety of aquatic invertebrates that salamanders eat.
Two types of studies were used to evaluate the potential of atrazine to affect primary
productivity. Laboratory studies were used to determine whether atrazine may cause
direct effects to aquatic plants. In addition, the threshold concentrations, described in
Section 4.2, were used to further characterize potential community level effects to Barton
Springs salamanders resulting from potential effects to aquatic plants. A summary of the
laboratory data for aquatic plants is provided in Section 4.1.4.1. A description of the
threshold concentrations used to evaluate community-level effects is included in Section
4.2.
4.1.4.1 Aquatic Plants: Laboratory Data
Numerous aquatic plant toxicity studies have been submitted to the Agency. A summary
of the data for freshwater vascular and non-vascular plants is provided below. Section
A.4.2 and Tables A-40 and A-41 of Appendix A include 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 ECso value for
four different species of freshwater algae is 1 |ig/L, based on data from a 7-day acute
study (MRID # 000235-44). Vascular plants are less sensitive to atrazine than
freshwater non-vascular plants with an ECso value of 37 |ig/L, based on reduction in
duckweed growth (MRID # 430748-04).
Comparison of atrazine toxicity levels for three different algae endpoints suggests that the
endpoints in decreasing order of sensitivity are cell count, growth rate and oxygen
production (Stratton, 1984). Walsh (1983) exposed Skeletonema costatum to atrazine and
concluded that atrazine is only slightly algicidal at relatively high concentrations (i.e.,
500 & 1,000 ug/L). Caux et al. (1996) compared the cell count ICso and fluorescence
and concluded that atrazine is algicidal at concentrations affecting cell counts.
73
-------
Abou-Waly et al. (1991) measured growth rates on days 3, 5, and 7 for two algal species.
The pattern of atrazine effects on growth rates differs sharply between the two species.
Atrazine had a strong early effect on Anabaena flos-aquae followed by rapid recovery in
clean water (i.e., EC50 values for days 3, 5, and 7 are 58, 469, and 766 ug/L,
respectively). The ECso values for Selenastrum capricornutum continued to decline from
day 3 through 7 (i.e., 283, 218, and 214 ug/L, respectively). Based on theses results, it
appears that the timing of peak effects for atrazine may differ depending on the test
species.
It should be noted that rrecovery 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 of recovery. Thus, before recovery can be
considered as an uncertainty, an agreed upon interpretation is needed. The Agency
believes that recovery has a simple and straight forward interpretation: a return to pre-
exposure levels for the affected population., not for a replacement population of more
tolerant species. Further research is needed to quantify the impact that recovery and
resistance would have on aquatic plants.
4.1.5 Freshwater Field Studies
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 effects.
Atrazine has been the subject of many mesocosm and microcosm studies in ponds,
streams, lakes, and wetlands. The durations of these studies have ranged from a few
weeks to several years at exposure concentrations ranging from 0.1 ug/L to 10,000 ug/L.
Most of the studies have focused on atrazine effects on phytoplankton, periphyton, and
macrophytes; however, some have also included measurements on animals.
As described in the 2003 IRED for atrazine (U.S. EPA, 2003a), 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 ug/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 reviewed as part of the 2003 IRED is included in Section A.2.8a and Tables A-
22 through A-24 of Appendix A. Given the large amount of microcosm and mesocosm
and field study data for atrazine, only effects data less than or more conservative than the
74
-------
10 |ig/L aquatic community effect level identified in the 2003 IRED were summarized
from the open literature search that was completed in February 2006. Field study data for
aquatic-phase amphibians, including frogs and salamanders, are summarized in Section
4.1.2 and discussed in greater detail in Sections A.2.3 and A.2.4 of Appendix A. Based
on the selection criteria for review of new open literature, all of the available studies
show effects levels to freshwater fish, invertebrates, and aquatic plants at concentrations
greater than 10 |ig/L.
Community-level effects to aquatic plants that are likely to result in indirect effects to the
rest of the aquatic community, including the Barton Springs salamanders, are evaluated
based on threshold concentrations. These screening threshold concentrations, which are
discussed in greater detail in Section 4.2 and Appendix B, incorporate the available
micro- and mesocosm data included in the 2003 IRED, as well as additional information
gathered following completion of the 2003 atrazine IRED (U.S. EPA, 2003b and 2003e).
4.2 Community-Level Endpoints: Threshold Concentrations
In this endangered species assessment, direct and indirect effects to the Barton Springs
salamander are evaluated in accordance with the screening-level methodology described
in the Agency's Overview Document (U.S. EPA, 2004). If aquatic plant RQs exceed the
Agency's non-listed species LOG (because the salamander does not have an obligate
relationship with any one particular plant species, but rather relies on multiple plant
species), based on available ECso data for vascular and non-vascular plants, risks to
individual aquatic plants are assumed.
It should be noted, however, that the indirect effects analysis in this assessment is unique,
in that the best available information for atrazine-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
associated with potential aquatic community-level effects (via aquatic plant community
structural change and subsequent habitat modification) to the Barton Springs salamander.
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). This information, which represents the best scientific data available,
is described in further detail below and in Appendix B.
As previously mentioned in Section 2.3, 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
75
-------
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 Barton Springs salamander, 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 LOG), rather than actual thresholds to be used as an LOG (U.S. EPA, 2003e).
The following threshold concentrations for atrazine were identified (U.S. EPA, 2003e):
• 14-day average = 38 ug/L
• 30-day average = 27 ug/L
• 60-day average =18 ug/L
• 90-day average =12 ug/L
Effects of atrazine on aquatic plant communities that have the potential to subsequently
pose indirect effects to the Barton Springs salamander 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 (U.S. EPA 2003e) are used to evaluate
potential indirect effects to aquatic communities for the purposes of this endangered
species assessment. 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
76
-------
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 Barton Springs salamander. 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 screening threshold concentrations is
provided in Figure 4.2. Further information on threshold concentrations is provided in
Appendix B.
1 ay affect,
not likely to
adversely affect
May affect, buT~
not likely to
, adversely affect'
Refine EECs based on monitoring data. Do refined EECs
exceed the threshold concentrations above?
i
Y
1
f J^~~
-
-— -^.
"Likely to
adversely affect'
Figure 4.2. Use of Threshold Concentrations in Endangered Species Assessment
77
-------
4.3 Use of Probit Slope Response Relationship to Provide Information on the
Endangered Species Levels of Concern
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.
4.4 Incident Database Review
A number of incidents have been reported in which atrazine has been associated with
some type of environmental effect, with variable levels of certainty that atrazine caused
the effects, ranging from unlikely to highly probable. As of the writing of the 2003 IRED
(U.S. EPA, 2003a), 109 incidents were listed in the Ecological Incident Information
System (EIIS) files under atrazine: 4 cases were listed as highly probable, 40 as probable,
50 as possible, 13 as unlikely, and 2 as unrelated. Atrazine alone is not very toxic to the
birds, mammals, and aquatic animals cited in most of these incidents. In none of these
cases has evidence been provided that firmly demonstrates that atrazine has produced the
reported effects. Atrazine residues in fish tissue were measured in only one incident
78
-------
reported as a fish kill (# 1004021-004); however, many chemicals were identified and
high profenofos levels were found. Therefore, the organophosphate was determined to be
responsible for the large fish kill. In many cases, the inference of these reported incidents
to atrazine is likely due to the widespread use of atrazine and the proximity of the
atrazine application and timing to the occurrence of the incident.
Between October 26, 2000 and June 9, 2006, 8 incidents were listed in the EIIS involving
the use of atrazine: 6 cases are listed as possible and 2 are listed as unlikely. The effects
of these incidents ranged from major fish kills to minor burning of garden plants adjacent
to a field treated with atrazine. Of these incidents, 5 were caused by drift, 1 by runoff
and 2 because of misuse.
Of the 6 cases that were listed as "Possible," all were terrestrial and, therefore, not
relevant to this assessment. In the two cases listed as "Unlikely," one resulted in the
death of 50-60 bass, 2,000 crappie and 300-400 bluegills (IN: #1013987-001). Three
chemicals, including terbufos, atrazine and acetochlor, were used in a product suspected
to be present in the runoff. Tests were conducted in the two affected ponds and terbufos
was the only chemical listed as being detected in both. It is not clear if atrazine and
acetochlor were measured in the pond water analysis. However, it is likely that terbufos
was responsible for the fish kill because it has a greater lethality to fish than atrazine and
acetochlor.
One of the two reported "misuse" incidents caused substantial damage to aquatic animals
(TN: # 1016990-001) resulting in the death of 2,000 bluegill sunfish, 400 catfish, and a
snake. This incident was credited to the dumping of 4 to 5 gallons of a product suspected
of containing atrazine into a one half acre pond.
Based on the available incident information, supporting data is not available to clearly
demonstrate that atrazine is the cause of the observed aquatic effects (i.e., death to fish).
In addition, the best available toxicity information shows that atrazine is not directly toxic
to freshwater fish (and aquatic-phase amphibians) at environmentally relevant
concentrations (see Sections 4.1.1 and 4.1.2). Further information on the atrazine
incidents reported in the 2003 IRED (U.S. EPA, 2003 a) and a summary of uncertainties
associated with all reported incidents are provided in Appendix F.
5. Risk Characterization
Risk characterization is the integration of the exposure and effects characterizations to
determine the potential ecological risk from varying atrazine use scenarios within the
action area and likelihood of direct and indirect effects on the Barton Springs salamander.
The risk characterization provides an estimation and a description of the likelihood of
adverse effects; articulates risk assessment assumptions, limitations, and uncertainties;
and synthesizes an overall conclusion regarding the effects determination (i.e., "no
effect," "likely to adversely affect," or "may affect, but not likely to adversely affect") for
the Barton Springs salamander.
79
-------
5.1 Risk Estimation
Risk was estimated by calculating the ratio of the estimated environmental concentrations
(EECs; see Table 3.5) and the appropriate toxicity endpoint (see Table 4.1). This ratio is
the risk quotient (RQ), which is then compared to pre-established acute and chronic
levels of concern (LOCs) for each category evaluated (Appendix G). Screening-level
RQs are based on the most sensitive endpoints and modeled surface water concentrations
from the following scenarios for atrazine:
• residential granular use @ 2 Ib ai/A; 2 applications with 30 days between
applications (assumes 1% over-application of atrazine granules to impervious
surfaces)
• residential liquid use @ 1 Ib ai/A; 2 applications with 30 days between
applications (assumes 1% over-spray of atrazine to impervious surfaces)
• turf granular use @ 2 Ib ai/A; 2 applications with 30 days between applications
• turf liquid use @ 1 Ib ai/A; 2 applications with 30 days between applications
• rights-of-way liquid use @ 1 Ib ai/A; 1 application (assumes 1% over-spray of
atrazine to impervious surfaces)
• fallow/idle land use @ 2.25 Ib ai/A; 1 application
In cases where the screening-level RQ exceeds one or more LOCs, additional factors,
including Barton Springs salamander life history characteristics, refinement of the EECs
using available monitoring data, and consideration of community-level threshold
concentrations, are considered and used to characterize the potential for atrazine to affect
the Barton Springs salamander. Risk estimations of direct and indirect effects of atrazine
to the Barton Springs salamander are provided in Sections 5.1.1 and 5.1.2, respectively.
As previously discussed in the effects assessment, the toxicity of the atrazine degradates,
including HA, DEA, DIA, and DACT, is assumed to be less than the parent compound;
therefore, RQ values were not derived for the degradates.
5.1.1 Direct Effects
Direct effects associated with acute and chronic exposure to atrazine in Barton Springs
are not expected to occur for the Barton Springs salamander. Risk quotients used to
estimate direct effects to the Barton Springs salamander are provided in Table 5.1 below.
Risk quotients were calculated only for the use that resulted in the highest EEC (granular
residential use) because none of the acute or chronic LOCs were exceeded. These risk
quotients are further characterized in Section 5.2.1.
80
-------
Table 5.1. Summary of Direct Effect RQs for the Barton Springs Salamander
Effect to
Barton
Springs
Salamander
Acute Direct
Toxicity
Chronic Direct
Toxicity
Surrogate
Species
Rainbow
trout
Brook trout
Toxicity
Value Qig/L)
LC50 = 5,300
NOAEC = 65
EEC Qig/L)
Peak: 41.2
60-day: 1.2
RQ
0.008
0.018
Probability of
Individual
Effect
1 in 1.7E+08
(1 in 1,870 to 1
in 5.82E+15)3
Not calculated
for chronic
endpoints
LOC
Exceedance
and Risk
Interpretation
Nob
Nob
a Based on a probit slope of 2.72 with 95% confidence intervals of 1.56 and 3.89 (MRID# 000247-16).
b RQ < acute endangered species LOC of 0.05.
5.1.2 Indirect Effects
Pesticides have the potential to exert indirect effects upon listed species by inducing
changes in structural or functional characteristics of affected communities. Perturbation
of forage or prey availability and alteration of the extent and nature of habitat are
examples of indirect effects.
In conducting a screen for indirect effects, direct effects LOCs for each taxonomic group
(i.e., freshwater fish, invertebrates, and aquatic plants) are employed to make inferences
concerning the potential for indirect effects upon listed species that rely upon non-listed
organisms in these taxonomic groups as resources critical to their life cycle (U.S. EPA,
2004). This approach used to evaluate indirect effects to listed species is endorsed by the
Services (USFWS/NMFS, 2004b). If no direct effect listed species LOCs are exceeded
for non-endangered organisms that are critical to the Barton Springs salamander's life
cycle, the concern for indirect effects to the Barton Springs salamander is expected to be
minimal.
If LOCs are exceeded for freshwater invertebrates that are prey items of the Barton
Springs salamander, there is a potential for atrazine to indirectly affect the salamander 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 analyzed 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 (U.S. EPA, 2004).
As an herbicide, indirect effects to the Barton Springs salamander from potential effects
on primary productivity of aquatic plants are a principle concern. If plant RQs fall
81
-------
between the listed species and non-listed species LOCs, a no effect determination for
listed species that rely on multiple plant species to successfully complete their life cycle
(termed plant-dependent species) is concluded. If plant RQs are above non-listed species
LOCs, this could be indicative of a potential for adverse effects to those listed species
that rely either on a specific plant species (plant species obligate) or multiple plant
species (plant-dependent) for some important aspect of their life cycle (U.S. EPA, 2004).
Based on the information provided in Appendix D, the Barton Springs salamander relies
on multiple plant species, including aquatic moss, pondweed, arrowhead, water primrose,
cabomba, and other aquatic plants for cover and as a source of habitat and food for the
variety and abundance of aquatic invertebrates that salamanders eat.
In summary, the potential for indirect effects to the Barton Springs salamander was
evaluated using methods outlined in U.S. EPA (2004) and described below in Sections
5.1.2.1 and 5.1.2.2, respectively.
5.1.2.1 Evaluation of Potential Indirect Effects via Reduction in Food Items
(Freshwater Invertebrates
Potential indirect effects from direct effects on animal food items (i.e., freshwater
invertebrates) were evaluated by considering the diet of the Barton Springs salamander
and the distribution of the sensitivities of the prey organisms to atrazine. Barton Springs
salamanders feed on a wide range of freshwater aquatic invertebrates including ostracods,
copepods, chironomids, snails, amphipods, mayfly larvae, leeches, and adult riffle
beetles. The most prevalent invertebrates found in stomach and fecal samples from a
limited number of adult and juvenile Barton Springs salamanders were ostracods,
amphipods, and chironomids (USFWS, 2005). However, data on the relative percentage
of each type of aquatic invertebrate in the salamander's diet are not available. The RQs
used to characterize potential indirect effects to the Barton Springs salamander from
direct acute and chronic effects on freshwater invertebrate food sources are provided in
Tables 5.2 and 5.3, respectively. Acute and chronic RQs are based on the most sensitive
toxicity endpoint for the midge (ECso = 720 ug/L) and the scud (NOAEC = 60 ug/L),
respectively.
82
-------
Table 5.2. Summary of RQs Used to Estimate Indirect Effects to the Barton Springs
Salamander via Direct Acute Effects on Dietary Items
Indirect Effect
to Barton
Springs
Salamander
Reduced Food
Supply via
Acute Direct
Toxicity to
Invertebrates
Surrogate
Food Item
/ Toxicity
Value
(HS/L)
Midge
EC50=720
Use (appl.
method; rate; #
appl.; interval
between appl.)
Residential
(granular; 2 Ib
ai/A; 2 appl.; 30 d
interval)
Residential
(ground liquid; 1
Ib ai/A; 2 appl.;
30 d interval)
Turf (granular; 2
Ib ai/A; 2 appl.;
30 d interval)
Turf (ground
liquid; 1 Ib ai/A;
2 appl.; 30 d
interval)
Fallow/Idle land
(aerial liquid;
2.25 Ib ai/A; 1
appl.)
Rights-of-Way
(liquid; 1 Ib ai/A;
1 appl.)
Peak
EECs
(jig/L)
41.2
26.6
22.4
16.2
7.5
6.2
RQ
0.057
0.037
0.031
0.023
0.010
0.009
Probability of
Individual
Effect3
1 in 4.55E+07
1 in 6.72E+09
lin6.29E+10
lin3.53E+12
1 in 1.46E+18
1 in 8.97E+18
LOC
Exceedance
and Risk
Interpretation
Yesb
Noc
Noc
Noc
Noc
Noc
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 invertebrates studies
that may serve as food items for the salamander; 95% confidence intervals could not be calculated based on
the available data (Table A-18; Taylor et al., 1991; MRID# 452029-17).
b RQ > acute listed species LOC of 0.05. Further evaluation of the range of freshwater invertebrate species
sensitivity to atrazine and dietary requirements of the Barton Springs salamander is completed in Section
5.2.2.
c RQ < acute listed species LOC of 0.05.
For freshwater invertebrates, acute RQs exceed the acute risk to the listed species LOC of
0.05 for the residential granular use (2 Ib ai/A) only. Acute RQs for the other modeled
atrazine uses are less than the listed species LOC. Because the listed species LOC is
exceeded for the residential granular use, atrazine use related to residential granular
applications has the potential to indirectly affect the Barton Springs salamander via
reduction in the availability of sensitive aquatic invertebrate food items. However, this
analysis was based on the most sensitive aquatic invertebrate endpoint of freshwater
species tested in laboratory studies and did not consider the range of aquatic invertebrate
species sensitivity to atrazine or the specific dietary requirements of the Barton Springs
83
-------
salamander. Therefore, additional characterization of the potential for atrazine to affect
freshwater invertebrate food items of the Barton Springs salamander is presented as part
of the Risk Description in Section 5.2.2.
Table 5.3. Summary of RQ and LOC Used to Estimate Indirect Effects to the
Barton Springs Salamander via Direct Chronic Effects on Dietary Items
Indirect Effect
to Barton
Springs
Salamander
Reduced Food
Supply via
Chronic Direct
Toxicity to
Invertebrates
Surrogate
Food Item
/ Toxicity
Value
(HS/L)
Scud
NOAEC =
60
Use (appl.
method; rate; #
appl.; interval
between appl.)
Residential
(granular; 2 Ib
ai/A;2appl.;30d
interval)
21-day
EECs
(jig/L)
2.5
RQ
0.04
LOC Exceedance and Risk
Interpretation
Noa
a RQ < chronic risk LOC of 1.0.
As shown in Table 5.3, the chronic LOC is not exceeded for freshwater invertebrates,
based on the use that results in the highest EECs (granular residential use). Therefore,
indirect effects to the Barton Springs salamander based on direct chronic effects to
dietary items are not expected to occur.
5.1.2.2 Evaluation of Potential Indirect Effects via Reduction in Habitat
and/or Primary Productivity (Freshwater Aquatic Plants)
Potential indirect effects from effects on habitat and/or primary productivity were
assessed using RQs from freshwater aquatic vascular and non-vascular plant data as a
screen. If aquatic plant RQs exceed the Agency's non-listed species LOC (because the
salamander relies on multiple plant species), potential community level effects are
evaluated using the threshold concentrations, as described in Section 4.2. Risk quotients
used to estimate potential indirect effects to the Barton Springs salamander from effects
on aquatic plants primary productivity are summarized in Table 5.4.
84
-------
Table 5.4. Summary of RQs Used to Estimate Indirect Effects to the Barton Springs
Salamander via Direct Effects on Aquatic Plants
Indirect Effect
to Barton
Springs
Salamander
Reduced
Habitat and/or
Primary
Productivity via
Direct Toxicity
to Aquatic
Plants
Use (appl. method;
rate; # appl.; interval
between appl.)
Residential (granular; 2
Ib ai/A; 2 appl.; 30 d
interval)
Residential (ground
liquid; 1 Ib ai/A; 2
appl.; 30 d interval)
Turf (granular; 2 Ib
ai/A; 2 appl.; 30 d
interval)
Turf (ground liquid; 1
Ib ai/A; 2 appl.; 30 d
interval)
Fallow/Idle land (aerial
liquid; 2.25 Ib ai/A; 1
appl.)
Rights-of-Way
(liquid; 1 Ib ai/A; 1
appl.)
Peak EECs
(jig/L)
41.2
26.6
22.4
16.2
7.5
6.2
Non-vascular
plant RQ
(EC50 = 1
Hg/La)
41.2
26.6
22.4
16.2
7.5
6.2
Vascular
plant RQ
(EC50 = 37
Hg/L")
1.11
0.72
0.61
0.44
0.20
0.17
LOG Exceedance
and Risk
Interpretation
Yesc
Yesd
Yesd
Yesd
Yesd
Yesd
aBased on 1-week EC50 value of 1 ug/L for four species of freshwater algae (MRID # 000235-44).
b Based on 14-day EC50 value of 37 ug/L for duckweed (MRID # 430748-08).
0 RQ > non-listed species LOG of 1.0 for both non-vascular and vascular plants. Direct effects to non-
vascular and vascular aquatic plants are possible. Further evaluation of the EECs relative to the threshold
concentrations (for community-level effects) is necessary.
d RQ > non-listed aquatic plant species LOG of 1.0 for non-vascular plants; RQ < non-listed plant species
LOG of 1.0 for vascular plant. Direct effects to non-vascular aquatic plants are possible. Further
evaluation of the EECs relative to the threshold concentrations (for community-level effects) is necessary.
Based on the results shown in Table 5.4, LOCs (RQ > 1.0) for direct effects to aquatic
non-vascular plants are exceeded for all modeled atrazine use scenarios; LOCs for direct
effects to aquatic vascular plants are exceeded only for the residential granular use
scenario. Therefore, atrazine has the potential to indirectly affect the Barton Springs
salamander via direct effects on both non-vascular and vascular aquatic plants for the
residential granular scenario, and via direct effects on non-vascular aquatic plants for all
modeled use scenarios. However, this screening-level analysis was based on the most
sensitive EC50 value from all of the available freshwater aquatic plant toxicity
information. No known obligate relationship exists between the Barton Springs
salamander and any single freshwater plant species; therefore, listed species RQs using
the NOAEC/ECos values for aquatic plants were not derived. Further analysis of the
time-weighted EECs relative to their respective threshold concentrations is necessary to
determine whether effects to individual plant species would likely result in community-
level effects. This analysis is presented as part of the Risk Description in Section 5.2.3.
85
-------
5.2 Risk Description
The risk description synthesizes an overall conclusion regarding the likelihood of adverse
impacts leading to an effects determination (i.e., "no effect," "may affect, but not likely
to adversely affect," or "likely to adversely affect") for the Barton Springs salamander.
If the RQs presented in the Risk Estimation (Section 5.1) show no indirect effects and
LOCs for the Barton Springs salamander are not exceeded for direct effects, a "no
effect" determination is made, based on atrazine's use within the action area. If,
however, indirect effects are anticipated and/or exposure exceeds the LOCs for direct
effects, the Agency concludes a preliminary "may affect" determination for the Barton
Springs salamander.
Following a "may affect" determination, additional information is considered to refine
the potential for exposure at the predicted levels based on the life history characteristics
(i.e., habitat range, feeding preferences, etc) of the Barton Spring salamander and
potential community-level effects to aquatic plants. Based on the best available
information, the Agency uses the refined evaluation to distinguish those actions that
"may affect, but are not likely to adversely affect" from those actions that are "likely to
adversely affect" the Barton Springs salamander.
The criteria used to make determinations that the effects of an action are "not likely to
adversely affect" the Barton Springs salamander 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. "Take" in this
context means to harass or harm, defined as the following:
• 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.
• Likelihood of the Effect Occurring: Discountable effects are those that are
extremely unlikely to occur. For example, use of dose-response
information to estimate the likelihood of effects can inform the evaluation
of some discountable effects.
• Adverse Nature of Effect: Effects that are wholly beneficial without any
adverse effects are not considered adverse.
86
-------
A description of the risk and effects determination for each of the established assessment
endpoints for the Barton Springs salamander is provided in Sections 5.2.1 through 5.2.3.
5.2.1 Direct Effects to the Barton Springs Salamander
Respective acute and chronic RQs of 0.008 and 0.018 (based on the modeled EECs from
the residential granular scenario assuming 1% overspray and 30% impervious surfaces)
are well below the Agency's acute and chronic risk LOCs for all modeled uses of atrazine
within the action area. Using an upper bound assumption of residential granular use
EECs, based on 10% overspray and 10% impervious surfaces (peak EEC = 67.1 ug/L and
60-day EEC = 2.0 ug/L; see Table 3.9), also results in respective acute and chronic RQs
of 0.013 and 0.03 that are less than the Agency's LOCs. As previously discussed, direct
effects to the Barton Springs salamander were based on freshwater fish data, which are
used as a surrogate for aquatic-phase amphibians.
The probability of an individual event to the Barton Springs salamander was calculated
for the acute RQ of 0.008, based on the dose response curve slope from the acute toxicity
study for the rainbow trout of 2.72 (MRID # 000247-16). The corresponding estimated
chance of an individual acute mortality to the Barton Springs salamander at an RQ level
of 0.008 (based on the acute toxic endpoint for surrogate freshwater fish) is 1 in 170
million. 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 default values for the rainbow trout dose
response curve slope estimate (95% C.I.: 1.56 to 3.89) were used to calculate upper and
lower estimates of the effects probability associated with the acute RQ. The respective
lower and upper effects probability estimates are 1 in 1,870 (0.05%) and 1 in 5.82E+15
(~1.7E-14%). Given the low probability of an individual mortality occurrence and acute
and chronic RQs that are well below LOCs, atrazine is not likely to cause direct adverse
effects to the Barton Springs salamander.
Further lines-of-evidence that atrazine is unlikely to cause direct adverse effects to the
Barton Springs salamander are provided by the information in the open literature. As
previously discussed, the Agency has concluded that there is currently insufficient
evidence to confirm or refute the hypothesis that atrazine exposure may impact gonadal
development in amphibians (U.S. EPA, 2003d). Further examination of the available
open literature data for aquatic-phase amphibians (discussed in Section 4.1.2) shows that
exposure to atrazine does not cause direct acute and/or chronic mortality at
environmentally relevant concentrations similar to the upper bounds of the modeled
EECs. Reported sublethal effects to aquatic-phase amphibians show reduced weight and
length at metamorphosis for frogs and salamanders at atrazine exposure concentrations of
approximately 200 and accelerated metamorphosis in salamanders at concentrations of
approximately 184 ug/L; however, no effects to growth or time to metamorphosis have
been reported at concentrations of < 68 ug/L, similar to the upper bound of modeled
EECs for atrazine uses within the action area. Therefore, direct effects to the survival,
growth, and reproduction of Barton Springs salamanders are unlikely to occur.
87
-------
As discussed in Section 4.1.1.3, several open literature studies raise concern about
sublethal effects of atrazine on endocrine-mediated functions in freshwater fish, which
are used as a surrogate for aquatic-phase amphibians. However, the significance of these
effects is difficult to quantify because they are not quantitatively linked to changes in
survival, growth, and reproduction of individuals (i.e., the assessment endpoints for the
Barton Springs salamander). In addition, differences in habitat and behavior of the tested
species compared with the Barton Springs salamander suggest that the results may not be
relevant to this assessment. Furthermore, there is uncertainty associated with
extrapolating effects observed in the laboratory to more variable exposures and
conditions in the field. Further details on potential atrazine-related sublethal effects to
fish are provided in Appendix A.
A review of the available aquatic incidents shows that only two incidents involving fish
kills have been reported from 2000 through 2006. One of the two incidents was reported
as "unlikely" (#10139876-001) and the other was reported as a "misuse" (#1013550-003).
Based on all reported aquatic incidents for atrazine, none were reported in Texas and
none of the incidents reported effects to aquatic-phase amphibians. Further information
on all of the reported aquatic incidents for atrazine is provided in Section 4.4 and
Appendix F. Uncertainties related to the use of incident information from the Ecological
Incident Information System (EIIS) are also discussed in Appendix F.
In summary, the Agency concludes a "no effect" determination for direct effects to the
Barton Springs salamander, via mortality, growth, or fecundity, based on all available
lines of evidence.
5.2.2 Indirect Effects via Reduction in Food Items (Freshwater Invertebrates)
The results of the screening-level risk assessment for the Barton Springs salamander
suggest the potential for direct acute adverse effects to freshwater invertebrates, based on
the residential granular use of atrazine at 2 Ib ai/A (assuming 1% overspray and 30%
impervious surfaces). The acute RQ of 0.057 exceeds the listed species of 0.05;
therefore, atrazine use related to residential granular applications has the potential to
indirectly affect the Barton Springs salamander via reduction in the availability of
sensitive aquatic invertebrate food items.
However, this analysis was based on the lowest LCso value of 720 jig/L for the midge
(Chironomus spp.). Consideration of all acute toxicity data for the midge shows a wide
range of sensitivity within and between species of the same genus (2 orders of
magnitude) with values ranging from 720 to >33,000 |ig/L. Although the midge is a
component of the Barton Spring salamander's diet, this species reportedly consumes a
wide range of freshwater invertebrates that also include ostracods, copepods, snails,
amphipods, mayfly larvae, leeches, and adult riffle beetles. Available acute toxicity
values for other freshwater invertebrates that are included in the Barton Spring
salamander's diet (i.e., amphipods, leeches, and snails) are 5,700 jig/L and higher.
88
-------
The potential for atrazine to elicit indirect effects to Barton Springs salamanders via
effects on food items is dependent on several factors including: (1) the potential
magnitude of effect on freshwater invertebrate individuals and populations; and (2) the
number of prey species potentially affected relative to the expected number of species
needed to maintain the dietary needs of the Barton Springs salamander. Together, these
data provide a basis to evaluate whether the number of individuals within a prey species
is likely to be reduced such that it may indirectly affect the Barton Springs salamander.
Table 5.5 presents acute RQs and the probability of individual effects for dietary items of
the Barton Springs salamander including midges, amphipods, leeches, and snails. The
species sensitivity distribution of all acute toxicity data for freshwater aquatic
invertebrates tested is represented in Figure 4.1. This analysis considers only acute risk
to aquatic invertebrate food items as chronic risk quotients for invertebrates were less
than the Agency's LOG. Even at the upper bound of EECs (21-day EEC of 4.2 ug/L
from Table 3.9) based on assumptions of 10% overspray and 10% impervious surfaces
for the residential granular use scenario), the chronic RQ of 0.07 is well below the LOG.
Table 5.5. Summary of RQs Used to Assess Potential Risk to Freshwater
Invertebrate Food Items of the Barton Springs Salamander
Barton Springs
Salamander Food
Item Species
Midge
Amphipod
Leech
Snail
Acute
Toxicity
Value Range
Oig/L) (No. of
Studies)
720 - >33,000
(5)
5,700 - 14,900
(3)
>16,000 (1)
>16,000 (1)
RQ Range
(based on
an EEC of
40 uff/U>
<0.01 -
0.057
0.01
Probability of
Individual
Effect*
Up to 1 in
4.55E+07
1 46E+18
Risk Interpretation
Atrazine may affect sensitive food
items, such as the midge; however
the low probability of an individual
effect to the midge is not likely to
indirectly affect the Barton Springs
salamander via reduction in midge
prey items.
Based on low probability of
individual effects and RQs that are
well below acute LOCs, atrazine is
not likely to indirectly affect the
Barton Springs salamander via
reduction in amphipod, leech, or
snail prey items.
*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 invertebrates studies that may serve as
food items for the salamander; 95% confidence intervals could not be calculated based on the available data
(Table A-18).
As shown in Table 5.5, the listed species LOG is exceeded for the midge (RQ = 0.057),
based on the LCso value of 720 ug/L. However, acute RQs based on the other acute
toxicity data for the midge are <0.04, less than the acute risk to endangered species LOG.
Sufficient dose-response information was not available to allow for an estimation of the
probability of an individual effect on the midge. Therefore, the probability of an
individual effect was calculated using a probit dose response curve slope of 4.4; this is
89
-------
the only slope for technical grade atrazine reported in available ecotoxicity data for
freshwater invertebrates that are a component of the Barton Springs salamander's diet
(amphipod; MRID # 452029-17). Based on a probit slope of 4.4, the probability of an
individual mortality to the midge at an RQ of 0.057 is approximately 1 in 45.5 million
(2.2E-08%).
Acute LOCs are not exceeded for the other dietary items of the Barton Springs
salamander including the amphipod, leech or snail, based on the residential granular use
EEC (assuming 1% overspray and 30% impervious surfaces). In addition, acute RQs
based on the upper bound residential granular peak EEC of 67.1 ug/L (assuming 10%
overspray and 10% impervious surfaces) are also less than acute LOCs for these food
items in the salamander's diet.
Based on the non-selective nature of feeding behavior in the Barton Springs salamander
and low magnitude of anticipated individual effects to all evaluated prey species, atrazine
is not likely to indirectly affect the Barton Springs salamander via a reduction in
freshwater invertebrate food items. This finding is based on insignificance of effects
(i.e., effects to freshwater invertebrates are not likely to result in "take" of a single Barton
Springs salamander) and discountability (i.e., the effect to freshwater invertebrates is
extremely unlikely to occur given the estimated individual event probability of 1 in 45.5
million). Therefore, the effects determination for the assessment endpoint of indirect
effects on the Barton Springs salamander via direct effects on prey (i.e., freshwater
invertebrates) is "may affect, but not likely to adversely affect."
5.2.3 Indirect Effects via Reduction in Habitat and/or Primary Productivity
(Freshwater Aquatic Plants)
Direct adverse effects to non-vascular aquatic plants are possible, based on all modeled
atrazine uses within the action area. In addition, direct effects to vascular plants are
possible, based on the residential granular use of atrazine. Based on these direct effects,
atrazine may indirectly affect the Barton Springs salamander via direct effects on aquatic
plants. Therefore, the time-weighted EECs (for 14-day, 30-day, 60-day, and 90-day
averages) were compared to their respective time-weighted threshold concentrations to
determine whether potential effects to individual plant species would likely result in
community level effects. As discussed in Section 4.2, concentrations of atrazine from the
exposure profile at a particular use site and/or action area that exceed any of the
following time-weighted threshold concentrations indicate that changes in the aquatic
plant community structure could be affected:
• 14-day average = 38 ug/L
• 30-day average = 27 ug/L
• 60-day average =18 ug/L
• 90-day average =12 ug/L
90
-------
A comparison of the 14-, 30-, 60-, and 90-day EECs for the Barton Springs salamander
with the atrazine threshold concentrations representing potential aquatic community-level
effects is provided in Table 5.6.
Table 5.6. Summary of Modeled Scenario Time-Weighted EECs with Threshold
Concentrations for Potential Community-Level Effects
Use Scenario
Res.
(granular)
(1%OS;30%
IS) 7(10%
OS; 10% IS)
Res
(liquid)
Turf
(granular)
Turf
(liquid)
Rights-of-
way
Fallow/
Idle land
14-day
EEC
3.57
6.0
2.5
2.0
1.7
1.1
1.0
Threshold
Cone.
38
30-day
EEC
1.9 /
3.2
1.5
1.2
1.0
0.8
0.7
Threshold
Cone.
(jig/L)
27
60-day
EEC
(jig/L)
1.2 /
2.0
1.0
0.8
0.7
0.6
0.6
Threshold
Cone.
18
90-day
EEC
1.0 /
1.5
0.8
0.7
0.6
0.6
0.6
Threshold
Cone.
12
OS = overspray
IS = impervious surfaces
Based on the results of this comparison, predicted 14-, 30-, 60-, and 90-day EECs for all
modeled atrazine use scenarios (including the upper bound residential granular use EECs
assuming 10% overspray and 10% impervious surfaces) are well below the threshold
concentrations representing community-level effects. Although atrazine use may directly
affect individual aquatic plants in Barton Springs, its use within the action area is not
likely to adversely affect the Barton Springs salamander via indirect community-level
effects to aquatic vegetation. This finding is based on insignificance of effects (i.e.,
community-level effects to aquatic plants are not likely to result in "take" of a single
Barton Springs salamander) Therefore, the effects determination for the assessment
endpoint of indirect effects on the Barton Springs salamander via direct effects on habitat
and/or primary productivity of aquatic plants is "may affect, but not likely to adversely
affect."
91
-------
6. Uncertainties
6.1 Exposure Assessment Uncertainties
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 particular, peak exposures are generally an order of magnitude above
the highest detection found in any of the four springs. 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 exercise conservatively assumes that the spring and atrazine application
site are adjacent. In reality, there are likely to be processes at work which cannot be
accounted for in the modeling that will reduce the predicted exposures. In addition, the
impact of setbacks on runoff estimates has not been quantified, although these buffers,
especially those that are well-vegetated, are likely to result in significant reduction in
runoff loading of atrazine.
6.1.1 Modeling Assumptions
Overall, the uncertainties addressed in this assessment cannot be quantitatively
characterized. However, given the available data and the tendency to rely on
conservative modeling assumptions, it is expected that the modeling results in an over-
prediction in exposure. In general, the simplifying assumptions used in this assessment
appear from the characterization in Section 3.2.7 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 exposure over-estimation that cannot be quantified, but
can be qualitatively described. For instance, modeling for each use site assumes (with the
exception of the rights-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 conservative assumption in lieu of actual data.
6.1.2 Impact of Vegetative Setbacks on Runoff
Unlike spray drift, tools are currently not available to evaluate the effectiveness of a
vegetative setback on runoff and loadings. The effectiveness of vegetative setbacks is
highly dependent on the condition of the vegetative strip. For example, a well-
established, healthy vegetative setback can be a very effective means of reducing runoff
and erosion from agricultural fields. Alternatively, a setback of poor vegetative quality
or a setback that is channelized can be ineffective at reducing loadings. Until such time
as a quantitative method to estimate the effect of vegetative setbacks of various
conditions on pesticide loadings becomes available, the aquatic exposure predictions are
likely to overestimate exposure where healthy vegetative setbacks exist and
underestimate exposure where poorly developed, channelized, or bare setbacks exist.
92
-------
6.1.3 PRZM Modeling Inputs and Predicted Aquatic Concentrations
In general, the linked PRZM/EXAMS model produces estimated aquatic concentrations
that are expected to be exceeded once within a ten-year period. The Pesticide Root Zone
Model (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 simulates 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.
Uncertainty associated with each of these individual components adds 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 are actually treated with atrazine
may be lower than the Agency's default assumptions including use of the maximum
allowable application rate, treatment of the entire crop, and the estimated 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.
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
93
-------
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, and is therefore,
considered as protective of the Barton Springs salamander.
6.2.2 Use of surrogate species effects data
Guideline toxicity tests are not available for salamanders; therefore, freshwater fish are
used as surrogate species for aquatic-phase amphibians including salamanders. The
available open literature information on atrazine toxicity to aquatic-phase amphibians
shows that acute and chronic ecotoxicity endpoints for aquatic-phase amphibians are
generally about 3 to 4 times less sensitive than freshwater fish. Therefore, endpoints
based on freshwater fish ecotoxicity data are assumed to be protective of potential direct
effects to aquatic-phase salamanders including the Barton Springs salamander, and
extrapolation of the risk conclusions from the most sensitive tested species to the Barton
Springs salamander is likely to overestimate the potential risks to those species. 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. In
addition, the Agency's LOCs are intentionally set very low, and conservative estimates
are made in the screening level risk assessment to account for these uncertainties.
6.2.3 Acute freshwater invertebrate toxicity data for the midge
The initial acute risk estimate for freshwater invertebrates was based on the lowest
toxicity value from Chironomus studies, which 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 freshwater invertebrates may represent
an overestimation of potential direct risks to freshwater invertebrates and indirect effects
to the Barton Springs salamander via a reduction in available food.
6.2.4 Extrapolation of long-term environmental effects from short-term
laboratory tests
The influence of length of exposure and concurrent environmental stressors to the Barton
Springs salamander (i.e., urban expansion, habitat modification, decreased quantity and
quality of water in Barton Springs, predators, etc.) will likely affect the species response
to atrazine. Additional environmental stressors may decrease the Barton Spring
salamander's sensitivity to the herbicide, although there is the possibility of
additive/synergistic reactions. 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. However, as previously discussed, the
Agency's LOCs are intentionally set very low, and conservative estimates are made in the
screening level risk assessment to account for these uncertainties.
94
-------
6.2.5 Use of threshold concentrations for 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
Barton Springs salamander. 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 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.
The threshold concentrations are predictive of potential atrazine-related community-level
effects in aquatic ecosystems, such as Barton Springs, where the species composition
may differ from those included in the micro- and mesocosm studies. Although it is not
possible to determine how well the responses observed in the micro- and mesocosm
studies reflect the Barton Springs aquatic community, estimated high-end atrazine
exposure concentrations in the action area (from modeled EECs) are predicted to be
between 10 to 30 times lower than the community-level threshold concentrations,
depending on the modeled atrazine use and averaging period. 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, including the Barton Springs salamander. Additional uncertainties
associated with use of the screening thresholds to estimate community-level effects are
discussed in Section B.8 of Appendix B.
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.
Sufficient dose-response information was not available to estimate the probability of an
individual effect on the midge (one of the dietary food items of the Barton Springs
salamander). Acute ecotoxicity data from the midge was used to derive RQs for
freshwater invertebrates. Based on a lack of dose-response information for the midge,
the probability of an individual effect was calculated using the only probit dose response
curve slope value reported in available freshwater invertebrate ecotoxicity data for
technical grade atrazine. Therefore, a probit slope value of 4.4 for the amphipod, which
is also a component of the Barton Springs salamander's diet, was used to estimate the
probability of an individual effect on the freshwater invertebrates. It is unclear whether
the probability of an individual effect for freshwater invertebrates other than amphipods
would be higher or lower, given a lack of dose-response information for other freshwater
invertebrate species. However, the assumed probit dose response slope for freshwater
95
-------
invertebrates of 4.4 would have to decrease to approximately 1 to 2 to cause an effect
probability ranging between 1 in 10 and 1 in 100, respectively, for freshwater
invertebrates.
7. Summary of Direct and Indirect Effects to the Barton Springs Salamander
In fulfilling its obligations under Section 7(a) (2) of the Endangered Species Act, the
information presented in this endangered species risk assessment represents the best data
currently available to assess the potential risks of atrazine to the Barton Springs
salamander. A summary of the risk conclusions and effects determination for the Barton
Springs salamander, given the uncertainties discussed in Section 6, is presented in Table
7.1.
Table 7.1. Effects Determination Summary for the Barton Springs Salamander
Assessment Endpoint
Effects determination
Basis for Determination
Survival, growth, and
reproduction of Barton
Springs salamander
individuals via direct
effects
No effect
No acute and chronic LOCs are exceeded.
Indirect effects to Barton
Springs salamander via
reduction of prey (i.e.,
freshwater invertebrates)
May affect, but not likely
to adversely affect
Acute LOCs are exceeded based on the most sensitive
ecotoxicity value for the midge; however RQs for other dietary
items (amphipods, leeches, snails) are less than LOCs. Based
on the non-selective nature of feeding behavior in the Barton
Springs salamander and low magnitude of anticipated
individual effects to all evaluated prey species, atrazine is not
likely to indirectly affect the Barton Springs salamander via a
reduction in freshwater invertebrate food items. This finding is
based on insignificance of effects (i.e., effects to freshwater
invertebrates are not likely to result in "take" of a single Barton
Springs salamander) and discountability (i.e., the effect to
freshwater invertebrates is extremely unlikely to occur given
the estimated individual event probability of 1 in 45.5 million).
Indirect effects to Barton
Springs salamander via
reduction of habitat and/or
primary productivity (i.e.,
aquatic plants)
May affect, but not likely
to adversely affect
Although atrazine use may directly affect individual vascular
and non-vascular aquatic plants in Barton Springs, its use
within the action area is not likely to adversely affect the
Barton Springs salamander via indirect community-level
effects to aquatic vegetation. Predicted 14-, 30-, 60-, and 90-
day EECs for all modeled atrazine use scenarios within the
action area are well below the threshold concentrations
representing community-level effects. This finding is based on
insignificance of effects (i.e., community-level effects to
aquatic plants are not likely to result in "take" of a single
Barton Springs salamander).
96
-------
8. References
Abou-Waly, Hoda, M. M. Abou-Setta, H. N. Nigg and L. L. Mallory. 1991.
Growthresponse of freshwater algae, Anabaenaflos-aquae and Selenastrum
capricornutum to Atrazine and hexazinone herbicides. Bull. Environ. Contam.
Toxicol. 46:223-229.
Allran, J. W. and Karasov, W. H. 2001. Effects of Atrazine on Embryos, Larvae, and
Adults of Anuran Amphibians. Environ. Toxicol.Chem. 20: 769-775.
EcoReference No.: 59251.
Armstrong, D. E., C. Chester and R. F. Harris. 1967. Atrazine hydrolysis in soil. Soil Sci.
Soc. Amer. Proc. 31:61-66.
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.
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.
Beliles, R. P. and W. J. Scott, Jr. 1965. Atrazine safety evaluation on fish and wildlife
Bobwhite quail, mallard ducks, rainbow trout, sunfish, goldfish): Atrazine: Acute
toxicity in rainbow trout. Prepared by Woodard Res. Corp.; submitted by Ciba-
Geigy Corp., Greensboro, NC. (MRID No. 000247-16).
Boone, M. D. and James, S. M. 2003. Interactions of an Insecticide, Herbicide, and
Natural Stressors in Amphibian Community Mesocosms. Ecol.Appl. 13: 829-841.
EcoReference No.: 81455.
Brune, G. 1981. Springs of Texas, Volume I. Branch-Smith, Inc., Fort Worth, Texas.
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).
Caux, Pierre-Yves, L. Menard, and R.A. Kent. 1996. Comparative study of the effects
of MCPA, butylate, atrazine, and cyanazine on Selenastrum apricornutum.
Environ. Poll. 92(2):219-225.
Chippindale, P.T. 1993. Evolution, phytogeny, biogeography, and taxonomy of Central
Texas spring and cave salamanders, Eurycea and Typhlomolge (Plethodontidae:
Hemidactyliini). Dissertation, University of Texas at Austin, Austin, Texas.
97
-------
City of Austin (COA). 2003a. Unpublished Land Use Data.
http://www.ci.austin.tx.us/landuse/. Accessed 15 February 2005.
City of Austin. 2003b. Land Use Data (ftp://coageoid01.ci.austin.tx.us/GIS-
Data/Regional/coa gis.html).
City of Austin Parks and Recreation Department (P ARD) Integrated Pest Management
Plan(IPM). 2005. April 27, 2005.
Davis, B. 2006. County Extension Agent, Texas Cooperative Extension (Hays County)
Personal Communication.
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.
DeLong-Amaya, A. 2006. Ladybird Johnson Wildflower Center, Personal
Communication.
Diana, S. G., Resetarits, W. J. Jr., Schaeffer, D. J., Beckmen, K. B., and Beasley, V. R.
2000. Effects of Atrazine on Amphibian Growth and Survival in Artificial
Aquatic Communities. Environ.Toxicol.Chem. 19: 2961-2967. EcoReference
No.: 59818.
Field, M. 2004. Forecasting Versus Predicting Solute Transport in Solution Conduits for
Estimating Drinking-Water Risks. Acta Carsologica, XXXIII/II, pp 115-150.
Forson, D. and Storfer, A. 2006. Effects of Atrazine and Iridovirus Infection on Survival
and Life-History Traits of the Long-Toed Salamander (Ambystoma
macrodactylum). Environ.Toxicol.Chem. 25: 168-173. EcoReference No.: 82033.
Garcia, E. 2006. Natural Resources Conservation Service - District Conservationist
(Travis County), Personal Communication.
Giddings, J.M., T.A. Anderson, L.W. Hall, Jr., R.J. Kendall, R.P. Richards, K.R.
Solomon, and W.M. Williams. 2000. 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.
Gucciardo, L. S. 1999. The Use of Anuran Larvae to Determine Chronic and Acute
Toxicological Effects from Exposure to Atrazine and Metolachlor. Ph.D. Thesis,
Iowa State Univ., Ames, IA 164 p. EcoReference No.: 78286.
Hauwert, N., Johns, D. Hunt, B., Beery, J., Smith, B., and Sharp, J.M., Jr. 2004. Flow
Systems of the Edwards Aquifer Barton Springs segment Interpreted from tracing
and associated field studies.: So. Texas Geol. Soc. /Austin Geol. Soc. Symposium
on the Edwards Aquifer, San Antonio, TX, in press.
98
-------
Hecker, M., Park, J. W., Murphy, M. B., Jones, P. D., Solomon, K. R., Van der Kraak,
G., Carr, J. A., Smith, E. E., Du Preez, L., Kendall, R. J., and Giesy, J. P. 2005.
Effects of Atrazine on CYP19 Gene Expression and Aromatase Activity in Testes
and on Plasma Sex Steroid Concentrations of Male African Clawed Frogs
(Xenopus laevis). Toxicol.Sci. 86:273-280. EcoReference No.: 79287.
Hoberg, J. R. 1993. Atrazine technical: Toxicity to duckweed, (Lemnagibba). SLI Rep.
No. 93-4-4755. Prepared by Springborn Laboratories, Inc., Wareham, MA.;
submitted by Ciba-Geigy Corporation, Greensboro, NC. (MRID No. 430748-04).
Johnson, I. C., Keller, A. E., and Zam, S. G. 1993. A Method for Conducting Acute
Toxicity Tests with the Early Life Stages of Freshwater Mussels. In: W.G.Landis,
J.S.Hughes, and M.A.Lewis (Eds.), Environmental Toxicology and Risk
Assessment, ASTMSTP 1179, Philadelphia, PA 381-396.
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).
Kaul, M., Kiely, T., Jones, A., and Widawsky, D. 2006. Atrazine, Simazine, Carbaryl,
Metolachlor, Diazinon, and Prometon Usage for Blanco, Hays, and Travis
Counties in Texas to Support the Barton Springs Endangered Species Assessment
(D322226, D322266 & D322267).
Macek, K. J., K. S. Buxton, S. Sauter, S. Gnilka and J. W. Dean. 1976. Chronic toxicity
of atrazine to selected aquatic invertebrates and fishes. US. EPA, Off. Res. Dev.,
Environ. Res. Lab. Duluth, MN. EPA-600/3-76-047. 49 p. (MRID # 000243-
77).
Maher, B. 2005a. The Case of the Anomalous Springs. Using all the clues to solve a
karst mystery. USGS presentation to the ESI France-UT Karst Workshop;
February 8, 2005.
Maher, B. 2005b. Research Hydrologist, U.S. Geological Survey, Texas District.
Personal Communication.
Majewski, M.S., and P.D. Capel. 1995. Pesticides in the Atmosphere - Distribution,
Trends, and Governing Factors: Ann Arbor Press, Chelsea, Michigan, 228 p.
Majewski, M.S., W.T. Foreman, and D.A. Goolsby. 2000. Pesticides in the atmospheres
of the Mississippi River Valley, Part I - Rain. Sci. Tot. Environ. 248:201-212.
99
-------
Markwardt, D. 2006. Vegetation Management Specialist. Texas DOT, Maintenance
Division, Personal Communication.
Mason, J. 2006. Vegetation Management Specialist. Texas DOT, Maintenance
Division, Personal Communication.
McVeety, E. 2006. Pesticide Applicator at Zilker Park, City of Austin Parks and
Recreation Department, Personal Communication.
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.
Perez, C. , 2006, District Conservationist (Hays County), Personal Communication.
Pipkin, T. and M. Freeh, editors. 1993. Barton Springs eternal. Softshoe Publishing,
Austin, Texas.
Rohr, J. R., Elskus, A. A., Shepherd, B. S., Crowley, P. H., McCarthy, T. M.,
Niedzwiecki, J. H., Sager, T., Sih, A., and Palmer, B. D. 2003. Lethal and
Sublethal Effects of Atrazine, Carbaryl, Endosulfan, and Octylphenol on the
Streamside Salamander (Ambystoma barbouri). Environ.Toxicol.Chem. 22:
2385-2392. EcoReference No.: 71723.
Rohr, J. R., Elskus, A. A., Shepherd, B. S., Crowley, P. H., McCarthy, T. M.,
Niedzwiecki, J. H., Sager, T., Sih, A., and Palmer, B. D. 2004. Multiple Stressors
and Salamanders: Effects of an Herbicide, Food Limitation, and Hydroperiod.
EcolAppl. 14: 1028-1040. EcoReference No.: 81748
Schulz, A., F. Wengenmayer and H. M. Goodman. 1990. Genetic engineering of
herbicide resistance in higher plants. Plant Sci. 9:1-15.
Shay, K. 2006. Water Quality Education Manager, City of Austin Watershed Protection
and Development Review Board, Personal Communication.
Slade, R. J., Jr., Dorsey, M. E., Stewart, S. L. 1986. Hydrology and Water Quality of the
Edwards Aquifer Associated with Barton Springs in the Austin Area, Texas, U.S.
Geological Survey Water Resources Investigations Report 86-4036.
Stratton, G. W. 1984. Effects of the herbicide atrazine and its degradation products,
alone and in combination, on phototrophic microorganisms. Bull. Environ.
Contam. Toxicol. 29:35-42. (MRID # 45087401).
Taylor, E.J., SJ. Maund, and D. Pascoe. 1991. Toxicity of four common pollutants to
the freshwater macroinvertebrates Chironomus riparius Meigen (Insecta:
100
-------
Diptera) and Gammaruspulex (L.) (Crustacea: Amphipoda). Arch. Environ.
Contain. Toxicol. 21:371-376. (MRID 452029-17).
Torres, A. M. R. and L. M. O'Flaherty. 1976. Influence of pesticides on Chlorella,
Chlorococcum, Stigeoclonium (Chlorophyceae), Tribonema, Vaucheria
(Xanthophyceae) and Oscillatoria (Cyanophyceae). Phycologia 15(l):25-36.
(MRID #000235-44).
U. S. Department of Agriculture (USDA), Natural Resources Conservations Service
(NRCS). 2000. Conservation Buffers to Reduce Pesticide Losses. Natural
Resources Conservation Service. Fort Worth, Texas. 21pp.
USDA Soil Conservation Service. 1986. Urban Hydrology for Small Watersheds.
Technical Release 55, 2nd edition, June 1986.
U.S. Environmental Protection Agency (EPA). 1998. Guidance for Ecological Risk
Assessment. Risk Assessment Forum. EPA/630/R-95/002F, April 1998.
U.S. EPA. 2003a. Interim Reregi strati on Eligibility Decision (IRED) for Atrazine.
Office of Pesticide Programs. Environmental Fate and Effects Division. January
31, 2003. http://www.epa.gov/oppsrrdl/REDs/0001.pdf
U.S. EPA. 2003b. Revised Atrazine Interim Reregi strati on (IRED). Office of Pesticide
Programs. Environmental Fate and Effects Division. October 31, 2003.
http://www.epa.gov/oppsrrdl/REDs/0001.pdf
U.S. EPA. 2003c. Ambient Aquatic Life Water Quality Criteria for Atrazine - Revised
Draft. Office of Water, Office of Science and Technology, Health and Ecological
Criteria Division, Washington, D.C. 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 (LOG) 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 Pesticide Programs. Washington, D.C. January 23, 2004.
U.S. EPA. 2005. Pesticide Root Zone Model (PRZM) Field and Orchard Crop
Scenarios: Guidance for Selecting Field Crop and Orchard Scenario Input
101
-------
Parameters, Version II. Water Quality Tech Team, Environmental Fate and
Effects Division, Office of Pesticide Programs, U. S. Environmental Protection
Agency. Washington, D.C. 20460. April, 2005.
U.S. EPA. 2006a. 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. 2006b. 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 (USFWS). 2005. Barton Springs Salamander (Eurycea
sosoruni) Recovery Plan. Southwest Region, USFWS, Albuquerque, New
Mexico.
U.S. Fish and Wildlife Service (USFWS) and National Marine Fisheries Service
(NMFS). 1998. Endangered Species Consultation Handbook: Procedures for
Conducting Consultation and Conference Activities Under Section 7 of the
Endangered Species Act. Final Draft. March 1998.
USFWS/NMFS. 2004a. 50 CFR Part 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).
U.S. Geological Survey (USGS), National Mapping Division, Rocky Mountain Mapping
Center. 2003. Edwards Aquifer Land Use / Land Cover. Denver, Colorado.
Accessed online at http://www.tceq.state.tx.us/gis/lulc.html.
Walsh, G. E. 1983. Cell death and inhibition of population growth of marine unicellular
algae by pesticides. Aquatic Toxicol. 3:209-214. (MRID # 45227731).
Ward, D., 2006, Travis County Highway Department, Personal Communication.
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. 456223-04).
Zinn, N. and A. Jones. 2006. Application Timing and Refined Usage Data for Atrazine,
Simazine, Metolachlor, and Prometon in Three Texas Counties (Blanco, Hays,
and Travis), D322539.
102
------- |