UNITED STATES ENVIRONMENTAL PROTECTION AGENCY

WASHINGTON D.C., 20460

OFFICE  OF 

PREVENTION, PESTICIDES AND 

TOXIC SUBSTANCES

PC Code: 058001

DP Barcodes: 311127, 311128

DATE:  25 October 2006

MEMORANDUM	

SUBJECT:     Azinphos Methyl—OPP Response to Stakeholder Comments
Regarding Ecological Risks for the Proposed Decision for “Group 3”
Uses

FROM:  	Colleen Flaherty, Biologist (ERB 3)

		R. David Jones, Chemist (ERB 4)

		Environmental Fate and Effects Division (7507C)

	

THRU:	Daniel Rieder, Branch Chief (ERB 3)

		Elizabeth Behl, Branch Chief (ERB 4)

		Environmental Fate and Effects Division (7507C)

TO:		Katie Hall, Risk Manager (RRB 2)

		Special Review and Reregistration Division (7505C)

	Attached please find the Office of Pesticide Program’s (OPP’s)
response to stakeholder comments on the proposed decision regarding the
ecological risk assessment for the use of azinphos methyl on apples,
blueberries (low- and high bush), Brussels sprouts, cherries (sweet and
tart), grapes, nursery stock, parsley, pears, pistachios, and walnuts
(i.e. “Group 3 uses”). 

 Bayer CropScience and Makhteshim Agan of North America 

(EPA-HQ-OPP-2005-0061-0162)

Summary

1. Comment (page 2 of 17): “The agency relies on screening level
modeling to establish predicted environmental (exposure) concentrations
for AZM in the aquatic systems.  These estimates are extremely
conservative by design.”

OPP response:  The Agency’s estimates are conservative by design, but
not extremely so. The models themselves (PRZM and EXAMS) are
approximately best estimators; the conservatism comes dominantly from
five sources – scenario selection, hydrologic setting, chemical input
parameter selection, use pattern, and EEC return frequency. Tier 2
scenarios are selected by best professional judgment for each crop to
represent a location that would be expected produce EECs higher than 90%
of sites where a particular crop is grown. The hydrologic setting is a
pond high in the watershed where a single field dominates the land use
in the watershed. This watershed serves as a surrogate for a number of
small vulnerable water bodies that may be found at the top of a
watershed such as prairie potholes, playa lakes, vernal pools, some
wetlands, and zero and first order streams. Generally, best estimate
values are used for most chemistry input parameters. However, metabolic
degradation rates have high background variability, and the upper 90%
confidence bound on the mean value is used rather than the means as the
input parameters. The maximum application rate allowed on the label is
used for the application pattern, and in some cases, this can be
substantially greater than the typical use pattern for a given crop. In
the case of azinphos methyl, the typical use rates are consistent with
the maximum label rates.  Finally, a return frequency of 1 in 10 years
is used for the EEC. These conservatisms are expected to compensate for
the variability that occurs in the environment across the landscape and
through time, as well as a range of management practices, and
uncertainties due to a limited data set. We expect these exposures to
occur in some places at some times. It is unknown what proportion of
water bodies that border treatment areas exceed known effect levels and
how quickly azinphos methyl concentrations drop below effects thresholds
when traveling downstream.

2. Comment (page 2-3 of 17): “…the most recent (and relevant) data
provide(s) estimates that are far below the agency’s screening level
predictions...more weight should be given (to) these data as they offer
more realistic exposure estimates.”

OPP response: Monitoring data and modeling predictions have distinct
temporal and spatial differences, and these data cannot truly be
compared.  However, recent monitoring data indicate that actual surface
water concentrations are not inconsistent with model predictions.  

For example, the United States Geological Survey NAWQA program has
collected 120 samples at discharge of the Granger Drain in Washington
from 1999 through 2004 and analyzed these for azinphos methyl. The
number of samples analyzed per year varied from 4 (2000) to 30 (2003).
The frequency of detection above the detection limit varied from 0 in
2000 to 80% is 2001. The lack of azinphos methyl detections in 2000 is
likely to be related to the limited number of samples in that year. In
three of the four years with detection, there was at least one sample
over the acute level of concern (LOC) for aquatic invertebrates (0.08
ppb). In the remaining year, 1999, the maximum measured concentration
was 0.076 ppb, just below the LOC. Because of the limited number of
samples, the measured samples are unlikely to have captured the actually
maximum concentration that occurred. For example, with weekly sampling
there is only a one in seven chance of collecting a sample on the day
when the peak concentration occurred. Furthermore, NAWQA uses a
multi-analyte method for pesticide analysis. The analytical efficiency
varies for each pesticide in the method. Unfortunately, azinphos methyl
is not efficiently measured with this analytical method. Only 13% of the
pesticide in a water sample is recovered on average (USGS NAWQA, 1999).
As a consequence, all reported detections of azinphos methyl are marked
with an ‘E’ for estimated, indicating that these values are of lower
reliability than other measurements made by the program. It also
indicates that actual concentrations in water are 8 to 10 times greater
than the measurements indicate. This would also result in the maximum
measured value of 0.179 ppb approximately 1.4 ppb, which is fairly close
to the EEC estimated for azinphos methyl use on apples in Washington,
4.1 ppb, particularly considering the difference in return frequency and
hydrologic setting for modeling versus the sampling location.

Also, USGS NAWQA data for Zollner Creek, OR detected concentrations up
to about 4 µg/L in June of 2005.  This level is within the range of
1-in-10 year peak PRZM/EXAMS modeled exposures for the use of AZM on
apples, which is from about 3 to 9 µg/L. 

3.  Comment (page 3 of 17):  “…more weight should be given to these
(higher tiered field level toxicity) studies as they were conducted
under more realistic conditions and thus provide more realistic
estimates of aquatic toxicity.”

OPP response:  The ecological risk assessment of azinphos methyl is
based on a screening-level deterministic approach.  As described in the
Overview of the Ecological Risk Assessment Process document, at a
screening level, it is OPP policy to rely on the most sensitive species
tested to evaluate potential effects to non-target animals at a national
level.  The assessment endpoints for the azinphos methyl ecological risk
assessment were defined as reduced survival and/or reproductive
impairment for aquatic and terrestrial species from direct acute and
chronic exposures.  These data were obtained from registrant-submitted
studies and from the open literature.  Surrogate test species used in
toxicity studies are not selected based on their sensitivity to
pesticides, but rather on their ability to thrive under laboratory
conditions.  Therefore, OPP’s endpoint selection may not be
representative of the most sensitive species that may be exposed to
azinphos methyl, but rather it is the most sensitive species tested in
the laboratory.  

Field effects data can be used as additional lines of evidence in
ecological risk assessments (as in case of the azinphos methyl). 
However, field studies are limited in the ability to account for the
myriad combinations of physical, chemical, and biological variables that
may affect organism response to pesticides in the environment. 
Consequently, field studies cannot conclusively validate screening risk
assessment predictions, but they can allow inferences on the
reasonableness of the assessment predictions. In the case of azinphos
methyl, the available aquatic field studies suggest that community-level
zooplankton effects occur at concentrations between 1 and 4 µg/L. 
These studies, which were described in detail in the ecological risk
assessment (D307568), support the laboratory toxicity data, which
suggests that individual aquatic invertebrate mortalities can occur at
levels below 1 µg/L.

4.  Comment (page 3 of 17): “…the agency should strongly consider
exposure estimates from surface water monitoring and toxicity estimates
from aquatic field studies…the data demonstrate risk that is
acceptable and more predictive than the screening level assessment
conducted by the agency.”

OPP response:  Surface water monitoring data were considered and are
discussed in the assessment. The available monitoring data for azinphos
methyl show exposures above the Agency’s threshold of concern.
Furthermore, surface water monitoring data tends to underestimate
exposures, particularly acute exposures, because infrequent sampling
makes it unlikely that the highest concentrations that occur in any year
will be captured in any given sample.  In addition, the duration of most
monitoring studies is typically only a few years, and thus, information
to characterize less frequent events is limited.  The hydrologic setting
of most sampling sites in a monitoring program is downstream of
relatively more vulnerable sites in the same watershed. In most cases,
monitoring data serves as a lower bound for the potential exposures (and
consequent risk) and is useful on that basis.

5.  Comment (page 3 of 17): “…further weight should be given to the
lack of aquatic incidents entered into the agency’s tracking
system…The agency repeatedly cites dated aquatic incidents around the
use of AZM to substantiate the ecological risk assessment and proposed
decision but does not provide adequate weight to the recent trends in
this area.”

OPP response:  As discussed previously (DP barcode 307569), this
“trend” of decreasing adverse ecological incidents in the EIIS
database is not unique to azinphos methyl and is likely a function of
reduced reporting rather than a drastic decrease in pesticide risk to
fish and wildlife.  Further, since there is no nationwide, rigorous
incident monitoring program in place, reported incidents can only be
used as a line of evidence to support a risk conclusion that a given
taxon may be at risk from a particular pesticide use.  Given the
haphazard nature of incident reporting, lack of incident data cannot
suggest lack of ecological risk.  

In the case of azinphos methyl, based on a deterministic ecological risk
assessment, risk to aquatic animals was concluded, and aquatic incidents
were used to support this conclusion.  The reported “fish kill”
incidents linked to azinphos methyl use suggest that when this
organophosphate insecticide is applied to various crops, exposures in
water bodies adjacent to or immediately downstream of the treated area
may be high enough to elicit mortality effects.

6.  Comment (page 3 of 17):  “Similar to trends in the aquatic area,
model predications (sp.) indicate dietary exposure via residues in
excess of mortality thresholds…these estimates are overly predictive
compared to likely dietary exposure residues.”

OPP response:  As mentioned in the ecological risk assessment for the
“Group 3” azinphos methyl uses (DP barcode 307568), field studies by
Johnson et al. (1989, MRID 41139701) and Sheeley et al. (1989, MRID
41195901) show that there is concordance between the predicted residues
based on the Kenaga nomogram and actual measured residues of azinphos
methyl in the field.  In some cases, measured residues from these
studies actually exceed those predicted by the Kenaga nomogram. 
Residues on apple tree foliage were measured within 24 hours of spray
blast applications, and mean-measured residues were 199 (82-393) and 236
(105-476) ppm, for Washington and Michigan, respectively.  In
Washington, measured residues after the second and third application
were 312 ppm (123-564 ppm) and 328 ppm (122-611 ppm), respectively.  In
Michigan, residues measured after the second and third applications were
429 ppm (111-1499 ppm) and 536 ppm (208-1747 ppm), respectively. 
Predicted residues based on the Kenaga nomogram range from 45-713 ppm
for the “upper-bound” estimate and from 21-253 ppm for the mean
estimate.  

7.  Comment (page 3 of 17):  “…the agency uses dated incident
reports to substantiate possible risk to terrestrial organisms…”

OPP response:  See above response to Comment #5.

8.  Comment (page 3-4 of 17):  “The agency also cites terrestrial
field studies performed in apple orchards to substantiate a position on
potential exposure and risk.  The agency should consider that these
studies performed were level 1 field studies which are designed to
provide only qualitative information about effects…They did not
include control or reference sites so it is not possible to directly
assess whether mortality rates at AZM treated sites was greater than
background rates.  The numbers of vertebrate mortalities found per unit
of search effort on treatment sites in the Michigan study cited were
similar to control sites of other avian field studies conducted by the
registrants.  The numbers of vertebrate mortalities found per unit
search effort in the Washington study was greater than control sites in
other studies, however, most of these mortalities occurred at 2 sites
where rodenticides were applied and were attributed by the authors to
the rodenticide applications rather than to AZM.”

OPP response:  The field studies conducted in apple orchards in Michigan
(Sheeley et al. 1989, MRID 41195901) and Washington (Johnson et al.
1989, MRID 41139701) demonstrated that some birds and small mammals are
at risk for acute mortality effects following azinphos methyl spray
applications at the current rate.  In both studies, the effects of
Guthion 35WP applications on wildlife were assessed by comparing the
number of casualties found before and after treatment, by examining
circumstances under which casualties were found, and residue levels in
carcasses.  The pre-treatment search served as the control.

In the Michigan study, two casualties were recorded pre-treatment and 27
were recorded post-treatment.  Of the 27 post-treatment mortalities
(tabulated below), 14 were considered highly likely to have been
treatment related, six were possibly treatment related, and seven were
not treatment related.  Most carcasses were found within the orchards
(38%) or along their perimeter (45%), but 17% were located in adjacent
areas outside the orchards. 

In the Washington study, 173 casualties were recorded, including 59
birds of 14 species, 109 mammals of seven species, and five reptiles of
two species.  Of these, 162 (94%) were found after treatments began. 
American robins and California quail accounted for 34% and 20%,
respectively, of the total avian casualties.  Meadow voles comprised 82%
of the mammalian casualties.  Only 40 of the 173 casualties were
analyzed for tissue residue, and 21 (53%) were considered treatment
related based on the detection of residue in carcasses.  Additionally,
117 other casualties might have been treatment related based on the
circumstances and/or time frames under which carcasses were found.  Only
35 casualties were considered not treatment related.  Of the carcasses
recovered, 46% were found along orchard perimeters, 41% in orchard
interiors, and 13% in areas adjacent to the orchards.

Several other terrestrial field studies (Edge et al., 1996; Peterson,
1996; Schauber et al., 1997; Matz et al., 1998) have been previously
described in the ecological risk assessment for the Group 3 azinphos
methyl uses.  These studies also indicated adverse effects to
terrestrial wildlife following azinphos methyl exposure.  

Section 1. Screening Level Exposure Modeling and Monitoring Data

9.  Comment (page 4 of 17):  “With the extensive amount of monitoring
data available on AZM, the risk assessment can and should be based on
the actual concentrations seen in the environment…the monitoring data
are comprehensive enough (ca. 21,000 samples), in terms of number of
monitoring sites, location of the sites and length of sampling periods,
to be used in the risk assessment.”

  

OPP response:  We agree that the monitoring data are extensive. As
discussed in the response to comment #4, the available monitoring data
for azinphos methyl show exposure above the Agency’s threshold of
concern. Furthermore, surface water monitoring data tends to
underestimate exposures, particularly acute exposures, because
infrequent sampling makes it unlikely that the highest concentrations
that occur in any year will be captured in any given sample.  In
addition, the duration of most monitoring studies is typically only a
few years, and thus, information to characterize less frequent events is
limited.  The hydrologic setting of most sampling sites in a monitoring
program is downstream of relatively more vulnerable sites in the same
watershed. In most cases, monitoring data serves as a lower bound for
the potential exposures (and consequent risk) and is useful on that
basis.

10. Comment (page 4 of 17): “Literature studies have shown that the
standard modeling approach over-predicts residues by one to four orders
of magnitude (Jackson et al. 2005, Jones 2005)”

OPP Response: Both of these papers fail to account for the differences
in hydrologic setting, duration of monitoring versus modeling
assessment, frequency of sampling, and the relative vulnerabilities of
modeled versus monitored sites.  The studies make inappropriate
comparisons of modeled exposures and monitoring data estimates. 

The estimates from Tier 2 modeling are by design, expected to be
somewhat higher than those found in the environment. However, EFED does
not agree with the conclusions reached by Jackson et al. (2005) and
Jones (2005) that the modeling is unrealistic.  The basis for
comparisons between modeling and monitoring in this article is unclear,
and the endpoints derived from the modeling and the reported monitoring
results are wholly different and are not comparable. The paper did not
provide adequate detail about (1) the differences in the endpoints for
modeling and monitoring, (2) the assumptions and inputs used in
modeling, or (3) the design of the monitoring study.  It did not
describe the nature of the EPA models, or how these models were used.
These details are important in interpretation of results, in supporting
both the author’s recommendations to develop adjustment factors and
the author’s judgment on the validity of the conceptual model.  

Additionally, the authors’ description of how EPA’s endpoints (i.e.
1-in-10 year annual peaks at a 90% site) are derived was inadequate and
misleading. For a reasonable comparison to be made, key input parameters
used in modeling (application rates, application frequencies,
application dates, and percentage cropped area adjustment factors)
should represent actual pesticide use and usage in each of the
watersheds monitored. Similarly, monitoring intervals and duration
should match those used in modeling. This could not be determined
because the authors had not provided the information necessary to
interpret the monitoring study results, and such data are usually not
available. Additionally, the study authors compared a representative
monitoring dataset to the 1-in-10 year return frequency model estimates
and did not provide details of pesticide usage in the watershed.  As a
result, statistical analysis of the data cannot provide meaningful
results. Given the uncertainties in the overall methodology, the
authors’ conclusions about the adequacy of EPA’s conceptual model
for water modeling are not supported by their analysis.

11. Comment (page 5 of 17): “…the ca. 21,000 samples taken and
analyzed for AZM are significant value to the risk assessment.”

OPP Response: NAWQA data through 2000 was considered in ecological risk
assessment supporting the IRED. Other sources of monitoring data were
also considered and described in the assessment. For example, STORET is
a database of different data collected at different times for different
purposes. In most cases, it is difficult to establish the context (e.g.
why it was collected, how it was collected, sample QA/QC, environmental
conditions, etc.) for which the data was collected as this data is
usually not maintained in the database. Without context, it is not
possible to use the data in risk assessment. Additionally, the Pilot
Reservoir Monitoring Study was designed to establish the potential for
exposure to pesticides in reservoir-based drinking water. It was
targeted to reservoirs vulnerable to pesticide contamination, but not to
azinphos methyl use areas.

The assessment and conclusions of Makhteshim and Bayer CropScience are
based on an unscientific assessment of the monitoring data. Placing all
identified data regardless of source into a single distribution of
values results in conclusions that are not meaningful, for the following
reasons:

Aquatic organisms do not typically experience exposures from many
different streams, but only one stream at a time. 

The analysis does not account for differences in sampling frequency and
duration at each sampling site.

The analysis does not account for the different total number of samples
at each site.

The analysis does not account for differences in analytical
methodologies (e.g. levels of detection, extraction efficiency) from
study to study.

The analysis ignores differences in hydrologic setting from study to
study and from site to site.

Since none of these factors have been accounted for, the combined
distribution of these monitoring values has no scientific value.  Hence,
any comparisons to this monitoring analysis are not useful
scientifically.  

12. Comment (page 5 of 17):  “…special attention should be given to
the recent water monitoring and invertebrate toxicity study conducted by
Chelan County in Washington State...”

  

OPP response:  EPA has evaluated the monitoring data available from the
Chelan County Conservation District.  It would be difficult to conclude
that these samples represent high exposure conditions for azinphos
methyl.  Downstream sites in this study are at least 2nd and 3rd order
streams and do not represent the most vulnerable locations in the
watershed. It is not clear from the study the proximity of the various
orchards in the watershed to the creeks sampled. Surprisingly, while
there were detections in Mission Creek, the application data show no
applications of azinphos methyl to Mission Creek during June. Only
Yaksum Creek had applications of azinphos methyl made during the
sampling window. 

It is not surprising that there were no significant mortality effects on
Daphnia pulex.  Laboratory data indicate that the LC50 for daphnids is
approximately 1 µg/L.  The detected levels in this study ranged from
less than 0.02 µg/L to 0.19 µg/L.

Section II. Aquatic Toxicity

13.  Comment (page 12 of 17): “…there is a low probability of
exposure exceeding the threshold value (1 ppb) derived from higher tier
aquatic studies.”

  

OPP response:  The study referenced in the comment was evaluated by EPA
and the “threshold value” cited (1 µg/L for aquatic organisms) was
determined to not be useful for EPA’s risk assessment, as it is not
adequately protective for azinphos methyl. The field studies that the
registrants refer to suggest that a single exposure of azinphos methyl
at 4 µg/L can elicit population-level effects in aquatic ecosystems. 
Most of the remaining azinphos methyl uses allow more than one
application per season, and the magnitude of the population declines and
the time to recover would surely increase with increasing applications. 
The EFED ecological risk assessment used a benchmark toxicity value of
0.16 µg/L for freshwater invertebrates based on the LC50 for the scud.

In addition, chronic toxicity studies indicate that significant
reproductive effects can occur at levels well below 1 µg a.i./L.  In a
21-day Daphnia magna chronic toxicity study, significant effects on
survivorship, length, and fecundity (mean number of young per adult per
reproductive day) were observed at a LOAEC of 0.40 μg a.i./L. 

Further, we disagree with the conclusion of low probability of exposure
exceeding 1 μg a.i./L.  The analysis method for monitoring data used to
draw that conclusion was substantially flawed as discussed above in the
response to comment #11.

Section III. Incidence Reports

14. Comment (page 12 of 17): “What is not clearly presented is the
significant reduction in both aquatic and terrestrial incidences
reported in association to labeled uses.”

OPP response:  See above response to Comment #5.

Farm Worker Justice (EPA-HQ-OPP-2005-0061-0165)

This stakeholder has submitted a document entitled “Do Buffers Make
Good Neighbors? Minimizing Pesticide Exposure at the Ag-Urban
Interface” by Dr. Allan Felsot, an environmental toxicologist from
Washington State University.

Dr. Felsot has proposed using the AgDrift Model to estimate vegetated
buffer strips around dwellings to reduce exposure to populations living
in the vicinity of orchards. In general outline, we agree that Dr.
Felsot’s approach is sound, and we are considering a similar approach
to explore potential mitigation measures for the azinphos methyl risks.
A list of specific comments on Dr. Felsot’s approach follows.

1. Comment (page 6): “…I have concluded that simulations of orchard
spraying with AgDrift is conservative enough to allow its use for
designing no-spray buffer zones to protect bystanders.”

OPP response:  The orchard airblast component of AgDrift is a based on a
regression analysis of the measured drift from about 40 drift trials in
orchards. The AgDrift output is the mean estimator across these drift
trials, and thus, is a best estimate of the drift from orchards, not a
conservative one. Multiplying the drift estimate by three is used as a
surrogate for 90% of the drift curves.

2. Comment (page 6): “Although the creators of the AgDrift model claim
that wind is not a very important factor in drift from orchards… the
light winds blowing across the transect may have dampened drift away
from the orchard.”

OPP response:  It would be better to say there is no statistically
significant association of wind speed on drift magnitude found in the
spray drift data set. It is likely that this association was masked by
other confounding factors during the drift trials. We would agree with
Dr. Felsot that wind speed and direction are factors in spray drift from
air blast application in orchards.

3. Comment (page 8): “The effect of not spraying outside rows…the
no-spray buffer zone drops…”

OPP response:  We agree that not spraying the outer rows can reduce
spray drift, though it may provide refugia for the pests. 

4. Comment (page 9): “Buffers: one strategy among many”

OPP response:  We agree that there a large number of other drift
reduction strategies available including shelter belts (lines of dense
trees at the edge of the orchard) and alternative pesticides.

American Farm Bureau Federation (EPA-HQ-OPP-2005-0061-0166)

1. Comment (page 1): “Relevant portions of the ecological data and
modeling are unrealistic and significantly overestimate risk.”

OPP response:  In the case of azinphos methyl, the deterministic
risk-quotient based approach is supported by various aquatic and
terrestrial monitoring data, field studies, and adverse ecological
incidents.  

2. Comment (page 1): “…some relevant ‘lower risk’ incident data
may not have (been) fully considered.”

OPP response:  It is unclear which data are being referred to in this
comment.  Any additional data that stakeholders can provide regarding
this subject will be considered in future assessments. EPA has evaluated
and considered all available incident data in our assessment.

3. Comment (page 1): “…EPA should more thoroughly explore potential
risk mitigation measures that may allow for the continued use of AZM.”

OPP response:  Any mitigation proposals that stakeholders can provide
will be considered in future assessments.

Northwest Horticultural Council (EPA-HQ-OPP-2005-0061-0168)

1. Comment (page 4): “The outcomes predicted by the conservative
screening model used by the agency appear to overestimate real world
values…because…this model has never been adequately validated across
the range of environmental and site conditions where AZM is used.”

OPP response:  The two components of Tier 2 exposure analysis, PRZM and
EXAMS, have been repeatedly validated (see list below) and are
approximately best estimators of transport of pesticides in the
environment.  Tier 2 modeling is dominantly conservative because of
choices of scenario, return frequency, and chemical input parameters,
which are used to account for variation in the environment and the
uncertainties due to limited data sets.

Allen, B. Walter, Jr., M. Craig Barber, Sandra Bird, Lawrence, A. Burns,
J. Mark Cheplick, Dudley R. Hartel, Charles A. Kitner, Foster L. Mayer,
Luis Suarez, and Steven E. Wooten. 1991. PIRANHA: Pesticide and
Industrial Chemical Risk Analysis and Hazard Assessment, Version 2.0. p
21-35. Environmental Research Laboratory, Office of Research and
Development, U. S. Environmental Protection Agency, Athens, GA.

Durborow, T. E., N. L Barnes, S. Z. Cohen, G. L. Horst, A. E. Smith.
2000. Calibration and Validation of Runoff and Leaching Models for Turf
Pesticides, And Comparison with Modeling Results. Chapter 12 in Fate and
Management of Turfgrass Chemicals. ACS Symposium Series 743. American
Chemical Society.

Jones, R. David, Sidney Abel, William Effland, Robert Matzner, and
Ronald Parker. 1998. An Index Reservoir for Use in Assessing Drinking
Water Exposure. Chapter IV in Proposed Methods for Basic-Scale
Estimation of Pesticide Concentration in Flowing Waters and Reservoirs
for Tolerance Reassessment. Presented to the FIFRA Scientific Advisory
Panel, July 29-30, 1998.    HYPERLINK
"http://www.epa.gov/scipoly/sap/1998/index.htm" 
http://www.epa.gov/scipoly/sap/1998/index.htm 

Jones, Russell L., and Mark H. Russell (eds.). 2001. FIFRA Environmental
Modeling Validation Task Force: Final Report. American Crop Protection
Association. Washington, DC.   HYPERLINK
"http://femvtf.com/femvtf/Files/FEMVTFbody.pdf" 
http://femvtf.com/femvtf/Files/FEMVTFbody.pdf 

Kolset, Knut and Anders Heiberg. 1988. Evaluation of the ‘Fugacity
(FEQUM) and the ‘EXAMS’ Chemical fate and Transport Models: A Case
Study on the Pollution of the Norrsundet Bay (Sweden). Wat. Sci. Tech.
20:1-12.

Melancon, Susan, M., James M. Pollard, and Stephan C. Herd. 1986.
Evaluation of SESOIL, PRZM and PESTAN in a Laboratory Column Leaching
Experiment. Environmental Toxicology and Chemistry 5:865-878.

Mueller, Thomas C., Ronald E. Jones, Parshall B. Bush, and Phillip
Banks. 1992. Environmental Toxicology and Chemistry 11:427-436.

Office of Pesticide Programs. 2002. Revised OP Cumulative Assessment:
Appendix E1: Comparisons of Estimated Regional OP Distributions with
Occurrences in Ambient Water from the USGS NAWQA Program.   HYPERLINK
"http://www.epa.gov/pesticides/cumulative/rra-op/III_E_WaterAppendix/III
_E_1.pdf" 
http://www.epa.gov/pesticides/cumulative/rra-op/III_E_WaterAppendix/III_
E_1.pdf 

Parrish, Rudolph S., Charles N. Smith, and J. Mark Cheplick. 1989.
Testing Pesticide Leaching Models: An Agricultural Field Study.
Environmental Research Laboratory, Athens, GA. 

Parrish, Rudolph S., Charles N. Smith, and Fred K. Fong. 1992. Tests of
the Pesticide Root Zone Model and the Aggregate Model for Transport and
Transformation of Aldicarb, Metolachlor, and Bromide. Journal of
Environmental Quality 21(4):685-697.

Smith, C. N., R. S. Parrish, R. F. Carsel, A. S. Donigian, jr, and J. M.
Cheplick. 1988. Validation Status of the Pesticide Leaching   SEQ
CHAPTER \h \r 1 and Groundwater Transport Model. Environmental Research
Laboratory, Office of Research and Development. U. S. Environmental
Protection Agency. Athens, GA.

Walker, Allan, R. Calvet, A. A. M. Del RE, W. Pestemer, and J. M.
Hollis. 1995. Evaluation and Improvement of Mathematical Models of
Pesticide Mobility in Soils and Assessment of Their Potential to Predict
Contamination of Water Systems. Blackwell Wissenschafts-Verlag GmbH,
Berlin

2. Comment (page 4): “Data collected in Washington state in 2003 and
2004 do not confirm the values used in the agency’s September 29, 2005
ecological risk assessment.”

OPP response:  As described above in the response to comment #12 from
Bayer CropScience and Makhteshim Agan of North America, it would be
difficult to conclude that these samples represent high exposure
conditions for azinphos methyl.  Downstream sites in this study are at
least 2nd and 3rd order streams and do not represent the most vulnerable
locations in the watershed. It is not clear from the study the proximity
of the various orchards in the watershed to the creeks sampled.
Surprisingly, while there were detections in Mission Creek, the
application data show no applications of azinphos methyl to Mission
Creek during June. Only Yaksum Creek had applications of azinphos methyl
made during the sampling window. 

California Almond Board (EPA-HQ-OPP-2005-0061-0172)

1.  Comment (page 5):  “…we have doubts about relying on the very
conservative ecological risk assessment exclusively, especially as more
recent water quality monitoring data do not back up the numbers
calculated in the risk assessment.”

OPP response:  As described above in the response to Bayer CropScience
and Makhteshim Agan of North America comment #2, monitoring data and
modeling predictions have distinct temporal and spatial differences, and
these data cannot truly be compared.  However, recent monitoring data
indicate that actual surface water concentrations are not inconsistent
with model predictions.  

For example, the United States Geological Survey NAWQA program has
collected 120 samples at discharge of the Granger Drain in Washington
from 1999 through 2004 and analyzed these for azinphos methyl. The
number of samples analyzed per year varied from 4 (2000) to 30 (2003).
The frequency of detection above the detection limit varied from 0 in
2000 to 80% is 2001. The lack of azinphos methyl detections in 2000 is
likely to be related to the limited number of samples in that year. In
three of the four years with detection, there was at least one sample
over the acute level of concern (LOC) for aquatic invertebrates (0.08
ppb). In the remaining year, 1999, the maximum measured concentration
was 0.076 ppb, just below the LOC. Because of the limited number of
samples, the measured samples are unlikely to have captured the actually
maximum concentration that occurred. For example, with weekly sampling
there is only a one in seven chance of collecting a sample on the day
when the peak concentration occurred. Furthermore, NAWQA uses a
multi-analyte method for pesticide analysis. The analytical efficiency
varies for each pesticide in the method. Unfortunately, azinphos methyl
is not efficiently measured with this analytical method. Only 13% of the
pesticide in a water sample is recovered on average (USGS NAWQA, 1999).
As a consequence, all reported detections of azinphos methyl are marked
with an ‘E’ for estimated, indicating that these values are of lower
reliability than other measurements made by the program. It also
indicates that actual concentrations in water are 8 to 10 times greater
than the measurements indicate. This would also result in the maximum
measured value of 0.179 ppb approximately 1.4 ppb, which is fairly close
to the EEC estimated for azinphos methyl use on apples in Washington,
4.1 ppb, particularly considering the difference in return frequency and
hydrologic setting for modeling versus the sampling location.

As another example, USGS NAWQA data for Zollner Creek, OR detected
concentrations up to about 4 µg/L in June of 2005.  This level is
within the range of 1-in-10 year peak PRZM/EXAMS modeled exposures for
the use of AZM on apples, which is from about 3 to 9 µg/L. 

2. Comment (page 5):  “…AZM is never applied in March when there is
significant rainfall.  Applications occurred in mid-May through
mid-August (2000-2003), and now occur primarily in July (2004).”

OPP response: OPP has reviewed data from California DPR on the timing of
pesticide applications since 2002 and agrees that since that time, the
vast majority of applications have been made in June through August to
control navel orange worm.  To characterize these applications, EPA has
conducted additional modeling runs simulating applications on May 15 and
July 15, as suggested by the California Almond Board (see Appendix). 
For comparison purposes, the simulation for a March 15 application was
rerun using the orchard option to generate spray drift. This is intended
to show the potential for drift when spray blast is not directed over
the canopy as stated on the label. (See D307569 for details). The
one-in-ten-year peak EEC for the March application is 6.3 ppb and is
reduced to 3.2 ppb for the May application and 3.1 for July application.
This modification to the application scheme substantially reduces the
estimated exposures; however, acute and chronic risks to aquatic animals
continue to exceed the Agency’s levels of concern.

While these additional simulations do more accurately reflect current
azinphos methyl usage for control of the navel orange worm, the previous
simulation modeling using March 15 as an application date is appropriate
for another pest, the peach twig borer.  The recommended application
timing of azinphos methyl for control of this pest is “post-bloom,”
which is mid-March, as documented in the ecological risk assessment for
the Group III uses (D307568).

3. Comment (page 5): “EPA claims that they cannot rely on the more
recent incident reports because incidents are underreported…EPA cannot
have it both ways.”

OPP response: As discussed in the above response to Bayer CropScience
and Makhteshim Agan of North America comment #5, since there is no
nationwide, rigorous incident monitoring program in place, reported
incidents can only be used as a line of evidence to support a risk
conclusion that a given taxon may be at risk from a particular pesticide
use.  Lack of incident data cannot suggest lack of ecological risk.  

In the case of azinphos methyl, based on a deterministic ecological risk
assessment, risk to aquatic animals was concluded, and aquatic incidents
were used to support this conclusion.  The reported “fish kill”
incidents linked to azinphos methyl use suggest that when this
organophosphate insecticide is applied to various crops, exposures in
water bodies adjacent to or immediately downstream of the treated area
may be high enough to elicit mortality effects.

4. Comment (page 5): “The Regional Water Quality Board has had several
detections in the San Joaquin Valley, but they are well below any
toxicological levels of concern.”

OPP response:  OPP has not done a full review of these particular
monitoring data.  However, it appears that the AZM detections in San
Joaquin valley cited in this comment are incorrect.  According to
California State Water Resources Control Board, AZM was detected in July
2005 at concentrations of 0.165 and 0.33 parts per billion (µg/L) –
not parts per trillion as suggested by the CA Almond Board in this
comment.  These exposures do, in fact, exceed the Agency’s water
quality criterion for azinphos methyl (0.01 µg/L).

5. Comment (page 6): “EPA does not mention that the pyrethroids may
have ecological issues of their own, especially with regard to aquatic
species.”

OPP response:  EPA acknowledges that there are potential ecological
risks associated with any pesticide, including synthetic pyrethroids.  A
comparative ecological risk assessment for azinphos methyl and all
possible alternatives has not been performed.  However, the Agency’s
understanding of the ecological risks associated with the remaining uses
of azinphos methyl is that the pesticide poses both acute and chronic
risks to aquatic and terrestrial animals and that at least some of the
alternatives, e.g., phosmet, are less toxic and less persistent and
mobile than azinphos methyl, and as such, pose less of a risk to the
environment.

6. Comment (page 6): “…OP-consuming enzyme…This may be one way of
reducing both the concern with run-off and for terrestrial non-target
species.”

OPP response:  An OP-consuming enzyme sounds like it has the potential
to be a promising risk mitigation option.  However, the feasibility of
this option in the context of azinphos methyl application to various
fruit and nut crops is unknown.  It is also unknown to what extent
aquatic and terrestrial exposures may be reduced as a result of
application of this OP-consuming enzyme in conjunction with azinphos
methyl.  Any references that can be provided regarding this subject
would be considered in future assessments.

U.S. Apple Association (EPA-HQ-OPP-2005-0061-0163)

1. Comment (page 7): “The model overestimates the amount of drift into
the farm pond because the wind speed will not be 10 mph all the time and
the wind will not blow in the direction of the pond for every
application.  Growers prefer applications with minimal wind speed for
most effective target coverage.” 

OPP response: OPP acknowledges that assuming that the wind always blows
directly at the water body and that the wind speed is 10 mph will
overestimate exposure. An improved spray drift modeling capability that
can account for wind speed and direction effects on aquatic exposure is
currently be developed. Note that wind speed effects cannot be simulated
for air blast applications since there was no statistically significant
effect of wind speed in the air blast drift trials used to develop
AgDrift model.

2. Comment (page 7): “The model overestimates the amount of spray
drift because fine droplets will be carried a farther distance from the
target.  Additionally, growers are directed by the label to use larger
droplet sizes and will likely follow the recommendation.”

OPP response: For apples, the model only considers droplet sizes for the
aerial application allowed on the 24(c) label in Idaho. The portion of
AgDrift that simulates air blast application does not use droplet size
spectrum as an input parameter. Consequently, this label recommendation
does not affect the simulation results for this application method.

3. Comment (Page 7): “The model overestimates the drift calculation
because an orchard canopy with full vegetation is dense and will capture
a much higher percentage of the application than a dormant tree. 
Additionally, dormant applications are prohibited.”

OPP response: In recognition of the statements on the label prohibiting
application to dormant orchards and disallowing spray over the top of
the canopy, the orchard used in AgDrift to simulate airblast
applications was changed from ‘sparse orchard’ to ‘orchard’ for
the additional simulations performed in response to the comments on the
Group 3 risk assessment (D307577).  Thus, spray drift is not
overestimated in the latest assessment.

4. Comment (Page 8): “By not using buffers in its water run off model,
the model greatly overestimates the actual pesticide runoff into water
sources.”

OPP response: Based on recommendations from the Natural Resources
Conservation Service (USDA NRCS, 2000), OPP only considers reduction in
pesticide loading from buffers that are constructed and maintained to
control runoff. Buffers that are not maintained tend to develop a
‘hump’ at the front edge of the buffer as runoff that slows down
drops it sediment. After a few years, runoff begins to flow along the
leading edge of the buffer rather than across it. At a low point, the
runoff then moves across the buffer as concentrated flow, resulting in
little reduction of pesticide movement across the strip. Consequently,
OPP does not consider buffers for reduction of runoff unless buffers are
constructed and maintained according to the above referenced NRCS
document. 

5. Comment (Page 8): “The model overestimates the amount of pesticide
runoff by as much as 83 percent by not considering the effect of buffers
with vegetative cover that reduce pesticide loading even though buffers
with vegetative cover are common in apple production.”

OPP response: See the response to comment 4.

6. Comment (Page 9): “The model overestimates the amount of pesticide
that would run off into water sources because some orchards are planted
on soils that are not prone to high runoff potential.”

 

OPP response:  Tier 2 simulations use scenarios that are more vulnerable
than most sites on which a particular crop can be grown. This is
necessary in order to assure that there is a margin of protection at
most sites when the pesticide is used.

7. Comment (Page 9): “The model uses a rainfall value that is 6 times
greater than the actual rainfall in majority of Washington orchards. 
Using this value will lead the modeler to predict a greater amount of
pesticide entering waterways.”

OPP Response: OPP does not routinely base risk assessments on
“typical” use scenarios because a large fraction of the sites (50%,
if typical means median) will be more vulnerable and would not be
adequately protected. The Oregon apple scenario is more vulnerable than
most sites in the Pacific Northwest used to grow apples and is thus
expected to be protective of aquatic life in those areas. However, to
further characterize the potential risks, a scenario for Washington
apples was simulated using weather data from Yakima (D307577). The peak
1-in-10 year EEC for the Washington scenario was 4.8 ppb compared to 5.2
ppb for the Oregon scenario. In both cases, all LOCs for aquatic life
were exceeded.

8. Comment (Page 9): “Despite the conservative choices already made
for the model’s inputs, EPA makes the drift value even more
conservative by multiplying the generated value by three.  Tripling the
drift value is unreasonable and unnecessary.”

OPP Response:  The spray blast model in AgDrift is a regression model,
and as such, predicts the mean estimate across the trials that were used
in the regression calculation. Consequently, about half of the time, the
spray drift deposition will be higher; the other half of the time, it
will be lower. Multiplying the pond deposition accounts for this
background variability and assures that the screening estimate will be
conservative.

9. Comment (Page 9): “The agency’s preliminary decision document
indicates soil particle runoff is the primary pathway for azinphos
methyl entry into its model farm pond, and that this element is a
significant driver in the calculated aquatic risk.  Important factors in
determining the amount of runoff would be the amount of rainfall, the
distance the particles would have to travel to the pond and the ease at
which the particles travel the distance to the pond.  In this regard,
the agency’s approach seems particularly conservative and
unrealistic.”

OPP Response:  The Agency assessment indicates that azinphos dominantly
moves dissolved in runoff, not on the sediment suspended in the runoff.

10. Comment (Page 9): “In the Western scenario, the agency uses a site
with an average annual rainfall of 36 inches, when the primary apple
production region receives annual rainfall of 6 inches.”

OPP response:  See the response to comment #7.

11. Comment (page 9):  “Regarding the distance soil particles would
travel to the pond, the model assumes applications are made at the edge
of the pond and residues are on the edge of the pond poised for rain to
wash the particles into the hypothetical pond, even though the label
states applications must be made 25 feet from permanent bodies of
water.”

OPP Response: See the response to comment #4.

12. Comment (page 9): “Regarding the ease at which soil particles will
move, the agency’s model assumes the ground is bare and free of
vegetation when in reality; grass covers the orchard floor, except for
the herbicide strip directly under trees, in nearly all apple orchards. 
EPA maintains it uses “high runoff potential soils in its model so the
simulation will account for the most sensitive areas where there are
apple orchards in general.”  USApple contends that nearly all apple
orchards are covered in grass and other dense vegetation at the
perimeter of orchards, and this vegetation will prevent soil erosion and
movement of soil particles and pesticides into bodies of water.”

OPP Response: OPP considered the reduction of movement with eroded
sediment due to the vegetation maintained on the orchard floors in the
western orchards. OPP agrees that the erosion from the Pennsylvania
apple scenario is excessive as the C factors for the scenario were set
assuming no cover below the orchard which is very uncommon in eastern
apple orchards. This scenario is currently being revised, and a more
appropriate cover will be used in the new scenario. While new
simulations have not been run, a scenario with more appropriate erosion
would likely have 20 to 30% less azinphos methyl loaded to surface
water.  Such a decrease in EECs would still result in RQs that exceed
the Agency’s LOCs.  Also, see the response to comment #9.

13. Comment (page 9): “Extensive research has been done on the
movement of pesticides using various vegetative barriers.  Attached
below in Table 3 are thirty studies that conclude that vegetative
buffers do prevent movement of pesticides.”

OPP Response: See the response to comment #4.

14. Comment (page 10):  “Attached for your consideration is a study
titled “Pesticide Use and Toxicity Assay in Mission, Bender and Yaksum
Creeks” prepared in 2003 by the Chelan County Conservation
District.”  

OPP response: EPA has given special attention to the monitoring data
available from the Chelan County Conservation District.  Based on a
cursory analysis, it would be difficult to conclude that these samples
represent high exposure conditions for azinphos methyl.  Downstream
sites in this study are at least 2nd and 3rd order streams and do not
represent the most vulnerable locations in the watershed. It is not
clear from the study the proximity of the various orchards in the
watershed to the creeks sampled. Surprisingly, while there were
detections in Mission Creek, the application data show no applications
of azinphos methyl to Mission Creek during June. Only Yaksum Creek had
applications of azinphos methyl made during the sampling window. 

It is not surprising that there were no significant mortality effects on
Daphnia pulex.  Laboratory data indicate that the LC50 for daphnids is
approximately 1 µg/L.  The detected levels in this study ranged from
less than 0.02 µg/L to 0.19 µg/L.

U.S. Apple Association, September 22, 2006 

1. Comment (pages 3-4): “In the Western scenario, the agency uses a
site with an average annual rainfall of 36 inches, when the primary
apple production region receives annual rainfall of 6 inches.”

OPP response: OPP does not routinely base risk assessments on
“typical” use scenarios because a large fraction of the sites (50%,
if typical means median) will be more vulnerable and would not be
adequately protected. The Oregon apple scenario is more vulnerable than
most sites in the Pacific Northwest used to grow apples and is thus
expected to be protective of aquatic life in those areas. However, to
further characterize the potential risks, a scenario for Washington
apples was simulated using weather data from Yakima (D307577). The peak
1-in-10 year EEC for the Washington scenario was 4.8 ppb compared to 5.2
ppb for the Oregon scenario. In both cases, all LOCs for aquatic life
were exceeded.

2. Comment (page 4): “The modelers go beyond that to use an annual
value of 39.7 inches to capture the 90th percentile of the maximum
annual peak concentration in 30 years to calculate the risk.”

OPP response: OPP does use the 1-in-10 year return frequency EECs at a
site that is intended to be more vulnerable than 90% of sites used to
grow a particular crop on a national basis. U. S. Apple is also correct
that simulations generally use 30 years of simulation to estimate the
1-in-10 year peaks.

 

3. Comment (page 4): “In the Eastern Scenario, the agency’s model
estimates the 90th percentile of annual erosion to be 121 tons/ha. (49
tons/acre).  The U.S. Department of Agriculture’s (USDA) National
Resources Conservation Service (NRCS) estimates average annual erosion
from Pennsylvania orchards from 2.1 tons/acre in average soils to 5.9
tons/acre in highly erosive soil in Adams County, Pennsylvania using its
widely accepted universal soil erosion model.  Additionally, the 2003
NRCS National Resources Inventory Survey estimates New England/Mid
Atlantic average soil erosion to be 3.2 tons/acre.”

OPP response:  OPP agrees that the erosion from the Pennsylvania apple
scenario is excessive as the C factors for the scenario were set
assuming no cover below the orchard which is very uncommon in eastern
apple orchards. This scenario is currently being revised, and a more
appropriate cover will be used in the new scenario. While new
simulations have not been run, a scenario with more appropriate erosion
would likely have 20 to 30% less azinphos methyl loaded to surface
water.  Such a decrease in EECs would still result in RQs that exceed
the Agency’s LOCs.  

4. Comment (page 4): “The orchards visited by EPA staff during the
week of Sept. 11, 2006 clearly demonstrate that virtually continuous sod
cover crops in these orchards are playing a major role in reducing
run-off.  We would argue that a continuous cover crop in orchards and
between orchards and the water body offer significant protection when
compared to the agency’s model which greatly underestimates this
protection.”

OPP response:  OPP agrees that cover crops reduce onsite runoff. The
scenarios for the western orchards considered the sod cover in
estimating exposure.  However, azinphos methyl aquatic exposures in the
West are dominated by spray drift, and the benefit of cover crops under
the orchard in the west is reduced due the decreased rainfall that
occurs during the growing season.

5. Comment (page 4): “EPA’s STORET (STOrage and RETrieval Database)
and U.S. Geological Survey (USGS) monitoring data indicate azinphos
methyl concentrations are mostly 1 ppb or less, with two peak values of
less than 4 ppb between 1990 and 2005.  After 2001, the highest
concentrations from actual monitoring data were 0.75 ppb in Oregon and
less than 0.05 ppb in Pennsylvania.  However, EPA’s predictive model
estimates the concentrations to be 9.9 ppb in Oregon and 15.1 ppb in
Pennsylvania.”

OPP response:  As discussed in the response to Bayer CropScience and
Makhteshim Agan of North America comment #4, the available monitoring
data for azinphos methyl show exposures above the Agency’s threshold
of concern. Furthermore, surface water monitoring data tends to
underestimate exposures, particularly acute exposures, because
infrequent sampling makes it unlikely that the highest concentrations
that occur in any year will be captured in any given sample.  In
addition, the duration of most monitoring studies is typically only a
few years, and thus, information to characterize less frequent events is
limited.  The hydrologic setting of most sampling sites in a monitoring
program is downstream of relatively more vulnerable sites in the same
watershed. In most cases, monitoring data serves as a lower bound for
the potential exposures (and consequent risk) and is useful on that
basis.

6. Comment (page 4): “USApple believes the real concentrations are
likely to diminish over time because of the industry’s voluntary
mitigation, which proposes to reduce total annual applications from 8
lbs. to 6 lbs. of formulated product.”

OPP response:  As described in a previous evaluation of risk mitigation
proposals (D307577), the mitigation practices suggested by the U. S.
Apple Assoc. (2 applications, 60 foot buffer) along with consideration
of different spray drift scenario reduce the EECs from the 15.1 to 8.8
µg L-1 in Pennsylvania, and from 9.9 to 3.6 µg L-1 in Oregon for the
1-in 10 year peak concentration, reductions of 42 and 67% respectively.
However, in neither case does the reduction lower the risk below the
level of concern for acute risk of 0.3 µg L-1 for fish and 0.15 µg L-1
for freshwater aquatic invertebrates. In addition to the 60 ft buffers
suggested by U. S. Apple Association, a 100 ft buffer was also
simulated, with 1 in10 year peak EECs of 6.5 and 3.1 µg L-1 for
Pennsylvania and Oregon respectively. In fact, assuming no drift in the
lower exposure Oregon scenario still results in a 1 in 10 year peak EEC
of 2.1 µg L-1. So, while these suggested changes do substantially
reduce the estimated exposure for the use of azinphos methyl, they do
not mitigate the risk below the level of concern.

7. Comment (page 5):  “Attached for your consideration is a study
titled “Pesticide Use and Toxicity Assay in Mission, Bender and Yaksum
Creeks” prepared in 2003 by the Chelan County Conservation
District.”  

OPP response: EPA has evaluated the monitoring data available from the
Chelan County Conservation District.  It would be difficult to conclude
that these samples represent high exposure conditions for azinphos
methyl.  Downstream sites in this study are at least 2nd and 3rd order
streams and do not represent the most vulnerable locations in the
watershed. It is not clear from the study the proximity of the various
orchards in the watershed to the creeks sampled. Surprisingly, while
there were detections in Mission Creek, the application data show no
applications of azinphos methyl to Mission Creek during June. Only
Yaksum Creek had applications of azinphos methyl made during the
sampling window. 

It is not surprising that there were no significant mortality effects on
Daphnia pulex.  Laboratory data indicate that the LC50 for daphnids is
approximately 1 µg/L.  The detected levels in this study ranged from
less than 0.02 µg/L to 0.19 µg/L.

8. Comment (page 5):  “Monitoring of the Sulphur Creek Wasteway
indicate that maximum concentrations of azinphos methyl ranged from
0.023 ppb in 2003 to 0.14 ppb in 2005.”

OPP response: It is interesting that the median and maximum
concentrations increased over this period, though we believe that a
single site and three years of sampling with 18 to 31 samples per year
is insufficient to characterize the exposures associated azinphos methyl
in any general sense. Also, if that the monitoring point was at the
mouth of the wasteway then the watershed was over 100,000 acres, and the
measurements would hardly represent a vulnerable case.

9. Comment (page 5): “An analysis of the Lower Neel (sp) Creek in
Oregon during the period of summer azinphos methyl use indicates that
average concentrations of azinphos methyl were reduced from
approximately 0.33 ppb to 0.02 ppb in 2005.”

OPP response: It is difficult to determine the relevance of this
information for the azinphos methyl aquatic exposure characterization
since no ancillary data were provided.References

Flaherty, C.M. and R.D. Jones. 2005. Azinphos-methyl Insecticide:
Ecological Risk Assessment for the Use of Azinphos-methyl on Almonds,
Apples, Blueberries (Low- and Highbush), Brussels Sprouts, Cherries
(Sweet and Tart), Grapes, Nursery Stock, Parsley, Pears, Pistachios, and
Walnuts. September 29. DP Barcode 307568.

Flaherty, C.M. and R.D. Jones. 2006. Azinphos Methyl—EFED Response to
Stakeholder Comments on the Ecological Risk Assessment for “Group 3”
Uses. May 30. DP Barcode 307569.

Jones, R. David. 2006. Evaluation of Risk Mitigation Proposals from the
U. S. Apple Assoc. for Use of Azinphos methyl. Internal EPA memorandum
to Katie Hall dated July 12, 2006. D307577.

USDA NRCS. 2000. Conservation Buffers to Reduce Pesticide Losses.
Natural Resources Conservation Service. Fort Worth TX. 21 pp.

USGS NAWQA. 1999. PESTICIDES ANALYZED IN NAWQA SAMPLES: 

Use, Chemical Analyses, and Water-Quality Criteria.   HYPERLINK
"http://ca.water.usgs.gov/pnsp/anstrat/" 
http://ca.water.usgs.gov/pnsp/anstrat/ 

Appendix – Aquatic Exposure Model Input File Names tc \l1 "Appendix A
– Aquatic Exposure Model Input File Names 

Input files archived for azinphos methyl applied to pome fruits.



File Name	

Date	

Description



W23232.dvf	

	

weather for CA almonds scenarios (Sacramento, CA)



Caalmond0C.txt	

June 17, 2004	

PE4 scenario file for CA almonds, unirrigated



Pond298.exv	

August 29, 2002	

standard pond
捳湥牡潩映牯攠慸獭܇倍㑅猠浩汵瑡潩⁮湩異⁴楦敬⁳
⸨婐⁒硥整獮潩⥮܇」㠵〰‱䅃愠浬湯⁤㌰ഇ敓瑰浥敢
⁲㐱‬〲㘰ഇ䅃愠浬湯獤‬灳慲⁹汢獡ⱴ㈠‵瑦戠晵敦
Ⱳ㌠ㄯ‵灡ݰഇ㔰〸㄰䌠⁁污潭摮〠ܴ匍灥整扭牥ㄠⰴ
㈠〰ܶ䌍⁁污潭摮ⱳ猠牰祡戠慬瑳‬㔲映⁴畢晦牥‬⼵
㔱愠灰܇

^„Ð`„ࡰ摧絹Ú

hƒ

2

R

u

±

²

…

†

…

†

‰

“

”

•

hƒ

•

h6

hƒ

hˆ

hƒ

h*

hˆ

hƒ

h«

hƒ

h«

hƒ

hƒ

hÉ

h9

hƒ

  hƒ

hƒ

  hƒ

 hƒ

  hƒ

 hò

  hò

hƒ

hƒ

hƒ

hƒ

hƒ

hƒ

㄀Ĥ摧攄Àᘀ

 hò

  hò

hƒ

hƒ

hƒ

hƒ

hs

  h

^„Ð`„0ýgds

gds

 hò

  hò

hs

hs

hs

hs

hs

hs

hs

hs

hs

hs

hs

hs

hs

hs

hs

hs

hs

hs

ਁ䐃愀϶D瑹㘣.܀058001 CA almond 05	

September 14, 2006	

CA almonds, 25 ft buffer, 7/15 app



 http://www.epa.gov/oppfead1/endanger/consultation/ecorisk-overview.pdf 

 www.waterboards.ca.gov/centralvalley/tentative/0509/irr-lands/info-rpt-
program/info-program-staff-rpt.pdf

 Federal docket reference number unassigned at time of this publication

 PAGE   

 PAGE   21 

