UNITED STATES ENVIRONMENTAL PROTECTION AGENCY

WASHINGTON, D.C. 20460

OFFICE OF

PREVENTION, PESTICIDES, AND

TOXIC SUBSTANCES

			PC Code:	 098301

			DB barcode: 299865

MEMORANDUM						

March 2, 2007

Subject:	EFED Response to Public Comments for Aldicarb IRED: 

	Docket ID# Opp-2005-0163-0176 

To:	Bob McNally, Branch Chief

	Ann Overstreet, Team Leader

	Sherrie Kinard, Chemical Review Manager

Special Review Branch

Special Review and Reregistration Division

From:		Jonathan Angier

		Jeannette Martinez

		Donna Randall	

Environmental Risk Branch II 

Environmental Fate and Effects Division 

Through:	Dana Spatz, RAPL

Tom Bailey, Ph.D, Branch Chief, ERB II, 

Environmental Fate and Effects Division 

Attached are  EFED’s responses to Bayer Crop Science’s refined
terrestrial and aquatic risk assessment (submitted March, 2006) and
public comments (submitted January, 2007) on the EFED risk assessment
dated September 2006.



Bayer Crop Science presented refined terrestrial and aquatic risk
assessments which used a revised (“refined”) PRZM-EXAMS model
currently used by EFED.  The PRZM-EXAMS model as revised by BCS utilized
putative region-specific soil degradation values, different soil
incorporation efficiencies, and different rainfall pattern assumptions.
The refined risk assessment submitted by BCS also incorporated a new
terrestrial model, GranPARAM, which claims to account for expected bird
behavior (specifically, ‘preferences’ for consuming grit over
granules).  In our response, the Agency will focus on the major
assumptions and supporting data Bayer used in developing their
assessment and model, and critically evaluate whether or not those
assumptions are adequately supported for use in an ecological risk
assessment, and ultimately for a risk management decision.

(1) Agency’s comments regarding BCS’ “Refined Wildlife Risk
Assessment”

As modeled in GranPARAM, birds ingest granules and grit by choice only,
either randomly or non-randomly. The fact that birds incidentally ingest
granules by foraging on soil invertebrates or insects is not accounted
for in the model. According to the Wildlife Exposure Factor Handbook (US
EPA, 1993. Vol. I of II), incidental soil ingestion varies greatly among
types of birds, and for some species makes up a large percentage of
their daily food intake. Some examples of percent ingestion of soil in
diet (dry weight) for birds that visit aldicarb treated fields are: <2%
for mallard duck, 8.2% for Canada goose, and 10.4 % for American
woodcock. Incidental soil ingestion is too important a pathway to be
neglected when assessing risk to birds from aldicarb use. The Agency’s
deterministic model accounts for incidental soil ingestion, among other
exposure pathways. 

Furthermore, GranPARAM does not account for the fact that granules
dissolve in water, thus, becoming available to birds via drinking water
uptake (puddles formed on fields containing dissolved aldicarb), nor
does it account for food sources such as soil invertebrates and insects
carrying dissolved aldicarb residues on their membranes and exoskeleton.
Other important routes of exposure ignored by the model include
ingestion of contaminated plant material. The Agency’s deterministic
model indirectly considers all of these exposure pathways with the LD50
ft2 calculation. 

Although granular aldicarb also poses risk to mammals, GranPARAM does
not address risks to this group of organisms.  Acute and chronic
exposure via drinking water uptake (puddle water) and food sources, such
as mentioned for birds above, must also be addressed for mammals. In
addition to these routes of exposure, incidental soil ingestion is also
a factor that must be accounted for. Some examples of percent ingestion
of soil in mammalian diets are: 2.4% for meadow vole, 2.8% for red fox,
and 9.4 % for raccoon.

According to BCS’ sensitivity analysis, the most important input
parameter in GranPARAM is the attractiveness of aldicarb granules
compared to grit (1 to 1000). This preference factor was derived from a
laboratory study with one bird species, the house sparrow, where birds
were given a choice to select food from trays with differently colored
granules (one of them containing gypsum and, therefore, resembling
aldicarb formulations). The Agency questions whether the house sparrow
and its preference for grit over aldicarb granules are suitably
representative of all relevant avian species. Food –stuff preference
is an important issue for the Agency. Furthermore, birds are curious
creatures and when presented with the opportunity in the field, may
instinctively peck at food-like objects on the ground, which is
significant given that a single granule of aldicarb will kill a small
bird.

BCS proposed a risk categorization scheme which has not been vetted by
the Agency or Services as acceptable in the risk assessment process. 
Further, an adjective risk categorization scheme for risk is deemed
unnecessary according to EFED science policy (Ecological Risk
Description, April 24, 2003) even where the likelihood of integration of
exposure and effects can be summarized based on quantitative
distributions. Agency LOCs have been vetted before the SAP, and have
relevance to risk managers who relate them to risk management
determinations of unreasonable adverse effect.  Likewise the Services
have an understanding of how LOCs relate to risk determinations, whereas
the above arbitrary risk categorizations have not been determined to
relate to current regulatory interpretation.

   

(2) Agency’s comments regarding BCS’ “Refined Aquatic Risk
Assessment”

Our response to Bayer CropScience’s (2006) aquatic probabilistic risk
assessment (PRA) focuses on the key issue that most accounts for the
differences in conclusions drawn by the Agency in its ecological risk
assessment (May 2006) and those drawn by Bayer in its aquatic PRA. 
Although there are differences between the methods EFED would use in
performing an aquatic PRA and those used by Bayer, these differences
have much less influence on the results and conclusions than the choice
of aquatic and soil metabolism inputs for the PRZM/EXAMS modeling.  The
choice of inputs by Bayer CropScience results in considerably different
exposure distributions than those used in the deterministic assessment
for the same crops, scenarios, application rates, application dates, and
incorporation efficiencies.  For example, as shown in the table below
with crops and scenarios modeled by both Bayer and EFED, the 90th
percentile peak and 21-d values calculated by Bayer CropScience were 8
percent lower (Pecan, Georgia, 99% incorporation, peak EEC) to 94.7
percent lower (Cotton, Mississippi, 99% incorporation, 21-d EEC) than
values determined by EFED.  Aquatic and soil half-life values used by
EFED follow the standard acceptable practice for calculation of these
parameters (PRZMEXAMS (ver. 3.12 and 2.98.04)) based on the available
number, type, and range of data 
(http://www.epa.gov/oppefed1/models/water/input_guidance2_28_02.htm).
For more of a discussion about regional aerobic soil half-lives, see
also response to comments below.

Comparison of 90th percentile peak and 21-d EEC values 

Crop and Scenario	Application Rate

(lb a.i./A)	Incorporation

Eff. %	Bayer CropScience (2006) PRA	EPA Deterministic (May and November,
2006)



	Peak	21-d	Aerobic soil half-life (days)	Peak	21-d	Aerobic soil
half-life (days)

Citrus FL	4.95	99	0.19	0.07	18.25	0.64	0.41	55

Citrus FL	4.95	85



9.6	6.0	55

Pecan GA	10.05	99	0.77	0.31	39.54	2.6	1.7	55

Pecan GA	10.05	85	11.4	4.7	39.54	12.0	11.4	55

Cotton MS (1 app)	4.05	99	0.34	0.16	18.25	4.0	3.0	55

Cotton MS (1 app)	4.05	85	7.1	3.4	18.25	28.0	26.6	55



	Bayer CropScience (2006)	EPA Deterministic (May and November 2006)

Aerobic aquatic metabolism	5 days	12 days

Anaerobic aquatic metabolism	5 days	24 days

CAM (FL citrus)	1	8



Bayer claims to have used region-specific soil degradation rates in the
PRZM/EXAMS model.  However, the Agency believes this may only be
significant for crops grown specifically on those soils evaluated. 
There is actually greater variability in soils within regions than there
is between ‘representative’ soils from different regions. Therefore,
variability in degradation rates based upon individual soils from a
region will also yield a wide range of degradation rates.  Unless it can
be shown that faster degradation rates can be applied to most soils in
that region upon which the relevant crops are grown, it is not
sufficiently protective to apply differing soil degradation rates for
different regions.

(3) Agency’s Response to Bayer’s Comments on the September 2006
Aldicarb Ecological Risk Assessment

Bayer Crop Science comment: “Available monitoring data and incident
reports were also examined to see if these lines of evidence supported
the predictions of the refined exposure and risk assessments. Only two
incidents in aquatic ecosystems involving aldicarb have been documented
since it was first used in 1970. Several factors call into question the
reliability of these incident reports. Both incidents occurred at a time
of year when aldicarb would not have been used for one of the crops
associated with the incidents. Aldicarb was one of several compounds
detected during the incidents. Thus, it cannot be conclusively
identified as the compound that caused the adverse effects.”

Agency response: Reporting of wildlife kills is voluntary in the United
States.  Therefore, it cannot be concluded that low numbers of
ecological incident reports in the Agency’s database indicate low
ecological risk. The incidents cited above may reflect enhanced
discharge of tainted groundwater into surface water, which is typically
observed soon after a storm event (e.g., the tailing limb of a storm
hydrograph).  This may explain why these incidents occurred “at a time
of year when aldicarb would not have been used for one of the crops
associated with the incidents” and why other compounds were also
detected. Toxicity tests performed by the registrant indicate the
potential for chronic reproductive effects in fish and aquatic
invertebrates, and the chronic EEC values as estimated using PRZM-EXAMS
for different crops and application rates suggest potential chronic risk
to these organisms (EFED ERA, 2006). Chronic risk quotients exceed the
Agency LOC of 1.0 for all crops and application rates modeled for
freshwater fish (and freshwater invertebrates and estuarine/marine
invertebrates).  Documented cases of chronic reproductive effects in
fish (and invertebrates) are not available because there is currently no
system for monitoring or assessing such effects in the field.  If the
registrant is aware of such monitoring efforts that have shown no
effects, we urge the registrant to submit that information to the Agency
for consideration.

BCS comment: Surface water monitoring data are available from the
National Water 

Quality Assessment Program (NAWQA), the California Surface Water
Monitoring Database, and the Beaver Creek, Tennessee monitoring study.
Of 5,530 samples collected by NAWQA, 13 (0.2%) had detectable
concentrations of aldicarb residues (aldicarb, aldicarb sulfoxide,
aldicarb sulfone). Twelve of these samples are suspect because only one
aldicarb carbamate compound was detected, or little to no use of
aldicarb occurred in the area near where the samples were taken. 

Agency response: The NAWQA sampling regimen is not targeted for aldicarb
detections, and samples are obtained far too infrequently (approximately
once every two weeks) to expect many detections; random periodic
sampling is not likely to detect aldicarb in surface water except by
chance.  Samples would have to be obtained frequently (especially during
and soon after application), and specifically during and after the first
moderate-to-large storm event following application.  Indeed, given the
sporadic, non-targeted nature of the NAWQA samples, a detection rate of
0.2% would indicate that aldicarb residues are in fact a potential issue
– and almost certainly no ‘peak’ values will have been included in
the NAWQA database. The presence of only one aldicarb degradate in a
sample in no way indicates an absence of aldicarb.  Not all forms
(parent, sulfoxide, sulfone) need to be present in every sample to
verify detection.  

BCS comment: The California Surface Water Monitoring Database contains
three 

samples with detectable concentrations of aldicarb residues. Aldicarb
and aldicarb sulfoxide concentrations of 0.12 and 0.28 µg/L,
respectively, were measured at the San Joaquin River Hill Ferry sampling
station in Stanislaus County. The aldicarb concentration of 0.12 µg/L
is well below the freshwater invertebrate LC10 for the most sensitive 1%
of species. Follow-up sampling at this site found no detectable
concentrations of aldicarb or aldicarb sulfoxide, and one detectable
concentration of 

aldicarb sulfone (0.05 µg/L). One sample collected at the Turlock
Irrigation District Drain #5 site showed a detectable concentration of
aldicarb sulfone (0.26 µg/L), but not aldicarb or aldicarb sulfoxide.

Agency response: Again, the presence of any of the 3 toxic compounds
indicates a positive detection.

BCS comment: BCS has several concerns about using data from the Beaver
Creek monitoring study, and it was demonstrated in the refined
assessment that Beaver Creek represents a worst case scenario for an
atypical environment (see Section 5.2.2 in Moore et al., 2006a). Samples
were predominantly taken from drainage ditches at the edge of
agricultural fields. These ditches go dry after most storm events (see
photographs in Figures 2 and 3). Samples were taken at the field edge
would typically show the highest concentration, samples further
downstream would produce lower concentrations because of dilution from
parts of the watershed not associated with treated agricultural fields.
Samples taken further downstream, in an environment capable of
supporting aquatic biota, would be more representative of a typical
environment. Such data are unavailable for the Beaver Creek watershed.

Agency response: To reiterate from previous Responses to Comments
(January 2005, May 2005, August 2005, November 2005, August 2006), the
fact that samples can be obtained from the Beaver Creek site only when
there is runoff from the treated field actually makes this a more
accurate benchmark for off-site concentrations than random stream
samples.  Although there would likely be some downstream dilution within
the receiving water body, a concentrated plume originating from a
treated site would retain some of its form (and chemical load) for some
distance downstream, with attenuation depending upon stream flow rates
and channel turbulence.  In addition, there may be successive aldicarb
inputs along that same stream channel from other sources within the same
watershed.

BCS comment: RQs for saltwater invertebrates exceeded 0.5 for cotton,
pecans and soybeans. Considering that these acute RQs were calculated
for worst-case exposure scenarios (maximum application rate, banded
application leaving 15% of product on the surface, high rainfall and
storm intensity) and used conservative toxicity benchmarks, it is not
surprising that refinements to the exposure and toxicity
characterizations would result in much lower estimates of acute risk. 

Agency response: The assertion that RQs were calculated for worst-case
exposure scenarios is misleading.  PRZM-EXAMS uses historical (30-year)
weather data, not “high rainfall and storm intensity” data, so
weather inputs utilized by PRZM-EXAMS reflect the variability in weather
(particularly rainfall) that actually occurred at or near the site over
a 30-year period.  This helps ‘average-out’ the effect of weather;
high rainfall and storm intensity in one year is likely countered by low
rainfall and intensity in another year over that same (30-year) time
period. Maximum application rates are used specifically in an attempt to
model upper bound exposure, since these application amounts are allowed
according to the label.  Banded applications were also used for the same
reasons, as they are permitted on the label for these crops.  However,
scenarios were also run that accounted for ‘average’ use rates, and
with higher incorporation rates (99%, vs. 85% for banded applications);
nevertheless, EECs obtained by EFED were still higher than those cited
by BCS (see Table, above).

BCS comments: The results of these [refined] risk assessments indicate
that aldicarb poses de minimis or low risk to aquatic life and wildlife
for the vast majority of areas where the pesticide is currently or could
potentially be used. The few scenarios that indicated elevated risk
(e.g., banded application to cotton, peanuts and pecans in southeastern
US region, banded application to potatoes in Pacific region) represent
minor uses for aldicarb. The record of wildlife and fish kill incidents
and field study results are consistent with the low level of risk
predicted by the refined assessments. 

Agency response: When all these uses are summed, they appear to account
for at least 2-3% of total uses.  While this may seem to be a small
percentage, these uses still need to be considered – especially as
they represent some of the most sensitive scenarios.

BCS comments: The Agency’s stated rationale for requiring these
[plant] studies is “because of potential phytotoxic effects (delayed
emergence and plant stand) identified by the registrant on the label.”
It is further explained that terrestrial and aquatic plant testing are
typically not required for non-herbicides unless there is evidence of
potential phytotoxicity at use rates based on label warning statements
or from other lines of evidence.

Agency response: There is evidence in the literature (Kennedy, 2002)
that application of aldicarb in excess of label rate (0.0138 kg
a.i./container) causes delayed emergence response in pearl millet. Such
excess concentrations may stem from concentrated runoff from fields
where applications occurred at or below the maximum label application
rate. Concentrated runoff, originating from a treated site (as
demonstrated at Beaver Creek site), may move off-field and receive
successive aldicarb inputs along the way before reaching non-target
sites where listed species occur. 

BCS comments: While there is potential for seedling emergence effects if
the highest rates are applied directly into the seed furrow of some crop
plants, this scenario is not applicable to exposures of non-target
plants. Nontarget non-crop plants will not have Temik® granules applied
into the seed furrow and the concentration in the soil surrounding seeds
and seedlings of non-target plants will be much lower.

	

Agency response: Aldicarb may dissolve and become available to plants
via puddles that form off-field, or groundwater that reemerges in
lowland, non-target plant habitats. Because aldicarb is systemic and may
cause hazard to endangered and threatened plants, the Agency needs the
Tier I plant studies to fully characterize risk to these taxa.

 PAGE   

 PAGE   3 

