Guidance for
Assessing Pesticide
Risks to Bees
Office of Pesticide Programs
United States Environmental Protection Agency
Washington, D.C. 20460
Health Canada Pest Management Regulatory Agency
Ottawa, ON, Canada
California Department of Pesticide Regulation*
Sacramento, CA
*Currently, due to resource limitations, the California Department of Pesticide Regulation does not conduct full
ecological risk assessments, but reserves the right to do so in the future.
June 19, 2014
Executive Summary
This document provides guidance to risk assessors for evaluating the potential risk of pesticides to bees,
particularly honey bees (
Apis mellifera). This guidance is not limited to identifying the risk assessment
process but includes consideration of the underlying data on which the process is based. For purposes of
brevity, this guidance refers to the
White Paper in Support of the Proposed Risk Assessment Process for
Bees1 submitted to the Federal Insecticide, Fungicide and Rodenticide Act (FIFRA) Scientific Advisory
Panel (SAP) for review and comment in September 2012. The White Paper describes the basic framework
of the risk assessment process and the data used to inform the various tiers of refinement that may be
required to support risk management decisions. This guidance also considers recommendations2 provided
by the FIFRA SAP in response to the White Paper, where such recommendations can be immediately
implemented. Additional recommendations from the FIFRA SAP that cannot be implemented at this time,
because the science supporting such efforts has not been sufficiently vetted, will be considered as the
science evolves.
The risk assessment process described in the White Paper identifies a tiered approach where data are
collected on individual bees that are representative of different life stages (larval/pupal versus adults) and
castes (
e.g., worker bees). While additional data may be available on other bee species and these data can
be included in the tiered risk assessment process as an additional line of evidence, the primary process relies
on honey bee data as a surrogate for both
Apis and non-
Apis bees. In this process, laboratory-based studies
of larval/pupal and adult honey bees provide data on individual bees that can be used as a surrogate for
other species of bees, including solitary species. At the semi-field and full-field levels, studies of the colony
can be used to represent effects to honey bees themselves and as a surrogate for other social bees. An
advantage of using honey bees is that the husbandry and life cycle of the species and its significance in
pollination services is well known and test protocols are available. As the science evolves, methods and
studies using non-
Apis bees may be considered and incorporated into the risk assessment.
The risk assessment process for bees is consistent with that used for other taxa, as described in the Overview
Document3, in that it consists of three phases (
i.e., problem formulation, analysis, and risk characterization)
and it is tiered. The first tier consists of a screening-level risk assessment that is intended to be sufficiently
conservative such that chemicals that pass the screen are considered to represent a relatively low risk of
adverse effects to bees. For those chemicals which do not pass the initial screen, refinements in exposure
estimates and/or mitigation measures may sufficiently reduce risk quotients (RQs) below levels of concern
(LOCs) such that further refinements are not needed. For chemicals where RQ values still exceed LOCs
and, depending on risk management needs, additional refinements in exposure and/or effects estimates can
be made based on studies with increasing levels of environmental realism. Although approaches for
estimating exposure and effects differ across aquatic and terrestrial systems as well as between plants and
1 USEPA. 2012. White Paper in Support of the Proposed Risk Assessment Process for Bees. Submitted to the FIFRA Scientific Advisory Panel
for Review and Comment September 11 – 14, 2012. Office of Chemical Safety and Pollution Prevention Office of Pesticide Programs
Environmental Fate and Effects Division, Environmental Protection Agency, Washington DC; Environmental Assessment Directorate, Pest
Management Regulatory Agency, Health Canada, Ottawa, CN; California Department of Pesticide Regulation
http://www.regulations.gov/#!documentDetail;D=EPA-HQ-OPP-2012-0543-0004
2 USEPA 2012. Final FIFRA SAP Pollinator Meeting Report. SAP Minutes No. 2012-06. A Set of Scientific Issues Being Considered by the
Environmental Protection Agency Regarding Pollinator Risk Assessment Framework. September 11 – 14, 2012, FIFRA Scientific Advisory Panel
Meeting. http://www.regulations.gov/#!documentDetail;D=EPA-HQ-OPP-2012-0543-0047
3 USEPA. 2004. Overview of the Ecological Risk Assessment Process in the Office of Pesticide Programs, U. S. Environmental Protection Agency.
Endangered and Threatened Species Effects Determinations. Office of Prevention, Pesticides and Toxic Substances Office of Pesticide Programs,
Washington DC. January 23, 2004. http://www.epa.gov/espp/consultation/ecorisk-overview.pdf
2
animals, the basic process of moving from a screening-level assessment with conservative assumptions to
more refined measures of exposure and effects is consistent across taxa.
The guidance follows the generic structure of the risk assessment process, as described in the White Paper
and Overview Document. While this guidance is not intended to be exhaustive, it provides staff with
sufficient information with which to ensure consistency in ecological risk assessments written in support
of new and existing pesticide registration decisions. The different levels of refinement described in the
guidance are not intended to be prescriptive; the specific set of data used in assessing potential risks of a
pesticide to bees ultimately depends on multiple lines of evidence and risk management objectives.
3
Acknowledgement
The following individuals and organizations contributed to this guidance document:
U.S. Environmental Protection Agency Office of Pesticide Programs (OPP)
Rueben Baris
Joseph DeCant
Frank Farruggia
Kristina Garber
Anita Pease
Keith Sappington
Mah Shamim
Thomas Steeger
Allen Vaughan
Christina Wendel
Health Canada Pest Management Regulatory Agency (PMRA)
Connie Hart
Wayne Hou
California Department of Pesticide Regulation (CALDPR)
Richard Bireley
4
Contents
1
Overview of Pollinator Risk Assessment Process ................................................................................ 6
1.1
Foliar Spray Applications ............................................................................................................. 6
1.2
Soil Application and Seed Treatment ......................................................................................... 10
2
Problem Formulation .......................................................................................................................... 13
3
Analysis Phase .................................................................................................................................... 15
3.1
Exposure Characterization .......................................................................................................... 15
3.1.1
Tier I Exposure Estimates ................................................................................................... 15
3.1.2
Refinements of Tier I Exposure .......................................................................................... 16
3.2
Effects Characterization .............................................................................................................. 19
3.2.1
USEPA Toxicity Testing Requirements for Bees ............................................................... 19
3.2.2
Additional Guidance for USEPA Pollinator Testing .......................................................... 19
3.2.3
PMRA Toxicity Testing Requirements for Bees ................................................................ 21
3.2.4
Tier I Effects Characterization ............................................................................................ 22
3.2.5
Tier II Effects Characterization ........................................................................................... 24
3.2.6
Tier III Effects Characterization ......................................................................................... 27
4
Risk Characterization .......................................................................................................................... 31
4.1
Risk Estimation ........................................................................................................................... 31
4.1.1
Calculation of Risk Quotients for Tier I Risk Assessment ................................................. 31
4.1.2
Levels of Concern for Tier I Risk Assessment ................................................................... 32
4.2
Risk Description .......................................................................................................................... 33
4.2.1
Use of Other Lines of Evidence .......................................................................................... 33
4.2.2
Synthesis of Risks among Tiers .......................................................................................... 34
4.2.3
Risk description on Sublethal Effects ................................................................................. 35
4.2.4
Use of Simulation Models ................................................................................................... 36
4.2.5
Uncertainties ....................................................................................................................... 36
4.2.6
Data Gaps ............................................................................................................................ 40
Appendix 1. Conceptual Models ................................................................................................................ 42
A1.1 Non-systemic, Foliar Spray Applications .................................................................................... 42
A1.2. Systemic, Foliar Spray Applications ........................................................................................... 43
A1.3 Systemic, Seed Treatment............................................................................................................ 44
A1.4 Systemic, Soil Application ......................................................................................................... 45
Appendix 2. Considerations related to quantifying residues of pesticides in pollen and nectar using
pesticide-specific studies .................................................................................................................... 46
Appendix 3. Bee REX ................................................................................................................................ 48
Appendix 4. Tier 3 Field Study Design Considerations ............................................................................ 57
5
1 Overview of Pollinator Risk Assessment Process
This section summarizes the overall risk assessment process for characterizing the risks of pesticides to
honey bees (
Apis mellifera), which are used as a surrogate species for other
Apis and non-
Apis bees and
other insect pollinators. It provides a brief overview of the key steps and decision points involved in the
risk assessment process. As such, it should be not used in isolation; rather it should be considered in
conjunction with the detailed risk assessment guidance described in the ensuing sections on Problem
Formulation (
Section 2), Analysis (
Section 3), and Risk Characterization (
Section 4). If acceptable data
are available for non-
Apis bees, that information should be considered as a line of evidence in determining
potential risks to bees.
An illustration of the decision-making process for assessing risks to honey bees from foliar spray
applications of pesticides is shown in
Figure 1 while that for pesticides applied via soil or seed treatments
is shown in
Figure 2. The overall approach is a tiered process whereby risks are first assessed using simple
and conservative exposure screening models to generate estimated environmental concentrations (EECs)
coupled with toxicity estimates derived from laboratory studies (Tier I) to calculate risk quotients (RQs)
for individual bees. Pending results of the Tier I risk estimation, consideration is given to collecting and
evaluating information at higher tiers (Tiers II and III), where results are based on effects to the colony, and
are considered more environmentally realistic but also requiring greater resources to conduct and
interpret. A step-by-step summary of the processes is provided below.
The process described below is not intended to be prescriptive and risk assessors should evaluate multiple
lines of evidence in determining which data to recommend and be cognizant of the timeline for risk
management decisions. Across all of the scenarios described below, an initial determination must be made
as to whether a reasonable potential exists for exposure of bees to the pesticide of concern. If there is no
exposure, then the likelihood of adverse effects (
i.e., risk) is presumed to be low and an additional
assessment is not warranted. In such cases, the risk assessor should provide an explanation for why the use
of the pesticide will not be likely to result in exposure of concern to bees. Also, and consistent with the
process used to assess risk to other taxa, the risk assessment process is intended to be iterative, and the risk
assessor should consider the effect that mitigation options may have on reducing exposure and thereby
mitigating the need for additional refinements.
1.1
Foliar Spray Applications
Step 1. Determine if Bees May Be Exposed. As part of Problem Formulation (
Section 2
), information on
the pesticide use characteristics (
Box 1, Figure 1), chemical properties and potential exposure routes are
evaluated to determine the need for conducting a bee risk assessment. Information provided with the
generic conceptual models for bee risk assessment should be consulted (
Appendix 1. Conceptual Models).
In general, outdoor spray applications are assumed to have a reasonable potential to result in exposure of
adult bees and their brood (eggs, larvae, pupae) to pesticides if they are applied to pollinator attractive crops
or drift to pollinator attractive plants during periods when bees are likely to be foraging. Exposure to bee
brood and other castes of bees in hives is expected when exposure to foraging bees is identified, as foragers
will bring residues back to the hive. Pre-bloom foliar application of pesticides to pollinator attractive crops
may also result in exposure to bees if the pesticide is persistent and may translocate to pollen and nectar
after spray application. In contrast, indoor uses are generally assumed not to have a reasonable potential
for exposure of bees to pesticides. Exceptions do occur, such as in greenhouses where bees are used for
pollination (
e.g., bumble bee pollination of tomatoes).
6
Step 2. Calculate Tier I Screening-Level Risks. If a reasonable potential for exposure to the pesticide is
identified, a screening-level risk assessment is conducted. This step involves a comparison of Tier I
estimated exposure concentrations (EECs) for contact and oral routes of exposure to adults and larvae
(
Boxes 3a,b,c; Section 3.1) to Tier I acute and chronic levels of effects to individual bees using laboratory-
based studies (
e.g., acute median lethal dose (LD50), chronic no observable adverse effect concentration
(NOAEC);
Section 1.1). The Tier I EECs can be estimated using the Bee-REX model (see
Appendix 3. Bee REX for additional details). The conservatism of the Tier I screening-level risk quotient (RQ) value
results primarily from the model-generated exposure estimates that, while intended to represent
environmentally relevant exposure levels, are nonetheless considered high-end estimates. The resulting
acute and chronic RQ values (
Boxes 4a,b,c) are then compared to the corresponding level of concern (LOC)
values for acute and chronic risk (
i.e., 0.4 and 1.0, respectively). Generally, if RQ values are below their
respective LOCs, a presumption of minimal risk is made, since the Tier I risk estimation methods are
designed to be conservative. Risk assessors should also consider other lines of evidence in making this
determination, as explained in
Section 3. It is also important that any uncertainties related to screening-
level estimates of exposure and/or toxicity are characterized in the risk discussion section of the risk
characterization.
Step 3. Refine Tier I Screening-Level Risk Estimates. If risk concerns are identified, the Tier I
assessment may be refined using additional data (
Box 6; Section 3.1.2). The initial Tier I risk estimation
is designed to produce conservative estimates of risk in order to minimize the occurrence of false negative
findings. Refinements to Tier I risk estimates may include consideration of pesticide-specific residue data
available from crop magnitude of residue studies that are relevant to bees or available studies that quantify
pesticide residues in pollen and nectar (
Box 9a). Refined RQ values, based on refined estimates of exposure
coupled with the Tier 1 effects endpoints from individual bees, are then compared to the aforementioned
LOCs to determine the potential for risk (
Box 7).
Step 4. Consider Uncertainties, Risk Mitigation Options and Need for Tier II Risk Estimation. If risks
are identified from Tier I, the risk assessor should consider the uncertainties associated with risk estimation,
information from other lines of evidence, and the impact of any risk mitigation options identified for the
pesticide of concern (
Box 8). These risk mitigation options may include reductions in application rates and
restriction of application methods, and recalculation of Tier I risk estimates as a result of reduced
environmental loading. Restrictions on the timing of pesticide applications and crops species may also be
considered to minimize exposure to bees. The risk assessor should also consider whether information on
pesticide exposure and effects collected using Tier II studies are needed (
e.g., residues in pollen and nectar
studies described in
Appendix 2. Considerations related to quantifying residues of pesticides in pollen and
nectar using pesticide-specific studies; semi-field tunnel studies as described in
Section 3.2.5). Tier II
studies may be used to identify more targeted risk mitigation options than those that could be identified
based on Tier I data. Measured residues in pollen and nectar (
Box 9a) from these studies may also be used
to refine risk estimates for Tier I as described previously. Tier II effect studies characterize pesticide effects
at the whole-colony level and therefore, reduce uncertainty associated with extrapolating effects on
individual bees under laboratory conditions (Tier I toxicity studies) to effects on the colony. It is important
to recognize that Tier II effect studies are conducted under semi-field conditions where the high-end
exposure at the colony level is expected. In Tier II studies other stressors may be present and potential
compensatory mechanisms of the colony may occur. Tier II studies should be designed to address potential
uncertainties identified in the Tier I assessment or elsewhere (
e.g., incident reports). Unlike Tier I,
characterization of risk in Tier II does not involve the calculation of RQ values
per se (
Box 10). Rather,
risks at the colony level are usually characterized in relation to pesticide application rate and/or measured
residue levels (
Section 3.2.6). Interpretation of such whole-colony effects studies is often much more
7
complex than Tier I studies, and relies on comprehensive considerations of whether adverse effects are
likely to occur at the colony level.
Step 5. Consider Uncertainties, Risk Mitigation Options and Need for Tier III Studies. Based on the
risks identified at lower-tier assessments, their associated uncertainties, and other lines of evidence, the risk
assessor should consider the impact of any risk mitigation options identified for the pesticide of concern
(
Box 11). The need for more refined information conducted at Tier III should be determined depending on
the nature of the estimated risks, the associated uncertainties, and available risk mitigation options. Risk
mitigation options may include reduced application rates, reduced application intervals, restrictions on
applications at or near bloom or off-labeling use on a particular crop. As an example, effects on the ability
of colonies to successfully emerge in the spring (
e.
g., produce sufficient brood and adult bees after over-
wintering) may be a concern for some pesticides/uses which are not typically addressed in lower tiers. Tier
III studies are full-field studies that are designed to mimic actual pesticide applications and exposure of
bees encountered in the environment (
Section 3.2.6; Box 12). Tier III field studies are usually highly
complex and require a high level of effort to design and conduct so as to address specific sources of
uncertainties and potential risks identified in lower tiers. Because of the length and complexity of these
studies, other factors affecting colony survival (
e.g., disease, pests, nutrition) may impact the successful
completion and interpretation of these studies. As with any field study, the design and conduct of such
studies is crucial to their interpretation and utility in risk assessment. Similar to risk characterization at Tier
II, risk characterization at Tier III considers multiple lines of evidence available from lower Tiers and other
information sources (
e.g., open literature) that meet the respective Agency’s standard for inclusion in risk.
Risk assessment conclusions are made based on the weight of evidence, available risk mitigation options,
and uncertainties in the available data and methods (
Section 4).
At any stage of the risk assessment process, EPA/PMRA/CDPR may determine that risk mitigation is
appropriate. The decision to implement risk mitigation is based on the existing analysis and does not
necessarily depend on completing all three tiers of the full risk assessment process.
8
Figure 1. Tiered Approach for Assessing Risk to Honey Bees from Foliar Spray Applications.
9
1.2
Soil Application and Seed Treatment
The risk assessment process for evaluating soil applications (
e.g., soil drench) and seed treatments is similar
to that described previously with foliar applications, except that risk from contact exposure is not evaluated.
For soil application, it is generally assumed that exposure of honey bees from direct contact with the
pesticide is minimal, given the nature of the application to bare soil, although exceptions may occur if
applications are made with bee-attractive weeds present. Contact exposure of non-
Apis bees (
e.g., solitary
and ground-nesting bees) may be significant with soil applications; however, the extent of this potential
exposure is uncertain. It is also noted that for seed treatments, exposure of bees to pesticides has been
documented via drift of abraded seed coat dust when planting under certain conditions; however, there are
multiple factors determining the extent to which dust-off occurs. Modeling tools have not been developed
to estimate exposure under these conditions. EPA and PMRA may determine that implementing best
management practices designed to mitigate this route of exposure may be appropriate. As discussed in the
preceding section, the decision to implement risk mitigation would be based on the existing analysis and
does not necessarily depend on completing all three tiers of the full risk assessment process.
Step 1. Determine if Bees May Be Exposed. As part of Problem Formulation (
Section 2), information on
the pesticide use characteristics (
Box 1, Figure 2), chemical properties and potential exposure routes is
evaluated to determine the need for conducting a bee risk assessment. Information provided with the
generic conceptual models for bee risk assessment should be consulted (
10
Appendix 1. Conceptual Models). In absence of information to indicate otherwise, it is assumed that soil-
applied and seed-treated pesticides are systemic and able to be transported to pollen and nectar. Therefore,
seed treatments and outdoor application to soils are generally assumed to have a reasonable potential to
result in exposure of bees, including both adult and immature stages of bees, to pesticides via consumption
of contaminated pollen and/or nectar.
Step 2. Calculate Tier I Screening-Level Risks. If a reasonable potential for exposure to the pesticide is
identified, a screening-level risk assessment is conducted. This step is identical to that described in Step 2
for assessment of foliar spray applications except that the exposure modeling tools differ. Different
methods are used to estimate pesticide residues in pollen and nectar from soil application and seed treatment
(see
Section 3.1.1). Otherwise, risk estimation, LOCs and consideration of multiple lines of evidence are
identical to those described previously in Step 2 for foliar pesticide applications.
All subsequent steps (
i.e.,
Step 3. Refine Tier I Screening-Level Risk Estimates; Step 4. Consider
Uncertainties, Risk Mitigation Options and Need for Tier II Refinements and Step 5. Consider
Uncertainties, Risk Mitigation Options and Need for Tier III Studies) are identical to those described
previously for foliar spray application of pesticides.
11
Figure 2. Tiered Approach for Assessing Risk to Honey Bees from Soil/Seed Treatments.
12
2 Problem Formulation
Problem formulation is a critical step in ecological risk assessment and is intended to articulate, among
other things, the protection goals around which the assessment is conducted. Relative to bees, the protection
goals include the maintenance of pollination services, hive product production and biodiversity (
Table 1).
These goals do not apply uniformly across
Apis and non-
Apis bees; however, they are considered protective
for social and solitary bees, and honey bees are generally used a surrogate for non-
Apis bees. These
protection/management goals in turn dictate assessment endpoints for which specific measurement
endpoints are identified. Ideally, problem formulations should articulate the protection/management goals
as well as the risk hypothesis including how these goals may be compromised due to the proposed or
existing use(s) of a pesticide. The risk hypothesis and conceptual model are used to depict the hypothesis
in terms of the source of the stress, route of exposure, receptor, and changes in the receptor attribute(s) of
concern. A number of generic conceptual models have been developed and these should be adapted to
reflect, where appropriate, the potential risks to bees that will be evaluated. The White Paper discusses and
Appendix 1. Conceptual Models of this guidance provides conceptual models (
e.g., foliar application of
non-systemic and systemic pesticides, soil-applied and seed-coated systemic4 pesticides) for honey bees
that can be readily adapted and integrated into conceptual models that include other taxa or included as a
stand-alone conceptual model.
Table 1. Protection goals and examples of associated assessment and measurement (population and individual)
endpoints for bees.
Example Measurement Endpoints
Protection Goal
Assessment Endpoints
Population level and
Individual Level
higher
Individual bee survival
Individual worker and
(solitary bees) and colony larval survival assays;
Contribution to Bee
Species richness1 and
strength and survival
larval emergence;
Biodiversity
abundance
(social bees)
queen
Species richness and
fecundity/reproduction
abundance1
Individual worker and
Population size2 and
Colony strength and
larval survival assays;
Provision of Pollination
stability of native bees
survival; colony
queen fecundity;
Services
and commercially
development
brood success;
managed
bees
worker bee longevity
Individual worker and
Quantity and quality of
larval survival assays;
Production of Hive
Quantity and quality of
hive products; including
queen
Products
hive products
pesticide residue levels on fecundity/reproduction;
honey/wax
larval emergence
1Use of honey bees as a surrogate for other insect pollinators has limitations; however, it is assumed that as with all surrogates, data on individual
organisms as well as colony-level data would provide some relevant information on the potential effects of a pesticide on both solitary bees as well
as “eusocial” taxa. In addition, protection of honey bees would contribute to pollinator diversity indirectly by preserving the pollination and
propagation of the many plants species pollinated by honey bees, which also serve as food sources for other pollinating insects.
2 For managed honey bees, population size can include numbers of colonies.
For most pesticides used in an agricultural setting, the predominant exposure routes are through diet (
i.e.,
consumption of nectar and pollen) and contact (
i.e., direct spray). Exposure due to the vapor phase of a
4 Depending on physical-chemical properties, systemic pesticides can move within the vascular (xylem and/or phloem) tissues to untreated tissues
of the plant or remain locally distributed through extracellular movement. Therefore, a pesticide may be xylem mobile, phloem mobile, both xylem
and phloem mobile, or locally systemic.
13
pesticide is relatively small compared to diet and contact, with the exception of fumigants. Additionally,
the importance of exposure through consumption of drinking water, relative to the dietary and contact routes
of exposure is under investigation5. For pesticides that are applied to seeds, exposure to dust from treated
seed during planting may also be of concern. The extent to which honey bees are exposed via contact with
abraded seed coat dust is determined by many factors including the physical‐chemical properties of the seed
coating, seed planting equipment, use of seed delivery agents (
e.g., talc or graphite), environmental
conditions (wind speed, humidity), existence of flowers nearby the sowing area, and hive location in
relation to sowing. As recognized by the FIFRA SAP, the production of dust during planting should be
minimized to the extent possible in order to minimize the exposure to bees; additional considerations and
assessment of risk should be completed on a chemical-specific basis.
The primary question risk assessors must address initially is whether, given the existing or proposed use(s)
and physical-chemical properties of a pesticide, exposure is likely to occur for bees from the treated crop
or resulting spray/dust drift to a blooming weed on the field or blooming plants near-field. If exposure is
not likely for bees, then the problem formulation should state and the conceptual model should depict why
the exposure is considered unlikely. Where exposure cannot be precluded, the problem formulation should
identify the available body of information on exposure and effects that will be considered to support the
screening-level risk assessment. If, during the risk assessment, further refinements are deemed necessary,
the risk hypothesis and conceptual model can be modified accordingly and the analysis plan revised to
identify additional exposure and effects data that will be considered thereby reflecting the iterative nature
of the risk assessment process.
Use of honey bees as a surrogate for other insect pollinators has limitations; however, it is assumed that, as
with all surrogates, data on individual organisms as well as colony-level data would provide some relevant
information on the potential effects of a pesticide on both solitary bees as well as social bees. In addition,
protection of honey bees would contribute to pollinator diversity indirectly by preserving the pollination
and propagation of the many plants species pollinated by honey bees, which also serve as food sources for
other pollinating insects. In evaluating potential risks specific to honey bees, the most important commercial
pollinators, the protection goals of preserving pollination services and production of hive products (
e.g.,
honey, wax) are readily assessed through the assessment of population size and the stability (
e.g., presence
of a queen, uniform brood pattern) of the colony and through direct and indirect measures of the quantity
and quality of hive products (
Table 1). As such, the sensitivity of individual larval or adult honey bees
based on laboratory-based acute and chronic toxicity studies serve as reasonable measurement endpoints
for screening-level assessments of potential adverse effects on colony strength, survival and capacity of the
colony to produce any products. While these measurement and assessment endpoints are tested using
managed honey bee colonies, they apply to feral honey bee colonies and, in the absence of data specific to
other bees, these measurement endpoints provide useful information for assessing the survival and
development of solitary bees and potential effects on bee species richness and biodiversity. To the extent
that data are available for other species such as the bumble bee (
e.g.,
Bombus terrestris), blue orchard bee
(
Osmia lignaria), alfalfa leafcutting bee
(
Megachile rotundata), these species may also be considered in
the risk assessment. As discussed in the White Paper, available information for the bumble bee, blue
orchard bee and the alfalfa leafcutting bee indicate that the screening-level risk assessment based on effects
to individual honey bees is likely to be protective for these three species; however, this is likely chemical
specific and available data must be considered in determining whether this assumption is supported for
chemical under consideration.
5 This investigation includes consideration of recommendations provided by the FIFRA SAP with regard to the model used for estimating the
relative importance of drinking water as a significant exposure route. An in-depth analysis of the model used to estimated pesticide
concentrations in puddles is currently being completed.
14
3 Analysis Phase
The analysis phase consists of the exposure characterization and the effects characterization relative to bees.
A tiered approach, from the most conservative at lower tiers (Tier I) to more realistic at higher tiers (Tiers
II and III) should be considered during the data requirement determinations and risk assessment. Steps that
should be considered within each characterization are depicted in the relevant decision tree in
Figure 1 and
Figure 2.
3.1
Exposure Characterization
3.1.1 Tier I Exposure Estimates
Contact and dietary exposure are estimated separately using different approaches specific for different
application methods.
Table 2 summarizes the methods used for deriving Tier I estimated environmental
concentrations (EECs) for contact and dietary routes of exposure for foliar, soil, seed treatments and tree-
trunk injections. These EECs are calculated using the Bee-REX model.
In the Tier I, pesticide exposures are estimated based on honey bee castes with known high-end
consumption rates. For larvae, food consumption rates are based on 5-day old larvae, which consume the
most food compared to other days of this life stage. For adults, the screening method relies upon nectar
foraging bees, which consume the greatest amount of food (pollen and nectar) compared to other adult
worker bees. It is assumed that this value will be comparable to the consumption rates of adult drones and
will be protective for adult queens as well. Although the queen consumes more food than adult workers
and drones, the queen consumes food that is assumed, based on currently available data, to contain orders
of magnitude less pesticide than that consumed by workers since the queen is only fed a processed food,
i.e., royal jelly. As described in the White Paper, nectar is the major food source for foraging honey bees
as well as nurse bees. Therefore, pesticide residues in nectar likely account for most of the exposures to
bees, and may represent most of the potential risk concerns for adult bees. However, if residues in pollen
are of concern, exposures to nurse bees, which consume more pollen than any other adult honey bees,
should be considered. This is the case especially when pesticide concentrations in pollen are much greater
than in nectar, or for crops that mainly provide pollen to bees and would be assessed on a case-by-case
basis. In fact, the Bee-Rex allows calculation of RQs for all types of bee castes.
For chemicals with no empirical data to represent the concentration of the chemical in pollen and nectar,
dietary exposure for Tier I risk assessment is estimated using generic residue data generated from other
chemicals as well as other plant parts. For foliar applications for dietary exposure, it is assumed that
pesticide residues on tall grass (from the Kenaga nomogram of T-REX which is incorporated into Bee-
REX) are a suitable surrogate for residues in pollen and nectar of flowers that are directly sprayed. For soil
applications, pesticide concentrations in pollen and nectar are assumed to be consistent with chemical
concentrations in the xylem of barley (calculated using the Briggs’ model). For seed treatments, pesticide
concentrations in pollen and nectar are based on concentrations in leaves and stems of treated plants (based
on the European and Mediterranean Plant Protection Organization (EPPO) default value discussed in the
White Paper), assumed to be 1 milligram per kilogram (mg/kg) or 1 part per million (ppm). More details
on these methods are available in the White Paper and in the T-REX user’s guidance. For tree trunk
injections, methods are still under development as discussed in the White Paper; one approach is where
application rates are converted into a tree-foliage weight basis (
Table 2).
The Tier I method is intended to generate “reasonably conservative” estimates of pesticide exposure to
honey bees, where reliable residue values (
i.e., measured residue levels in pollen and/or nectar) are not
15
available. As noted in
Table 2 (far right column), exposure estimates for foliar applications are derived
using the application rate (AR) being assessed. Model users should follow the input parameter guidance
contained in the Bee-REX user’s guide when running this model.
Table 2. Summary of contact and dietary exposure estimates used for foliar applications, soil treatments, seed
treatments and tree trunk injections of pesticides for Tier I risk assessments.
Measurement
Exposure Route
Exposure Estimate*
Endpoint
Foliar Applications
AR
Individual Survival (adults)
English*(2.7
µg a.i./bee)
Contact
ARMetric*(2.4
µg a.i./bee)
AR
Individual Survival (adults)
English *(110
µg a.i /g)*(0.292 g/day)
Diet
ARMetric *(98
µg a.i /g)*(0.292 g/day)
AR
Brood size and success
English *(110
µg a.i /g)*(0.124 g/day)
Diet
ARMetric *(98
µg a.i /g)*(0.124 g/day)
Soil Treatments
Individual Survival (adults)
Diet
(Briggs EEC)*(0.292 g/day)
Brood size and success
Diet
(Briggs EEC)*(0.124 g/day)
Seed Treatments
Individual Survival (adults)
Diet (1
µg a.i /g)*(0.292 g/day)
Brood size and success
Diet (1
µg a.i /g)*(0.124 g/day)
Tree Trunk Applications++
Individual Survival (adults)
Diet
(
µg a.i. applied to tree/g of foliage)*(0.292 g/day)
Brood size and success
Diet
(
µg a.i. applied to tree/g of foliage)*(0.124 g/day)
AREnglish = application rate in lbs a.i./A; ARMetric = application rate in kg a.i./ha
*Based on food consumption rates for larvae (0.124 g/day) and adult (0.292 g/day) worker bees and concentration in pollen and nectar.
++Note that concentration estimates for tree applications are specific to the type and age of the crop to which the chemical is applied.
In the Tier I approach, it is assumed that all chemicals applied as a soil drench, seed treatment, or trunk
injection may be systemically transported. This assumption may be refuted using data such as Log Kow
(Ryan
et al. 1988) and monitoring data (
e.g., crop rotation studies). Whether a chemical is transported
systemically in plants could potentially be confirmed using empirical data submitted to EPA, PMRA and
CADPR (
e.g., plant metabolism studies); however, it would be up to pesticide registrants/applicants to
submit sufficient data to demonstrate that a pesticide is not systemic.
3.1.2 Refinements of Tier I Exposure
In cases where RQs exceed the LOC (discussed below), estimates of exposure may be refined using
measured pesticide concentrations in pollen and nectar of treated crops, and further calculated for other
castes of bees using their food consumption rates (see
Table 3).
As discussed above in
Section 3.1.1, the most conservative (highest) exposure estimates for contact and/or
diet exposure routes are selected for the Tier I screening-level assessment. These exposure estimates are
based on adult and larval bees with the highest food consumption rates among bees. The Bee-REX tool
also calculates dietary exposure values and associated RQs for larvae of different ages, adult workers with
different tasks (and associated energetic requirements) and the queen. This is accomplished using the food
consumption rates provided in
Table 3. Those food consumption rates are based on work described in the
16
White Paper (USEPA, PMRA, CDPR 2012)6 and updated to reflect comments from the Scientific Advisory
Panel (SAP). Exposure values for other groups of bees within a hive along with their RQs can be used to
characterize risks of dietary exposures of different bees within the hive.
Empirical data may be used to refine conservative exposure estimates and reduce uncertainties associated
with the Tier I exposure assessment by providing direct measurements of pesticide concentrations resulting
from actual use settings. Studies investigating pesticide concentrations in pollen and nectar should be
designed to provide residue data for crops and application methods of concern.
Appendix 2. Considerations
related to quantifying residues of pesticides in pollen and nectar using pesticide-specific studies
includes
considerations related to quantifying residues of pesticides in pollen and nectar using pesticide-specific
studies.
6 USEPA, PMRA, CDPR (2012) White paper in support of the proposed risk assessment process for bees. United States Environmental Protection
Agency, Office of Pesticide Programs, Washington DC. Pest Management Regulatory Agency, Health Canada, Ottawa. California Department of
Pesticide Regulation, Sacramento, CA.
17
Table 3. Estimated food consumption rates of bees.
Life
Average age
Daily consumption rate (mg/day)
Caste (task in hivea)
Stage
(in days)a
Jelly
Nectarb
Pollen
Total
1 1.9
0 0 1.9
2 9.4
0 0 9.4
Worker
3 19
0 0 19
4 0
60 c 1.8d 62
5 0
120 c 3.6d 124
Larval
Drone 6+
0
130
3.6
134
1 1.9
0 0 1.9
2 9.4
0 0 9.4
Queen
3 23
0 0 23
4+ 141
0 0 141
Worker (cell cleaning and capping)
0-10
0
60f
1.3 - 12
g,h
61 - 72
Worker (brood and queen tending, nurse bees)
6-17
0
113 - 167f
1.3 - 12
g,h
114 - 179
Worker (comb building, cleaning and food handling)
11-18
0
60f 1.7
g 62
Worker (foraging for pollen)
>18
0
35 - 52f 0.041
g
35 - 52
Adult
292
Worker (foraging for nectar)
>18
0
0.041
g 292
(median)c
Worker (maintenance of hive in winter)
0-90
0
29f
2
g 31
Drone
>10
0
133 - 337 c 0.0002
c
133 - 337
Queen (laying 1500 eggs/day)
Entire lifestage
525
0
0
525
a Winston (1987)
b Consumption of honey is converted to nectar-equivalents using sugar contents of honey and nectar.
c Calculated as described in this paper.
d Simpson (1955) and Babendreier
et al. (2004)
e Pollen consumption rates for drone larvae are unknown. Pollen consumption rates for worker larvae are used as a surrogate.
f Based on sugar consumption rates of Rortais
et al. (2005). Assumes that average sugar content of nectar is 30%.
g Crailsheim
et al. (1992, 1993)
hPain and Maugenet 1966
18
3.2
Effects Characterization
3.2.1 USEPA Toxicity Testing Requirements for Bees
USEPA data requirements for pollinator testing are currently specified in Title 40 (Protection of the
Environment) of the Code of Federal Regulations, Part 158 (Data Requirements for Pesticides) Subpart G
(Ecological Effects) § 158.630 (Terrestrial and Aquatic Non-target Organism Data Requirements Table).7
When certain pesticide use patterns or triggers are met, current test requirements include the honey bee
acute contact toxicity test (OCSPP Guideline 850.3020)8, the honey bee toxicity of residues on foliage test
(OCSPP Guideline 850.3030)9 and field testing for pollinators (OCSPP Guideline 850.3040)10. The honey
bee acute contact toxicity test is required for pesticide technical grade active ingredients (TGAI) with
terrestrial, forestry and residential outdoor uses and is conditionally required for pesticides with aquatic
uses as a Tier I screen conducted under laboratory conditions. If the results of the honey bee acute contact
toxicity test indicates that a pesticide has a median acute lethal dose to 50% of the animals tested,
i.e., the
LD50 value, of less than (<) 11 micrograms (µg) per bee and the use pattern indicates that honey bees may
be exposed, then the toxicity of residues on foliage test is required as a laboratory-based test using the
technical end-use product (TEP). As specified in CFR40 § 158.63011, field testing of pollinators is required
if any of the following conditions are met:
Data from other sources (Experimental Use Permit program, university research, registrant submittals,
etc.)
indicate potential adverse effects on colonies, especially effects other than acute mortality (reproductive,
behavioral,
etc.); data from residual toxicity studies indicate extended residual toxicity; or data derived
from studies with terrestrial arthropods other than bees indicate potential chronic, reproductive or
behavioral effects.
Field studies are intended to represent real world conditions and are considered refined (Tier III) toxicity
tests. Pollinator field study designs have in the past varied considerably; therefore, study design elements
that are consistent with the specific hypothesis being testing in the field study should be identified in
advance and considered in the development of the study protocol.
Appendix 4. Tier 3 Field Study Design
Considerations
includes some generic study design elements.
3.2.2 Additional Guidance for USEPA Pollinator Testing
In addition to the honey bee adult contact toxicity study identified in the 40CFR158, several additional
studies are recommended to support the Tier I screen, as indicated in
Figure 1 and
Figure 2 and as discussed
in the White Paper. These studies include:
Acute oral toxicity to adult honey bees
Acute oral toxicity to larval honey bees
Chronic oral toxicity to adult honey bees
7 CFR40. 2011. Part 158, subpart G, §158.630 http://ecfr.gpoaccess.gov/cgi/t/text/text-
idx?c=ecfr;sid=fe712efed37d095118c7637457e011b3;rgn=div5;view=text;node=40%3A23.0.1.1.9;idno=40;cc=ecfr#40:23.0.1.1.9.7
8 USEPA. 1996
a. Ecological Effects Test Guidelines OPPTS 850.3020. Honey Bee Acute Contact Toxicity. EPA 712-C-96-147.
http://www.epa.gov/ocspp/pubs/frs/publications/OPPTS_Harmonized/850_Ecological_Effects_Test_Guidelines/Drafts/850-3020.pdf
9 USEPA 1996
b. Ecological Effects Test Guidelines OPPTS 850.3030. Honey Bee Toxicity of Residues on Foliage. EPA 712-C-96-148.
http://www.epa.gov/ocspp/pubs/frs/publications/OPPTS_Harmonized/850_Ecological_Effects_Test_Guidelines/Drafts/850-3030.pdf
10 USEPA 1996c. Ecological Effects Test Guidelines OPPTS 850.3040. Field Testing for Pollinators. EPA 712-C-96-150.
http://www.epa.gov/ocspp/pubs/frs/publications/OPPTS_Harmonized/850_Ecological_Effects_Test_Guidelines/Drafts/850-3040.pdf
11
Ibid CFR40 2011.
19
Chronic oral toxicity to larval honey bees.
Although EPA guidelines have not been developed for these studies, the Organization for Economic
Cooperation and Development (OECD) has developed a formal test guideline for an acute oral toxicity
study with adult honey bees (OECD 213)12 as well as a test guideline for acute oral toxicity study with
honey bee larvae (OECD 237)13. As noted in the White Paper, OECD also has developed a test guideline
for assessing acute contact toxicity with adult bees (OECD 214)14, that may provide sufficient data to fulfill
the 40CFR158 adult contact toxicity test requirement ordinarily fulfilled by EPA 850.3020. A chronic
(repeat dose) study guidance for assessing the oral toxicity to honey bee larvae is currently in development
by OECD, as is a guideline for assessing chronic (10-day) oral toxicity to honey bee adults. In situations
where formal guidelines do not exist, data recommendations are identified to address specific uncertainties
regarding the potential for adverse effects; these studies should be recommended as “special studies” and
reference the appropriate OECD test guideline and/or other applicable guidance. Typically, non-guideline
studies recommended as “special studies” (
e.g., acute adult oral and larval toxicity tests) are intended to
address specific uncertainties and a suitable rational for the study should be included in the data
recommendation.
Semi-field (
e.g., tunnel studies) and full-field studies have frequently been categorized as field pollinator
studies falling under EPA 850.3040. As discussed in the White Paper, OECD has developed guidance for
semi-field tests (
e.g., OECD 75)15 and efforts are underway to further refine and standardized methods for
conducting such tests. As discussed previously, Tier III studies of free-foraging bees are typically required
to address specific uncertainties identified in lower tier studies and/or based on a concordance of multiple
lines of evidence. Therefore, the methods used in these studies while described generically in EPA
850.3040 would likely be chemical/use specific and guidance is provided (see
Appendix 4. Tier 3 Field
Study Design Considerations) on elements to consider when recommending/evaluating such studies.
The decision to recommend bee toxicity data should consider multiple factors including the pesticide use
pattern, chemical properties, and the nature of uncertainties based on existing data for the chemical or
similarly structured chemicals. For example, the Tier I toxicity studies are generally limited to outdoor use
patterns, although some indoor uses in greenhouses may also warrant testing when plants are commercially
pollinated (
e.g., bumble bee pollination of tomatoes). As discussed in the White Paper and in this guidance,
if exposure is not considered likely, then effects testing may not be warranted beyond the screening-level
data collected at Tier 1 or may not be warranted at all if no exposure is considered likely. If a particular use
is determined not to result in exposure to bees, then a suitable rationale should be developed to support this
determination. However, toxicity data using the typical end use product (TEP) may be needed in addition
to data on technical grade active ingredient (TGAI) if there are data to indicate that a typical end-use product
is potentially more toxic than the technical grade active ingredient and bees may come directly in contact
with the intact TEP.
The need for additional data should be informed by whatever information has already been identified and
should consider multiple lines of evidence. This may include data that may be available for similarly
12 OECD. 1998
a. OECD Guidelines for the Testing of Chemicals. Honeybees, Acute Oral Toxicity Test. 213.
http://lysander.sourceoecd.org/vl=5988235/cl=12/nw=1/rpsv/cgi-bin/fulltextew.pl?prpsv=/ij/oecdjournals/1607310x/v1n2/s14/p1.idx
13 OECD. 2013. OECD Guidelines for Testing Chemicals. Honey bee (
Apis mellifera) larval toxicity test, single exposure.. http://www.oecd-
ilibrary.org/environment/test-no-237-honey-bee-apis-mellifera-larval-toxicity-test-single-exposure_9789264203723-en
14 OECD.1998
a. OECD Guidelines for the Testing of Chemicals. Test Number 214, Acute Contact Toxicity Test. http://www.oecd-
ilibrary.org/environment/test-no-214-honey bees-acute-contact-toxicity-test_9789264070189-en;jsessionid=43gvto47wnue9.delta
15 OECD 2007. Guidance document on the honey bee (
Apis mellifera) brood test under semi-field conditions. OECD Environment, Health and
Safety Publications. Series on Testing and Assessment No. 75. ENV/JM/MONO (2007) 22.
20
structured chemicals with common modes of action to achieve an economy of effort and limit, to the extent
possible, the need for animal testing that may ultimately prove to be redundant.
While this guidance articulates a tiered progression of exposure and effects data, the agencies retain the
right to ask for data in the order that is most relevant to risk management needs considering the multiple
lines of evidence that are available.
3.2.3 PMRA Toxicity Testing Requirements for Bees
The PMRA has been requiring both acute oral and contact honey bee adult toxicity studies when there is
potential for exposure for insect pollinators. Colony-level bee studies have also been required when there
is a potential for colony exposure and effects on the colony and developing brood. These acute studies,
together with colony-level studies, including Tier II semi-field and Tier III field studies, are incorporated
in the risk assessment for insect pollinators. Based on the SAP Pollinator Risk Assessment Framework,
additional studies are recommended to support the evaluation of insect pollinators where exposure is
considered likely. These include both larval bee toxicity and adult bee chronic oral toxicity studies. Insect
pollinator studies are generally not required for use site categories where pollinator exposure is not
considered likely. In general, Tier I studies should be conducted under controlled laboratory conditions
using the TGAI, while studies with the end-use products may be used to supplement the TGAI information.
Additionally, information on potentially toxic transformation products may be required. Tier II and III
studies should be conducted under progressively more realistic conditions using typical end-use products.
A tiered approach is used for these data requirements; when a risk is identified at a lower tier, higher-tier
studies are required to further assess the risk. Additionally, if other available information, such as scientific
research and published literature, indicates potential for adverse effects on insect pollinators and/or
colonies, then higher-tier studies or specific studies may be required to address the potential for concern.
Data submitted to PMRA for other non-target arthropods (predators and parasitoids) may also be considered
in risk assessment for insect pollinators.
Generally, the following studies may be required to assess the risk to insect pollinators:
Tier I insect pollinator studies:
o Bee adult acute contact toxicity
o Bee adult acute oral toxicity
o Bee larvae toxicity
o Bee adult chronic oral toxicity
Higher Tier and other insect pollinator studies:
o Residue study for insect pollinators
o Semi-field study for insect pollinators
o Field study for insect pollinators
o Bee toxicity of residues on foliage
o Other insect pollinator studies
21
3.2.4 Tier I Effects Characterization
3.2.4.1 Acute Oral and Contact Studies
Acute toxicity testing with both adult and larval bees examines the short-term effect of the test material
after a short-term exposure; chronic toxicity testing typically examines multiple exposures (repeat or
continuous dosing) with an extended period of observation time. Acute and/or chronic oral and contact
toxicity studies should be conducted according to available study guidelines described in the previous
sections. In cases where established guidelines are not available either in North America or through the
OECD, proposed study protocols should be reviewed and approved prior to the initiation of the definitive
study. Ideally, these acceptable protocols should be made available for consideration by other risk assessors
attempting to address similar uncertainties.
Typically, the primary measurement endpoint derived from the acute oral and acute contact toxicity studies
is the median lethal dose for 50% of the organisms tested (
i.e., LD50).
A. mellifera is used as a surrogate
for assessing risks to bees. However, depending on the circumstances triggering the data requirements, risk
assessors should be flexible in determining which species and studies are most relevant.
Any biological effects and abnormal responses, including sublethal effects, other than the mortality should
be reported during the acute contact and oral studies according to the requirement of the study guidelines.
To the extent that these measurement endpoints are dose responsive, consideration should be given to
developing a median effect dose for 50% of the organisms tested (
i.e., ED50) in addition to a median lethal
dose (LD50), if the data permit such an analysis. Alternatively, the risk assessment should note where
sublethal measurement endpoints appear to exhibit a trend across treatment levels.
Although it may be possible to extract both a no-observed adverse effect level (NOAEL) and an EDx or
LDx value from the laboratory-based acute studies, it is important to note differences in hypothesis-based
test designs needed to reliably support NOAEL versus regression-based study designs needed to support
EDx and LDx estimates. Hypothesis-based endpoints require sufficient replication to test for treatment
effects; whereas, regression-based estimates do not. The dose-response relationship derived from Tier I
studies can provide useful information in terms of slope of the dose-response curve.
3.2.4.2 Chronic Toxicity Studies
Chronic effects on adult and/or larval bees are considered in Tier I of the new assessment framework. Such
effects observed in guideline toxicity tests when protocols are sufficiently vetted and/or effects that have
been reported in the open literature that could potentially affect the colony should be considered in
determining which, if any, higher-tier tests are required and how the study should be designed. Studies
under development include the adult 10-day toxicity test and the repeat-dose larval toxicity test. Appendix
O of the European Food Safety Authority (EFSA) guidance document on bee risk assessment (EFSA
2013)16 discusses study design elements for the 10-day adult toxicity test and the OECD is currently
evaluating a proposed guidance for repeat-dose testing with honeybee larvae that extends through
emergence of adult bees. For the 10-day adult study, EFSA recommends the derivation of an LC50 and
NOAEC while the draft OECD repeat-dose larval toxicity study recommends the derivation of an NOAEC;
however, as discussed above, consideration should be given to the study design in determining the most
appropriate endpoint to derive.
16 EFSA. Guidance on the risk assessment of plant protection products on bees (
Apis mellifera,
Bombus spp. and solitary bees. EFSA Journal
2013;11(7):3295, 266 pp. doi:10.2903/j.efsa.2013.3295. Available online: www.efsa.europa.eu/efsajournal
22
3.2.4.3 Toxicity Studies Involving Exposure to Vapor Phase Pesticides
For a limited number of compounds (
e.g., fumigants), inhalation toxicity studies may be required as a
special study. Since current acute toxicity testing with young adult bees is typically conducted using a
small cage where bees are exposed through the diet (spiked sugar solution), it would be possible to conduct
an inhalation study using a similar test design where bees are fed untreated sugar solution, but the cage is
contained within a hermetic or flow-through container containing differing concentrations of the pesticide
in air. Because this is a modification of the existing protocol, registrants should submit the proposed study
protocol for review and approval by EPA and/or PMRA prior to conduct of the study. The protocol should
contain a rationale for the proposed air concentrations and durations of exposure.
3.2.4.4 Toxicity of Residues on Foliage to Honey Bees
Based on the contact LD50 value, the pesticide is classified as practically non‐toxic (LD50 ≥11
μg/bee),
moderately toxic (10.9 > LD50 >2
μg/bee), or highly toxic (<2
μg/bee). Unless the pesticide is determined
to be practically non‐toxic, EPA would then typically require a study on the toxicity of residues on foliage
to honey bees (OSCPP 850.3030; USEPA 2012)17.
The current OCSPP guideline study 850.3030 (USEPA 2012) evaluates the toxicity of residues on foliage
to honey bees. In this study, a formulated product of a chemical is applied to a bee attractive plant (
e.g.,
clover or alfalfa) at the maximum application rate with the minimum application interval. The crop is then
harvested in a manner that provides different age residues (
e.g., 0, 3, 8, 24, 48, 72 and 96 hours after
application). The harvested foliage is brought back to the laboratory where it is placed into cages. Adult
bees are placed in each cage with the foliage and allowed to come into contact with the foliage. Food and
water are provided
ad libitum. The bees exposed to weathered foliage are monitored for a period of time
until mortality declines to below 25%. The measurement endpoint derived from this study is the residual
time to 25% mortality (RT25), which is defined as the time needed to reduce the “residual toxicity” of the
test substance, as measured by mortality, to 25% of the test organisms. The time period determined by this
toxicity value is considered to be the time that the test substance is expected to remain toxic to bees in the
field from the residual contact exposure of the test substance on vegetation at an expressed rate of
application (typically expressed in terms of lbs a.i./A). The results of the foliar residue study may be useful
in the characterization of effects and risks of a pesticide to honey bees through a contact exposure route
with treated foliage and can be used to inform risk mitigation measures related to timing of pesticide
application (
e.g., how long a pesticide may be expected to present a risk from contact with foliage to
foraging honey bees after its application). Currently, the residue on foliage study is not used to derive RQs
as part of the Tier I risk assessment process; however, the test may provide useful information for
characterizing the extent (duration of time) to which residues may remain toxic to bees and may be helpful
in identifying particular formulations with extended residual toxicity. It is important to note that RT25
values are specific to formulations and may not be predictive across multiple formulations of a particular
active ingredient Although it may be inferred from the White Paper that an RT25 >8 hrs represents a
chemical with an extended residual toxicity (ERT), ERT has not been formally defined by EPA
17 USEPA. 2012. Ecological Effects Test Guidelines OCSPP 850.3030 Honey Bee Toxicity of Residues on Foliage. EPA 712‐C‐
018. January 2012. http://www.regulations.gov/#!documentDetail;D=EPAHQ‐OPPT‐2009‐0154‐0017
23
3.2.5 Tier II Effects Characterization
As discussed in the White Paper, the decision to move to more refined effects testing and to transition from
laboratory-based studies with individual bees (
i.e., Tier I) to colony-based studies (
i.e., Tier II or Tier III)
depends on whether Tier I LOCs are exceeded, the availability of data, and the nature of uncertainties that
warrant further testing. The Tier II studies are typically considered “semi-field” studies where small
colonies (referred to as nuclei colonies or “nucs”)18 are enclosed in tunnels, along with pesticide-treated
crops. Tier II studies may also include feeding studies in which whole colonies are tested; however, the
colonies are not confined to enclosures. Typical semi-field studies are usually conducted under conditions
that represent the worst-case exposure scenario of proposed uses to the entire colony (over the duration of
the study) or designed to address specific uncertainties with respect to effects on the colony; whereas, the
feeding studies are usually conducted with diets spiked with known concentration of test chemical using
colonies that are not confined to enclosures (
i.e., free-foraging bees).
The Tier II study designs may be amenable to additional treatment levels and replication, thus facilitating
the quantification of an application rate-response (tunnel study) or dose-response (feeding study)
relationship at the colony level and determination of a NOAEC. This information may be particularly useful
and transferable from the test crop/concentration to other crops where residue concentrations in pollen and
nectar are available in conjunction with associated application rates.
3.2.5.1 Tunnel Design Studies
Semi-field tunnel studies provide a means of recording a number of measurement endpoints,
e.g., adult and
larval survival, larval/pupal development, queen fecundity, worker bee behavior, to estimate various
attributes regarding the whole colony. The design of a Tier II tunnel study should be flexible in addressing
the specific risks identified at the Tier I risk assessment for either foliar application, seed or soil treatments,
or other exposure routes of concerns with reasonable modification. Although a general study guidance for
conducting a semi-field study is still under development, the OECD 75 guidance document on honey bee
brood testing19, and the European and Mediterranean Plant Protection Organization (EPPO) 170 describe
basic semi-field study elements that should be considered.
In a tunnel study there is typically a pesticide exposure period in the tunnel and an extended observation
period when test bees are allowed to freely forage from the landscape. While typically colonies can only
be maintained in enclosures for a limited exposure time (~10 days), these colonies may be monitored
following their removal from the enclosure to evaluate extended effects resulting from the exposure period
or delayed exposure from ingestion of stored pollen/nectar. Each of these phases of the study presents a
number of concerns. Some of the most important considerations include the following topics:
Bees in the tunnel: Food resource: It is important to ensure to the extent possible that bees have sufficient
forage and that the source of pollen and nectar is not depleted during the study. In tunnel studies, small
hives, referred to as “Nucs”, are used so as not to overwhelm the foraging capacity contained within the
enclosure. Because of the limited bloom period for most plants used in enclosure studies and because of
the stress on bees confined to a limited foraging area, the exposure duration of these studies is usually about
10 days, including acclimation period.
18 Nuclei colonies have been characterized as consisting of approximately 3000 brood cells respectively 750 cm’ with brood in all
stages, 1 good comb with honey and pollen and approximately 6000 worker bees; the ratio of brood to food (pollen/nectar)
should not exceed 4:1 (OECD 2007).
19 OECD. 2007. Guidance document on the honey bee (
Apis mellifera L.) brood test under semi-field conditions. Series on
Testing and Assessment No. 75. ENV/JM/MONO(2007)22.
24
Bees in the tunnel: Stress: Bees are intended to fly freely, and confinement to a tunnel may cause stress
for the bees. Reducing confinement stress caused by the tunnels is another reason the duration of tunnel
studies are generally limited to 10 days.
Behavior of test bees: Honey bee foragers are intended to fly freely in a large area. Limited tent space
may have an impact on the behavior of those bees. Therefore, caution should be taken in the interpretation
the results that are related to the behavior of the test bees.
Acclimation period: Long acclimation periods (
i.e., >3 days) within the tunnel should be avoided as they
can further limit the exposure period; however, if acclimation periods are not sufficient, bees may not begin
foraging efficiently until after residues on treated plants have started to decline. Therefore, initial mortality
counts for bees that have not been adequately acclimated to the enclosure may not be reliable estimates of
treatment-related effects. On the other hand, for seed treatment and/or soil application of a systemic
pesticide, residues in pollen and/or nectar are expected at some point in time post application. In these
cases, information on application timing and maximum resulting residues are necessary to ensure
appropriate exposures (
e.g., peak concentrations) are included during the in-tunnel phase of the study. Such
exposure considerations may negate the ability to include a pesticide-free acclimation period in the tunnels.
Climate and Success: Since these studies are conducted outside, initiation of the study may have to be
delayed due to inclement weather where it may not be feasible to move ahead with applications of the test
material.
Free Foraging Site Location: Depending on the objective of study where an extended observation period
may be required, the study report must also describe potential forage sites for the colonies and once the
hives are removed from the tunnel enclosures and allowed to free forage, the sites should ideally be removed
from agricultural areas where exposure to pesticides may take place and confound the study.
Semi-field studies also provide an opportunity to collect exposure data. Measurement of residues in pollen
and nectar as well as on/in foliage is recommended to provide data to confirm the level of potential exposure
to test hives as well as to refine RQs as part of the Tier I risk characterization. Therefore, consideration
should be given to recommending residue data in pollen and nectar of treated plants in semi-field studies
provided that such measurements do not interfere with the exposure to bees. Depending on the crop treated,
samples of pollen and/or nectar may not be readily collected from treated plants and it may be necessary to
consider alternative means of collecting these samples. One option may be to use the confined bees
themselves to collect samples of pollen/nectar; otherwise, it may be necessary to use intact flowers as an
indirect means of determining residues in pollen/nectar combined. Although the endpoint of a semi-field
study can be represented in terms of an application rate, measuring pesticide residues in pollen and nectar
allows for a more direct understanding of the concentrations to which the test bees are exposed and allows
for transferability of the effect results of the study to other crops, for which empirical residue data may be
available for pollen and nectar. Additionally, pollen and nectar (stored in hives as bee bread and honey or
uncapped honey) may be sampled within the hive to further characterize the exposure to colony. For
systemic compounds, application timing represents an uncertainty as movement of the compound to pollen
and nectar varies with the plant species and cultivation techniques.
As alluded to previously, semi-field studies conducted in tunnels/enclosures have limitations as the colonies
can typically be maintained in the enclosures for only limited amounts of time before food reserves are
exhausted and bees are stressed by the confined conditions. In addition, the foraging is not entirely natural
because the tunnels may alter the foraging behavior of the bees. These stresses may impact the function of
25
the colonies over time and result in adverse impacts, such as reduced brood production or decreased food
storage, from the enclosure itself. As with all higher tier studies, the statistical power of tunnel studies is
often low due to high variability as well as low replication; however, as with any study, if sufficient
resources are available, the number of replicates may be increased to increase the statistical power.
Tier II tunnel studies provide relevant toxicity information, given the constraints noted above, from a
limited exposure duration to characterize risk associated with specific use patterns based on the selected
tunnel design. Not all crops bloom for extended periods, and some exposure periods will be short in
duration as in the case of a foliar application with a relatively short foliar dissipation half-life. So the tunnel
study is most appropriate for assessing scenarios in which a short exposure period (
e.g., short bloom period
or rapid degradation/dissipation) is anticipated. Multiple application rates may be tested for a specific crop
to provide a NOAEC. This application rate may then be compared to proposed application rates to
characterize the potential risk from exposure to a specific application use pattern or determine the utility of
various mitigation options. In addition, the measurement of residues in pollen and nectar associated with
the various application rates tested in the tunnel study may provide information relevant to other crops when
data are available from targeted pollen and nectar residue studies. It is important to note, however,
differences between the crop tested in the tunnel and the actual uses (
e.g., the proposed crop uses on the
label) to which the tunnel study results are intended to represent; these differences may include bee
attractiveness, pollen and/or nectar production, bloom duration,
etc., in order to identify if a comparison is
appropriate.
While tunnel studies have limited value in addressing uncertainties related to long-term chronic exposure
and long-term chronic effects, they can provide valuable information on shorter exposure durations
including effects from all exposure routes (
e.g., contact and oral for foliar applications), and on actual
residue exposure levels specific to application methods, rates and test crops under consideration.
3.2.5.2 Feeding Design Studies
The methodology described by Oomen
et al. 199220, and the extended feeding field study design proposed
in the SAP White Paper, may also be considered a useful study for assessing the potential effects of
pesticides on bees at the colony level. Rather than restricting bees to tunnel enclosures with a treated crop,
colonies are unrestricted and fed food sources spiked with known concentration of pesticides. The amount
of pesticide can be monitored to provide an estimate of an overall amount of pesticide dose consumed in
hives.
The extended feeding study design offers some advantages over the tunnel design in that the duration of
exposure can be extended (
e.g., weeks or months compared to 10 days in tunnels) due to the lack of
confinement stress caused by the tunnels. The exposure may also be timed with a dearth period in the
available food resource to maximize the potential for exposure to treated diet. This design may be especially
applicable to pesticides whose use pattern and fate properties suggest that prolonged exposure of bees may
be likely.
Some limitations of the feeding design include the fact that bees are free to forage on sources other than the
spiked diet, which may introduce uncertainty in the exposure assessment. Currently available data from
these studies suggests that the colony will completely consume the provided sucrose solutions, but freely
foraged nectar and pollen also enter the hives. In the feeding study test bees are exposed through routes
20 Oomen, P. A. A. DeRuijter and J. Van der Steen. 1992. Method for honey bee brood feeding tests with insect growth-
regulating insecticides. Bul OEPP/EPPO Bulletin 22: 613 – 616.
26
defined specifically by which spiked food sources are provided, while bees in the field are likely exposed
concurrently via multiple exposure routes, including both pollen and nectar depending on crop species. It
is noted that feeding studies may also include spiked pollen in addition to spiked sucrose. Test bees are
also typically exposed to a range of chosen test concentrations. Such concentrations may not represent
exposure resulting from proposed application methods/use patterns. Differences in the field exposure routes
and residue levels from the tested exposure routes and test concentrations may contribute to the
uncertainties associated with effects observed in the study. In addition, excessive sucrose and or pollen
supply may have impact on the foraging behavior of foraging bees and cause uncertainty in effects on
colony development. Similar to the tunnel studies, colonies in the field feeding studies also have high
variability between colonies that decreases the statistical power of the tests; however, if sufficient resources
are available, the number of replicates may be increased.
Given the above limitations, the feeding study designs may provide useful information for the
characterization of risk associated with various crop use patterns. The study design can incorporate multiple
treatment levels of residues in spiked food to obtain a dose response and a NOAEC at the colony level for
the specific route of dietary exposure (
e.g., pollen, nectar, or both) employed in the study. The dose
response from the extended feeding study may then be compared to residue levels measured in specially
designed pollen and nectar residue studies that are tied to specific application rates. When evaluating these
comparisons specifically for nectar, the risk assessor should consider the sugar content of the feeding
solution relative to the sugar content of the nectar in a given crop as the amount of pesticide per amount of
sugar may differ. In the absence of crop specific information, the Tier I exposure assumption of an average
value of 30% for nectar may be assumed in order to provide a more refined comparison based on residue
levels normalized by sugar content. These comparisons of the dose response with measured residue levels
may provide colony-level information for the risk characterization of bee exposure to a given pesticide in
a specific crop. The extended feeding study design also offers some advantages over the tunnel design in
that the duration of exposure can be extended (
e.g., weeks or months compared to 10 days in tunnels) due
to the lack of confinement stress caused by the tunnels.
3.2.6 Tier III Effects Characterization
As indicated in the White Paper, full-field studies represent the highest level of refinement for pollinator
studies since they are intended to reflect the potential effects of a pesticide on bee colonies under actual
chemical use conditions. Tier III studies may be considered when pollinator risks cannot be excluded at
lower tiers of the risk assessment. These studies are intended to address specific uncertainties,
i.e., risk
hypotheses, which have been identified through lower tier studies and/or through the open literature under
reasonable worst case exposure scenarios in the field. While guidance is available for field pollinator testing
(
e.g., OCSPP Guideline 850.304021, EPPO 170), this information is relatively generic and is intended to
provide information on study design elements that should be considered in conducting field pollinator
studies. Additional general guidance is provided in the Field testing for Terrestrial Wildlife guidance
document (OCSPP Guideline 850.2500)22. Because full-field studies are intended to address specific
uncertainties, the study protocols should be reviewed and approved by the risk assessors prior to the conduct
of the study. Similar to the semi-field study protocols intended to address specific uncertainties, approved
study protocols should be made available to other risk assessors who may be confronted with similar
uncertainties in the future.
21
Ibid USEPA.1996c.
22 USEPA 2012. Ecological Effects Test Guidelines OCSPP 850.2500: Field Testing for Terrestrial Wildlife. Office of Chemical
Safety and Pollution Prevention (7101). EPA 712-C-021. January 2012. http://www.regulations.gov/#!documentDetail;D=EPA-
HQ-OPPT-2009-0154-0014
27
Although field pollinator testing has historically focused on honey bees, these studies are not limited to the
eusocial
A. mellifera and could include other bees (
e.g., bumble bees and solitary species) that are either
run concurrent with or
in lieu of honey bees. The rationale for recommending an alternative test species
should be included in a proposed protocol and reviewed by the risk assessors. Given the variability
associated with field-scale studies, plot size and attractiveness of the study crop should be considered. The
use of extinction curves (
i.e., the decline in the number of foraging bees relative to distance from their
colony) for estimating foraging distances of test species as a means of determining appropriate plot sizes
could be useful provided such data are available.
Appendix 4. Tier 3 Field Study Design Considerations
and the White Paper identify multiple study design considerations for full-field testing, but the risk
hypothesis should be a primary consideration.
The duration of field studies must be weighed against the extent to which confounding effects may limit
the utility of information obtained from the study. Consideration should be given to the type of pollinator
effects to be addressed in the field study. For chronic effects, a longer study duration including an
overwintering component may be considered. However, as outlined in the White Paper, multiple factors
(
e.g., nutrition, disease, pests) can impact colony survival, and longer study durations can influence the
extent to which these other factors may confound the study results. Frequent manipulations of the colony
to collect measurement endpoints can also stress the colony and affect the extent to which it is vulnerable
to disease/pests. Therefore, the decision to collect such information should be weighed against the
importance of these data in addressing the risk hypothesis.
As with semi-field studies, the full-field studies should include measures of exposure that can be used to
link residues in pollen/nectar (and foliage) back to specific application rates. Measured residue values
provide a means of ensuring that exposure actually occurs for durations that are expected to occur under
actual use conditions of the pesticide under evaluation. These data also provide a means of tracking the
movement of residues into various compartments of the intact colony and determining the extent (
i.e.,
magnitude and duration) to which other castes and/or life stages of the bees may be exposed. In addition,
measures of foraging activity on the treated crop can provide the risk assessor with information to determine
the extent to which the bees utilize the treated crop. Similar to semi-field studies, samples of pollen and/or
nectar may not be readily collected from treated plants and it may be necessary to consider alternative
means of collecting these samples (
e.g., use confined bees to collect pollen/nectar). In addition, as bees are
likely to forage on plants other than the target crop in the field, pollen composition (pollen palynology) and
residue analysis of pollen collected by free-foraging bees can be used to estimate the extent to which bees
are foraging on the treated and untreated crop (
i.e., level of exposure dilution) in the study. Such exposure
dilution and associated observed effects can allow for a more direct understanding of the actual level of
exposure at which to test hives, and potential transferability of the effect observed in the study to other
crops for which empirical residue data and foraging preference may be available.
As discussed in the White Paper, it is incumbent on the risk assessor to evaluate whether statistically
significant measurement endpoints are also biologically significant in Tier III and in the Tier II studies
discussed previously. Similarly, high variability in some measurement endpoints may make it difficult to
detect a statistically significant effect; however, a sufficient trend and magnitude of response may be
reported to support the conclusion that the effect is biologically significant. The utility of these effects
should be discussed in the risk characterization section of the risk assessment.
In the preceding sections, studies conducted at the screening level (Tier 1) and at high levels of refinement
(Tiers II and III) are discussed.
Table 4 provides an overview of strengths and limitations of various bee
toxicity studies in the context of overall bee risk assessment objectives. However, strengths and weaknesses
associated with any study depend on the purpose of the study and the specific hypotheses being tested.
28
Table 4. Expected endpoint and consideration of the strengths and limitations of various bee toxicity studies.
Primary
Study Name
Strengths
Limitations
Endpoints
Tier I
Mortality, Contact
Only 1 exposure route
Adult, Acute Contact
LD
considered
50
NOAECs are
Quantifiable test doses
typically not
Dose-response curve is
generated
generated
Acute exposure only
Mortality, Oral
Some sublethal effects
Measurement of
Adult, Acute Oral
LD50
can be measured
sublethal effects is
often limited in scope
Effects are assessed at
the individual level
Only 1 exposure route
Quantifiable test doses
considered
NOAEC and/or dose-
Measurement of
response curve can be
sublethal effects is
Adult, Chronic Oral
Mortality, NOAEC
generated
often limited in scope
Some sublethal effects
Effects are assessed at
can be measured
the individual level
Test is currently under
development
Quantifiable test
Larval Acute (single
Mortality, Larval
concentrations
Actual consumed
dose)
LD50/NOAEC
NOAEC and/or dose-
dose may vary
response curve can be
Assessment of effects
generated
through pupation is
Contact and oral
currently difficult
exposure routes are
Effects are assessed at
Larval Chronic (repeat Adult
included
the individual level
dose)
Emergence/NOAEC Larval effects are
Test is currently under
important for some
development
MOAs (insect growth
(chronic)
regulators)
Contact exposure
Acute exposure only
through residues on
Mortality, Residual
Actual dose is not
Foliar Residue
foliage
toxicity and/or RT
quantified
25
Can assess pesticide
Effects are assessed at
residual toxicity
the individual level
Tier 2
Colony strength
Brood pattern and
Multiple exposure routes
Short-term exposure
Semi-field, Tunnel
development
(contact, oral) related to
only (usually 7-10
Foraging activity
pesticide use methods
days in tunnel)
29
Primary
Study Name
Strengths
Limitations
Endpoints
Worker mortality
Minimizes influence of
Foraging may not be
and behavior
outside exposure to other
natural
Food storage and
chemicals
Stress on colonies
consumption
Some behavioral
from tunnel
Queen health
endpoints can be
confinement
quantified
Replication and
Standard test protocol
statistical power are
(OECD)
often low
Colony-level effects can
Usually based on a
be related to application
surrogate crop
rate and/or residues
Oral route only
Colony strength
Long-term exposure can
Brood pattern and
be assessed
Consumed dose may
differ from that
development
Colony-level effects can
encountered in the
Foraging activity
be related to
field
Food storage and
concentration in diet
Semi-field, Feeding
consumption
Protocols have not
Greater replication can be
Worker mortality
been standardized
achieved vs. tunnel or full
and behavior
field studies
Confounding
Foraging activity
influences of off-site
Greater control over
Queen health
foraging and exposure
exposure vs. full field
studies
Foraging may not be
natural
Tier 3
Practical constraints
Colony strength
may limit ability to
Brood pattern and
assess “high end”
development
Most environmentally
realistic of crop/pesticide
exposure scenarios
Foraging activity
exposure conditions
Full Field
Food storage and
Replication and
statistical power often
(Experimental)
consumption
Can resolve specific
low
Worker mortality
uncertainties raised from
and behavior
lower tiers
Confounding
Queen health
influences of off-site
foraging and exposure
Costly
Colony strength
Brood pattern and
Exposure may be
development
Most environmentally
difficult to quantify
realistic of crop/pesticide
Full Field
Food storage and
exposure conditions
Causal linkages may
(Monitoring)
consumption
be confounded by
Worker mortality
Can incorporate multiple
other stressors
and behavior
crop exposure scenarios
Costly
Queen health
30
4 Risk Characterization
Risk characterization is the final phase of the risk assessment process and consists of two parts,
i.e., risk
estimation and risk description. Risk estimation involves integration of exposure and effects information
to estimate the likelihood of adverse effects on the ecological receptors as a result of exposure. In Tier I,
risk estimation involves the calculation of RQs. Risk description includes an interpretation of the risks in
the context of uncertainty and sensitivity of risk estimates to underlying assumptions and quality of data.
In addition, risk description considers how risk can be mitigated through restrictive label language and/or
best management practices (BMPs). A weight-of-evidence approach should be used in the overall risk
characterization. The risk characterization should also discuss data gaps and whether uncertainties could
be readily addressed through additional data.
4.1
Risk Estimation
4.1.1 Calculation of Risk Quotients for Tier I Risk Assessment
Table 5 lists the exposure and effect estimates used in quantifying risk in Tier I. As discussed in the White
Paper, risks are quantified for individual bees in Tier I; however, RQ values are not derived for colony-
level effects. As is the case with other taxa, the RQ value is calculated by dividing the exposure estimate
by the effect endpoint. It is important that the measurement endpoint units are reported or can be converted
to units that are consistent with exposure estimates such that a risk quotient can be reasonably derived for
Tier 1 screening RQ values.
31
Table 5. Summary of exposure and effect estimates used in deriving risk quotients for Tier I risk assessments.
Measurement
Exposure
Acute Effect
Chronic Effect
Exposure Estimate+
Endpoint
Route
Endpoint
Endpoint+++
Foliar Applications
Individual
AR
Acute contact
Survival
Contact
English* (2.7
µg a.i./bee)
None
AR
LD
(adults)
Metric* (2.4
µg a.i./bee)
50
AR
Chronic adult oral
Individual
English * (110
µg a.i /g) (0.292
g/day)
Acute oral
NOAEL
Survival
Diet
AR
LD
(effects to survival or
(adults)
Metric * (98
µg a.i /g) (0.292
50
g/day)
longevity)
AREnglish * (110
µg a.i /g) (0.124
Chronic larval oral
Brood size and
g/day)
Diet
Larval LD
NOAEL (effects to adult
success
50
ARMetric *(98
µg a.i /g) (0.124
emergence, survival,)
g/day)
Soil Treatments
Chronic adult oral
Individual
Acute oral
NOAEL
Survival
Diet
(Briggs EEC) * (0.292 g/day)
LD
(effects to survival or
(adults)
50
longevity)
Chronic larval oral
Brood size and
Diet
(Briggs EEC) * (0.124 g/day)
NOAEL (effects to adult
success
Larval LD50
emergence, survival,)
Seed Treatments
Chronic adult oral
Individual
Acute oral
NOAEL
Survival
Diet (1
µg a.i /g) * (0.292 g/day)
LD
(effects to survival or
(adults)
50
longevity)
Chronic larval oral
Brood size and
Diet (1
µg a.i /g) * (0.124 g/day)
NOAEL (effects to adult
success
Larval LD50
emergence, survival,)
Tree Trunk Applications++
Chronic adult oral
Individual
(
µg a.i. applied to tree/g of foliage) Acute oral
NOAEL
Survival
Diet
* (0.292 g/day)
LD
(effects to survival or
(adults)
50
longevity)
Chronic larval oral
Brood size and
(
µg a.i. applied to tree/g of foliage)
Diet
Larval LD
NOAEL (effects to adult
success
* (0.124 g/day)
50
emergence, survival,)
AREnglish = application rate in lbs a.i./A; ARMetric = application rate in kg a.i./ha
+Based on food consumption rates for larvae (0.124 g/day) and adult (0.292 g/day) worker bees and concentration in pollen and nectar.
++Note that concentration estimates for tree applications are specific to the type and age of the crop to which the chemical is applied.
+++To calculate RQs for chronic effect, NOAEC can be used as the effect endpoint to compare with the exposure estimate in concentration.
4.1.2 Levels of Concern for Tier I Risk Assessment
Based on the process described in the White Paper, risk estimates,
i.e., RQ values are only calculated on
individual adult and larval bees during the screening-level, Tier I, stage. RQ values are then compared to
Levels of Concerns (LOCs). The LOCs for acute and chronic exposure are 0.4 and 1.0, respectively. If an
RQ exceeds its LOC, then the chemical use being assessed poses a potential risk to insect pollinators. Since
RQ values are calculated for individual bees, they can be considered applicable to solitary as well as social
bees although they may be conservative estimates of risk for the latter.
32
The LOC of 0.4 identified in the White Paper for assessing potential risks to individual bees was considered
by the FIFRA SAP to be highly conservative. The value is based on the median probit dose-response slope
across acute contact and oral toxicity studies. While the SAP correctly noted that the 10% level of mortality
resulting from RQ values equal to an LOC of 0.4 has not been demonstrated as detrimental to intact
colonies, this LOC was based on an effect level that would be consistent with background (
i.e., control
mortality) in laboratory-based studies. This LOC is intended to be conservative and serve as a reasonable
screen for determining whether higher-tier testing is needed. As discussed in the White Paper and as
depicted in
Figure 1 and Figure 2, the RQ values estimated using screening-level exposure models can be
refined using measured residue levels prior to determining whether additional colony-level (Tier II) data
are needed. The LOC of 0.4 only applies to interpreting RQ values from laboratory-based acute toxicity
studies of individual bees and does not apply to the higher-tier semi-field and full-field studies on whole
colonies. The LOC of 0.4 also applies to interpreting RQ values for acute larvae toxicity since limited
available data of dose-response slopes for larvae prevent determining an LOC using the same approach as
for adults.
Although laboratory-based chronic toxicity studies with adult and larval bees are still under development,
protocols have been drafted for conducting such studies and these data may be available for review. As
discussed in the White Paper, chronic RQ values are to be compared to an LOC of 1.0,
i.e., risk is evaluated
based on whether EECs exceed the NOAEC from chronic toxicity studies with individual bees.
For those chemicals where RQ values exceed LOCs even after exposure estimates have been refined using
measured residue values in pollen and nectar, more refined testing on honey bee colonies may be needed
using Tier II semi-field studies and, depending on the nature of remaining uncertainties, Tier III full-field
studies. Higher-tier studies with whole colonies are used to provide a more realistic characterization of
potential adverse effects to colonies since the study design is intended to reflect actual exposure conditions.
The risk assessor should use semi- and full-field studies to determine whether effects reported in laboratory-
based studies on individual bees are apparent at the level of the whole colony and nature, magnitude and
duration of these effects considering potential routes of exposure, and the biological relevancy of effects
must be gauged, as well as sublethal effects that may not manifest in Tier I studies. To the extent possible,
available estimates of exposure to colony bees through measured residues in pollen and nectar coming into
the colony through the labeled use of the pesticide should be characterized relative to the reported effects.
4.2
Risk Description
4.2.1 Use of Other Lines of Evidence
The risk description phase of the Risk Characterization provides an opportunity to discuss additional lines
of evidence and uncertainties regarding potential risks to bees beyond the RQ values calculated in the risk
estimation.
All studies and relevant information are used to evaluate the likelihood and/or extent of risks to bees under
various pesticide use scenarios. Risk assessments must provide a transparent description of assumptions
and uncertainties surrounding the assessment and the lines of evidence considered. To the extent possible,
the risk assessment should reference more detailed discussions regarding uncertainties surrounding these
lines of evidence than contained in other documents such as the Overview Document, the White Paper and
this guidance. It is not necessary to include all of the uncertainties noted in this document in a risk
assessment; however, if the risk assessor thinks that an uncertainty is of particular concern for the chemical
being assessed, a discussion of such uncertainty should be included in the risk assessment. In characterizing
uncertainty, the risk assessor should consider the full weight-of-evidence. Application of a weight-of-
evidence analysis is an integrative and interpretive process routinely used to evaluate ecological toxicity
33
data23 in a manner that takes into account all relevant scientific information. The analysis should consider
whether the existing data provide relevant, robust and consistent evidence (
e.g., agreement within and
among the outcomes of laboratory and semi-/full-field studies) that a chemical has the potential to adversely
affect bees and at what level of biological organization (
i.e., individual bees or at the colony level) and
under what conditions (
e.g., at what application rate, exposure duration). As part of this analysis, the risk
assessor should consider the biological plausibility of the effect, the coherence, strength and consistency of
the body of information as evaluated across all of the relevant information that are available.
Incident data from the EPA Ecological Incident Information System (EIIS), PMRA Pesticide Incident
Reporting Program (IRP) and other aggregate databases (
e.g., the EPA Incident Data System) should be
considered in the context of the certainty index applied to these data.
Open literature studies that have met the applicable agency standards for inclusion in ecological risk
assessments should be discussed in the risk description as well. Such studies may provide useful
information in considering the overall weight of evidence for the risk characterization. Careful
consideration should be given to the relevance of endpoints derived from open literature studies in relation
to the assessment endpoints of concern.
Additionally, the risk description section is a good place to incorporate potential mitigation options and
describe how specific mitigations could impact risk conclusions. Through the engagement of risk
managers, risk assessors can incorporate multiple lines of evidence and alternative exposure scenarios into
the risk description.
4.2.2 Synthesis of Risks among Tiers
The risk description should provide a synthesis of the various levels of refinement for both exposure and
effects that were considered and discuss the extent to which semi-field and, when available, full-field
studies of whole colonies provide information consistent with laboratory-based studies with individual
bees. For example, if acute mortality was observed for larval bees under laboratory conditions, did semi-
field studies identify effects in the developing brood of colonies? For social bees while the risk to individual
bees quantified in Tier I are important, a critical question is whether adverse effects occur at the level of
the whole colony.
The Tier I toxicity testing with honey bees typically involves summer bees as opposed to winter bees. This
is primarily due to the logistical constraints of collecting suitable numbers of winter bees without
irreparably damaging the colony from which they were derived. Worker bees can be potentially replaced
by the queen in summer whereas they cannot in winter. The assessment should note this distinction and
that reduced survival in winter bees, which cannot be replaced and which must survive for a longer duration
in order to ensure that the colony successfully overwinters, can be more detrimental to the colony survival.
Therefore, the timing when potential exposure to bees may occur in the field should be considered during
the course of risk assessment.
Compared to worker bees in hives that are many and typically replaced, there is only one queen in each
hive. The loss of the single queen can result in the loss of the entire colony although colonies may attempt
to supersede a queen that is failing. Effect on queens may be tested in Tier I level by modifying the
laboratory toxicity study protocol that is currently designed for worker bee individuals. As part of the
evaluation of whole colonies, consideration should be given to whether there are any data with which to
determine whether the queen is functioning properly. If no direct measurements of toxicity to the queen
23USEPA. 1998. Guidelines for Ecological Risk Assessment. Published on May 14, 1998, Federal Register 63(93): 26846 – 26924.
http://www.epa.gov/raf/publications/pdfs/ECOTXTBX.PDF
34
bee are available, the condition of the brood may serve as an indirect measure of queen performance. Brood
pattern (
i.e., spotty versus uniform), presence of eggs, and relatively uniform development of the larvae and
pupae can be useful indicators. Another useful indicator of queen performance can be whether there are
any supersedure (presence of queen cells intended to replace an old or failing queen) in the brood comb and
queen replacement; colonies in which the queen is failing may attempt to replace (supersede) the queen.
Additionally, an over-abundance of drone cells may be indicative of an unfertilized queen that will lead to
colony loss.
Studies of the whole colony should avoid over manipulation of the colony, general observations of colony
activity and health should be included as part of the assessment. Colonies confined to enclosures in a Tier
II study will likely exhibit increased stress from a variety of factors linked to the limited space within the
enclosure. These factors should be described even if they are not considered treatment-related as they can
affect the power of the study to detect treatment effects. Additional descriptions of pre- and post-application
levels of pest and diseases should be assessed as an indicator of bee health.
4.2.3 Risk description on Sublethal Effects
In addition to the measurement endpoints that have direct linkages to assessment endpoints and that were
used quantitatively in the risk estimation section of the risk characterization, the risk description should
also include, if available, sublethal endpoints (
e.g., behavioral effects, proboscis extension reflex) that are
associated with the chemical under evaluation. However, these sublethal endpoints often lack information
on subsequent effects on survival, growth or reproduction. Although the FIFRA SAP recommended an
increased use of sublethal endpoints, until suitable linkages have been developed between sublethal
measurement endpoints to assessment endpoints, their use in risk assessment should remain qualitative; this
is consistent with the process used for other taxa as described in the Overview Document as well as in the
open literature guidance.
Observation of sublethal effects is required for Tier I laboratory toxicity studies. However, these studies
are tested using individual bees. The ability of these studies to test social interactions between bees (such
as tropholaxis (
i.e., the transfer of food between bees) and flight and/or foraging behavior is limited.
However, such sublethal effects information should be considered as part of the weight of evidence in
deciding whether to proceed with higher tier testing and in the overall risk characterization.
It is important that sublethal effects are described even when their association with apical endpoints (
e.g.,
colony survival) may not be apparent since the absence of such an effect may be the result of limitations in
study design (
e.g., insufficient study duration). For example, radiofrequency identification (RFID
technology) has been employed to mark and track bee movement and can help to quantify foraging activity;
however, situations where marked declines in foraging activity may not have been reflected in adult and/or
larval survival estimates could indicate that there were compensatory mechanisms (
e.g., sufficient food
stores in the comb) that obscured the effect.
The extent to which forage bee and/or hive bee behavior is affected by pesticide treatments should be
discussed in the context of whether other measurement endpoints were affected. While impaired behavior
can affect proper functioning of the hive, it is frequently difficult to detect such effects during short-term
studies. Similarly, longer-term studies can be impacted by a number of factors (
e.g., weather, pests, disease,
and available forage) that can increase variability in measurement endpoints and make it difficult to detect
statistically significant effects at the colony level. To the extent possible, where behavioral effects are
reported either in studies submitted in response to testing requirements or in acceptable open literature
studies, an effort should be made to determine whether these effects are evident in other endpoints that
would most likely be impacted. For example, impaired foraging success may affect the quantity of food
35
reserves in the comb and may be accompanied by decreased brood production by the queen due to a
reduction in available food.
When reviewing single studies, all available information should be considered in evaluating the dynamics
of colony development and potential interactions among many of the measured endpoints.
4.2.4 Use of Simulation Models
Different applications of colony simulation models for pesticides are currently under evaluation. Some of
these potential applications include: estimating exposures and characterizing risks of bees exposed to
pesticides when little data are available, and completing a sensitivity analysis of lethal and sublethal
measurement endpoints to determine which endpoints are critical to the survival of a colony (this could be
used to inform toxicity test study designs or regulatory LOCs for risk assessments). The White Paper
discussed simulation models and, while the SAP agreed that such models may prove useful, none of the
models available were deemed suitable at this time. However, the SAP recommended that components of
existing models (
e.g., analytical subroutines) may be appropriate for use at this time. EPA and PMRA are
working with researchers in government, academia and industry both domestically and internationally to
evaluate and adapt models for their use in risk assessments intended for regulatory purposes.
4.2.5 Uncertainties
The primary sources of uncertainties for the risk assessment are related to estimating pesticide exposures
to bees and effect of those exposures to bees. With respect to dietary exposure, the first source of uncertainty
may be related to the extent to which the amount of food consumed by bees for the Tier I exposure estimate
represents pesticide concentration in bee food sources. With respect to contact exposure, there is uncertainty
as to the extent that residues on leaves and even soil may be available to bees for uptake. There is also
uncertainty as to the extent to which bees may be exposed to pesticide residues through various sources of
water, including puddle and plant exudates, and whether that water is ingested, used to dilute honey, or
used to cool the colony. The second source of uncertainty is related to differences in bee biology and their
foraging behaviors that may directly impact exposure routes and the extent to which bees forage on the
treated crop. All these sources of uncertainty may be reflected in the study designs at different tiers.
4.2.5.1 Uncertainties: Tier I exposure estimate
In the Tier I assessment, dietary exposure is estimated based on upper-bound food consumption rates for
honey bees. These rates are estimated based on laboratory studies conducted under controlled conditions.
Uncertainties may exist in relation to bee physiology changes that occur naturally, for example difference
between summer bees and winter bees. However, the upper-bound food consumption rates using in the
screening-level assessment are expected to be conservative.
The Brigg’s model is currently used to estimate Tier I EECs resulting from a soil application. There are
five notable limitations to using the modified Briggs’ model approach. The first is that this methodology
is based on empirical data from only one species of plant. The second limitation is that the data set used to
derive elements of the model is based on a limited number of chemicals that represent only two classes of
pesticides (
i.e.,
O-methylcarbamoyloximes and substituted phenylureas). The third limitation is that this
approach is based on data from non-ionic organic chemicals and may have limited utility for ionic chemicals
that whose transport may not be predicted well using Kow and Koc. The fourth limitation of the Briggs’
model is that it is based on passive transport of chemicals into xylem, therefore, this approach does not
directly estimate pesticide concentrations in plants that are the result of phloem transport. The fifth
limitation involves the use of estimated pesticide concentrations in vegetative plant matrix (
i.e., shoots) as
a surrogate for nectar and pollen. As additional data become available, the utility of the Brigg’s model for
36
providing screening-level estimates of exposure from soil applications will be assessed and the model will
either be revised or more appropriate means of estimating exposure will be adopted.
4.2.5.2 Uncertainties: Use of residue data
Use of chemical-specific pollen and nectar residue data reduces the uncertainties associated with the
methods discussed above; however, these data should be used with caution. Care should be taken to ensure
that the available empirical data are representative of the registered/proposed uses of the chemical of interest
and do not under-represent exposures to bees. Samples collected from free-foraging bees that may forage
on non-targeted crops could underestimate the actual concentration in pollen and nectar from a treated crop
as the bees may not have exclusively foraged on the treated crop.
4.2.5.3 Uncertainties: Agronomic Practices
One of the most important considerations within the agronomic practices is the use of managed pollinators
for crop production. For some crops, growers will bring in managed bees to augment the pollination
services of local bees if the crop requires pollination and wild bee populations are insufficient for adequate
pollination. These commercially managed bees may include honey bees, bumble bees, blue orchard bees,
alfalfa leafcutting bees,
etc. When commercially managed bees are used to pollinate a crop, the potential
for exposure and the magnitude of that exposure to the pollinating bees may be greatly increased.
Depending on the physical-chemical property of a pesticide and use methods, agronomic practices, such as
irrigation, may affect the translocation of systemic pesticides and have effects on the residues in bee food
sources.
The agronomic practices may also be related to the extent to which a particular registered use may be
applied across the landscape of the use. Different use patterns may occupy varying spatial extents of
coverage.
4.2.5.4 Uncertainties: Pollination Biology
Uncertainties on the exposure of pesticides to pollinators in the field are also associated with plant/crop
pollination biology. Risk assessment usually encompasses a wide variety of crops or plants that have unique
pollination characteristics. These plants may include ornamental annuals or perennials, trees or bushes
covered under forestry uses, or annual or perennial crops. The pollination biology of each of these plants
is important to consider when developing a description of the potential risk to bees. In the problem
formulation stage of the assessment, information on the attractiveness of plants to pollinators covered by
the proposed uses should be considered to determine whether exposure may occur and the scope of a risk
assessment. At the risk description phase of the assessment, information on the pollinator attractiveness of
the plant as well as acreages and application methods will help the risk assessor to determine the spatial
and temporal aspect of risk to the bee pollinators identified in the problem formulation as well as potential
mitigation solutions.
The pollination biology of plants relates to the intrinsic characteristics of the plant itself. These
characteristics include the following considerations:
Bee visitation to the flowers of the ornamental, forestry tree, or crop: Not all plants produce flowers
that are attractive sources of forage for honey bees, bumble bees, or solitary bees. Conversely, flowers of
37
some plant species last only short period. The short blooming duration reduces the potential exposure of
flower visitors.
Harvest period of the crop: While a crop or plant may produce flowers that are attractive to bees, some
of these crops are harvested prior to bloom.
Bloom period of the crop: Different plant species will bloom at different times of the year. In addition,
the length of the bloom period can differ between plants. Some plants bloom within a specific, relatively
narrow window of time called a determinate bloom period. Other crops may produce blossoms
continuously over the course of the growing season or for an extended period of time, which is called
indeterminate bloom. Indeterminate blooming crops provide a much longer window of potential exposure
to pollinating bees.
Pollen versus nectar as the food source of the bees: Not all flowers produce a nectar reward for bees,
while for some species of plants, flowers may not be the only source of nectar.(
e.g., cotton). Depending on
the species of plant, some flowers do not produce pollen that bees will collect. For example, some varieties
of citrus produce only minimal quantities of pollen, while at the same time provide a rich source of nectar
for pollinating bees. On the other hand, corn is a grass monocot that produces copious amounts of pollen
that bees have been shown to use to varying extents, while it produces virtually no nectar. The type of food
source produced by the plant may be an important consideration for the specific crop.
Bee diversity: In terms of the types of bees, honey bees and bumble bees are colonial while there are a
variety of bees, both managed and wild, that are solitary and, depending on the plant, their foraging
strategies may differ substantially; therefore, potential exposure may differ.
4.2.5.5 Uncertainties: Differences in Bee Life History
As noted in the White Paper and as discussed in the SAP’s response, there is uncertainty regarding the
extent to which any risk assessment process that relies on data on a specific species,
e.g.,
A. mellifera, can
be considered representative of an entire taxon or multiple taxa. This is especially true for honey bees,
which are a highly social (eusocial) species, where the colony/hive is dependent on the collective tasks of
multiple castes and function as a “superorganism”; whereas, the majority of other bee species, particularly
those species native to North America, are solitary.
4.2.5.6 Uncertainties: Differences in Pests/Pathogens/Nutrition/Management
Multiple factors can influence the strength and survival of bees whether they are solitary or social. These
factors, including disease, pests (
e.g., mites), nutrition, bee management practices, can confound the
interpretation of studies intended to examine the relationship of the test chemical to a receptor (
i.e., larval
or adult bee). Therefore, most studies attempt to minimize the extent to which these other factors impact
the study; however, higher tier studies afford less control over these other factors, and their role may become
increasingly prominent as the duration of the study is extended. Although studies attempt to minimize the
confounding effects of other environmental factors, there is uncertainty regarding the extent to which the
effects of a chemical may be substantially different had these other factors been in place.
4.2.5.7 Uncertainties: Uncertainty in Study Designs
Data from the toxicity of residues on foliage study are used qualitatively to characterize the length of time
that residues remain toxic to bees. The results of the guideline study may result in precautionary label
38
statements similar to those discussed in the EPA Label Review Manual24 or in guidance documents intended
to reduce the potential effects of pesticides on bees (
e.g., Riedl
et al. 2006)25.
Studies in which colonies are provided a source of spiked food (
i.e., either spiked sugar solution or spiked
pollen) are referred to as “feeding studies”. Feeding studies are used as a means of examining the potential
effects of pesticides on whole bee colonies that are free foraging; however, there are uncertainties associated
with these studies. Since bees are provided a specific diet,
e.g., spiked sugar solution as proposed by Oomen
et al. 199226, there is uncertainty whether the specific diet may afford limited nutritional diversity to the
test bees. Although the quantity of food consumed is recorded, consumption rates within treatments can
vary widely, and there is uncertainty as to the extent to which the test material is actually consumed versus
stored in the comb. The richness of other food sources in the test area may also influence the actual
consumption of spiked test chemicals. Although the colonies are typically positioned away from
agricultural areas, the extent to which the free foraging bees may have access to plants that may be
contaminated with other pesticides is uncertain. Additionally, these studies may frequently have large
numbers of colonies, belonging to different treatment groups, in close proximity where bees may potentially
cross between treatments through robbing and/or scavenging activity in impaired colonies; this activity may
be enhanced in areas where alternative forage may be sparse. Finally, failing colonies may be robbed of
resources by bees from other treatment/controls groups resulting in exposure to different doses of the test
material than planned.
Tier II tunnel studies contain challenges inherent to the design itself. In these studies, a mesh enclosure is
placed over the colonies and contains within its boundaries a specified area of a test crop to serve as a food
source. The enclosures can impact the performance of the colonies and ultimately limit the development
of the colonies either from the stress of the tunnels placed on the colonies or from inadequate forage for the
bees within the tunnels. The availability of food source and impact on the behavior of test bees limit the
exposure duration of a tunnel study. Although a Tier II study should be carefully designed, by considering
the test hive size, exposure duration and adequacy of feed sources, to minimize the effect of tent enclosure
itself, the reviewer should be attentive to the performance of the control colonies within the tunnels to
ascertain if stress placed on the colonies has a notable effect on colony performance that may impact the
ability to derive meaningful results from the study data.
Evaluating Tier II or full-field Tier III studies can be challenging because the study environment is relatively
uncontrolled and a number of factors can impact the study. In the case of tunnel studies these factors can
include the weather, disease/parasites, extent to which bees feed on the treated crop as opposed to simply
storing the food, and the potential for contaminated or inadequate forage once the colonies are placed
outside of the tunnel for extended evaluation. In the case of full-field studies, confounding factors can also
include the weather, disease/parasites, the extent to which bees feed on the treated crop as opposed to simply
storing the food, in addition to the alternative (untreated or treated with other pesticides) forage areas and
the distance that worker bees can forage even during the exposure phase of the study. When confounding
effects are encountered in these studies, it is incumbent on the reviewer to determine whether any
information is contained in the study, given the complexity and expense of these studies. The SAP
accurately noted that the longer a study is conducted, the greater the opportunity for other factors to affect
study colonies; appropriate study designs and replication can provide a means of accounting for these other
factors particularly where studies are conducted over an extended spatial-temporal framework.
Direct measures of effects on specific castes (
e.g., queens) may be limited due to inadequate numbers of
test organisms or logistical constraints to isolating particular elements of the colony for study. Therefore,
24 USEPA. 2012. Label Review Manual. http://www.epa.gov/oppfead1/labeling/lrm/
25 Riedl, H. E. Johansen, L. Brewer and J. Barbour. 2006. How to Reduce Bee Poisoning from Pesticides. A Pacific Northwest Extension Publication.
Oregon State University, University of Idaho, and Washington State University. http://extension.oregonstate.edu/catalog/pdf/pnw/pnw591.pdf
26
Ibid Oomen
et al. 1992.
39
potential effects or lack thereof are frequently inferred based on other colony-level measures. For example,
queen performance may be indirectly assessed through evaluation of egg laying, supersedures, and the
extent of drone cell production. The risk assessment should be transparent in acknowledging the
uncertainties associated with making such inferences.
Due to the high cost and large variation of field conditions associated with a field study, design of an
appropriate field study can be challenging. Additionally due to concerns of using toxic reference
compounds in open field studies, these studies do not typically include toxic reference compounds such as
those used in laboratory-based and semi-field studies. Field studies also do not typically have high numbers
of replicates; therefore, the statistical detection power of a study may be low However, full-field studies
are conducted under more realistic exposure conditions (actual application rates) and are expected to reflect
the potential impact of a pesticide in the field. Although the risk assessment considers the full weight of
evidence across all of the studies evaluated, results from Tier III studies provide the most realistic
understanding of effects on colonies from the proscribed use of the pesticide.
4.2.6 Data Gaps
One of the largest challenges facing risk assessors is the extent of variability and uncertainty associated
with estimating the potential likelihood and magnitude of adverse effects to bees. The decision to
recommend additional studies should be based on a reasonable risk hypothesis that has considered all of
the available lines of evidence for a study.
As indicated in this guidance, the screening-level risk assessment can be refined using chemical-specific
residue data. Residue studies to support screening-level assessments should focus on a small number of
pollinator-attractive crops that may serve as a good source of pollen or nectar for foraging bees and will be
used to be representative of the broader list of crop uses. Additional considerations for crop selection could
include the acreage of the crop that may be potentially treated in an area, or the percent of the crop treated
as well as application methods. In examining other lines of evidence, risk assessors should determine
whether exposure data on similarly structured chemicals may be of utility and/or whether data required by
other regulatory authorities may be accessible. While a number of uses may be proposed, it may not be
practical and/or reasonable to attempt to collect residue data on all crops in order to refine risk estimates.
While the data used in residue studies are typically used to refine screening-level RQ values, test data can
be useful in putting colony-level data into context particularly when similar data are collected during the
conduct of semi- and full-field level studies.
Based on the property of chemicals of interest, proposed uses, and available lines of evidence, a tiered
approach should be adopted in determining what effect data are necessary in order to conduct an appropriate
risk assessment for pollinators. Potential exposure routes and exposure duration, as well as level of
exposure should be considered to determine the types of studies required. For example, if multiple
exposures are expected, a chronic study may be needed. Higher tier studies are required when risk concerns
cannot be addressed at lower tiers. In addition and if possible, higher tier studies should be designed and
conducted to address specific risk concerns identified at lower tiers or through other available information
(see
Section 3.2.1 for details for data requirement).
As noted in this guidance, the need for additional data and the order of data recommendations are not
dictated by a regimented sequence. Rather, multiple lines of evidence should be considered in conjunction
with an understanding of risk management needs. An effort should be made to determine whether data are
available for similarly structured chemicals with a common mode of action; these data may be able to
address some uncertainties and reduce the need for additional testing. Similarly though, data for other
compounds may inform decisions to deviate from the typical sequence of studies to recommend special
studies to address particular uncertainties. At each step of the process, potential risk mitigation measures
40
and management options should be considered that may reduce the need for additional studies since the
likelihood of exposure has been sufficiently mitigated.
41
Appendix 1. Conceptual Models
A1.1 Non‐systemic, Foliar Spray Applications
For non-systemic pesticides applied via foliar spray, dominant exposure routes of foraging bees include
direct deposition of spray droplets onto bees, deposition onto plant surfaces (leaf, flower, pollen, nectar,
extra-floral nectaries) followed by contact and/or ingestion, and inhalation of gaseous phase chemical (for
highly volatile pesticides;
Figure A1.1).
Figure A1.1 Generic Conceptual Model of Non-Systemic, Foliar-Applied Pesticides for Honey Bee Risk
Assessment. Dashed lines represent routes of exposure that are not considered to be major.
42
A1.2. Systemic, Foliar Spray Applications
Foliar applications of systemic pesticides are likely to result in many of the same routes of exposure to
honey bees
as described previously for non-systemic, foliar-applied pesticides, with several important
exceptions. First, deposition onto plant surfaces and soil will lead to translocation of the pesticide to other
plant tissues, potentially contributing to higher quantities of pesticide residues in pollen and nectar. For
persistent systemic pesticides, the exposure window could include longer periods of time (red arrows,
Figure A1.2) compared to similar applications of non-systemic pesticides. Second, pesticide residues in
plant exudates (guttation fluid, honey dew) also become a potentially relevant route of exposure.
Figure A1.2. Generic Conceptual Model of Systemic, Foliar-Applied Pesticides for Honey Bee Risk Assessment.
Red depicts systemic pathways. Dashed lines represent routes of exposure that are not considered to be major.
43
A1.3 Systemic, Seed Treatment
Major exposure routes of honey bees to systemic pesticides used as seed treatments include pollen, nectar,
exudates (
e.g., guttation fluid), and honey dew resulting from translocation from the seed to growing plant
tissues
(Figure A1.3). Another important route of exposure includes contact with abraded seed coat dust
during planting.
Figure A1.3. Generic Conceptual Model of Systemic, Seed Treatment-Applied Pesticides for Honey Bee Risk
Assessment. Red depicts systemic pathways. Dashed lines represent routes of exposure that are not considered
to be major.
44
A1.4 Systemic, Soil Application
Systemic pesticides are also applied soil applications (
Figure A1.4). Exposure of honey bees to pesticides
via these applications are expected to result primarily from translocation to plant tissues (pollen, nectar,
exudates, and honey dew). For soil applications, there is potential exposure via runoff and subsequent
translocation into plants adjacent to the treated field.
Figure A1.4. Conceptual Model of Soil-Applied Systemic Pesticides for Honey Bee Risk Assessment. Red
depicts systemic pathways. Dashed lines represent routes of exposure that are not considered to be major.
45
Appendix 2. Considerations related to quantifying residues of pesticides
in pollen and nectar using pesticide‐specific studies
Empirical data may be used to refine conservative assumptions and reduce uncertainty associated with the
Tier I exposure assessment by providing direct measurements of pesticide concentrations resulting from
actual use settings. Studies investigating pesticide concentrations in pollen and nectar should be designed
to provide residue data for crops and application methods of concern.
It is preferable that residues are measured from pollen and nectar samples collected directly from a crop.
Otherwise they may be measured from pollen and nectar samples taken from bees that have collected the
samples (
e.g., during Tier II tunnel studies). Measures should be taken to avoid degradation during
sampling. The use of honey bees to collect samples may be desirable for crops that have flower structures
that complicate the direct collection of pollen and nectar by humans. However, when honey bees are used
for sampling pollen/nectar, caution should be taken to ensure that the residue data are representative of the
target crop; otherwise, residue data may underestimate the target crop exposure. Chemical degradation in
pollen and nectar may occur on/in bees after being collected. Bees should be limited from collecting pollen
and nectar from sources other than the crop of interest.
Additional considerations may include the collection of vegetative portions of the treated crop ensuring the
peak residue concentration is captured. These data, taken over a period of time, may reflect the
accumulation/depuration curve of the chemical and may provide information about the transport and
potential duration/concentration measures in the plant.
It is not necessary to have residue data for each crop for which a pesticide is registered/proposed. For
refinement purposes, reliance on data for a select number of crops that adequately represent the diversity
pollinator-attractive crops and registered uses is typically considered sufficient. In this approach, individual
crops are used as surrogates for other crops. In selecting the surrogate crops, the nature of the crop (
i.e.,
whether it blooms within a specified (determinant) period or is indeterminant), its attractiveness to bees,
application method, and site selection should be considered for selecting exposures that would yield a
reasonably conservative estimate for exposure through pollen and nectar (
e.g., higher magnitude, duration
and spatial extent of residues). Regarding the number and nature of residue studies to recommend, risk
assessors should consider how these data will be used in the risk assessment. To the extent possible, studies
should focus on those scenarios that are believed to represent high-end exposure potential (application
method/crop/soil/location). As with any study, consideration must be not only be given to the resources
involved in conducting the study but also in reviewing the studies, as well as its contribution to the entire
risk assessment.
The following considerations are relevant to evaluating the breadth and design of residue study protocols:
-Attractiveness of the studied crop to bees
-Spatial extent of the crop (
e.g., expected use acreage)
-Duration and timing of blooming period in relation to bee pollination/foraging
-Pesticide application rates, methods and timing
-Influence of soil factors and agronomic practices on residues in the crop
-Influence of soil hydric/meteorological/and transpiration conditions on crop residues
-Diversity of crop biology and physiology within a crop group, and
-The temporal nature of pesticide residues in pollen and nectar (
e.g., time to peak residue occurrence
and the potential for season to season accumulation of pesticide residues)
Although the aforementioned list of considerations is substantial, it most likely will not be necessary to
quantify the effect of all such factors on residues in each surrogate crop. Insight into the nature of pesticide
46
residues in crops may be obtained from magnitude of residue, plant metabolism and rotational crop studies
conducted for human health assessment. Furthermore, crop residue studies can be structured so that results
from earlier studies can help inform the design and need for subsequent studies.
47
Appendix 3. Bee REX
The Bee-REX model is a screening level tool that is intended for use in a Tier I risk assessment to assess
exposures of bees to pesticides and to calculate risk quotients. This model is individual-based, and is not
intended to assess exposures and effects at the colony-level (
i.e., for honey bees).
The Tier I exposure method is intended to account for the major routes of pesticide exposure that are
relevant to bees (
i.e., through diet and contact). Exposure routes for bees differ based on application type.
In the model, bees foraging in a field treated with a pesticide through foliar spray could potentially be
exposed to the pesticide through direct spray as well through consuming contaminated food. For honey
bees foraging in fields treated with a pesticide through direct application to soil (
e.g., drip irrigation),
through seed treatments, or through tree injection, direct spray onto bees is not expected. For these
application methods, pesticide exposure through consumption of residues in nectar and pollen are expected
to be the dominant routes. Foraging honey bees may also be exposed to pesticides via contact with dust
from seed treatments or via consumption of water from surface water, puddles, dew droplet formation on
leaves and guttation fluid; however, the Bee-REX tool does not include quantification of exposures via
these routes.
Appendix 3, Table 1 summarizes the exposure and effect estimates used in developing the Tier 1 screening-
level risk quotients (RQs) for individual adult bees and larvae for foliar spray applications, soil applications,
seed treatments and tree trunk applications. The appropriate exposure values and effect endpoints would be
used to derive these RQs. The resulting RQ values would then be compared to the LOCs for acute and
chronic exposures, which are 0.4 and 1.0, respectively.
48
Appendix 3, Table 1. Summary of exposure and effect estimates used in deriving risk quotients for
Tier I risk assessments.
Measurement
Exposure
Acute Effect
Chronic Effect
Exposure Estimate+
Endpoint
Route
Endpoint
Endpoint+++
Foliar Applications
Individual
AR
Acute contact
Survival
Contact
English* (2.7
µg a.i./bee)
None
AR
LD
(adults)
Metric* (2.4
µg a.i./bee)
50
AR
Chronic adult oral
Individual
English * (110
µg a.i /g) (0.292
g/day)
Acute oral
NOAEL
Survival
Diet
AR
LD
(effects to survival or
(adults)
Metric * (98
µg a.i /g) (0.292
50
g/day)
longevity)
AREnglish * (110
µg a.i /g) (0.124
Chronic larval oral
Brood size and
g/day)
Diet
Larval LD
NOAEL (effects to adult
success
50
ARMetric *(98
µg a.i /g) (0.124
emergence, survival,)
g/day)
Soil Treatments
Chronic adult oral
Individual
Acute oral
NOAEL
Survival
Diet
(Briggs EEC) * (0.292 g/day)
LD
(effects to survival or
(adults)
50
longevity)
Chronic larval oral
Brood size and
Diet
(Briggs EEC) * (0.124 g/day)
NOAEL (effects to adult
success
Larval LD50
emergence, survival,)
Seed Treatments
Chronic adult oral
Individual
Acute oral
NOAEL
Survival
Diet (1
µg a.i /g) * (0.292 g/day)
LD
(effects to survival or
(adults)
50
longevity)
Chronic larval oral
Brood size and
Diet (1
µg a.i /g) * (0.124 g/day)
NOAEL (effects to adult
success
Larval LD50
emergence, survival,)
Tree Trunk Applications++
Chronic adult oral
Individual
(
µg a.i. applied to tree/g of foliage) Acute oral
NOAEL
Survival
Diet
* (0.292 g/day)
LD
(effects to survival or
(adults)
50
longevity)
Chronic larval oral
Brood size and
(
µg a.i. applied to tree/g of foliage)
Diet
Larval LD
NOAEL (effects to adult
success
* (0.124 g/day)
50
emergence, survival,)
AREnglish = application rate in lbs a.i./A; ARMetric = application rate in kg a.i./ha
+Based on food consumption rates for larvae (0.124 g/day) and adult (0.292 g/day) worker bees and concentration in pollen and nectar.
++Note that concentration estimates for tree applications are specific to the type and age of the crop to which the chemical is applied.
+++To calculate RQs for chronic effect, NOAEC can be used as the effect endpoint to compare with the exposure estimate in concentration.
49
1.1. Toxicity inputs
Estimated exposure concentrations are integrated with available toxicity data in order to characterize risks
of a pesticide to honey bees. In doing so, Tier I estimated exposures and toxicity endpoints are compared
based on the same exposure routes. For instance, estimated exposures through direct spray onto foraging
bees are combined with toxicity endpoints from contact toxicity test, while estimated dietary exposures are
matched with oral toxicity data.
For acute exposures, the endpoints are LD50 values for adult and larval bees exposed to a single dose of a
pesticide and observed for several days. For adult bees, the contact LD50 value should be derived from a
study conducted in a manner consistent with guideline 850.202027 or OECD 214. The oral exposure
endpoint for adult bees should be from a study that is similar to the OECD 21328. For larvae, the acute oral
LD50 should be from a study that is conducted in a manner consistent with OECD 23729. A larval chronic
study is under development and endpoint for chronic toxicity should be generated in a manner consistent
with the potential guidance.
At this time, standard guideline methods are not available for laboratory-based chronic tests of bees. If a
scientifically-valid no-observed adverse effect level (NOAEL) or concentration (NOEAC) from a chronic
study with adult bees or larvae is available, the model user can enter these into the Bee-REX tool in order
to calculate chronic RQs. It should be noted that although it may be possible to extract both a NOAEL or
NOEAC and an ECx or LCx value from the laboratory-based studies, it is important to note differences in
hypothesis-based test designs needed to reliably support NOAEL/NOAEC versus regression-based study
designs needed to support ECx and LCx estimates. When a NOAEC, or ECx or LCx are used as endpoints,
the exposure estimates in concentrations should be used. In this case, RQs can be generated outside of the
BeeREX tool by dividing the concentration based estimates of exposure by the appropriate endpoint.
It should be noted that the tool uses toxicity data for worker bees to calculate RQ values for queens and
drones. There is uncertainty in using worker bees as surrogates for the royal cast due to the size differences
in the bees (
i.e., queens and drone adults are larger than workers).
1.2. Food consumption rates
As discussed in the effects characterization, oral toxicity data are necessary for adult and larvae in order to
characterize the risks of a pesticide. It is important that both exposure and toxicity endpoints are in the same
unit for RQ calculation; otherwise unit conversion must be conducted. When toxicity endpoints are
expressed as concentration (
e.g., NOAEC), exposure estimates should be in the same concentration units,
and conversion using food consumption rate is not needed.
When toxicity endpoints are expressed on a dose basis (
i.e., μg a.i./bee), it is necessary to convert estimated
concentrations of pesticides in food (expressed as mg a.i./kg) into doses. The major nutritional requirements
for honey bees are met through consumption of nectar, honey, pollen and bee bread as well as royal jelly
and brood food. Pesticide doses received by bees can be calculated using nectar and pollen consumption
rates for larval and adult worker bees. For larvae, the proposed total food consumption rate is 124 mg/day,
which is based on the total daily consumption of pollen and nectar by larvae during the last days in the life
27 USEPA. 2012. Ecological Effects Test Guidelines. OCSPP 850.3020 Honey Bee Acute Contact Toxicity. EPA 712-C-019. January 2012.
http://www.regulations.gov/#!documentDetail;D=EPA-HQ-OPPT-2009-0154-0016
28 OECD. 1998. OECD Guidelines for the Testing of Chemicals. Test Number 213: Honey bees, Acute Oral Toxicity Test. http://www.oecd-
ilibrary.org/environment/test-no-213-honey bees-acute-oral-toxicity-test_9789264070165-en;jsessionid=5p2ngklfmv8p4.epsilon
29 OECD. 2013. OECD Guidelines for the Testing of Chemicals. Test Number 237: Honey bee (Apis mellifera) larval toxicity test, single exposure.
http://www.oecd-ilibrary.org/environment/test-no-237-honey-bee-apis-mellifera-larval-toxicity-test-single-exposure_9789264203723-en
50
stage. For adult worker bees, the proposed food consumption rate is 292 mg/day, based on nectar
consumption rates of nectar foraging bees, which are expected to receive the highest dietary exposures
among different types of worker bees. In addition, it is likely that these food consumption rates are
protective of drones and queens.
The Bee-REX tool calculates dietary exposure values for larvae of different ages, adult workers with
different tasks (and associated energetic requirements) and royal castes. The most conservative RQ is
selected for the Tier I screen. All of the RQs can be used for risk characterization purposes. Those food
consumption rates, which are provided in
Appendix 3, Table 2, are based on work described in Appendix
1 of USEPA, PMRA and CADPR 201230 and updated by Garber (unpublished)31.
The tool also has the ability to override estimated concentrations with empirical data for pesticide
concentrations in pollen, nectar, jelly and bee bread. When pesticide concentrations in pollen are an order
of magnitude greater or more than in nectar, the most conservative adult may be the nurse bee, which
consumes 1.3-12 mg pollen/day, combined with 113-167 mg nectar/day (
Appendix 3, Table 2). The
BeeREX tool uses food consumption rates that are meant to represent typical bees within a worker group
or royal caste. For nurse bees, the food consumption rates of 9.6 mg pollen/day and 140 mg nectar/day were
used to represent central tendencies in food consumption. The pollen consumption rate reflects the highest
average daily consumption rate measured for nurse bees given the limited availability of data (only two
empirical studies were located). If the refined RQ values (derived using empirical concentration data) are
close to the LOC, the risk assessor may explore the influence in the variability in concentrations and in food
consumption rates on risk conclusions (
e.g., by bounding the exposure using high and low food
consumption values combined with high and low pesticide concentrations).
30 USEPA, PMRA, and CADPR (2012) White paper in support of the proposed risk assessment process for bees. United States Environmental
Protection Agency, Office of Pesticide Programs, Washington DC. Pest Management Regulatory Agency, Health Canada, Ottawa. California
Department of Pesticide Regulation, Sacramento, CA.
31 Garber, K.V. manuscript in preparation: Estimation of food consumption rates of honey bees (
Apis mellifera) by caste and worker task for use in
assessing dietary risks of pesticides.
51
Appendix 3, Table 2. Estimated food consumption rates of bees.
Life
Average age
Daily consumption rate (mg/day)
Caste (task in hivea)
Stage
(in days)a
Jelly
Nectarb
Pollen
Total
1 1.9
0 0 1.9
2 9.4
0 0 9.4
Worker
3 19
0 0 19
4 0
60 c 1.8d 62
5 0
120 c 3.6d 124
Larval
Drone 6+
0
130
3.6
134
1 1.9
0 0 1.9
2 9.4
0 0 9.4
Queen
3 23
0 0 23
4+ 141
0 0 141
Worker (cell cleaning and capping)
0-10
0
60f
1.3 - 12
g,h
61 - 72
Worker (brood and queen tending, nurse bees)
6-17
0
113 - 167f
1.3 - 12
g,h
114 - 179
Worker (comb building, cleaning and food handling)
11-18
0
60f 1.7
g 62
Worker (foraging for pollen)
>18
0
35 - 52f 0.041
g
35 - 52
Adult
292
Worker (foraging for nectar)
>18
0
0.041
g 292
(median)c
Worker (maintenance of hive in winter)
0-90
0
29f
2
g 31
Drone
>10
0
133 - 337 c 0.0002
c
133 - 337
Queen (laying 1500 eggs/day)
Entire lifestage
525
0
0
525
aWinston (1987)
bConsumption of honey is converted to nectar-equivalents using sugar contents of honey and nectar.
c Calculated as described in this paper.
dSimpson (1955) and Babendreier
et al. (2004)
e Pollen consumption rates for drone larvae are unknown. Pollen consumption rates for worker larvae are used as a surrogate.
fBased on sugar consumption rates of Rortais
et al. (2005). Assumes that average sugar content of nectar is 30%.
gCrailsheim
et al. (1992, 1993)
hPain and Maugenet (1966)
52
In the proposed approach for representing food consumption rates of larvae and adults, it is assumed that
exposures through consumption of nectar and pollen are conservative representations of potential exposures
through consumption of honey and bee bread, respectively. This approach is likely to be conservative
because it assumes that pesticides do not degrade while honey and bee bread are stored in the hive. For bees
that consume honey, it is assumed that the estimated pesticide exposures can be related back to the original
concentration in nectar by accounting for the amount of sugar consumed by bees. It is also assumed that
pollen and nectar consumption rates and resulting exposures are protective of exposures of bees to
pesticides through consumption of royal jelly and brood food. This is supported by work by Davis and
Shuel 1988 32 and Kamel
et al. (unpublished) 33 demonstrating that pesticide concentrations in food
consumed by nurse bees are 2-4 orders of magnitude higher than concentrations measured in royal jelly.
Based on these observations, for calculating RQ for queens, concentrations in royal jelly are 100X less
compared to the residue in nectar or pollen, whichever is greater.
1.3. Estimating pesticide concentrations in pollen and nectar
In an ideal situation, the Tier I exposure estimates for honey bees would be based on residue values
measured directly in nectar and pollen of flowers sprayed with pesticides. This cannot be achieved at this
time because there is an insufficient amount of data that may be used to adequately describe the distribution
of pesticide residues that occur in pollen and nectar relative to pesticide application rate. As an alternative,
the proposed method relies on upper-bound pesticide residue values plant leaves, which will serve as a
surrogate for pollen and nectar.
Methods for estimating dietary exposures to bees differ in the nature of the estimated concentrations in
pollen and nectar. For foliar spray applications, the approach involves the use of the tall grass residue value
from the T-REX model (v. 1.5) as a surrogate for pesticide concentrations in nectar and pollen. For soil
treatments, the method is based on a modification to a plant-soil uptake model developed by Briggs
et al.
1982 and 1983, which is designed to estimate pesticide concentrations in plant shoots; the concentrations
in plant shoots are proposed as a surrogate for concentrations in pollen and nectar (following systemic
transport). For seed treatments, the Tier I exposure method is based on the International Commission for
Plant-Bee Relationships’ (ICP-BR) 1 mg a.i./kg concentration to represent an upper-bound concentration
in nectar and pollen. For tree injections and trunk drenches, the method is a simplistic approach that
considers the mass of the pesticide applied to a tree and the mass of the leaves of the tree.
At the Tier I level, both acute and chronic exposure estimates are represented by the highest single day
exposure value. Although a time weighted estimated exposure value may be more representative of the
exposure used in a chronic toxicity test, exposure occurring over at the highest residue estimate could
potentially be sufficient to elicit effects while assume no degradation occurs. Therefore, in the Tier I
approach, chronic exposure is conservatively represented by the highest single day estimated exposure.
1.3.1. Foliar sprays
For dietary exposure, the method includes the use of the upper-bound residue for tall grass used in the T-
REX model (i.e., 110 ug a.i./g per 1 lb a.i./A or 98 ug a.i./g per 1 kg a.i./ha) as a surrogate for pesticide
concentrations in pollen and nectar of flowers that are directly sprayed with a pesticide. This exposure
32 Davis, A.R. and Shuel, R.W. (1988) Distribution of C(14)-Labeled carbofuran and dimethoate in royal jelly, queen larvae and nurse honeybees.
Apicologie, 19 (1), 37-50.
33 Kamel, A.; Dively, G.; Hawthorn, D. and Pettis, J. Unpublished research examining the fate of imidacloprid and its metabolites in honey bee
hives.
53
concentration would then be converted to a dietary dose received by adult and larval worker bees using
pollen and nectar consumption rates for these two life stages, which are 0.292 and 0.124 g/day, respectively.
In order to quantify contact exposures due to direct spray, the proposed upper bound value is 2.7
μg a.i./bee
per 1 lb a.i./A, or 2.4
μg a.i./bee per 1 kg a.i./ha, based on data published by Koch and Weisser (1997). As
with the dietary exposure, the contact exposure value can be adjusted to account for application rate.
1.3.2. Soil treatments
For soil treatments, it is assumed that bees will be exposed via dietary consumption of pollen and nectar
that are contaminated as a result of systemic transport of pesticides from soil. For these application types,
it is assumed that honey bees will not be directly exposed through contact because they are not expected to
be present on the surface of the soil. The method for estimating dietary exposures to bees resulting from
soil treatments is based on an empirically based model developed by Briggs
et al. 1982 and 1983, with
modifications (referred to here as “the Briggs’ Model”). This model relates the Log Kow of a chemical to its
concentration in plant shoots, which can be used as a surrogate for concentrations in nectar and in pollen).
In comparison to the dietary-based exposure values proposed for foliar spray applications (
i.e., based on
the tall grass upper-bound value), the Briggs’ model generates exposure values that are usually two orders
of magnitude lower. In using the Briggs’ model, the approach is to use Equation 1, with the 95th percentile
TSCF value that is specific to an assessed pesticide’s Log Kow (calculated using Equation 2). It is assumed
that the resulting value is equivalent to pesticide concentrations in pollen and nectar of crops receiving soil
treatments of the pesticide. The estimated concentration in pollen and nectar may be converted to dietary-
based exposures for adult and larval bees using the consumption rates for pollen and nectar (
i.e., 292 and
120 mg/day, respectively).
Equation 1.
10 . ∗
.
0.82 ∗
∗
∗
∗
∗
Where: Cstem = concentration in stems (
µg a.i./g plant)
Csoil
= concentration in soil (
µg a.i./g soil)
foc
= fraction of organic carbon in soil
θ
= soil-water content by volume (cm3/cm3)
ρ
= soil bulk density (g-dw/cm3)
Koc
= soil organic carbon-water partitioning coefficient (cm3/g-oc or L/kg-oc)
TSCF =Transpiration Stream Concentration Factor
Equation 2.
0.0648 ∗
0.241 ∗
0.5822
For the parameters in
Equation 1 that define the properties of the soil, conservative values were chosen to
maximize the concentration in soil pore water and thus maximize the amount of chemical available for
uptake into stems. Values were chosen to be consistent with standard scenarios used to run PRZM. For the
fraction of organic carbon in soil (foc), a value of 0.01 is used. A value of 1.5 g-dw/cm3 is selected to
represent bulk density (ρ). Soil water content (θ) is set to 0.2 cm3/cm3.
Equation 1 can be used to calculate
the concentration of a chemical in stems using the parameter values above and the application rate. If it is
assumed that this application rate is homogenously distributed throughout the top 6 inches (15 cm) of the
treated soil (based on the assumption that the majority of the pesticide will remain in this portion of the
soil), a rate of 1 lb a.i./A is equivalent to 0.50
µg a.i./g-soil (using the bulk density), a rate of 1 kg a.i./ha is
equivalent to 0.45
µg a.i./g-soil (using the bulk density).
54
In the Tier I approach, it is assumed that all chemicals may be systemically transported. This assumption
may be limited based on Log Kow (Ryan
et al. 1988). Therefore, it could be limited to the bounds of the
Log Kow values used in the empirical data set generated by the Briggs’ model (
i.e., for chemicals with Log
Kow<5; which is the bound on the data set used by Briggs). Whether a chemical is transported systemically
in plants could potentially be confirmed using empirical data submitted to EPA and PMRA (
e.g., plant
metabolism studies); however, it would be up to pesticide registrants to submit sufficient data to
demonstrate that a pesticide is not systemic.
There are five notable limitations to using the modified Briggs’ model approach. The first is that this
methodology is based on empirical data from only one type of plant. The second limitation is that the data
set used to derive Equation 1 is based on a limited number of chemicals that represent only two classes of
pesticides. The third limitation is that this approach is based on data from non-ionic organic chemicals and
may have limited utility for ionic chemicals that whose transport may not be predicted well using Kow and
Koc. The fourth limitation of the Briggs’ model is that it is based on passive transport of chemicals into
xylem, therefore, this approach does not directly estimate pesticide concentrations in plants that are the
result of phloem transport. The fifth limitation involves the use of estimated pesticide concentrations in
vegetative plant matrix (
i.e., shoots) as a surrogate for nectar and pollen.
At this time, the modified Briggs method represents the best available approach for estimating pesticide
concentrations in pollen and nectar of crops on soil treated with pesticides. The use of this approach was
supported by the FIFRA SAP. As more data become available (
e.g., measured pesticide residues in pollen
and nectar from crops on treated soil), EPA, PMRA and CADPR will reevaluate the modified Briggs
method.
1.3.3. Seed treatments
The Tier I exposure method for seed treatments is based on the EPPO 2010 screening value of 1 mg a.i./kg
in pollen and nectar of plants to which seed treatments were made. For these application types, it is assumed
that bees will only be exposed through the diet. In the Tier I approach, it is assumed that all pesticides that
are applied to seeds are systemic, and therefore can be transported into pollen and nectar that may be
consumed by honey bees. This approach may be used for all pesticides that are applied via seed treatment,
by assuming that the upper-bound exposure value for pollen and nectar is 1 mg a.i./kg. This value can be
multiplied by nectar and pollen consumption for adult and larval worker bees to determine the upper-bound
doses potentially received by bees. This method can be applied to all chemicals that are applied to seeds,
with no need for adjustment based on application rate or chemical properties.
1.3.4. Tree trunk applications
The method involves estimating the concentration in the vegetative part of the treated tree (excluding woody
parts) by dividing the mass of the pesticide applied to the tree by the mass of tree vegetation represented
primarily by leaves, but also by flowers. The application rate should be entered as the total mass of pesticide
applied per tree.
The mass of the leaves of the tree must be estimated based on the expected characteristics of the tree to
which the pesticide may be applied. It is expected that the mass of leaves and flowers of a tree will vary by
species and by age of the tree, therefore a standard equation is not provided here. There are various sources
available through government publications, extension offices and scientific literature that provide equations
55
for estimating the mass of tree leaves. For example, Jenkins
et al. 200334 provides equations for various tree
species that may be used to estimate the mass of foliage for hard and softwood species. Alva
et al. 200335
provide data on masses of citrus tree leaves.
This mass of leaves should be entered as a wet weight value. In some cases, it may be necessary to convert
from dry weight to wet weight. In that case a standard assumption of 80% water content in leaves may be
used (this is consistent with the T-REX approach for grass and leaves).
This approach assumes that the applied pesticide is homogenously distributed in the tree’s leaves and
flowers and is not present in other parts of the tree. One major uncertainty associated with this approach is
the estimate of the vegetative mass of the tree, which can vary greatly based on species, age, time of the
year and geography. In addition, in this approach, it is assumed that 100% of the active ingredient is taken
up into the tree and moved into the leaves and flowers. It is unlikely that all of the pesticide mass applied
to the tree actually enters solely to the leaves.
1.4. Using empirical exposure data
If empirical data are available for a chemical, these data may be entered into the Bee-REX tool to calculate
RQs. User-entered empirical exposure values will automatically over-ride estimated values. This can be
accomplished by selecting “yes” to the question “Are empirical residue data available?” Once that is done,
the user should enter the available values for pollen, nectar and jelly.
34 Jenkins, J., D. Chojinacky, L. Heath, and R. Birdsey. 2003. Comprehensive database of diameter-based biomass regressions for North American
tree species. General Technical Report NE-319, Forest Service, United States Department of Agriculture.
35 Alva, A.K.; Fares, A. and H. Dou. 2003. Managing citrus trees to optimize dry mass and nutrient partitioning. Journal of Plant Nutrition, 26 (8):
1541-1559.
56
Appendix 4. Tier 3 Field Study Design Considerations
Tier III studies conducted under full-field conditions where bees are free foraging are intended to address
specific uncertainties/risks that have been identified in lower-tier studies. The design of these studies will
depend on the specific questions that need to be answered; therefore, it is not possible to define a single
study design or specific design elements that must be incorporated into every full-field study. Below are
elements that the risk assessor should consider; however, these are not intended to be prescriptive. It is
incumbent on the review team to ultimately identify the study design elements that should be considered
by the pesticide registrant/applicant in developing a study protocol that is responsive to the Tier III study
requirement. It should also be noted that the data package submitted by the registrant may already include
a series of bee studies spanning multiple tiers; it such cases, the reviewer should ensure that the studies
address the uncertainties identified by review team to determine whether additional studies may be
necessary.
Chronic Field Pollinator Study Design Considerations
Application Conditions
Maximum application rate
Minimum reapplication interval
Maximum number of applications
Use of formulated end-product
Application method
o Foliar
o Soil treatment
o Seed treatment
o Combination
Suitable weather
o Avoid applications when rain and/or high winds are predicted.
Season
Test crop
Attractive to test bees
Long bloom period to address concerns identified at lower tiers
Large area to ensure majority of foraging on test crop
Follow standard [local] agriculture practices
Colonies
Package bees/new equipment to limit incidence of disease; if older colonies are used, they should
be as pest/disease free as possible.
o Colonies should not be used if they have received any chemical treatments within last 4
weeks. Colonies suspected of having American foulbrood should not be utilized. Other
disease treatments should be reported.
o Beekeeper standard practice for maintain colony health during study
All treatments must be uniform across study colonies.
Queen-right (healthy queen present); sister queens for each replicate.
Acclimation period: 2 months minimum to establish representative age distribution in newly
established hives;
57
Homogeneous colony strength, brood pattern as close as possible
If existing colonies are to be used; broad spectrum residue analysis in hive products (honey/nectar,
pollen, wax); must document low incidence of diseases/parasites.
Size of the colonies may vary depending on the focus of the study and when the study is initiated.
Typically, each hive should at least 10,000 bees to cover 10 frames and include at least 5 brood
frames. Excessive food storage should be avoided.
Colonies can be positioned in plots when test crops are blooming enough to minimize test bees
foraging on plants other than the test crop,
e.g., 20-25% bloom
Study Design Considerations
Historically there has been difficulty in controlling the extent to which the free-foraging bees utilize the
treated crop or that treatment groups cross-over. Sufficiently large field plots, if feasible, will overcome
the cross-over issue between plots and problems of ensuring exposure to treated crops due to competing
vegetation.
Mean honey bee foraging distance 1.5 – 3 km with extreme distances of 10 km; average surface
area range 7 – 100 km2 (Medrzycki
et al. 201336).
o EFSA 201337 recommends minimum of 2 ha to provide sufficient flowers and support
exclusive foraging; Medrzycki
et al. 2013 recommends minimum of 5 ha.
Suitable crop that is representative of actual use; good source of both pollen and nectar, (
e.g.,
phacelia (
Phacelia tanacetifolia), oilseed rape, mustard, buckwheat)
o Pollinator-attractive
Account for crops/alternative forage within 3 km of colonies.
Distance of treated crop from other nectar producing plants is essential to ensure exposure and must be
documented.
Pollen traps should be used to demonstrate extent to which bees have foraged on treated crop.
Pollen identification (palynological analysis) may be used to ensure origin of pollen
Pollen/nectar collection for residue analyses
o Collected by bees and sampled using pollen traps (corbicular pollen)
o Collected directly from plants
o Sampling of nectar forager honey stomachs
o Sampling comb pollen/nectar
Minimum number of replicate colonies: 6 - 10 per treatment (Medrzycki
et al. 2013); the number of
replicates per treatment will depend on the targeted magnitude of effects and desired statistical power.
Study duration should assess at least two brood cycles (42 days) to ensure brood is exposed to residues
stored in the colony (EFSA 2013).
Measurement Endpoints: depend on the risk hypothesis tested and the nature of uncertainties identified in
lower-tier tests. Possible measurement endpoints may include.
Adult Forage Bees
36 Medrzycki, P. H. Giffard, P. Aupinel, L. P. Belzunces, M-P. Chauzat, C. Classen, M. E. Colin, T. Dupont, V. Girolami, R. Johnson, Y.
LeConte, J. Lückmann, M. Marzaro, J. Pistorius, C. Porrini, A. Schur, F. Sgolastra, N. S. Delso, J van der Steen, K. Wallner, C. Alaux, D. G.
Biron, N. Blot, G. Bogo, J-L Burnet, F. Delbac, M. Diogon, H. El Alaoui, B. Provost, S. Tosi and C. Vidau. 2013. Standard methods for
toxicology research in Apis mellifera. Journal of Apicultural Research 52(4): http://dx.doi.org/10.3896/IBRA.1.52.4.14
37 EFSA..Guidance on the risk assessment of plant protection products on bees (
Apis mellifera,
Bombus spp. and solitary bees. EFSA Journal
2013;11(7):3295, 266 pp. doi:10.2903/j.efsa.2013.3295. Available online: www.efsa.europa.eu/efsajournal
58
Adult bee survival/longevity
Adult bee foraging activity (visual counts of returning foragers; mark-and-recapture; calibrate Dead
Bee Dead Zone traps)
Queen status over the course of the exposure
Colony health (disease/pest incidence)
Colony Strength
Brood (quantify number of eggs; larvae, capped cells, pollen, honey/nectar cells)
Monitoring of brood in a minimum of two staggered cohorts, mid-way and late in the exposure
period
Adult longevity: measured by using 30 newly-emerged adult bees from each colony (minimum n=6
colonies/treatment) in a controlled laboratory cage experiment monitoring daily mortality.
Newly-emerged bee weights
Hive weight
Other potential endpoints
Overwintering Success
Fitness measure: Pathogen challenge (
e.g., Nosema exposure) to newly emerged bees
Assess the ability of colonies to re-queen themselves by removing all queens and determining the
success of each colony in rearing a replacement queen.
Documenting Exposure
Residue analysis in pollen/nectar collected from crop
Residue analyses in pollen/nectar collected from bees
Residue analysis in pollen/nectar/honey collected from hive
Residue analyses in bee carcasses
Residue analyses in wax
Foliar Residue analysis
Measure total residues of concern (parent + degradate(s))
Suitable control bees (residue analyses to demonstrate lack of exposure). Utility of mark-and-recapture to
document drift of bees from treated colonies.
59
