TECHNICAL DESCRIPTION
AND USER'S GUIDANCE DOCUMENT
FOR THE
TERRESTRIAL INVESTIGATION MODEL (TIM)
Environmental Fate and Effects Division
Office of Pesticide Programs
Office of Chemical Safety and Pollution Prevention
U.S. Environmental Protection Agency
Washington, DC
Version 3.0 BETA
March 25, 2015
Page 1 of 77
-------
Acknowledgements
US EPA Terrestrial Investigation Model Workgroup
Edward Odenkirchen, Ph.D., Chair, Office of Pesticide Programs (OPP), Environmental Fate and
Effects Division (EFED)
Brian Anderson, M.E.M., OPP, EFED
Timothy Barry, Sc.D., National Center for Environmental Economics, Office of the
Administrator
Kristina Garber, M.S., OPP, EFED
Nick Mastrota, Ph.D., OPP, EFED
Former Refined Risk Assessment Team Members
Ingrid Sunzenauer, M.S., Retired (Former chair, Refined Risk Assessment Implementation
Team)
Edward Fite, M.S., Retired (Former chair, Terrestrial Investigation Model Workgroup)
Kathryn Gallagher, Ph.D., Office of Water (Former chair, Surface Water Assessment Model
Workgroup)
Christopher Salice, Ph.D., Texas Tech University (Former member of TIM Workgroup)
Government Contractors (ARCADIS US, Inc.)
Philip Goodrum, Ph.D.
Tim Negley, M.S.
Carolyn Meyer, Ph.D.
Mike Chase, B.S.
Jane Staveley, M.S.P.H.
The TIM workgroup would like to thank Matthew Etterson for developing the Graphical User
Interface for the current version of TIM. The TIM Workgroup would also like to acknowledge
the contributions of Chuck Peck and Faruque Khan in developing the spray drift exposure
methodology. In addition, the workgroup appreciates the contributions from scientists in EFED,
and the MCnest team from the Office of Research and Development (Matthew Etterson and
Richard Bennett) who have improved this document through peer review, data analysis, and
quality assurance/quality control efforts.
Page 2 of 77
-------
Table of Contents
1. Introduction 6
1.1. Purpose of Document 6
1.2. When to Use TIM 6
1.3. Overview of Terrestrial Investigation Model (v.3.0) 7
1.4. The Evolution of TIM 10
1.5. Model Executable 13
1.6. Quality Assurance and Quality Control 16
1.7. Organization of This Document 16
2. Avian Species 17
2.1. Diet Composition 17
2.2. Body Weight (BW) 18
2.2.1. Adults 18
2.2.2. Juveniles 19
2.3. Home Range Size 19
2.4. Frequency on Field and Residency Status 20
2.5. Fidelity Factor 21
2.6. Taxonomy 21
2.7. Species 21
2.7.1. Generic 21
2.7.2. Custom 22
3. Modeling Bird Behavior: Feeding and Location 30
3.1. Bimodal Feeding Model to Describe Feeding Behavior 30
3.2. Markov Chain Model to Describe Adult Movement during Feeding Periods 32
3.3. Juvenile Locations 34
3.4. Accounting for Off-Field Exposures Due to Spray Drift 34
3.4.1. Determining an Individual Bird's Distance from the Edge of the Field 36
3.4.2. Determining the Fraction of Exposure Compared to Field 37
4. Estimating Pesticide Exposure through Diet 38
4.1. Pesticide Concentrations on Food Items (Ck(t)) 40
Page 3 of 77
-------
4.1.1. Pesticide Residues on Food items at the Time of Application (Ck(t=o)) 40
4.1.2. Pesticide Residues on Food Items After First Application (Ck(t)) 40
4.1.3. Contaminated Fraction on Food Items (FCk) 41
4.2. Total Daily Food Intake Rate (TDIR) 41
4.3. Food Matrix Adjustment Factor (FMA) 43
5. Estimating Pesticide Exposure through Inhalation 44
5.1.1. Pesticide Concentration in a Volume of Air (Cair(drops)) 47
5.1.2. Calculation of Inhaled Air Volume (Vinhaiation) 47
5.1.3. Fraction of Applied Pesticide Spray (Frespired) 47
5.2. Calculating Pesticide Exposure through Inhalation of Vapor Phase Pesticide 48
5.3. Relating External Inhalation Dose to Oral Dose Equivalents 49
6. Estimating Pesticide Exposure through Dermal Contact 50
6.1. Dermal Exposure through Direct Interception 51
6.2. Dermal Exposure through Dislodgeable Pesticide Residues on Foliage 52
6.2.1. Dislodgeable Foliar Residue Adjustment Factor (Fdfr) 52
6.2.2. Surface Area of Bird that Contacts Foliar Residues 53
6.2.3. Rate of Foliar Contact (Rfoiiar contact) 53
6.3. Relating External Dermal Dose to Oral Dose Equivalents 53
7. Estimating Pesticide Exposure through Drinking Water 54
7.1. Pesticide Concentrations in On-field Puddles 55
7.2. Pesticide Concentration in Dew from Contaminated Forage 56
7.3. Drinking Water Intake Rate (DWIR) 57
8. Establishing Sensitivity and Mortality of Individuals 58
8.1. Determining Survival and Mortality 58
8.2. Establishing an Individual Threshold for Mortality (Tmortaiity) 58
8.3. Avian Acute Oral LD50 59
8.4. Slope 60
8.5. Metabolism (Fretained) 60
9. Model Results 60
9.1. Model_results.txt Output File 60
Page 4 of 77
-------
9.2 Dead_per_hour.txt 64
10. Uncertainties 64
10.1. Exposure Routes Not Considered 64
10.2. Avian species 65
10.2.1. Diet and Feeding 65
10.2.2. Body Weight (BW) 66
10.2.3. Frequency on Field and Residency 66
10.3. Modeling Bird Behavior: Influence of the Fidelity Factor 66
10.4. Dietary Exposure 68
10.5. Inhalation Exposure 69
10.6. Dermal Exposure 70
10.7. Drinking Water Exposure 71
10.8. Determining Mortality 72
10.8.1. Toxicity Data 72
10.8.2. Elimination 74
10.9. Other Considerations 74
11. References 74
Appendices
A. User's Guidance on Input Parameters
B. Example Input File for TIMv.3.0 (with Parameter Descriptions in /* */)
C. Parameters Used in TIMv.3.0
D. Avian Data to Support Generic and Custom Species
E. Initial Pesticide Residues on Arthropods
F. Approach for Calculating Juvenile Dietary Exposure
G. Equation for Calculating Home Range for Insect Eating Birds
H. Dermal Toxicity Estimation
I. Overview and History of Tiered Risk Assessment Framework
J. Method for Deriving Species Sensitivity Distributions
Page 5 of 77
-------
1. Introduction
1.1. Purpose of Document
The purpose of this document is to provide technical information on version 3.0 of the Terrestrial
Investigation Model (TIM v.3.0). TIM derives quantitative estimates of the probability (or
likelihood) and magnitude of mortality to birds (of the same species) exposed to the simulated
pesticide. This document describes how TIM derives joint distributions of exposure and toxicity
to calculate the risk of mortality to birds.
1.2. When to Use TIM
Conceptually, the ecological risk assessment framework includes four levels, which differ in
level of information, effort and assumptions. (See Appendix J for more details). When assessing
the Tier 1 level of risks from acute and chronic exposures of birds, EFED uses the T-REX
model, which provides a conservative estimate of exposure through diet. Tier I assessments are
based on risk quotients (RQs), which are calculated by dividing a point estimate of exposure by a
point estimate representing effects (e.g., LD50). After RQs are calculated, they are compared to
levels of concern (LOCs) in order to determine whether a pesticide use poses a risk to birds and
mammals through acute and chronic dietary exposure.
For probabilistic assessments, EFED uses TIM, a refined risk assessment model (Tiers II-IV, see
Appendix I) that focuses on acute exposures to birds. If acute RQs exceed LOCs for non-listed
or listed species, and the risk manager needs additional information on the risk posed by a
pesticide use, TIM may be used to quantify the probability and magnitude of mortality, to
characterize uncertainties and to explore mitigation options that may be implemented in a
manner specified on pesticide labels. The decision to run TIM and the purpose of the specific
analysis should follow discussions with the risk manager. The analysis carried out by the risk
assessor should be tailored to address the questions of the risk manager and to quantify the
influence of uncertainty associated with the existing dataset.
In addition to quantifying the probability and magnitude of mortality to a species of bird, TIM
may also be used to provide information for the MCnest (Markov Chain nest productivity)
model, a refined risk assessment model for estimating the chronic impact of pesticides on the
reproductive success of bird populations. In this case, the MCnest model may be run when
chronic avian RQs generated by the T-REX model exceed the chronic LOC. OPP/EFED is
currently working with ORD to integrate the TIM and MCnest models so that exposure estimates
generated by TIM (for individual birds and their offspring) may be used for determining the
potential decrease in fecundity associated with a pesticide exposure. Together, the outputs from
TIM and MCnest may be used to parameterize a population model for a species to determine
potential population-level impacts associated with declines in survival of adults and their
offspring and declines in fecundity.
Page 6 of 77
-------
1.3. Overview of Terrestrial Investigation Model (v.3.0)
TIM (v.3.0) is a multimedia exposure/effects model that can be used to address avian mortality
levels from acute pesticide exposure in generic or specific species over a user-defined exposure
window. This time frame corresponds to one growing season of the treated crop or a single sub-
annual pesticide application window. The spatial scale is at the field level, but specific field
dimensions are undefined. It is assumed that the field and surrounding area meet habitat and
dietary requirements for the modeled species. During the simulation, birds use the treated field
and edge habitat to meet their requirements for food and water. TIM also accounts for exposure
via dermal and inhalation routes for birds on the field or for adjacent habitat that receives spray
drift. It is expected that the relative importance of these routes of exposure will vary based on the
properties of the pesticide, its use, as well as the characteristics of the simulated bird species.
Risk, expressed as a function of exposure (dose) and toxicity, is assessed for liquid spray
applications of a pesticide made to vegetation or bare ground in the field. Pesticide application
methods that may be modeled in TIM v.3.0 include: aerial, ground broadcast, airblast, ground
banded and ground in furrow. For all of these application methods, exposure can be assessed on
the treated field and edge habitat where spray drift is transported. The model does not currently
account for exposures due to seed treatments or granular formulations. Additional model
limitations are described in Section 10.
As described in this technical manual, TIM relies upon distributions of parameter values in order
to consider variability in bird behavior, body size and exposure. Values for individual birds are
randomly selected from these distributions. Although there are inherent uncertainties associated
with this model through assumptions and lack of information (see Section 10), the model does
not account for uncertainty in each simulation. The impact of uncertainty on the probability
associated with mortality should be explored by the model user through selection of alternative
input parameters. For instance, the avian oral LD50 has a major impact on the model's results.
Thus, the model user should run scenarios using the best estimate of the LD50 as well as upper
and lower confidence bounds in order to understand the range of probabilities associated with
levels of mortality of interest for the assessment.
The major parameters addressed in the model are:
• Concentration estimates in/on vegetation, arthropods, water, and air for oral, inhalation,
and dermal routes of exposure;
o Concentration estimates in the model are distributions of residues as a function of
application rates and degradation/dissipation;
• Proportions of diet composed of each food type, defined by the food habits of defined
generic or selected specific species;
• Frequency of feeding on the treated field, determined in hourly time steps;
• Hourly ingestion rates of food and water as a function of body weight;
• Hourly inhalation rates of air as a function of body weight;
• Hourly dermal residue transfer rates from contaminated vegetation as a function of body
weight;
• Exposures in adjacent habitats receiving spray drift deposition from the treated field;
• Acute toxicity dose-response relationship based either on a specific species, when data
are available, or on inter-species extrapolations.
Page 7 of 77
-------
Each ran of TIM (v.3.0) involves simulating pesticide exposure for a user-defined number of
birds, with a minimum of 10,000. This number was selected because it is sufficiently large to
capture the variability in model outputs and will allow for stable results upon repeat of the
simulation. For each individual bird simulated, a random selection of values is made for the
major exposure parameters. These parameters are then used to establish pesticide dose through
diet, drinking water, inhalation and dermal exposure routes. The pesticide dose from each
exposure route is converted to an oral-dose equivalent. This conversion is accomplished using
relative toxicity data for the different routes of exposure.
Equation 1.1 depicts the approach for determining the total body dose (Dtotai(t)) of a pesticide at
time t. The pesticide body load is modeled over the simulation period at hourly time steps, with
the body load at any step consisting of any newly ingested dose {i.e., Ddiet(t) +Ddrinking(t) +
Dinhaiation(t) + Ddermai(t)) plus the remaining fraction of doses ingested in previous time steps {i.e.,
Dtotai(t-i)), accounting for the fraction of pesticide retained after elimination {i.e., Fretained). The
relative contributions of each exposure pathway are not equivalent, varying based on the
properties of the chemical, application, bird's behavior and time step. Doses account for the
location of the bird relative to the treated field. In the case that the bird is not located on the field
at the simulated time step, the bird receives a fraction of the on-field exposure, with that fraction
depending upon the spray drift deposition at the bird's location relative to the edge of the treated
field.
Equation l.l. ^total(t) ^diet(t) ^drinking ^ ^inhalation (t) ^dermal (t) ^'total (7-1) ^retained
The status of an individual bird (dead or alive) for each time step is assigned by comparing the
estimated body load (Dtotai(t)) to an unique threshold (Tmortaiity) that is randomly selected for that
bird from the dose/response curve obtained from the acute oral LD50 test. If the internal dose is
below the threshold, the bird is considered to be alive at that time step and the bird survives to
the next hour, where the process is repeated {i.e., if Dtotai(t) < Tmortaiity, bird survives to t+1). If the
dose exceeds the threshold, the bird is considered dead, and is no longer included in the
simulation {i.e., if Dtotai(t) > Tmortaiity, bird dies during t). As long as the bird is alive, the bird
continues to the next step until the body load is greater than the threshold or the user-defined
model duration is reached.
This procedure is repeated using Monte Carlo sampling methodology, and after multiple
iterations of individuals, a probability density function of percent mortality is generated. Figure
1.1 provides a conceptual diagram of TIM (v.3.0).
Page 8 of 77
-------
Figure 1.1. Conceptual Diagram of TIM (v.3.0)
Exposure routes modeled in TIM v.3.0 include dietary, inhalation (spray droplets and volatilized
pesticide), dermal (direct spray and contact with vegetation) and drinking water (dew and
puddles). Specific exposure routes included in the model differ by application method. (See
Table 1.1). All of the exposure routes are included in the model simulation for aerial and airblast
applications. Exposure routes for ground applications depend upon the height of the crop. If the
crop is >0.152 m (6 in), it is assumed that the birds can hide under the crop when the tractor
applies the pesticide, and therefore all exposure routes are included. When the crop height is
Page 9 of 77
-------
<0.152 m, it is assumed that the birds on the field will be flushed by the tractor and thus will not
be on the field at the time of the application. As noted by the FIFRA Scientific Advisory Panel
(SAP) in 2004, "Birds certainly could, and probably usually do, move out of the way of
application machinery. If for no other reason than that the application equipment is noisy." It is
also assumed that the birds land in an area beyond the spray drift exposure area {i.e., >304 m
from the edge of the field) and do not receive exposures via inhalation of droplets or direct spray
(dermal). Therefore, for ground applications where crop height is <0.152 m, relevant exposure
routes include dietary, drinking water, inhalation of volatilized pesticide and dermal contact with
treated vegetation. For pesticides applied to the ground via banded and in furrow methods,
pesticide exposure through drinking water, inhalation and dermal contact are assumed to be
negligible and thus are not included, meaning that dietary exposure is the only route for these
application methods. It is also assumed that there is no spray drift transport for banded and in
furrow applications.
Table 1.1. Summary of Exposure Routes Considered by A
Ground
Ground
Exposure route
Aerial
broadcast
Airblast
broadcast
(crop height
> 0.152 m)
broadcast
(crop height
<0.152 m)
Ground
banded
Ground
in furrow
Diet
Yes
Yes
Yes
Yes
Yes
Yes
Inhalation -spray
Yes
Yes
Yes
No
No
No
Inhalation - volatiles
Yes
Yes
Yes
Yes
No
No
Dermal - spray
Yes
Yes
Yes
No
No
No
Dermal - contact with foliage
Yes
Yes
Yes
Yes
No
No
Drinking - puddle
Yes
Yes
Yes
Yes
No
No
Drinking - dew
Yes
Yes
Yes
Yes
No
No
jplication JV
ethod.
In addition, the Graphical User Interface (GUI) includes pathway switches that allow the user to
turn off any of the exposure routes and spray drift exposure. The user may choose to turn off a
pathway switch if the route of exposure is not relevant to the application scenario, to explore risk
mitigation options {e.g., spray drift reduction) or to explore the model's sensitivity to a specific
pathway. For example, if the application is applied preplant (crop height is 0) and the field is
tilled {i.e., there are no weeds), inhalation, dermal and dew exposure routes may be turned off for
aerial and ground applications. If weeds may be on the field, the user should consider leaving
these exposure routes on. The model user may also wish to turn off a pathway based on the life
history of a specific species that is being simulated. For example, if the species does not drink
water, the dew and puddle switches could be turned off. If the species is an aerial feeder, and is
not expected to be beneath the crop canopy, the dermal and inhalation pathways could be turned
off.
1.4. The Evolution of TIM
TIM has evolved over time in response to FIFRA SAP comments and recommendations. Table
1.2 provides an overview of the evolution of the key features of TIM, from its early stages to
present.
Page 10 of 77
-------
Table 1.2. Components of Different Versions of TIM.
Model Component
version 1.0
(2-21-2001)1
version 2.0
(3-10-2004)2
version 3.0
Species considered
Focal
Generic
Generic or focal
(specific)
Duration of exposure
7 days
User defined
User defined
Time step
User defined
Hourly
Hourly
Exposure routes considered
Dietary
Drinking water
Dietary
Drinking water
Inhalation
Dermal
Dietary
Drinking water
Inhalation
Dermal
Spray drift
No
No
Yes
Feeding pattern
Unimodal
Bimodal3
Bimodal3
Serial correlation between
foraging events
None
Yes
Yes
Dew
Organic-carbon
based equilibrium
Organic-carbon
based equilibrium
Octanol-water
based equilibrium
Drinking water exposure from
puddles
Addressed using PRZM4
outputs
Addressed using complex
model comparable to
PRZM4 conceptual model
Addressed using
equilibrium partitioning
approach
Number of pesticide
applications modeled
1
1
Up to 5
User has ability to turn on/off
exposure routes
Yes
No
Yes
Model platform
Excel and
Crystal Ball
C executable
with Excel GUI
C executable
with Matlab GUI
Presented to FIFRA SAP in 2001
2Presented to FIFRA SAP in 2004
3The bimodal distribution simulates morning and evening feeding patterns of birds.
4Pesticide Root Zone Model.
In 2001, EPA presented two case studies {i.e., terrestrial and aquatic) to the SAP for review.
These case studies used probabilistic methods to assess the ecological risk from a generic
chemical (ChemX). The terrestrial case study (USEPA, 2001) was based on the characterization
of exposure and effects by defining the distributions of the major variables and combining these
into joint distributions to estimate the probability and magnitude of effects. TIM v. 1.0 was
utilized as a species-specific model, which addressed acute mortality levels over a defined
exposure window. The spatial scale was at the single treated field where the field and
surrounding areas were assumed to meet the habitat requirements for each focal species. The
temporal scale was for exposure during and immediately following a single application of a
pesticide. The major parameters addressed in the model were as follows:
• food habits;
• ingestion rates of food and water;
• frequency of feeding and drinking on sprayed field;
• distribution of residues on food and water;
• dissipation rates; and
• inter- and intra-species dose response variability.
For each run of the model, a random selection of values was made for the major exposure input
parameters to estimate a dose to an individual of a focal species. The likelihood (probability) of
Page 11 of 77
-------
mortality from this individual's estimated dose was calculated using the dose-response curve,
and a binomial probability approach was used to determine if an individual modeled bird
survives. This procedure was repeated using Monte Carlo simulations for a set of individuals to
generate an estimate of the percent of birds affected. After multiple iterations of sets of
individuals, a probability density function of percent mortality was generated.
Exposure pathways considered were from dietary and drinking water routes. Exposure was
divided into two equal time steps per day. At the onset of each time step, a binomial probability
function was used to determine if an individual modeled bird uses the treated field as a source of
food and water.
To estimate the acute toxicity of ChemX to focal species, an inter-species distribution-based
approach was used. The parameters of the distribution were determined from the available
toxicity values for the pesticide that is being assessed. For each focal species, three estimates of
its LD50 were made, 5th, 50th, and 95th percentile, to account for inter-species toxicity uncertainty.
Use of the slope of the dose response distribution addresses the establishment of a sensitivity
threshold for each modeled individual of a species (intraspecies variability).
To estimate the mortality distribution for a selected focal species, the likelihood of mortality for
the maximum estimated body burden, based on the external dose for the duration of the
exposure, was calculated from the dose response curve derived for the selected focal species.
The distribution of mortality of cohorts of individuals was determined through multiple iterations
using Monte Carlo simulations, thus, providing an estimate of the probability and magnitude of
effects.
Following the case study with ChemX, EPA considered the SAP's comments (SAP, 2001),
which strongly supported the EPA's efforts in developing both the aquatic and terrestrial case
studies. The terrestrial model was subsequently modified, and EPA returned to the SAP in April
2004 for an additional review of TIM (v.2.0). The major changes that were incorporated into the
revised model were in response to the SAP's comments (SAP, 2004) and included the following:
• establishment of generic birds that represent species occurring in and around agro-
environments;
o TIM (v.2.0) used generic attributes to represent the more vulnerable species, yet
retained the ability to address specific focal species, when appropriate;
• incorporation of inhalation and dermal exposures;
• incorporation of a 1-hour exposure time step to allow the inclusion of a bimodal feeding
pattern, as well as a higher resolution simulation of daily feeding behavior between
treated and untreated areas;
• incorporation of an algorithm (Markov Chain) to address serial correlation between
sequential foraging events; and,
• development of a model for estimating pesticide residues in on-field drinking water
sources (puddles) that accounts for a number of parameters that affect the formation of
puddles after a rainfall event {i.e., rainfall amount, rainfall duration, soil infiltration rates,
evaporation, degradation).
Page 12 of 77
-------
In reference to the inhalation and dermal models, limited data are available regarding these two
routes of exposure, which results in uncertainty in the estimates of risks. However, if these
routes of exposure are ignored or assumed to be minimal, the uncertainty in risk estimates is not
addressed. This is of concern because dermal and inhalation routes may contribute significantly
to total dose in some situations (Driver et al., 1991). The incorporation of the dermal and
inhalation exposure models provides an important initial step to evaluate the potential
significance of these routes of exposure relative to others in the overall risk estimates.
Following the 2004 SAP, TIM (v.2.0) was modified to TIM (v.3.0). Major modifications
reflected in latest version of the model (TIM v.3.0) and its documentation include:
• allowing for the assessment of up to five applications of a pesticide in a growing season;
• allowing for model run durations exceeding 15 days;
• review of avian census studies in multiple crops and locations in North America for
parameterization of generic and specific species (Appendix D);
• addition of generic species that consume 100% fruit, 100% grass or 100% broadleaf
forage;
• revised distribution of initial pesticide residues on arthropods (Appendix E);
• the respirable droplet size was increased from 7 to 100 |im;
• modification of dew exposure model that replaces the Koc-based equilibrium approach
(partitioning between total leaf and dew) with one that is based on Kow (partitioning
between dew and cuticle wax);
• simplified puddle model;
• exposure of off-field birds to pesticide from spray drift transport; and
• simulation of pesticide exposures to juvenile birds for use in MCnest model.
1.5. Model Executable
The TIM (v.3.0) executable is coded in C and has been compiled using Microsoft® Visual Studio
2010 C++ Express. The model code is composed of 9 files. The main file is titled "TIM3.0.C."
The names and purposes of the other files are summarized in Table 1.3. Details of the contents
of each of these files are contained within comments provided at the beginning of each file.
Table 1.3. TIM v.3.0
Code Files
Name*
Contents
Arrays.c
functions used to allocate and free memory for vectors and matrices
declarations.h
declares the functions that are used by TIM
default values.c
functions that call default values
Exposure.c
functions that estimate pesticide exposures to birds
macros.h
function-like macros
RANDOM.c
random numbers generated using different distributions
Report.c
functions that print to screen or to output files
Statistics.c
statistical functions
TIM3.0
main code for the Terrestrial Investigation Model (TIM) v3.0
*" .h" extension denotes a header file, ".c" extension denotes a source file.
The TIM executable is run using a GUI that was developed in Matlab®. See Appendix A for
user's guidance on how to install and run the model. TIM requires up to 97 input parameters that
Page 13 of 77
-------
define the model assumptions related to the simulated bird species, as well as pesticide specific
parameters related to the use, fate and toxicity. Appendix A provides detailed guidance to the
user on how to select input parameters, including information on default values, when no
chemical-specific or species-specific information are available. As described later, the model has
the ability to simulate generic bird species. In that case, no species-specific parameters are
needed. The GUI develops an input text file (TIM_inputs.txt) that is read by the executable.
Once TIM is executed, it generates several output files. The names and descriptions of these
output files are included in Table 1.4. The two output files that are generated specifically for the
TIM user {i.e., Model_Results.txt and Dead_per_hour.txt) are described in Section 9.
Table 1.4 lists the Quality Control (QC) files that may be generated by the model. These outputs
were used in the the QC test of the model code. These outputs are not needed by the model user.
The TIM v.3.0 executable also has the capability of estimating exposures to the juvenile
offspring of those exposed adults. The purpose of this exercise is to couple TIM and MCnest so
that they are using the same exposure profiles for adult and juvenile birds. While the focus of
TIM is to assess the risk of mortality to adult birds exposed to a pesticide, the focus of MCnest is
to determine potential impacts of a pesticide on the fecundity of a species. This technical manual
describes TIM's method for estimating exposures to adult birds. When relevant, the manual also
discusses approaches for determining exposures to juveniles that will be used by MCnest. The
juveniles simulated by TIM do not influence the model results generated by TIM. Table 1.4
identifies files that may be generated by the model if the MCnest switch is turned on. The output
files generated for MCnest are comma-delimited text files and thus can be easily converted into
Microsoft® Excel format if so desired. The MCnest output files include daily exposure values for
individual adult birds and their juvenile (hatchling) offspring. The MCnest files also include the
tolerances for those birds and information on whether or not the adults died during the TIM
simulation, and if so, when those deaths occurred.
Page 14 of 77
-------
Table 1.4. Summary of Output Files (.txt extensions) Generated by TIM v.3.0
File name**
Purpose of output file
Description
adult acquired doses
Input for MCnest
Daily acquired doses for juvenile birds* (each row represents daily doses for individual adult birds)
birddeath
Input for MCnest
Includes the individual threshold for each adult and its associated off spring (juveniles) as well as the
simulation day when the adult birds died
Dead_per_hour
TIM results for model user
Reports the number of dead birds for each hour of the simulation
(column 1 = hour, column 2 = # dead birds)
juvenile acquired doses
Input for MCnest
Daily acquired doses for juvenile birds* (each row represents daily doses for individual juveniles)
ModelResults
TIM results for model user
Includes input parameters and model results (number of birds killed, relative contributions of exposure
routes to mortality, probabilities)
QC section2
QC review of TIM code
Calculations from section 2 of manual (species parameters)
QC section3
QC review of TIM code
Calculations from section 3 of manual (feeding times, movement, spray drift)
QC section4
QC review of TIM code
Hourly dietary pesticide dose, section 4 of manual
QC section4 1 apl
QC review of TIM code
Concentrations on food items at time of application 1, section 4 of manual
QC section4 1 ap2
QC review of TIM code
Concentrations on food items at time of application 2, section 4 of manual
QC section4 1 ap3
QC review of TIM code
Concentrations on food items at time of application 3, section 4 of manual
QC section4 1 ap4
QC review of TIM code
Concentrations on food items at time of application 4, section 4 of manual
QC section4 1 ap5
QC review of TIM code
Concentrations on food items at time of application 5, section 4 of manual
QC section4 2
QC review of TIM code
Daily food intake rate, section 4 of manual
QC section4 HLapl
QC review of TIM code
Testing half-life for application 1, section 4 of manual
QC section4 HLap2
QC review of TIM code
Testing half-life for application 2
QC section4 HLap3
QC review of TIM code
Testing half-life for application 3
QC section5
QC review of TIM code
Hourly inhalation doses (for birds 1-10) section 5 of manual
QC section5 2
QC review of TIM code
Hourly vapor concentrations, section 5.2 of manual
QC section6
QC review of TIM code
Hourly dermal doses (for birds 1-10), section 6 of manual
QC section7
QC review of TIM code
Hourly drinking water doses (for birds 1-10), section 7 of manual
QC section8 death
QC review of TIM code
Summary of bird specific thresholds, section 8 of manual
QC_ section8_threshold
QC review of TIM code
Hourly total doses and thresholds for birds 1-10, section 8 of manual
*Total daily dose generated for MCnest is the sum of all hourly pesticide doses through all routes of exposure included in the model. This does not account for
elimination.
**The QC files were used in the Quality Control review of the executable and are not expected to be of interest to the model user. These files are generated only
when the QC switch is turned "on" in the input file
Page 15 of 77
-------
1.6. Quality Assurance and Quality Control
The quality assurance and quality control (QA/QC) review of TIM (v.3.0) involved three major
steps. The first step involved development of this document, which represents the technical
documentation describing the model, its equations, assumptions, parameters and uncertainties.
The technical documentation for TIM (v.3.0) was developed by compiling documentation from
SAP meetings held in 2001 and 2004, where TIM versions 1 and 2 were presented. Model
components were checked against their original sources (e.g., publications in the literature). All
model parameters and their underlying assumptions were described and unit balances were
verified for each equation.
The second step involved a review of the model executable. This QC review was completed
using only the executable and text input files. Version 3.0 of the executable reflects changes
made to version 2.1 (September 16, 2008). Parameter values included in the input file were
documented (see Appendix B). The model executable was run in order to verify that all input
values were read correctly, that switches were turned on/off based on user inputs and that default
parameters were correctly assigned when generic bird species were selected. Correct contents of
output files were verified. The variables, arrays, matrices, and functions of the model code were
described using comments embedded within the code. The functioning of the model code was
verified by two separate approaches: first, the code was reviewed and compared to the parameter
and equations provided in the technical manual; and secondly, the model calculations for each
section of the technical manual (Sections 2-9) were output to QC files (Table 1.4) and compared
to independent calculations carried out in Microsoft® Excel (where the user input parameter
values and equations of the technical manual were implemented). If an error was identified, the
code was modified to correct the error. The code was then recompiled and checked again. The
QA/QC review confirms that the final version of the model executable correctly calculates the
equations described in this technical manual.
The third step involved verification of the GUI. In this approach, model outputs generated by the
stand alone executable and GUI for the same input files were compared. Since the model
outputs were equal, it was concluded that the GUI is operating correctly.
1.7. Organization of This Document
This Technical Guidance Document is organized into 10 sections, beginning with an
introductory section (Section 1). Section 2 provides a review of how TIM represents avian
species. Section 3 discusses the determination of a bird's location on or off the treated field and
the fraction of on field exposure a bird receives when off-field (due to spray drift). Sections 4, 5,
6 and 7 describe how TIM estimates hourly pesticide exposures through dietary, inhalation,
dermal and drinking water routes, respectively. Section 8 discusses establishing sensitivity and
mortality of individuals, and Section 9 discusses the model's results. Section 10 discusses the
uncertainties associated with TIM v.3.0. The ten appendices of this document include:
- user's guidance for running the model and determining input parameters (Appendix A);
example input file (Appendix B);
summary of the parameters used in TIM v3.0 (Appendix C);
Page 16 of 77
-------
summary of avian census field studies that support the parameterization of generic and
custom species (Appendix D);
analysis of empirical data from studies measuring initial pesticide residues on arthropods
(Appendix E);
description of the basis for the food intake equation for juveniles (Appendix F);
information used to calculate the home range for insect eating birds (Appendix G);
data used to generate the equation to estimate an acute dermal toxicity endpoint
(Appendix H; from USEPA, 2004);
overview and history of the tiered risk assessment framework (Appendix I); and
method for deriving species sensitivity distributions that may be used to establish toxicity
endpoints and explore uncertainty (Appendix J).
2. Avian Species
In TIM v.3.0, an avian species is represented by several parameters, including diet, body weight
(BW), home range, frequency on field (FOF) and fidelity factor (determined by residency status).
BW is a particularly important parameter because it is used in the allometric equations that
derive intake rates for all exposure routes {i.e., food consumption, inhalation rate, surface area
and drinking water consumption).
The model user has the choice between using a generic species or a specific/custom species,
which can be parameterized by the user to represent an avian species of interest. The generic
species can be used to identify groups of species that may be at risk. In addition, generic species
that predominate agricultural areas, such as small to medium sized insectivores and omnivores,
may be simulated to consider potential indirect effects to predatory birds or mammals. Specific
species and species of interest, including federally-listed endangered and threatened species, may
be simulated as a refinement in the assessment.
BW, diet information and frequency on field data for specific species occurring in agricultural
areas in North America are provided in Appendix D. Data from these species were used to
derive default parameters for the generic species, including BW, diet and FOF. Data from these
studies were also used to derive default parameters for 56 specific species. It should be noted that
the parameters selected to represent species are generally representative of the breeding season.
During times when birds are not nesting, diets and FOF may change. The model user should
consider whether alternative assumptions for diet and FOF are necessary if simulating a pesticide
application before or after the breeding season.
2.1. Diet Composition
In TIM v.3.0, birds consume terrestrial food items, including grass, broad leaf plants, fruit, seeds,
and arthropods {e.g., insects, spiders, millipedes). Of the commonly observed species (n = 117)
in the avian census studies described in Appendix D, the majority have diets that are
predominantly insects {i.e., insectivores) or seeds {i.e., granivores), or the species have diets
composed of multiple food times {i.e., omnivores). Some of the species predominantly consume
Page 17 of 77
-------
other food items, including fruit, nuts, plant matter, small animals (mammals, birds, amphibians
and reptiles), aquatic invertebrates and nectar (Table D28).
Specific species (referred to as "custom species") can be parameterized by the model user so that
the diet is represented by a combination of food items and the sum of these food items is 100%.
The section below includes recommended input parameters to represent 56 species1 that are
known to use agricultural areas and their adjacent habitats. These species are also part of the
library used by MCnest. As described below, a bird's home range is determined by its diet. The
user must define whether a modeled species is an insectivore, herbivore, granivore or omnivore.
As a general rule, a species may be defined as insectivore, herbivore or granivore if >70% of its
diet is represented by the appropriate food item. If no food item represents >70% of the diet, the
species is identified as an omnivore.
For generic species that are insectivores, herbivores, granivores and frugivores, 100% of the
adult bird's diet consists of a single food item. For omnivores, the diet is distributed equally
between the available food items (i.e., 20% insects, 20% seeds, 20% fruit, 20% grass and 20%
broadleaf). For generic species, it is assumed that 100% of the juveniles' diet consists of
arthropods.
2.2. Body Weight (BW)
2.2.1. Adults
To represent the distribution of BWs of birds in a simulated field, TIM requires input values for
mean, standard deviation (SD), minimum (min) and maximum (max) weight values. These
values are used to generate a beta distribution of BWs that represent the BWs of the simulated
birds. The minimum and maximum values rescale the distribution. A description of the beta
distribution can be found in USEPA 2007a. The upper and lower bounds of the beta distribution
(i.e., alpha and beta, respectively) are calculated according to Equations 2.1 and 2.2,
respectively. A beta distribution was chosen because of its ability to retain meaningful central
tendency measures in the face of a distribution with finite limits on both ends.
Equation 2.1. a = (mean — min) * z Where: z =
('min—max)* sd.2
Equation 2.2. /? = —(mean — max) * z
EFED's screening level model for birds, T-REX (USEPA, 2012a), uses three generic BWs to
represent small, medium and large-sized birds (i.e., 20, 100 and 1000 g, respectively). These
values are consistent with species that have been documented as visiting agricultural fields.
Consistent with T-REX, TIM v.3.0 also has three generic BW distributions (Table 2.1), with the
mean values set to the small, medium and large BWs. The standard deviations are based on the
1 Three of these species, the American kestrel, bobolink and mallard, are not included in Appendix D because they
were not observed in the available avian census studies.
Page 18 of 77
-------
average coefficient of variance in BWs of birds that have been documented as visiting
agricultural fields (CV = 7.3%). The 90th percentile of the minimum and maximum BW values
for individuals within species are 66% and 152% of the mean, respectively. These percentages
are used to determine the minimum and maximum values of the beta distributions for the generic
species used in TIM (Section D.3.3 of Appendix D).
Table 2.1. BW Parameters Used for Generic Birds
BW Distribution Parameter (in g)
Small bird
Medium bird
Large bird
Mean
20
100
1000
Standard Deviation (=mean * 0.073)
1.5
7.3
73
Minimum (=mean * 66%)
13
66
660
Maximum (=mean * 152%)
30
152
1520
As discussed in Appendix D, for the commonly observed species in the available avian census
studies, BWs range 3.2-2943 g, with a mean of 103 g. The majority of species {61%) have mean
BWs <50 g. When the mean BWs are distributed, the 80th percentile is 97 g, suggesting that the
majority of the birds commonly found on agricultural fields and their adjacent habitats would be
represented by the 20 and 100 g (mean) BWs of the generic species used in TIM. Few birds
would be represented by the large generic bird category {i.e., 1000 g).
2.2.2. Juveniles
In simulating juveniles, BW of an individual juvenile is set to 0.5 times the BW of its parent.
This assumption is based on an analysis that indicates that the food intake rate of juveniles is
highest when they weigh approximately 0.5 times the BW of their parents. (See Appendix F for
details).
2.3. Home Range Size
The area (A) of the home range in square meters (where the factor of 10,000 is used to convert
hectares to square meters) is determined according to Equations 2.3-2.6, depending upon the
diet of the simulated bird {i.e., granivores, herbivores, omnivores, or insectivores). Equations for
granivores, herbivores and omnivores are from Mace and Harvey (1983). Since no equation is
available for frugivores, the herbivore equation is used as a surrogate. The equation for
insectivores was generated in Microsoft® Excel, using home range data from Schoener (1968)
(Appendix G). This equation was based on data for species that are identified in field surveys
associated with agricultural fields/orchards and adjacent habitats that are described in Appendix
D. The R2 values associated with each of these equations ranges 0.27-0.51, indicating that there
is variability associated with the home ranges of different species used to generate these
equations. This leads to uncertainty in the home range prediction for a simulated species.
Equation 2.3. A = 0.05 * BW112 * 10,000 (Granivores; R2 = 0.37)
Equation 2.4. A = 0.003 * BW123 * 10,000 (Herbivores; R2 = 0.38)
Equation 2.5. A = 0.004 * BW1,33 * 10,000 (Omnivores; R2= 0.27)
Equation 2.6. A = 0.003 * BW1,64 * 10,000 (Insectivores; R2 = 0.51)
Page 19 of 77
-------
2.4. Frequency on Field and Residency Status
Frequency on field (FOF) is the amount of time in a simulation that a bird spends on the treated
field. TIM requires input values for mean, minimum and maximum FOF values in order to
generate a beta pert distribution of FOF values for the simulated species of birds. A standard
assumption of 4 is used for the height of the mode. A description of the beta pert distribution is
available in USEPA (2004). For each simulated bird, a unique FOF value is selected from this
distribution.
There are multiple factors that can influence the FOF of individuals of a species. Some of these
factors include foraging preference, range, composition of edge habitat and time of season (Best
et al., 1990; Boutin el al., 1996). Therefore, the crop being studied and the geographic location
of the study site can influence the species observed as well as their frequencies on the field
relative to edge habitats. In order to account for some of these variables, 26 avian census studies
in agricultural fields and edge habitats were considered from 9 different crops (alfalfa, apples,
cabbage, citrus, corn, cotton, grapes, potatoes and soybeans) in different geographic locations in
North America (Alabama, Arizona, California, Florida, Kansas, Illinois, Iowa, Mississippi,
Nebraska, New Mexico, Oklahoma, Ontario, Texas and Wisconsin). These studies are described
in Appendix D. These studies reported observations of individuals of the same bird species
within an agricultural field and its edge habitat. The percent of the total number of individuals
observed at one time period that were observed within the agricultural field is used as a surrogate
for the mean FOF of individual birds within a species. On-field observations can be highly
variable for some species among different avian census studies (e.g., range of 1-89% for the red-
winged blackbird (Agelaiusphoeniceus)\ range of 3-100% for brown-headed cowbird
(Molothrus ater)). In TIM, the mean FOF values for generic species were set to represent a
reasonable high-end value based on these observed values. For specific species, the model user
may use a default mean FOF value (and range) or select values based on species-specific data.
In TIM, avian species are distinguished as field and edge residents. These classifications impact
the FOF values used to represent the species time on the treated field. In order to derive FOF
values for the generic species in TIM, it is necessary to define the residency of the commonly
observed species in avian census studies. Residency is based on the nesting habits of a species.
For agricultural fields species that build their nests on the ground in grassland areas are defined
as field residents, while other species are edge residents. For orchards and vineyards, field
residents are those that build their nests on the ground in grasslands and woodlands as well as
those that build their nests in the mid-story and canopy, while other species are edge residents.
The empirical on-field percent observations for field and edge residents are used to estimate
mean default FOF values for generic species using agricultural fields and orchards/vineyards.
These default means are based on 90th percentile estimates of available data. The data presented
in Appendix D represent mean values of FOF for individual species at specific sites. The 90th
percentile mean FOF values were selected as defaults for the generic species in order to present a
conservative mean FOF values. For field and edge residents using agricultural fields, the default
mean FOF values are 97% and 69%, respectively. The 90th percentile FOF values for field and
edge resident species using orchards and vineyards are similar (i.e., 87 and 85%, respectively)
and are based on a limited number of species and observations relative to the species observed on
Page 20 of 77
-------
agricultural fields. Therefore, default FOF values are not distinguished for field and edge
residents visiting vineyards and orchards, and the mean default value for both field and edge
species visiting orchards and vineyards is 87%. Since FOF values for individual species range 0-
100%, this range determines the minimum and maximum values of the beta pert distribution
used to describe FOF in TIM.
2.5. Fidelity Factor
The fidelity factor (Q) is the serial correlation between sequential foraging events. This
parameter represents the tendency of a bird to return to a specific area (field or edge) to feed. The
fidelity factor for field residents is 0.8. This value, which represents a relatively strong tendency
to return to a site to feed, was suggested by the SAP (SAP, 2001) as an appropriate scenario to
model. For edge species, the fidelity factor is 0.6. This value was selected because it is
somewhat lower than field resident species, but still allows for some tendency to return to the
same area to feed. Section 10.2 includes a discussion of the model sensitivity of this parameter.
2.6. Taxonomy
The majority of species observed on agricultural fields, orchards and vineyards in avian census
studies, described in Appendix D, are in the Passeriform order. In TIM, species are defined as
either "passerine" or "non-passerine." This distinction impacts the prediction of food and water
intake rates. Passerines have higher intake rates, resulting in higher dietary and drinking water
exposures relative to non-passerine species.
Since most bird species that visit agricultural areas are passerine and they generate higher
exposure values, it is assumed that all generic species are passerines. The model user may
distinguish between passerine and non-passerine species when modeling custom species.
2.7. Species
2.7.1. Generic
With the full combination of diet type, BW and residency status, a total of 30 generic species are
available to the user for modeling purposes (Table 2.2). The generic omnivore was established
with equal portions of each food item included in its diet.
Of the available generic species in Table 2.2, the highest estimated exposures are for small,
field-resident birds eating grass since field residents have the highest FOF and grasses have the
highest estimated initial pesticide residues. Small birds are assumed to be more sensitive to
pesticide exposure and are assumed to receive higher body burdens relative to larger birds.
Although this generic species may be useful as a screen, its representativeness may be limited to
small birds with small plants and grass in their diets for some period of time. Based on avian
survey data, the generic species that are most representative of species that occur on agricultural
areas and adjacent habitats are the small- and medium-sized insectivores, omnivores and
granivores. It is likely that feeding on short grass is not sustainable for a long period of time if
grass energy content value used in the model is assumed; however, young shoots are much
higher in protein and lipid than mature grasses.
Page 21 of 77
-------
2.7.2. Custom
Tables 2.3 - 2.5 contain input parameters that may be used to represent 56 different species that
are also part of the MCnest species library. With the exception of American kestrel, bobolink and
mallard, all of these species were observed in avian census studies discussed in Appendix D.
BW values for these species are based on Dunning (1984).
Dietary fractions were assigned based on the MCnest library, which includes diets of breeding
females and juveniles. In order to assign feeding categories for determination of the appropriate
home range equation, it was assumed that a diet of >70% of one food item would designate a
specific feeding category (e.g., insectivores have a dietary fraction for insects that is >70% of the
total diet). Omnivores did not have a food item that exceeded 70%.
Mean FOF values included in Table 2.3 may be used to represent the species. These values
represent the highest observed occurrence of a particular species on the agricultural field in the
available avian census studies discussed in Appendix D. The model user should exercise caution
when selecting the mean FOF because of the uncertainty associated with the limited number of
studies associated with these studies. Another uncertainty is the expectation that different
species could have different affinities for different crops (i.e., FOF would vary by crop for the
same species). Table 2.3 also includes the range of FOF values that may be representative of the
species given the available avian census data and the crops where the species was observed to
occur on field and in-edge habitats. The model user may choose to explore uncertainty associated
with the mean FOF value for a species by simulating the range of FOF values in separate model
runs.
Page 22 of 77
-------
Table 2.2. Generic Birds Available in TIM v.3.0
Bird
#
Example species
(Appendix D)
Mean BW (g)
FOF (%)
Fidelity
Description
Diet
Mean
SD
Range
Field crops
Orchards and
vineyards
Factor
(Q)
1
Small insectivore field resident
Dickcissel (Spiza americana)
20
1.5
13-30
97
0.8
2
Small insectivore edge resident
Tree swallow (Tachycineta bicolor)
69
0.6
3
Medium insectivore field resident
Killdeer (Charadrius vociferus)
100%
100
7.3
66-152
97
0.8
4
Medium insectivore edge resident
Northern flicker (Colaptes auratus)
arthropods
69
0.6
5
Large insectivore field resident
None
1000
73
660-1520
97
0.8
6
Large insectivore edge resident
None
69
0.6
7
Small granivore field resident
Horned lark (Eremophila alpestris)
20
1.5
13-30
97
0.8
8
Small granivore edge resident
American goldfinch (Carduelis tristis)
69
0.6
9
Medium granivore field resident
Mourning dove (Zenaida macroura)
100%
100
7.3
66-152
97
0.8
10
Medium granivore edge resident
Northern bobwhite (Colinus virginianus)
seeds
69
0.6
11
Large granivore field resident
None
1000
73
660-1520
97
0.8
12
Large granivore edge resident
None
69
0.6
13
Small herbivore field resident
None
20
1.5
13-30
97
0.8
14
Small herbivore edge resident
None
69
0.6
15
Medium herbivore field resident
None
100%
100
7.3
66-152
97
0.8
16
Medium herbivore edge resident
None
grass
69
87
0.6
17
Large herbivore field resident
None
1000
73
660-1520
97
0.8
18
Large herbivore edge resident
Canada goose (Branta canadensis)
69
0.6
19
Small frugivore field resident
Cedar waxwing (Bombycilla cedrorum)
(orchard)
20
1.5
13-30
97
0.8
20
Small frugivore edge resident
Cedar waxwing (Bombycilla cedrorum)
(field)
100% fruit
69
0.6
21
Medium frugivore field resident
None
100
7.3
66-152
97
0.8
22
Medium frugivore edge resident
None
69
0.6
23
Large frugivore field resident
None
1000
73
660-1520
97
0.8
24
Large frugivore edge resident
None
69
0.6
25
Small omnivore field resident
Vesper sparrow (Pooecetes gramineus)
20%
20
1.5
13-30
97
0.8
26
Small omnivore edge resident
Dark-eyed junco (Junco hyemalis)
arthropods,
69
0.6
27
Medium omnivore field resident
Blue jay (Cyanocitta cristata) (orchard)
20% seeds,
20% grass,
20%
broadleaf,
20% fruit
100
7.3
66-152
97
0.8
28
Medium omnivore edge resident
Blue jay (Cyanocitta cristata) (field)
69
0.6
29
Large omnivore field resident
None
1000
73
660-1520
97
0.8
30
Large omnivore edge resident
None
69
0.6
Page 23 of 77
-------
Table 2.3. Custom Species Parameters for Specific Species: Basic Species Information, Residency Status and Frequency on Field. Based on
Species (scientific name)
Passerine?
Altricial/
Precocial
Residency
(ag. field)
Residency
(orchard)
Mean
FOF
Range of
mean
FOF
Crops where a species was observed
(in field studies used for FOF)
American crow (Corvus brachyrhynchos)
Yes
Altricial
edge
field
0.74
0.03-0.74
Apples, cabbage, corn, potatoes,
soybeans
American goldfinch (Car due lis tristis)
Yes
Altricial
edge
field
0.84
0.02-0.84
apples, cabbage, corn, grapes, potatoes
American kestrel (Falco sparverius)
No
Altricial
edge
edge
0.69/0.89*
none
none
American robin (Turdus migratorius)
Yes
Altricial
edge
field
0.87
0.02-0.87
alfalfa, apples, cabbage, citrus, corn,
grapes, potatoes, soybeans
Ash-throated flycatcher (Myiarchus
cinerascens)
Yes
Altricial
NA
edge
0.23
0.23
citrus
Barn swallow (Hirundo rustica)
Yes
Altricial
edge
field
0.99
0.01-0.99
alfalfa, apples, cabbage, corn, cotton,
potatoes, soybeans
Black-capped chickadee (Poecile
atricapillus)
Yes
Altricial
edge
NA
0.48
0-0.48
alfalfa, cabbage, corn
Blue jay (Cyanocitta cristata)
Yes
Altricial
edge
field
0.67
0.01-0.67
alfalfa, apples, cabbage, corn, cotton,
potatoes
Blue-gray gnatcatcher (Polioptila caerulea)
Yes
Altricial
edge
NA
0.01
0.01
cotton
Blue-winged Teal (Anas discors)
No
Precocial
edge
NA
0.65
0.65
corn
Boat-tailed Grackle (Quiscalus major)
Yes
Altricial
edge
NA
0.68
0.68
corn
Bobolink (Dolichonyx oryzivorus)
Yes
Altricial
field
field
0.97/0.89*
none
none
Brewer's blackbird (Euphagus
cyanocephalus)
Yes
Altricial
edge
field
0.4
0-0.4
citrus, corn
Canada goose (Branta canadensis)
No
Precocial
edge
NA
1
100
corn
Carolina chickadee (Poecile carolinensis)
Yes
Altricial
edge
NA
0.08
0-0.08
alfalfa, cotton, potatoes
Carolina wren (Thryothorus ludovicianus)
Yes
Altricial
edge
NA
0.01
0-0.01
cotton, potatoes
Cassin's sparrow (Aimophila cassinii)
Yes
Altricial
field
NA
0.07
0.07
cotton
Cedar waxwing (Bombycilla cedrorum)
Yes
Altricial
edge
field
0.8
0-0.80
apples, cabbage, corn
Chipping sparrow (Spizella passerina)
Yes
Altricial
edge
field
0.15-0.88
0.15-0.88
alfalfa, apples, grapes, corn, soybeans
Common grackle (Quiscalus quiscula)
Yes
Altricial
edge
NA
0.97
0.02-0.97
alfalfa, cabbage, corn, cotton, potatoes,
soybeans
Common yellowthroat (Geothlypis trichas)
Yes
Altricial
edge
NA
0.46
0-0.46
cabbage, corn, cotton, potatoes
Dark-eyed junco (Junco hyemalis)
Yes
Altricial
edge
NA
0.11
0.11
alfalfa
Dickcissel (Spiza americana)
Yes
Altricial
field
NA
1
0.16-1
alfalfa, corn, cotton
Eastern bluebird (Sialia sialis)
Yes
Altricial
edge
field
0.79
0.76-0.79
apples, corn
Eastern kingbird (Tyrannus tyrannus)
Yes
Altricial
edge
field
0.45
0.01-0.45
alfalfa, corn, grapes
Eastern meadowlark (Sturnella magna)
Yes
Altricial
field
NA
0.89
0.88-0.89
alfalfa, corn
Page 24 of 77
-------
Eastern phoebe (Sayornis phoebe)
Yes
Altricial
edge
NA
0
0
corn
Field sparrow (Spizella pusilla)
Yes
Altricial
edge
NA
1
0.03-1
corn
Grasshopper sparrow (Ammodramus
savannarum)
Yes
Altricial
field
NA
0.56
0.15-0.56
alfalfa, corn
Great-tailed Grackle (Quiscalus mexicanus)
Yes
Altricial
edge
NA
0.35
0.35
corn
Horned lark (Eremophila alpestris)
Yes
Altricial
field
NA
0.88
0.36-0.88
corn, cotton, soybeans
House finch (Carpodacus mexicanus)
Yes
Altricial
edge
field
0.32
0.02-0.32
citrus, cotton
House sparrow (Passer domesticus)
Yes
Altricial
edge
NA
0.63
0.13-0.63
alfalfa, corn, soybeans
House wren (Troglodytes aedon)
Yes
Altricial
edge
NA
0.24
0-0.24
alfalfa, corn, potatoes
Killdeer (Charadrius vociferus)
No
Precocial
field
field
1
0.65-1
alfalfa, cabbage, corn, grapes, soybeans
Lark bunting (Calamospiza melanocorys)
Yes
Altricial
edge
NA
0.07
0.07
corn
Lark sparrow (Chondestes grammacus)
Yes
Altricial
edge
NA
0.79
0.33-0.79
corn, cotton
Mallard (Anas platyrhynchos)
No
Precocial
edge
edge
0.69/0.89*
none
none
Mourning dove (Zenaida macroura)
No
Altricial
edge
field
0.73
0-0.73
alfalfa, apples, citrus, corn, cotton,
potatoes, soybeans
Northern bobwhite (Colinus virginianus)
No
Precocial
edge
NA
0.53
0.01-0.53
alfalfa, corn, cotton, potatoes
Northern cardinal (Cardinalis cardinalis)
Yes
Altricial
edge
NA
0.68
0.01-0.68
alfalfa, corn, cotton, potatoes, soybeans
Northern flicker (Colaptes auratus)
No
Altricial
edge
NA
0.29
0-0.29
cabbage, corn, potatoes
Northern mockingbird (Mimus polyglottos)
Yes
Altricial
edge
edge
0.35
0.04-0.35
citrus, cotton, potatoes
Ovenbird (Seiurus aurocapillus)
Yes
Altricial
edge
NA
0
0
potatoes
Red-winged blackbird (Agelaius
phoeniceus)
Yes
Altricial
edge
edge
0.89
0.01-0.89
alfalfa, cabbage, citrus, corn, cotton,
grapes, potatoes, soybeans
Savannah sparrow (Passerculus
sandwichensis)
Yes
Altricial
field
field
0.87
0.14-0.87
alfalfa, apples, cabbage, corn, grapes
Tree swallow (Tachycineta bicolor)
Yes
Altricial
edge
NA
0.69
0-0.69
cabbage, cotton
Verdin (Auriparus flaviceps)
Yes
Altricial
edge
NA
0.57
0.57
cotton
Vesper sparrow (Pooecetes gramineus)
Yes
Altricial
field
NA
0.5
0.13-0.50
alfalfa, corn, soybeans
Western meadowlark (Sturnella neglecta)
Yes
Altricial
field
NA
0.33
0-0.33
alfalfa, corn
White-crowned sparrow (Zonotrichia
leucophrys)
Yes
Altricial
edge
NA
0.48
0.48
corn
White-winged dove (Zenaida asiatica)
No
Altricial
edge
NA
0.06
0.06
cotton
Willow flycatcher (Empidonax trailii)
Yes
Altricial
edge
NA
0
0
corn
Wood thrush (Hylocichla mustelina)
Yes
Altricial
edge
NA
0.01
0.01
potatoes
Yellow warbler (Dendroica petechia)
Yes
Altricial
edge
NA
0.29
0-0.29
corn, potatoes
Yellow-rumped warbler (Dendroica
magnolia)
Yes
Altricial
edge
NA
0.09
0.09
corn
*No data are available from avian census studies. Default values are usee
for fi el d/or chard.
Page 25 of 77
-------
Table 2.4. Custom Species Parameter Values for BW2
Species (scientific name)
Female BW (g)
Male BW (g)
Mean
SD
Min
Max
Mean
SD
Min
Max
American crow (Corvus brachyrhynchos)
438
32.0
289
666
458
33.4
302
696
American goldfinch (Carduelis tristis)
12.6
0.81
10
17.1
13.2
1.13
8.6
20.7
American kestrel (Falco sparverius)
120
9.2
79
182
111
9.3
73
169
American robin (Turdus migratorius)
77.3
0.38
63.5
103
77.3
0.38
63.5
103
Ash-throated flycatcher (Myiarchus cinerascens)
27.2
2.0
24
31
27.2
2.0
24
31
Barn swallow (Hirundo rustica)
18.6
1.49
13.4
23.4
18.6
1.49
13.4
23.4
Black-capped chickadee (Poecile atricapillus)
10.8
1.38
8.2
13.6
10.8
1.38
8.2
13.6
Blue jay (Cyanocitta cristata)
86.8
8.08
64.1
109
86.8
8.08
64.1
109
Blue-gray gnatcatcher (Polioptila caerulea)
6
0.13
4.8
8.9
6
0.13
4.8
8.9
Blue-winged Teal (Anas discors)
363
26
240
545
409
29.9
270
590
Boat-tailed Grackle (Quiscalus major)
119
8.7
102
132
214
15
175
253
Bobolink (Dolichonyx oryzivorus)
37.1
2.71
26.5
44.3
47
3.4
28.5
56.3
Brewer's blackbird (Euphagus cyanocephalus)
58.1
4.9
50.6
67
67.2
3.2
60
73
Canada goose (Branta canadensis)
3514
257
3062
3912
4181
305
3799
4727
Carolina chickadee (Poecile carolinensis)
9.8
0.59
6.5
14.9
10.5
0.72
6.9
16.0
Carolina wren (Thryothorus ludovicianus)
21
1.15
14
32
21
1.15
14
32
Cassin's sparrow (Aimophila cassinii)
18.9
1.51
14
23.5
18.9
1.51
14
23.5
Cedar waxwing (Bombycilla cedrorum)
33.1
1.07
28
40.2
30.6
1.72
25.5
39.6
Chipping sparrow (Spizella passerina)3
12.5
1.47
10.2
16.5
12.5
1.47
10.2
16.5
Common grackle (Quiscalus quiscula)
100
7.3
66
152
127
9.3
84
193
Common yellowthroat (Geothlypis trichas)
9.9
0.78
7.6
15.3
10.3
0.66
7.6
15.5
Dark-eyed junco (Junco hyemalis)
18.8
0.78
14.3
25.1
20.4
1.21
14.3
26.7
Dickcissel (Spiza americana)
24.6
1.8
16
37
29.3
2.1
19
45
Eastern bluebird (Sialia sialis)
31.6
0.92
21
48
31.6
0.92
21
48
Eastern kingbird (Tyrannus tyrannus)
39.5
1.85
35.8
40.8
39.5
1.85
35.8
40.8
Eastern meadowlark (Sturnella magna)
76
5.5
50
116
102
11.2
67
155
Eastern phoebe (Sayornis phoebe)
19.8
7.47
11.4
24.4
19.8
7.47
11.4
24.4
Field sparrow (Spizella pusilla)
12.5
1.47
10.2
16.5
12.5
1.47
10.2
16.5
Grasshopper sparrow (Ammodramus savannarum)
17
1.34
15
20.3
17
1.34
15
20.3
Great-tailed Grackle (Quiscalus mexicanus)
107
11.4
96
140
191
22.8
157
234
Horned lark (Eremophila alpestris)
30.8
2.2
20
47
31.9
2.3
21
48
House finch (Carpodacus mexicanus)
21.4
1.29
10
25.5
21.4
1.29
10
25.5
House sparrow (Passer domesticus)
27.4
2.24
20.1
34.5
28
1.55
20
34
House wren (Troglodytes aedon)
10.9
0.8
8.9
14.2
10.9
0.8
8.9
14.2
Killdeer (Charadrius vociferus)
101
7.37
87.7
121
92.1
10.4
83.9
109
Lark bunting (Calamospiza melanocorys)
37.6
3.66
29.5
51.5
37.6
3.66
29.5
51.5
Lark sparrow (Chondestes grammacus)
29
1.94
24.7
33.3
29
1.94
24.7
33.3
Mallard (Anas platyrhynchos)
1082
129
720
1580
1082
129
720
1580
Mourning dove (Zenaida macroura)
115
1.76
76
175
123
1.85
81
187
Northern bobwhite (Colinus virginianus)
178
13.0
117
271
178
13.0
117
271
Northern cardinal (Cardinalis cardinalis)
43.9
4.53
33.6
64
45.4
4.29
33.7
63.2
Northern flicker (Colaptes auratus)
129
7.67
106
164
135
6.37
114
160
Northern mockingbird (Mimus polyglottos)
48.5
3.5
36.2
55.7
48.5
3.5
36.2
55.7
Ovenbird (Seiurus aurocapillus)
19.4
1.22
14
28.8
19.4
1.22
14
28.8
Red-winged blackbird (Agelaius phoeniceus)
41.5
2.74
29
55
63.6
4.43
52.9
81.8
2 Body weights from Dunning (1984). If standard deviations, minimum and maximum values were not available, these values were
calculated by multiplying the mean by 0.073, 0.66 and 1.52, respectively. (See chapter 2).
3 Data not available in Dunning (1984) for chipping sparrow. Body weight information for field sparrow (same genus) used.
Page 26 of 77
-------
Species (scientific name)
Female BW (g)
Male BW (g)
Mean
SD
Min
Max
Mean
SD
Min
Max
Savannah sparrow (Passerculus sandwichensis)
19.5
2.29
13
30
20.6
1.35
14
31
Tree swallow (Tachycineta bicolor)
20.1
1.58
15.6
25.4
20.1
1.58
15.6
25.4
Verdin (Auriparus flaviceps)
6.8
0.69
5.5
8.5
6.8
0.69
5.5
8.5
Vesper sparrow (Pooecetes gramineus)
24.9
1.8
16
38
26.5
1.9
17
40
Western meadowlark (Sturnella neglecta)
89.4
6.5
59
136
106
7.7
70
161
White-crowned sparrow (Zonotrichia leucophrys)
32
2.18
27
35.5
32
2.18
27
35.5
White-winged dove (Zenaida asiatica)
153
13.2
125
187
153
13.2
125
187
Willow flycatcher (Empidonax trailii)
13.7
1.46
11.3
16.4
13.1
1.37
12
15.7
Wood thrush (Hylocichla mustelina)
47.4
4.17
39.2
57.7
47.4
4.17
39.2
57.7
Yellow warbler (Dendroica petechia)
9.2
0.59
7.4
16
9.8
0.68
7.9
12.8
Yellow-rumped warbler (Dendroica magnolia)
8.5
0.35
6.6
12.6
8.9
0.58
7
12.9
Page 27 of 77
-------
Table 2.5. Custom Species Parameters for Diet. Based on MCnest Species Library
Species (scientific name)
Feeding
4
category
Adult diet
Juvenile diet
insects
seeds
fruit
grass
broadleaf
insects
seeds
fruit
grass
broadleaf
American crow (Corvus brachyrhynchos)
Omnivore
0.28
0.54
0.18
0
0
0.165
0.835
0
0
0
American goldfinch (Car due lis tristis)
Granivore
0
1
0
0
0
1
0
0
0
American kestrel (Falco sparverius)
Insectivore
1
0
0
0
0
1
0
0
0
0
American robin (Turdus migratorius)
Insectivore
0.72
0
0.28
0
0
0.7
0
0.3
0
0
Ash-throated flycatcher (Myiarchus cinerascens)
Insectivore
1
0
0
0
0
1
0
0
0
0
Barn swallow (Hirundo rustica)
Insectivore
1
0
0
0
0
1
0
0
0
0
Black-capped chickadee (Poecile atricapillus)
Insectivore
0.9
0.05
0.05
0
0
1
0
0
0
0
Blue jay (Cyanocitta cristata)
Omnivore
0.4
0.6
0
0
0
0.4
0.6
0
0
0
Blue-gray gnatcatcher (Polioptila caerulea)
Insectivore
1
0
0
0
0
1
0
0
0
0
Blue-winged Teal (Anas discors)
Insectivore
0.91
0.09
0
0
0
1
0
0
0
0
Boat-tailed Grackle (Quiscalus major)
Insectivore
0.92
0
0.08
0
0
1
0
0
0
0
Bobolink (Dolichonyx oryzivorus)
Omnivore
0.57
0.43
0
0
0
1
0
0
0
0
Brewer's blackbird (Euphagus cyanocephalus)
Insectivore
0.82
0.18
0
0
0
1
0
0
0
0
Canada goose (Branta canadensis)
herbivore
0
0
0
1
0
0
0
1
0
Carolina chickadee (Poecile carolinensis)
Insectivore
0.9
0.05
0.05
0
0
1
0
0
0
0
Carolina wren (Thryothorus ludovicianus)
Insectivore
0.98
0.02
0
0
0
1
0
0
0
0
Cassin's sparrow (Aimophila cassinii)
Insectivore
1
0
0
0
0
1
0
0
0
0
Cedar waxwing (Bombycilla cedrorum)
Frugivore
0.2
0
0.8
0
0
0.2
0
0.8
0
0
Chipping sparrow (Spizella passerina)
Omnivore
0.38
0.62
0
0
0
0.2
0.8
0
0
0
Common grackle (Quiscalus quiscula)
Granivore
0.3
0.7
0
0
0
1
0
0
0
0
Common yellowthroat (Geothlypis trichas)
Insectivore
1
0
0
0
0
1
0
0
0
0
Dark-eyed junco (Junco hyemalis)
Omnivore
0.6
0.4
0
0
0
1
0
0
0
0
Dickcissel (Spiza americana)
Insectivore
0.7
0.3
0
0
0
1
0
0
0
0
Eastern bluebird (Sialia sialis)
Insectivore
0.93
0
0.07
0
0
1
0
0
0
0
Eastern kingbird (Tyrannus tyrannus)
Insectivore
0.855
0
0.145
0
0
1
0
0
0
0
Eastern meadowlark (Sturnella magna)
Insectivore
0.9
0.1
0
0
0
1
0
0
0
0
Eastern phoebe (Sayornis phoebe)
Insectivore
1
0
0
0
0
1
0
0
0
0
Field sparrow (Spizella pusilla)
Omnivore
0.5
0.5
0
0
0
1
0
0
0
0
Grasshopper sparrow (Ammodramus savannarum)
Omnivore
0.61
0.39
0
0
0
1
0
0
0
0
Great-tailed Grackle (Quiscalus mexicanus)
Insectivore
1
0
0
0
0
1
0
0
0
0
Horned lark (Eremophila alpestris)
Granivore
0.27
0.73
0
0
0
1
0
0
0
0
House finch (Carpodacus mexicanus)
Granivore
0.05
0.88
0.07
0
0
0.02
0.98
0
0
0
4 If species' diets are > 70% insects, they are identified as insectivores. This same cutoff applies to herbivores and granivores. If no dietary item represents > 70% of the overall
diet, the species is considered an omnivore.
Page 28 of 77
-------
Species (scientific name)
Feeding
4
category
Adult diet
Juvenile diet
insects
seeds
fruit
grass
broadleaf
insects
seeds
fruit
grass
broadleaf
House sparrow (Passer domesticus)
Granivore
0.04
0.96
0
0
0
0.68
0.32
0
0
0
House wren (Troglodytes aedon)
Insectivore
1
0
0
0
0
1
0
0
0
0
Killdeer (Charadrius vociferus)
Insectivore
1
0
0
0
0
1
0
0
0
0
Lark bunting (Calamospiza melanocorys)
Omnivore
0.64
0.36
0
0
0
1
0
0
0
0
Lark sparrow (Chondestes grammacus)
Omnivore
0.5
0.5
0
0
0
1
0
0
0
0
Mallard (Anas platyrhynchos)
Insectivore
0.72
0.28
0
0
0
0.9
0.1
0
0
0
Mourning dove (Zenaida macroura)
Granivore
0
1
0
0
0
0
1
0
0
0
Northern bobwhite (Colinus virginianus)
Granivore
0.2
0.8
0
0
0
0.9
0.1
0
0
0
Northern cardinal (Cardinalis cardinalis)
Omnivore
0.61
0.39
0
0
0
1
0
0
0
0
Northern flicker (Colaptes auratus)
Insectivore
0.9
0
0.1
0
0
0.9
0
0.1
0
0
Northern mockingbird (Mimus polyglottos)
Insectivore
0.85
0
0.15
0
0
1
0
0
0
0
Ovenbird (Seiurus aurocapillus)
Insectivore
1
0
0
0
0
1
0
0
0
0
Red-winged blackbird (Agelaius phoeniceus)
Insectivore
0.71
0.29
0
0
0
1
0
0
0
0
Savannah sparrow (Passerculus sandwichensis)
Insectivore
1
0
0
0
0
1
0
0
0
0
Tree swallow (Tachycineta bicolor)
Insectivore
1
0
0
0
0
1
0
0
0
0
Verdin (Auriparus flaviceps)
Insectivore
1
0
0
0
0
1
0
0
0
0
Vesper sparrow (Pooecetes gramineus)
Omnivore
0.56
0.44
0
0
0
1
0
0
0
0
Western meadowlark (Sturnella neglecta)
Insectivore
0.9
0.1
0
0
0
1
0
0
0
0
White-crowned sparrow (Zonotrichia leucophrys)
Omnivore
0.36
0.64
0
0
0
1
0
0
0
0
White-winged dove (Zenaida asiatica)
Granivore
0
1
0
0
0
1
0
0
0
Willow flycatcher (Empidonax trailii)
Insectivore
1
0
0
0
0
1
0
0
0
0
Wood thrush (Hylocichla mustelina)
Omnivore
0.65
0
0.35
0
0
1
0
0
0
0
Yellow warbler (Dendroica petechia)
Insectivore
1
0
0
0
0
1
0
0
0
0
Yellow-rumped warbler (Dendroica magnolia)
Insectivore
0.85
0
0.15
0
0
1
0
0
0
0
Page 29 of 77
-------
3. Modeling Bird Behavior: Feeding and Location
Hourly pesticide exposures through all pathways {i.e., feeding, inhalation, dermal and drinking)
are a function of the time relative to the feeding pattern and the presence or absence of the
simulated bird on the treated field. When a bird is on a treated field during a time step, it is
assumed that the bird may be exposed to the pesticide through any of the exposure pathways
considered in TIM. When the bird is off of the field, it is exposed to a fraction of the on-field
exposure that is based on the spray drift deposition relative to the bird's location with respect to
the edge of the treated field. TIMv.3.0 uses a bimodal feeding period to represent the feeding
behavior of birds during the day. In one day, a bird feeds during the morning and afternoon. A
Markov chain is used to model the movement of birds on and off of the field during feeding
hours. The location of an individual bird during non-feeding hours is different for field- and
edge-resident species. Edge species are assumed to be off of the treated field during non-feeding
hours, while field species are assumed to be located on the treated field during non-feeding
hours.
3.1. Bimodal Feeding Model to Describe Feeding Behavior
TIM incorporates a flexible, probability-based, algorithm to represent bird feeding behavior. The
bimodal feeding model is based on the assumption that birds have two distinct feeding periods
during the day (based on a recommendation of the 2001 SAP); i.e., a morning feeding period and
an afternoon feeding period (Figure 3.1). Each day is represented by a 24-hour clock, with hour
0 representing midnight to 1 am.
The beginning and ending times of both the morning and afternoon feeding periods are assumed
to vary randomly each day, within specified time windows, and vary from bird to bird. These
windows are based on sunrise and sunset times, as well as the heat of the day. Uniform
distributions are established for the morning start, morning end, afternoon start and afternoon
end times, using the minimum and maximum start/stop values entered by the model user. A
description of the uniform distribution is available in USEPA (2007a). TIM also assigns uniform
distributions to represent the mean of the morning and the afternoon feeding periods, using the
uniform distributions of the start and ending periods for each feeding time. For each bird, the
following times are assigned from their respective uniform distributions:
• morning start (ammin)
• morning end (ammax)
• morning mode (ammode)
• afternoon start (pmmin)
• afternoon end (pmmax)
• afternoon mode (pmmode)
Page 30 of 77
-------
0.5
o
0.4
L=
1)
=1
r^I-
0.3
E_
_L
0.2
0.1
0.0
f
/
-\
,
,
/
t
,
6 9 12 15 18 21 24
Hour of the Day
0,5
0.4
0.3
0.2
0.1
0.0
_L
9 12 15 18 21 24
0,5
0,4
0,3
0.2
0,1
0,0
0,5
0,4
0,3
0.2
0,1
0,0
{
_
\
.
V
,
,
12 15 18 21 24
1
1
;
i
>
_
\
i
V
i
i
/ i
,
12 15 18 21 24
Figure 3.1. Hypothetical Examples of the Avian Bimodal Feeding Pattern. X-axis is hour of
day; Y-axis is daily dietary fraction.
Each day of a simulation, the proportion of daily feeding is divided between the morning and
afternoon feeding periods (Figure 3.1). This distribution of feeding is determined by a variable
termed "Split" (S). Each day of a simulation, a different S value is selected for an individual bird
from a uniform distribution with minimum and maximum values established by the model user.
On a given hour that occurs during a feeding period, the proportion of the daily diet consumed
during that hour (HF(t)) is determined from beta pert (PP) distributions (See Vose, 1996;
Equations 3.1 and 3.2). These values are generated using the minimum, maximum and mode
feeding times selected for each bird, where time (t) is in hours. If t is outside of the morning and
afternoon feeding times of the simulated bird, HF(t) is equal to 0. Figure 3.2 illustrates several
random bimodal feeding patterns as they may be incorporated into TIM. A general description of
equations used to derive a beta pert distribution is provided in USEPA (2004).
Equation 3.1. HF(t) =S*0p{f, arnmm , ammode, amB
[Morning feeding time period]
Equation 3.2. HF(t) = (l - S)* f3 At \prnmm , pmmode , pmm3X) [Afternoon feeding time period]
Page 31 of 77
-------
0.5
0,4
0,4
0,3 -
0.2 -
0,1 -
0.0 -
0.5
0.4
0.3
0.2
0.1
0.0
12
IS
24
ill
£L
0.3
0.2
0.1
0.0
0.5
0.4
0.3
0.2
0.1
0.0
±1_
12
18
24
m
12
18
24
12
18
24
Figure 3.2. Examples of Hourly Feeding Fractions Used in TIM. X-axis is hour of day; Y-
axis is daily dietary fraction.
3.2. Markov Chain Model to Describe Adult Movement during Feeding Periods
The presence on-field, off-field parameter (st), is modeled as a first-order, two-state Markov
chain model. In 2004, the SAP agreed that the Markov chain allowed "a realistic characterization
of serial behavior." The Markov chain model is a statistical model for the persistence of binary
events, in this case, whether or not an individual bird is on or off the field in any particular hour
during feeding hours. In this application, two-state refers to the state X= 0, where the bird is off
the field, or state X= 1, where the bird is on the field. First-order means that the probability of
whether a bird is on the field or off the field in any hour depends only on the state (location) of
the bird in the previous hour. A first-order, two-state Markov chain is specified by four
transitional probabilities for a bird's state at time t+1, given the bird's state at time /,
, where:
Poo
PlO
Pol
Pu
• Poo = Prob {A', / = 0 A', = 0) = probability that a bird, now off the field, will remain off the field in the next hour;
• I'u] = Prob {A', / = 1 | A', = 0) = probability that a bird, now off the field, will be on the field in the next hour;
• Pi i = Prob {A', , = 1 | A', = 1) = probability that a bird, now on the field, will remain on the field in the next hour;
• Pio = Prob {A', / = 0 | A', = 1) = probability that a bird, now on the field, will be off the field in the next hour.
Page 32 of 77
-------
These transitional probabilities are illustrated in Figure 3.3. As usual in Markov Chain theory,
the rows of the transition matrix sum to unity, i.e., Poo + Poi = 1, and P]o + P\\ = 1.
Figure 3.3. The Two-state, First-order Markov Chain Model for Avian Location on or off a
Treated Field.
The location of a bird on or off of the treated field at time t (st) is determined considering the
bird's location in the previous hour and Poo and Pn. TIM estimates this value by first selecting a
random number (U) from a uniform distribution ranging from 0 to 1.
If the bird is OFF of the treated field in the previous time step (t-1), U is compared to Poo. IfU <
Poo, then the bird is off of the treated field at time t (i.e., st= 0). IfU > Poo, then the bird is on the
treated field at time t (i.e., st= 1). Therefore, the lower the value of Poo, the more likely it is that
the bird will move from the edge at time t-1 to the field at time t.
Likewise, if the bird is ON the treated field in the previous time step (t-1), U is compared to Pn.
IfU ^ Pn, then the bird is on the treated field at time t (i.e., st 1). IfLJ ^ P11, then the bird is off
of the treated field at time t (i.e., st= 0). Therefore, the higher the value of P11, the more likely it
is that the bird will stay on the field from time t-1 to time t.
The long run probability of a bird being on the field (FOF) is represented by Equation 3.3. P11 is
the probability that a bird on the field will remain on the field the following hour. This parameter
is treated as a random variable with a triangular distribution (Pn ~ Triangular(Pn Min, 1.0, Pn
mode)). A description of this distribution is available in USEPA 2007a. Given an estimate of
FOF for an individual bird (selected from a beta pert distribution), the minimum value that the
conditional probability Pn can assume is derived according to Equation 3.4. The mode of Pn is
derived using Equation 3.5. Note that O, the fidelity factor, is the fraction of the range of
permitted values for Pn. which helps specify the location of the mode. Q is discussed in more
detail below. With an estimate of Pn, Poi is calculated using Equation 3.6. Poo can be calculated
using Poi and Equation 3.7. P10 can be calculated using Pn and Equation 3.8.
Equation 3.3. FOF = ———
¦ 1 _i_ p _ p
r 11
Page 33 of 77
-------
Equation 3.4. Min of Pu -Max of
2* FOF - \\
, 0
FOF
Equation 3.5. mode Pu = Min of Pu +Q* APn
Where: min °f -^11
Equation 3.6. p = FOF*(l~pn)
01 1 -FOF
Equation 3.7. P00= 1 - Pc
01
Equation 3.8. Pw=l-Pn
As discussed below in Section 3.4, birds that are off field receive a fraction of the on-field
exposure that is determined by the spray drift deposition at their location relative to the edge of
the treated field. For birds that are located entirely within the edge habitat (i.e., FOF = 0), the
Markov Chain is not necessary for tracking movement. Rather, these birds move to random
locations within the edge habitat.
3.3. Juvenile Locations
During the simulation, juvenile locations depend upon whether the species is altricial or
precocial. Altricial birds are those that do not leave the nest until they are essentially adults (e.g.,
passerines are altricial). On the other hand, the young of precocial birds leave the nest when they
are hatchlings. Examples of precocial birds include waterfowl and upland game birds. In TIM
(v.3.0), it is assumed that altricial birds are located at their nest site and that the precocial birds
follow their parents, thus having the same location.
Whether a species is precocial or altricial impacts the exposures received by the juveniles.
Precocial juveniles receive similar exposures as the parents. Altricial juveniles do not receive
dermal contact or drinking water exposures. If altricial nests are located off of the treated field,
the juveniles receive only a fraction of the on-field exposure from inhalation and direct dermal
spray.
3.4. Accounting for Off-Field Exposures Due to Spray Drift
When birds are off of the field, spray drift transport and off-field deposition may result in
pesticide exposures that are a fraction of what they would receive on the treated field. This
fraction is determined based on the bird's location relative to the edge of the treated field. The
bird's location relative to the edge of the field is calculated by considering the area of the bird's
home range and the extent of overlap between the home range and the treated field.
Figure 3.4 depicts an example of three scenarios that may occur with overlaps between the home
ranges of simulated birds and the treated field and edge habitat. This figure depicts the treated
Page 34 of 77
-------
field, the spray drift zone and areas that do not receive pesticide exposure either because they are
beyond the spray drift zone (i.e., >304 m from the edge of the field), or the adjacent edge habitat
does not receive spray drift. In this example, the range of bird #1 overlaps with the treated field,
the spray drift transport area and the area receiving no pesticide exposure. The range of bird #2
overlaps with the spray drift area and the area receiving no pesticide exposure, but not the treated
field. The range of bird # 3 overlaps with the treated field and the edge habitat that does not
receive spray drift. Although not depicted in this figure, it is also possible that the home range of
a bird may overlap only with the treated field and the area receiving spray drift.
304 m
Figure 3.4. Example Overlap between Treated Field and Bird Home Ranges.
In this scenario, it is assumed that 100% of the edge habitat receives spray drift. This assumption
may be the case where there is a limited area of non-cropped habitat and other sides of the
treated field are represented by cropped fields. In this case, scenarios represented by bird #3
would not be included in a simulation. The model user may choose to alter this assumption by
entering an input parameter to represent the fraction of the edge habitat that receives spray drift.
For example, this assumption may be altered to represent fields with prevailing winds that reach
only a portion of habitat that is suitable for birds. If this is the case, the model randomly assigns a
proportion of the simulated birds' home ranges to overlap with edge habitats that do not receive
spray drift. This approach assumes that the birds are uniformly distributed and the proportion of
birds in the edge habitat receiving spray drift matches the model user's assumed fraction.
This approach also focuses on the risks to birds that feed on treated fields and in adjacent
habitats and assumes that all birds visit these areas at some point during the simulation. There are
Page 35 of 77
-------
no birds simulated that have home ranges entirely outside of the treated site and spray drift zone.
In the context of a population of birds, TIM assumes that all birds are being exposed to the
pesticide. TIM does not account for risks where a portion of the population is not exposed to the
pesticide.
3.4.1. Determining an Individual Bird's Distance from the Edge of the Field
The bird's distance from the edge of the field is determined based on the dimensions of the home
range (Section 2.3) and FOF (Section 2.4). In this approach, it is assumed that the home range is
a square with each side equal to the square root of the area. In Figure 3.4, bird #1 is d; bird #2 is
d', in meters. This section describes how TIM calculates a bird's distance from the edge of the
field using birds 1, 2 and 3 as examples.
A bird that has a frequency on field (FOF) value >0 has overlap between some portion of its
home range and the treated field (e.g., bird #1). Portions of the bird's range also overlap with the
area receiving spray drift deposition and possibly the area where it is assumed that there is no
spray drift deposition. For bird #1, d is represented by two segments: di, which is the portion of d
that overlaps with the treated field and di, which is the portion of d that is off of the treated field.
The farthest distance from the edge of the field that the bird will travel is di, which is calculated
by subtracting di from d. di is calculated using the area of the portion of the bird's range that
overlaps with the field. This area (A0veriaP) is calculated by multiplying the bird's home range (A)
by its FOF. Since the area of a rectangle is calculated by multiplying the lengths of its sides (i.e.,
d and di), di can be calculated by rearranging Equation 3.9 into Equation 3.10. When a bird is
located off of the treated field, its distance from the edge of the field at time t (dt) is determined
by randomly selecting a distance from a uniform distribution of values between 0 and di in
meters.
Equation 3.9. Aovenap = A* FOF = d *
A*F OF
Equation 3.10. = —-—
For birds that do not have home ranges overlapping with the treated field (e.g., bird #2), it is
assumed that some portion of their home range overlaps with the area receiving spray drift
deposition from the treated field. In this case, the FOF of the bird is 0. For these birds, some or
all of their home range may overlap with the area receiving spray drift from the field. The
distance between the bird's home range and the edge of the treated field (i.e., d3') varies by bird.
This value is selected as a random value from a uniform distribution, ranging from 0 to 303 m.
Because 304 m is the predictive limit of the AgDRIFT model (Tier 1) used to predict spray drift
deposition, a maximum of 303 m was selected as the farthest distance for the edge of the home
range. For birds with FOF = 0, at time t, the maximum distance a bird may be located away from
the treated field is calculated by adding d3' to the length of the home range (i.e., d'). During the
simulation, the bird's location at any time point is determined by selecting a random value from
the uniform distribution of values between d3' and d3'+d'.
Page 36 of 77
-------
In cases where the model simulation assumes that only a portion of the edge habitat receives
spray drift when birds are on the treated field, they receive pesticide exposure. When those birds
are off of the field, they receive no additional pesticide. For example, see Bird #3 in Figure 3.4.
When adults are not feeding, it is assumed that they are located in the center of their home
ranges. For altricial species, this is akin to sitting on the nest. The use of the center of the home
range as the birds' locations during non-feeding hours is a simplifying assumption that does not
explicitly take into account the particular land attributes. Although precocial birds do not use a
nest once the offspring have hatched, it is assumed that they will rest in the center of their home
range. It is assumed that during non-feeding hours, juveniles are in the same location as their
parents. During all hours of the simulation, altricial juveniles are located in the center of the
home range {i.e., on the nest).
For field residents, it is assumed that the birds are located on the field during non-feeding hours.
For edge residents, the bird's distance from the edge of the field during non-feeding periods
(dnon-feeding) is calculated using d and di or d3 (depending upon the bird's FOF). If the adult's
home range overlaps with the treated field {i.e., FOF>0), the nest location is calculated by
subtracting di from the central point of the home range {i.e., d/2) (Equation 3.11). If this value is
<0 (meaning that the non-feeding location would be on the field), dnon-feeding is assumed to be 1 m
because the species is an edge resident. If FOF = 0, meaning that the adult's home range does not
overlap with the treated field, dnon-feeding is determined by adding the distance between the edges
of the field and the center of the home range {i.e., d3) and d/2 (Equation 3.11).
Equation 3.11. IfFOF>0, dnon_^ee(nng — d^
If FOF—0, dnon_feeciing — (^3 +—
3.4.2. Determining the Fraction of Exposure Compared to Field
For birds that have home ranges that overlap with edge habitat that receives spray drift
deposition {e.g., birds 1 and 2 in Figure 3.4) when they are off of the field, the birds receive a
fraction of the on-field exposure (Ffieid). This fraction is determined using the spray drift
deposition that corresponds to the bird's randomly selected distance from the edge of the treated
field (dt). The pesticide dose received by the bird when it is off of the field is calculated by
multiplying the on-field exposure by Ffieid. When a bird is located beyond the spray drift area,
exposure is zero {i.e., if dt > 303 m, then Ffieid = 0). When the bird is on the field {i.e., St = 1),
Ffieid = 1.
Spray drift deposition at different distances from the edge of the field was calculated using the
Tier I AgDRIFT model (v. 2.1.1)5. Spray drift deposition differs by application method, droplet
spectra and release height. The standard equation used by the developers of AgDRIFT is
5 http://www.epa.gov/oppefedl/models/water/
Page 37 of 77
-------
calculated using Equation 3.12. This equation can be used to calculate Ffieid. In this equation, a
factor of 3.28 is used to convert the units of dt from meters to feet.
Equation 3.12. F/ieW =
An analysis of the deposition curves generated from AgDRIFT 2.1.1 yielded the following
parameters found in Table 3.1. For some application methods, the curves were split at a
particular distance in order to get a better fit to the data. The break point distances were
determined by visual observation.
Table 3.1. Parameters for Spray Drift Equations, Based on Application Method
Application
method
Droplet spectrum (distance from edge of field)
a
b
c
Aerial
Very Fine to Fine (< 43 m)
Very Fine to Fine (> 43 m)
Fine To Medium (< 16 m)
Fine To Medium (> 16 m)
Medium to Coarse
Coarse to Very Coarse
0.0204
0.0292
0.1187
0.0241
0.0721
0.1014
0.7278
0.8220
0.5699
0.8689
1.0977
1.1344
0.5001
0.6539
0.5000
0.1678
0.4999
0.4999
Ground*,
high boom
Very Fine to Fine
Fine to Medium/Coarse
0.1913
2.4154
1.2366
0.9077
1.0552
1.0128
Ground*,
low boom
Very Fine to Fine
Fine to Medium Coarse
1.0063
5.5513
0.9998
0.8523
1.0193
1.0079
Airblast, vineyard
Not applicable
0.1349
1.4405
0.0376
Airblast, orchard
Not applicable (<26 m)
Not applicable (> 26 m)
0.0414
6.7728
2.1054
1.2788
0.2223
27.027
*Equations generate 90th percentile deposition values.
The model user can simulate the effects of an infield spray drift buffer on potential risks
associated with decreased spray drift deposition in the edge habitat. In this approach, the length
of the in-field buffer (B; in m) is added to dt prior to calculating the spray drift deposition at the
bird's location (Equation 3.13). If dt + B >303 m, Ffieid is 0. In this approach, if the bird is on the
field, it still receives the full on-field exposure. Decreased exposure in the portion of the field
that is represented by the buffer is not quantified.
Equation 3.13. FfleU = aw((JB),3 28),,
4. Estimating Pesticide Exposure through Diet
In TIM v.3.0, an individual bird's pesticide dose through diet at time t (Ddiet(t)) is calculated
according to Equation 4.1. This equation considers several variables (Table 4.1), including
pesticide concentrations on individual food items (k), fraction of the diet attributed to each food
item (DFk), the fraction of each food item that is contaminated (FCk), the total daily ingestion
rate (TDIR), the fraction of that total daily intake rate that can be attributed to the time step that
is being considered (i.e., the hourly fraction; HF(t), described in Section 3.1), the bird's BW and
the food matrix adjustment factor (FMA). As discussed in Section 3, the dose is adjusted based
on the bird's location. If the bird is on the field, then Ffieid = 1. If the bird is off field and within
Page 38 of 77
-------
an area receiving spray drift, then Ffieidis assigned a value <1 based on the spray drift deposition
at the bird's location during that hour. The FMA is a constant that is intended to account for the
difference in dose-based and dietary-based toxicity of a chemical, where this difference can be
attributed to effects of the food on the toxicity of the chemical. The equations and assumptions
used to generate Ck(t), TDIR, and FMA are provided below. Uncertainties associated with
Equation 4.1 and the individual input parameters are described in Section 10.
Equation 4.1 Ddiem = Ffleld
(¦TDIR*HF(t))*Y,{Ck(t)*DFk*FCk)
BW*FMA
Table 4.1. Parameters Used for Equations in Section 4 to Estimate Pesticide Exposure
Concentrations through Diet.
Symbol
Definition
Variable
type*
Units
AEk
Assimilation efficiency
Random
none
BW
Body weight
Random
g/bird
Ck(t)
Pesticide concentration on food item k at time t
Random
|ig pesticide/g food
DFk
Fraction of diet attributed to food item k
Constant
none
Ddien 11
Estimated exposure concentration through diet for a pesticide
at time t
Random
|ig pesticide/g-bw
FCk
Fraction of food that is contaminated
Constant
none
F field
Fraction of on field exposure
Random
none
FIS
Food ingestion scale factor
Random
none
FMA
Food matrix adjustment factor
Constant
none
FMR
Field metabolic rate
Random
kcal/bird-day
GEk
Gross energy
Random
Kcal / g food (wet
wt)
MEk
Metabolizable energy of food item k
Random
kcal/bird-day
Mhioi,i
Total metabolizable energy
Random
kcal/g food
TDIR
Total daily intake rate (for food)
Random
g food/bird-day
* "Constant" indicates that the parameter is set to one value. "Random" indicates that the parameter's value varies
based on a distribution of possible values.
In TIM v.3.0, the diet of the simulated bird is defined by five categories of dietary food items:
arthropods, seeds, fruits, grass, and broadleaf forage. Each food item is assigned a fraction of the
total diet of the model bird, where the sum of all fractions totals 1. The fraction of each item in
the total diet (DFk) is dependent on the species being modeled. As described in Section 2, the
user may select one of the generic species for which dietary fractions are preset, or choose to
define a custom species and assign dietary fractions for each food category. The bird's pesticide
dose at time t is calculated by considering pesticide concentrations at that time (Ck(t)) for each
food item (k) in the bird's diet.
Page 39 of 77
-------
4.1. Pesticide Concentrations on Food Items (Ck(t))
4.1.1. Pesticide Residues on Food items at the Time of Application (Ck(t=o))
The initial pesticide residue on the food items (Ck(t=o)) is normalized to represent the |ig
pesticide/g food resulting from 1 lb pesticide/A. For plants (grass, broadleaf, seeds and fruit),
initial residue concentrations on plants are from Fletcher et al. (1994). Initial residue
concentrations on arthropods are based on an analysis of data from the scientific literature and
registrant-submitted studies (Appendix E). The mean and standard deviations from these data
sets are used in TIM (Table 4.2) to represent the initial pesticide concentration on different food
items that results from an application of 1 lb a.i./A.
Table 4.2. Mean and Standard Deviations (SD) of Initial Concentrations of Pesticide on
Different Food Items Relevant to TIM. Values are normalized to jig pesticide/g food (ppm)
Food Item (k)
Mean
SD
Source
Arthropods*
65
48
See Appendix E
Seeds
4.0
5.9
Fletcher et al. (1994)
Fruit
5.4
9.8
Fletcher et al. (1994)
Grass
84.8
60.3
Fletcher etal. (1994)
Broadleaf forage
45.0
56.7
Fletcher et al. (1994)
*Also referred to in TIM documentation as "insects." This food item is intended to represent terrestrial insects as
well as spiders, centipedes, millipedes, etc.
In TIM, the initial residue concentration of pesticide on each food item is transformed to be
representative of the specific pesticide application by multiplying the mean and standard
deviation of the initial residue concentration by the application rate of the pesticide (in lbs
a.i./A). For each food item, a lognormal distribution is derived using the transformed mean and
standard deviation of the initial residue concentration on the food items. A description of the
lognormal distribution is available in USEPA (2007a). Each bird in the simulation is assigned a
set of initial pesticide residue values on the five modeled food items that are normalized to 1 lb
a.i./A. These initial residue values are selected randomly from the five different distributions
based on the mean and standard deviations provided in Table 4.2. These values are converted to
the initial residues by multiplying by the application rate (Equation 4.2).
Equation 4.2. C^q — Cfc(normaiized) * Arate
4.1.2. Pesticide Residues on Food Items After First Application (Ck(t>)
In modeling exposures over multiple days, it is necessary to account for dissipation of pesticide
residues from avian food sources over time. In the case that dissipation half-life data are
available for different food items, TIM allows the model user to enter separate dissipation half-
life values for all food items. Dissipation half-life values should be obtained from the open
literature or from registrant-submitted studies. In the case that data are only available for the
foliar dissipation half-life, this value should be used to represent the dissipation half-life for all
food items. If no foliar dissipation half-life is available, then a default value of 35 days is used
Page 40 of 77
-------
based on the work by Willis and McDowel (1987), which represents a high-end foliar dissipation
value from that source (where the maximum tm = 36.9 days). The user's manual of T-REX
(USEPA 2012) includes guidance for selecting the appropriate foliar dissipation half-life.
To derive the pesticide residue dissipation half-life values for a food item, model users should
input chemical-specific, measured dissipation half-lives from available sources if suitable data
are available. The mean half-life value (fr/2k in days) for a food item (k) is used to calculate a
mean dissipation rate for that food item (rk in hours) using Equation 4.3. The half-life value is
converted from days to hours by multiplying by 24 (hours/day).
Equation 4.3. rk = ln(0.5)/(-^1/2i *24)
To calculate residues in wildlife food items at the first time steps after the first application of a
pesticide to the field (Ck(t)), the exposure model randomly selects an hour 0 residue
concentration from the distributions described in sections above and dissipates this residue using
Equation 4.4. TIM allows the user to simulate up to 5 pesticide applications. In the case that
multiple applications are simulated, residues from all applications are added to determine the
total residue value at time t.
Equation 4.4. = Q0 *e~rtt
4.1.3. Contaminated Fraction on Food Items (FCk)
For broadcast applications, the entire treated field is assumed to be exposed to the applied
pesticide and therefore each of the avian food items found on such fields are judged to be
contaminated with the pesticide {i.e., FCk values for all food items is assumed to be 1). For in-
furrow or banded spray applications, the pesticide application is assumed to be limited to the
portion of the treated field constituting the furrows or bands. For plant food items {i.e., seeds,
fruit, grass, broadleaf), FCk should be set at a fixed value representing the area proportion of the
treated field to which pesticide is directly applied. For example, for an in-furrow treatments with
40-inch row spacing and 2-inch wide furrows, FCk is 0.05 for seeds. For insects, FCk should be
set to 1 for all applications because insects are assumed to be mobile across the entire field; thus
the application of pesticide to only furrow or bands is not assumed to affect the fraction of the
food item assumed to be contaminated.
4.2. Total Daily Food Intake Rate (TDIR)
TDIR is calculated using Equation 4.5. The total daily intake rate (TDIR; g food/bird-day) for
food is calculated by considering the field metabolic rate (FMR; kcal/bird-day), the total
metabolizable energy (MEtotai; kcal/g food) of the food consumed by the model bird as well as a
scaling factor (Sf) that introduces variability in the total amount of food consumed within a day.
Equations 4.5-4.10 are used to derive TDIR. Table 4.1 provides descriptions of the variables
used to calculate TDIR.
FMR
Equation 4.5. TDIR = * SF * G
H MEtotal h
Page 41 of 77
-------
Sf is a random variable that is selected from a beta distribution that is established, assuming that
the mean is 1.0, and the minimum and maximum values are 0.9 and 1.1, respectively. The
scaling factor allows the daily intake rate of a bird to vary ±10%. This variability may be
attributed to factors such as physiological differences among individuals within a species or
availability of food.
G is the gorging factor. A value >1 is intended to account for an increase in feeding that may
occur when a bird is migrating or when excessive prey may be available. An appropriate value
should be selected to represent the increase in feeding of a species based on available data.
ECOFRAM (1999) suggests that total daily intake increases by a factor of 2-3 after starvation
due to poor weather. ECOFRAM (1999) also indicates that the upper limit of zinc uptake for
nutritional requirements is thought to be 5-fold of normal daily consumption. A value of 1
represents normal feeding (e.g., during the breeding season). It should be noted that by
increasing the amount of food consumed through the use of the gorging factor will result in the
amount of drinking water that is consumed by the bird. As a result, a larger fraction of the bird's
total daily water requirement will be met through water contained in the diet. See Section 7 for
details on how the drinking water intake rate is calculated.
Avian-specific food ingestion rates are based on the allometric equations of Nagy (1987) for
passerine and non-passerine birds, relating BW to the free-living metabolic rate. (FMR units are
expressed in kcal/bird-day, using Equations 4.6 and 4.7, respectively). Equation 4.6 results in
a higher FMR value compared to the FMR generated for non-passerine birds, using Equation
4.7. Since the majority of species observed with high frequency on agricultural fields are
passerines (Appendix D), the use of the FMR equation that is representative of passerines was
chosen for the generic birds. If the model user chooses to simulate a custom species that is not a
passerine, then Equation 4.7 is used by the model. For juveniles, Equation 4.8 is used to
determine FMR. See Appendix F for details.
Equation 4.6. FMR = 2.123 * BW0749 (Adult passerines)
Equation 4.7. FMR = 1.146 * BW0749 (Adult non-passerines)
Equation 4.8. FMR = 1.197 * BW0 782 (Juveniles)
The total amount of metabolizable energy in the food of a bird is determined by considering the
fractions of the different food items (DFk is unitless) that make up the bird's diet and the amount
of metabolizable energy in each food item (MEk in kcal/bird-day) (Equation 4.9).
Equation 4.9. MEtotal - I(I)I''k *MEk)
The metabolizable energy of food item k (MEk) is estimated based on values for the gross energy
(GEk) and assimilation efficiency (AEk) for that food item (Equation 4.10 from USEPA (1993)).
GEk (kcal/g food (wet weight)) is based on individual fresh food items and is independent of the
organism consuming the food. AEk (unitless) of fresh food items is that portion of gross energy
that can be assimilated by the bird. The AEk values are based on assimilation efficiencies of
Page 42 of 77
-------
individual food items by birds. The AEk values for seeds can be selected to represent passerine
and non-passerine birds.
Equation 4.10. MEk = GEk * AEk
Distributions of GEk and AEk are based on mean and standard deviation (SD) values from
USEPA (1993) (Tables 4.3 and 4.4). For each bird, GEk values are selected from lognormal
(truncated) distributions. In TIM, it is assumed that AEk values form a beta distribution, with
minimum and maximum values set to 0 and 1, respectively. In TIM, GEk and AEk values vary
for each bird, each day of the simulation.
Table 4.3. Mean and Standard Deviation (SD) Values for Gross Energy (Kcal / g food (wet
wt)) Content of Fresh Avian Food Items (from USEPA (1993)).
TIM Food Item (k)
Food item description in
USEPA 1993
GEk
Mean
SD
Arthropods
Grasshoppers, crickets, beetles
1.6
0.26
Seeds
Dicot seeds
4.6*
1.0*
Fruit
Pulp and skin of fruit
1.1
0.30
Grass
Young grasses
1.3
0.13**
Broadleaf forage
Dicot leaves
0.63*
0.074*
*Calculated using gross energy content on dry weight basis and water composition of food item.
**No SD is available for this food item; therefore, this value was calculated as 10% of the mean.
Table 4.4. Mean and Standard Deviation (SD) Values for Assimilation Efficiency (unitless)
of Fresh Avian Food Items (from USEPA (1993)). Values are based on assimilation
efficiency of each food item by birds.
TIM Food Item (k)
Food item description
in USEPA (1993)
AEk
Mean
SD
Arthropods
Terrestrial insects
0.72
0.051
Seeds
Wild seeds
0.75*
0.59**
0.090*
0.13**
Fruit
Fruit pulp, skin
0.64
0.15
Grass and broadleaf forage
Grasses, leaves
0.47
0.096
*Value is specific to passerines.
**Non-passerine birds.
4.3. Food Matrix Adjustment Factor (FMA)
The food matrix adjustment factor (FMA) is a constant that is intended to account for the
difference in dose-based and dietary-based toxicity of a chemical where this difference can be
attributed to effects of the food matrix on the toxicity of the chemical. It should be noted that the
use of the term "matrix" does not relate to mathematics, but rather to the food medium. This
parameter is useful because effects to birds from a chemical {i.e., the threshold for individual
simulated birds) are determined from available dose-based toxicity studies, where birds are
Page 43 of 77
-------
exposed to the pesticide through gavage. The simulated pesticide exposure, however, is based on
dietary exposure.
The default assumption for this parameter value is 1, meaning that the food matrix does not alter
the dose-based toxicity of the chemical. This default should only be altered by the user when
chemical-specific data are available to quantify the effects of the food matrix on the dose-based
toxicity of the chemical. A FMA value >1 indicates that the dietary matrix decreases the dose-
based toxicity of the chemical, while a FMA value <1 indicates that the dietary matrix increases
the dose-based toxicity of the chemical.
A pesticide specific FMA can be obtained from two methods. The first and most reliable method
is to compare the results of two acute, dose-based exposures of birds to the pesticide. In one
dose-based toxicity test, birds should be dosed with the pesticide via a typical carrier (e.g., corn
oil). In the other dose-based toxicity test, the birds should be dosed with the pesticide contained
in food. If a statistically significant difference is observed in the two LD50 values, the FMA can
be derived by dividing the LD50 resulting from the food dose by the LD50 obtained using the
typical carrier.
The second method for obtaining a pesticide-specific FMA is to convert an available dietary
LC50 value obtained from a sub-acute, dietary study to a LD50 value and compare that to the LD50
obtained from an acute oral toxicity study. There are two major uncertainties associated with this
approach that should be considered by the user. First, the LC50 value from the sub-acute dietary
toxicity study is influenced by food spillage, which may result in an overestimated LC50. Second,
birds involved in the sub-acute, dietary study (age 5-14 days) are younger than those involved in
the acute, dose-based study (age >16 weeks).
5. Estimating Pesticide Exposure through Inhalation
The inhalation exposure model considers two inhalation pathways: the direct inhalation of
airborne droplets immediately following pesticide application, and inhalation of vapor phase
pesticide from plant surfaces. For both inhalation routes, the exposure is expressed as an inhaled
dose at time t (Dspray(t) and Dvapor(t)), which is converted to an acute oral basis using an
equivalency factor (Fre; Equation 5.1; Section 5.3). As discussed already, for all routes of
exposure, the bird's dose is adjusted based on its location relative to the treated field using the
Fgeld parameter.
Equation 5.1. ^ inhalation(t) spray(t) ^Vapor(t) ,) re ^field
5.1. Calculating Pesticide Exposure through Inhalation of Airborne Droplets
Inhalation exposure from applied pesticide droplets is considered for the first exposure time step
immediately following the pesticide application for aerial, airblast and ground applied sprays.
The pesticide dose inhaled by the bird in airborne droplets from a spray application (Dspray(t)) is
estimated using Equation 5.2. (See Table 5.1 for parameter descriptions). This equation
accounts for the pesticide concentration in the volume of air under the release height (Cair(t)(droPs);
Page 44 of 77
-------
see Section 5.1.1), the volume of air respired by the bird during the time step (Vinhaiation; see
Section 5.1.2) and the fraction of droplets that can be respired by birds (Frespired; see Section
5.1.3), These factors considered together result in a mass of pesticide (|ig) respired by the bird in
1 hour. This number is converted to a dose basis by dividing by the BW of the bird.
t-, s — p\ Cair(t)(drops)*Vinhalation*^ respired
Equation 5.2. L)spray(t) — —
Exposure through inhalation from applied pesticide droplets is considered only for the first
exposure time step immediately following the pesticide application. It is assumed that a
suspended droplet will have either settled or cleared from the application area by the next time
step, which is 60 minutes after application. Since TIM v.3.0 allows the model user to simulate up
to 5 pesticide applications, a bird could potentially be exposed to a pesticide in airborne droplets
for a total of 5 (separate) hours of the simulation.
Page 45 of 77
-------
Table 5.1. Parameters Used for Equations in Section 5 to Estimate Pesticide Exposure
through Inhalation
Symbol
Parameter Description
Variable
Type*
Units
Arate
Application rate from label
Constant
lb a.i./A
Bvol
The volume-based biotransfer factor; function of Henry's law
constant and Log Kow
Constant
|ig/L fresh weight
leaf/ |ig/L air
BW
Body weight
Random
g/bird
Cair(drops)(t)
Pesticide concentration in a volume of air for the time step
immediately following the pesticide application
Constant
Hg/mL
Cair(t)(vol)
Concentration of the pesticide in air at time t (resulting from
volatilization); function of Mpesticide, mpiant, and Bvoi
Random
Hg/mL
CH
Height of crop
Constant
m
D
Fraction of hour where pesticide is applied
Constant
none
Dllllwhll loll' 1 1
Dose through inhalation for a pesticide at time t
Random
|ig pesticide/g-bw
Dspray(t)
Droplet Inhalation Dose
Random
|ig pesticide/g-bw
DVapor(t)
Volatilization inhalation dose; function of pesticide
concentration in air, volume of inhaled air, and body weight of
the bird
Random
|ig pesticide/g-bw
Fam
The ratio of avian to mammalian pulmonary membrane
diffusion rates from USEPA 2004
Constant
none
F field
Fraction of on field exposure
Random
none
Fre
The avian route equivalency factor
Constant
none
F respired
Volumetric fraction of droplet spectrum not exceeding the upper
size limit of respired particles for birds
Constant
none
H
Henry's law constant
Constant
atm-m3/mol
IS
Inhalation scale factor
Random
none
Kow
Octanol-water partition coefficient
Constant
none
LD50
Lethal dose sufficient to kill 50% of exposed individuals
Constant
mg/kg= (ig/g
-Mpesticide
The pesticide concentration on the treated field at time t
(accounting for dissipation); function of application rate
Random
mg
niplant
The mass of plant (crop) per hectare based on user input
Constant
kg
R
Universal gas constant (8.205 e 5)
Constant
atm-m3/mol-K
RH
Height of spray release
Constant
m
Rrate
Respiration rate
Random
mL/h
T
Air temperature
Constant
K
Vair
The volume of air in 1 ha to a height equal to the height of the
crop canopy
Constant
L
Vinhalation
Volume of air respired
Random
mL
Pplant
The density of the crop tissue assumed as fresh leaf (0.77)
Constant
kg/L
* "Constant" indicates that the parameter is set to one value. "Random" indicates that the parameter's value varies
based on a distribution of possible values.
Page 46 of 77
-------
5.1.1. Pesticide Concentration in a Volume of Air (Cair(drops))
The pesticide concentration in a volume of air (Cair(t)(droPs)) for the time step immediately
following the pesticide application is calculated according to Equation 5.3. (See Table 5.1 for
parameter descriptions). For all other time steps, Cair(t)(droPs) = 0.
T? „ s t n _ D*Arate*0.112
equation 3..J. Lair(tXdrops) — ^
Equation 5.3 uses the application rate of the pesticide (Arate), the release height (RH) of the
application and the fraction of the time step where the pesticide is being applied (D). For aerial
applications, it is assumed that D = 0.025 based on 90 s duration of direct spray inhalation and
for ground spray applications, D = 0.0083 based on 30 s duration of direct spray applications.
For ground and aerial applications, RH is assumed to be a constant value of 1 m and 3.3 m,
respectively. D (hours) is calculated by dividing the duration of the application (in minutes) by
60 minutes to give a fraction in hours, which is the duration of the time step of interest. In this
equation, the factor of 0.112 is used to convert the units of the application rate, which are lb
a.i./A, to the metric units needed to generate a concentration value expressed in |ig a.i./mL of air.
5.1.2. Calculation of Inhaled Air Volume (Vinhaiation)
For each bird, in any given exposure time step within the model where inhalation exposure is
calculated, a volume of inhaled air (Vinhaiation) is determined according to Equation 5.4. (See
Table 5.1 for parameter descriptions). Based on the recommendation of USEPA (1993), the
respiration rate (Rrate) is multiplied by 3 to adjust the laboratory derived Rrate value to represent a
field respiration rate. Vinhaiation is varied using the inhalation scale factor (Si), which is randomly
selected from a beta distribution that is established, assuming that the mean is 1.0 and the
minimum and maximum values are 0.9 and 1.1, respectively. This factor is intended to allow for
variability in the Vinhaiation from one hour to the next, which may be attributed to varying amounts
of activity from one time to the next. The Si can allow the volume of respired air of a bird to vary
by up to 10% of the maximum hourly value.
Equation 5.4. Vinhalation = 3 * Rmte * S}
The respiration rate (Rrate) is calculated using an allometric relationship from USEPA (1993) that
relates avian resting respiration rate to BW (Equation 5.5). Since this equation uses BW values
that are in kg, the BW of a bird is divided by 1000 to convert from g to kg. The original equation
for Rrate, as provided in USEPA (1993), generates values in units of mL/minute, which are in turn
converted to an hourly time step by multiplying by 60 (min/hour).
Equation 5.5. Rrate = 60* (284 * (BW/iqqq) )
5.1.3. Fraction of Applied Pesticide Spray (Frespired)
The inhalation exposure model for airborne pesticide application droplet considers exposure only
to those droplets that may enter the avian lung. The size of the spray droplet spectrum that can be
Page 47 of 77
-------
inhaled into the lungs is conservatively assumed to be up to 100 [j,m in diameter. The fraction of
applied pesticide spray (Frespired) is therefore assumed to be the fraction of the spray droplet
spectrum that is <100[j,m. This value varies based on the application scenario of the pesticide
being modeled, with variability attributed to nozzle types. Table 5.2 includes the default values
for Frespired that were determined using the Tier III aerial module of AgDRIFT for aerial and
ground spray applications (Teske et al., 2001). For airblast applications, droplet spectra are not
available in AgDRIFT. Therefore, for airblast applications, a default value of 0.28 is used for
Frespired, which is the most conservative value of the droplet spectra included in Table 5.2.
Table 5.2. Frespired Values for Different Droplet Spectra for Ground and Aerial
Droplet spectra
Frespired
Very fine to fine
0.28
Fine to medium
0.067
Medium to coarse
0.028
Coarse to very coarse
0.02
5.2. Calculating Pesticide Exposure through Inhalation of Vapor Phase Pesticide
Inhalation exposure for a vapor phase pesticide is calculated for every time step following the
first pesticide application. The on-field dose of volatilized pesticide inhaled by the bird is
estimated using Equation 5.6. (See Table 5.1 for parameter descriptions). This equation
accounts for the pesticide concentration in air as a result of volatilization from plant leaves
(Cair(txvoi)) and the volume of air respired by the bird during the time step (Vinhaiation; Equation
5.4). These factors considered together result in a mass of pesticide (mg) respired by the bird per
hour. This number is converted to a dose basis by dividing by the BW of the bird.
t, _ s r> Cair(t)(vol)*Vinhalation
Equation 5.6. Dvapom = 'gw
Air concentrations in treated agricultural fields are calculated using a two-compartment model
(Equation 5.7). These compartments include the crop foliage and the air that is between the
crop canopy and the soil of the treated field. The total pesticide mass applied to a 1-ha treated
field (Mpesticide; Equation 5.8) combined with dissipation between the time of application and
time t are used to estimate the total mass of pesticide available for partitioning between crop leaf
and canopy air. The density of the crop tissue (pPiant) assumed to be fresh leaf is 0.77 kg/L, based
on the Hazardous Waste Identification Rule (HWIR) Farm Food chain Model (USEPA, 1999).
The air compartment volume (Vair) is represented by a 1-ha area, with a height set at the top of
the canopy at time of application (Equation 5.9). The available pesticide residue is then
partitioned between the two compartments (air and leaf mass) through the application of the
volume-based biotransfer factor (Bvoi) developed for the HWIR model (Equation 5.10). It is
assumed that the air temperature (T) is a constant value of 298.1 K (equivalent to 25°C, 77°F). A
temperature of 25°C was chosen because Henry's law constant and octanol-water partition
coefficient (Kow) values for pesticides are frequently available at this temperature; however, the
relevance to the actual environment at the time of pesticide application is an uncertainty. The
total available residues establish an upper limit of available pesticide concentration in the air as a
result of volatilization from (treated) leaf surfaces. Variables are further described in Table 5.1.
Page 48 of 77
-------
Equation 5.7. Cairmvol) =-^^"rt
air+\ Pplant J
Equation 5.8. ^rate * 1-12 * 10
Equation 5.9. Vair = CH * 107
Equation 5.10. Log Bvol =1.065 * Log - Log
r>L)-v™
\RT J
Over time, dissipation of the pesticide is considered in the calculation of the pesticide
concentration in air. Degradation of the pesticide at every time step following a pesticide
application is based on the foliar dissipation half-life for broadleaf plants. The degradation rate
constant (r) used in Equation 5.7 is calculated using Equation 4.2. At each time step, pesticide
mass that remains from all previous applications (accounting for dissipation) is added.
5.3. Relating External Inhalation Dose to Oral Dose Equivalents
In cases where avian inhalation toxicity data are available for a specific chemical, they may be
used to derive the oral dose equivalence factor (Fre) using Equation 5.11.
Equation 5.11. F - LD^al^
L/^nfiHhcilcilioH.civum!
Generally, avian inhalation toxicity endpoints are expressed as concentration and specific
duration based on the exposure period of the test (e.g., 4-hour LCso in mg a.i./L). The user must
convert the LCso to a dose-based endpoint (i.e., LD50 in mg a.i./kg-bw) using Equation 5.12. It
should be noted that the BW and respiration rate used in this equation should be derived based
on the test species (BWtest and Rrate(test)). The variable h represents the duration of the exposure
period (in h) and is used to derive the total volume of contaminated air inhaled by the bird during
the study. This approach assumes that the birds that died during the study did so after the 4-hour
exposure period.
tti ,• c 11 r r> LCs0*Rrate(test)*h
Equation 5.12. i^sofinftaiation) —
When avian inhalation toxicity data are not available, TIM uses the relationship between rat
acute oral and acute inhalation LD50 values to establish a route equivalency factor. This factor is
applied to avian inhalation dose estimates to calculate an oral dose equivalent exposure for
subsequent comparison with avian oral dose acute toxicity endpoints. In order to account for
differences between the physiology of mammals and birds, the EPA evaluated the differences
between avian and mammalian respiratory physiology that might be considered in establishing a
more taxonomically appropriate route equivalency factor. USEPA (2004) includes a comparison
of basic aspects of avian and mammalian lung physiology and how these differences may
influence the bioavailability of inhaled pesticide through taxonomic differences in diffusion rate
across the pulmonary membrane. Based on pulmonary membrane diffusion rate estimates for
Page 49 of 77
-------
birds and mammals, USEPA (2004) indicates that the relative diffusion rates across the
pulmonary membrane (Fam) is between 2.4 and 3.4 times greater in birds than in mammals of
similar BWs (weight range 1 to 2,000 g). These differences in diffusion rate can be used to
modify the relationship of oral to inhalation toxicity endpoints in mammals to produce a route
equivalency factor Fre that would at least account for the expected higher diffusion rates across
avian pulmonary membranes (Equation 5.13; Table 5.1). In TIM v.3.0, values of 2.7, 2.9 and
3.3 are used to represent Fam for the small, medium, and large generic birds, respectively. These
values are based on the relative diffusion rates of chemicals across the pulmonary membranes of
birds and mammals (USEPA, 2004, Appendix D, Table Dl) and the mean BWs of these 3
generic birds {i.e., 20, 100 and 1000 g). The route equivalency factor is then multiplied by
estimated avian inhalation exposure doses {i.e., Dspray(t) and Dvapor(t)) to derive an estimate of the
equivalent oral dose. Using the oral equivalent dose to describe inhalation exposure allows the
available oral toxicity studies to describe potential risks resulting from estimated inhalation
exposures.
Equation 5.13. Fre =
^^50 (oral,mammal) (^"am)
50( inhalation,mammal)
6. Estimating Pesticide Exposure through Dermal Contact
The general dermal exposure model considers two pathways: direct interception of applied
material during pesticide application and incidental contact with dislodgeable pesticide residues
on treated foliage (Equation 6.1). As described below, these exposures are converted to an acute
oral basis using a dermal equivalency factor (Fred). Table 6.1 defines the input parameters used
in the equations included in Section 6.
Equation 6.1. ^dermal(t) ("^int ercept(t) contact (t)^} ^red Afield
Page 50 of 77
-------
Table 6.1. Parameters Used for Equations in Section 6 to Estimate Pesticide Exposure
Concentrations through Dermal Exposure.
Symbol
Parameter Description
Variable Type*
Units
Arate
Application rate from label
Constant
lb a.i./A
BW
Body weight
Random
g/bird
Cplant(t)
Concentration of the pesticide in crop foliage at time t
Random
mg/kg
DAF
Dermal absorption fraction
Constant
none
Dconhich 11
Incidental Dermal Contact Dose
Random
|ig pesticide/g-bw
Ddermal(t)
Dose through dermal exposure for a pesticide at time t
Random
|ig pesticide/g-bw
Dinlciccph I i
Intercepted Dermal Dose
Random
|ig pesticide/g-bw
DPR
Dislodgeable pesticide residues
Constant
mg/m2
Fat
Dislodgeable foliar residue adjustment factor
Constant
kg/m2
F field
Fraction of on field exposure
Random
none
Fred
Dermal route equivalency factor
Constant
none
Rfoliar contact
Rate of foliar contact (6.01)
Constant
cm2foliage/cm2body
surface (per hour)
SAtotal
Total surface area of bird
Random
cm2
TPR
Total pesticide residues
Constant
mg/kg
* "Constant" indicates that the parameter is set to one value. "Random" indicates that the parameter's value varies
based on a distribution of possible values.
6.1. Dermal Exposure through Direct Interception
Dermal exposure from applied pesticide droplets is considered for each time step representing a
pesticide application for aerial, airblast and ground applied sprays (See Section 1.4.2), The
dermal exposure dose from direct interception (Dintercept(t)) is calculated by considering the
pesticide application rate relationship to the surface area and BW of the bird (Equation 6.2;
Table 6.1). The dermal interception model assumes that pesticide deposition occurs in a manner
consistent with a horizontal surface in the treatment area. Surface area calculation of a bird for
the interception model assumes that the upper half of the bird in the field is exposed as a result of
either ground or aerial spray applications. Therefore, the total surface area of the bird is
multiplied by 0.5. The total surface area of a bird is calculated using the allometric equation for
relating BW to surface area (USEPA, 1993; Equation 6.3). The dermal adsorption fraction
(DAF) is used to account for pesticide specific data that define a fraction of the pesticide mass
present on the bird that is actually absorbed by the bird. These data may be submitted by the
registrant (non-guideline study) or obtained from the literature. When no data are available to
parameterize DAF, the default value is 1. In this equation, a factor of 11.2 is used to convert the
units of the application rate, which are lb a.i./A, to the metric units needed to generate a
concentration value expressed in |ig a.i./g-bw.
I. — (Arate*ll-2y(SAtotai*0.5)*DAF
Equation 6.2. 0/ntercept(t) —
Equation 6.3. SAtotal = 10 * BW0,667
Page 51 of 77
-------
6.2. Dermal Exposure through Dislodgeable Pesticide Residues on Foliage
During feeding hours, dermal contact with foliage is modeled using Equation 6.4. During non-
feeding periods for both edge and resident species, dermal exposure is assumed to be negligible
because birds are assumed to be relatively inactive; therefore, during non-feeding hours, DCOntact(t)
= 0. The dermal exposure doses from contact with dislodgeable pesticide residues on treated
foliage {i.e., incidental dermal contact dose) is calculated by considering the concentration of
pesticide on treated foliage, fraction of total residues that are dislodgeable, the rate of foliar
contact of the bird, the surface area of the bird that is contacted by dislodgeable foliar residues,
and BW of the bird (Equation 6.4; Table 6.1). Cpiant(t) is the same residue value used for the
broadleaf foliage concentration in the assessment of dietary exposure, which is described in
detail in Section 4.1. (Note that this value accounts for the fraction of contaminated foliage
FCbroadieaf). The dislodgeable foliar residue adjustment factor (Fdfr), surface area and rate of foliar
contact (Rfoiiarcontact) are discussed in detail below. In this equation, a factor of 0.1 is used to
generate DCOntact(t) value with units in |ig a.i./g-bw.
n a t\ Cpiant(t)*Fdfr*R foliar contact
-------
6.2.2. Surface Area of Bird that Contacts Foliar Residues
The dermal incidental contact model predicts transfer of pesticide residues from foliage to the
bird foot and lower leg. The model does not include transfer of residues to other areas of the bird
surface where feathers would provide a barrier to exposure of pesticides to the skin (Smith et al.
2007). The surface area calculation for dermal exposure of birds for the interception model uses
a point estimate of leg/foot surface area of 7.9 percent of the total body surface (USEPA, 1993).
Therefore, the total surface area of the bird (calculated using Equation 6.3) is multiplied by
0.079 to represent the surface area of the bird that contacts foliage. This value is different from
the fraction of the surface area exposed through direct spray {i.e., 0.5) because, in this case, only
the leg/foot of the bird is exposed to contaminated foliar residues, whereas for direct spray, it is
assumed that the upper half of the bird is exposed.
6.2.3. Rate of Foliar Contact (RfoUar contact)
The foliar contact rate is the surface area of vegetation that is contacted by a given surface area
of a bird over the course of a time step. Experimental measurements of such contact rates for
birds have not been identified in the literature to date. In the absence of data specific to incidental
foliar contact for birds, the model presently makes use of the data in USEPA (2004) to develop a
surrogate foliar contact rate. The model quantifies incidental contact exposure to the foot/lower
leg as it is assumed that incidental contact might be the most significant for birds as they move
about the foliage while foraging. Consequently, a surrogate value from the data in USEPA
(2004) (farm worker hands) was selected to represent a contact rate functionally equivalent to a
bird foot grasping vegetation. The range of mean contact values for the hand wash measurements
from farm workers, as they relate to foliar contact reported in USEPA (2004), is 11.9 to 5,050
cm2/hr. The model currently employs the value of 5,050 cm2/ hr. This value is not adjusted for
duration of contact, as the avian exposure model is based on an hourly time step. The total foliar
contact rate for farm workers' hands cannot be used without adjustment for the relative surface
area differences between farm workers and birds. A typical surface area value for adult male
hands of 840 cm2 was used to make this normalization. The result is a default value of 6.01 cm2
foliage/cm2body surface forRfoiiar contact.
6.3. Relating External Dermal Dose to Oral Dose Equivalents
The dermal route equivalency factor (Fred) is applied to estimated avian dermal exposures in
order to derive an estimate of the equivalent oral dose (Equation 6.6). In situations where avian
dermal and oral LD50 data are available for a pesticide, Fred is calculated by dividing the oral
LD50 by the dermal LD50. Since EPA does not have a data requirement for avian acute toxicity
testing via the dermal route, it is expected that a chemical-specific dermal LD50 will rarely be
available. In cases where a chemical-specific dermal LD50 value is not available, it can be
generated automatically by TIM using Equation 6.7 (Appendix H, reproduced from USEPA,
2004). This equation is based on available avian dermal and oral toxicity data. Although the data
set is limited to 25 chemicals (primarily organophosphate insecticides), it has the advantage of
being based on avian toxicity data for both routes of exposure. Therefore, there is no need to
extrapolate across taxa using mammalian toxicity data.
Page 53 of 77
-------
-¦71 r r r-> LD$q(avian oral)
Equation 6.6. Fred = - —
LD50(avian dermal)
Equation 6.7. log ££>5o(dermal) = 0-84 + 0.62 * log LD50(oml)
7. Estimating Pesticide Exposure through Drinking Water
In TIM, it is assumed that two sources of drinking water are available to birds on treated fields:
puddles and dew. Selection of water sources by birds is not well characterized in the scientific
literature. The most likely strategy is one of opportunistic exploitation of whatever water source
is immediately available (pers. comm. Louis Best, 2000). In TIM, it is assumed that puddles are
more likely to be immediately available and thus utilized as drinking water sources.
In the simulation, birds only drink during two hours of a 24-hour period, specifically, the last
hour of the morning and last hour of the afternoon feeding periods. In this approach, birds
consume water through their food throughout the morning and afternoon feeding periods. They
then make up the balance of their daily water requirement through drinking water from puddles
or dew. If a simulated bird's total daily water requirement is met by consuming food (based on
assigned diet discussed previously), it does not drink water from puddles or dew, and is,
therefore, not exposed to the pesticide through drinking water.
It is assumed that puddles are present at the time of each application and for 48 hours following
each application. If puddles are not present, the bird will consume dew during the morning and
no drinking water in the afternoon. Simulated birds can only consume dew on the last hour of the
morning feeding period because dew is not expected to be present in the afternoon.
The dose of pesticide ingested by a bird through drinking water (Ddrinking(t)) is calculated with
Equation 7.1. If t is within 48 hours of an application, the concentration in water is based on
puddles (i.e., CW(t) = CW(Puddie(t))). Otherwise, it is assumed that the puddles have dried up and that
the drinking water source is dew (i.e., CW(t) = CW(dew(t))). When the concentration in water is above
the limit of solubility in water, CW(t) is equivalent to the water solubility limit.
As noted above in this approach, it is assumed that the bird acquires its daily drinking water
during two hours of the simulation. The bird's daily drinking water rate, DWIR, is equally
distributed throughout the two hours when the bird consumes water. Therefore, during the last
hours of the morning and afternoon feeding periods, pesticide doses received by drinking is
calculated by multiplying DWIR by 0.5. During all other hours of the simulation, Ddrinking(t)is 0.
, n _ Cw{t)*DWIR*0.5 ^
Equation 7.1. ^drin/dngft) — * fyield
As discussed in Section 3, the dose is adjusted based on the bird's location. If the bird is on the
field, Fgeld = 1. If the bird is off field and within an area receiving spray drift, Ffieidis assigned a
value <1 based on the spray drift deposition at the bird's location during that hour.
Page 54 of 77
-------
Section 7 includes the equations for calculating Cw(Puddie(t)), CW(dew(t)) and the drinking water
intake rate (DWIR). Table 7.1 defines the input parameter values used in the equations provided
in this section.
Table 7.1. Parameters Used for Equations in Section 7 to Estimate Pesticide Exposure
Concentrations through Drinking Water.
Symbol
Parameter Description
Variable Type*
Units
A.rate
Application rate from label
Constant
lb a.i./A
BW
Body weight
Random
g/bird
Cw(dew)(t)
Concentration of the pesticide in dew at time t
Random
mg/L = iig/mL
Cplant(t)
Concentration of the pesticide in crop foliage at time t
Random
mg/kg = (ig/g
Cw(puddle)(t)
Concentration of the pesticide in puddle at time t
Random
mg/L = |ig/mL
Ddrinking(t)
Dose through drinking water for a pesticide at time t
Random
|ig pesticide/g-bw
DFk
Fraction of diet attributed to food item k
Constant
none
DWIR
Drinking water intake rate
Random
mL/day
dsoil
Depth of soil at equilibrium with water (in puddle)
Constant
cm
dw
Depth of puddle water
Random
cm
e
Base of natural logarithm (2.7182)
Constant
none
Fdfr
Dislodgeable foliar residue adjustment factor
Constant
kg/m2
F field
Fraction of on field exposure
Random
none
FlllXwater
Total daily water flux rate
Random
mL/day
foc(soil)
Fraction of organic carbon in soil
Constant
none
FWk
Fraction of water in a fresh food item k
Constant
none
Koc
Organic carbon:water partition coefficient
Constant
L/kg-oc
Hlwax
Mass of wax per surface area of leaf cuticle
Constant
kg/m2
r
Degradation rate constant
Constant
hour1
t
Time of simulation
Sequential value
hour
TDIR
Total daily intake rate (for food)
Random
g food/bird-day
Sw
Water adjustment scale factor
Random
none
Waterfood
Water from dietary items
Random
mL/day
Pb
Bulk density of soil
Constant
kg/L
Pp
Density of soil particles
Constant
kg/L
Pwater
Density of water (1)
Constant
kg/L
Osoil
Porosity of soil
Constant
none
* "Constant" indicates that the parameter is set to one value. "Random" indicates that the parameter's value varies
based on a distribution of possible values.
7.1. Pesticide Concentrations in On-field Puddles
Pesticide concentrations in puddles are estimated using a simple partitioning approach
(Equation 7.2; Table 7.1) that is based on the Tier I rice model (USEPA, 2007b), with
modifications. In this equation, the pesticide concentration in the water of the puddle is
dependent upon the pesticide application rate (Arate), mean organic carbon-water partitioning
coefficient of the pesticide (Koc; L/kg), and the puddle depth and soil properties. A factor of 11.2
is used to convert the units of the application rate, which are lb a.i./A, to the metric units needed
to generate a concentration value expressed in |ig a.i./mL.
Page 55 of 77
-------
Equation 7.2.
Cw(puddle)(t) —
(;
dw-\-dS0li soil~^~Pb*Koc *foc(soil
-<4rate*ll-2
Puddle depth is assumed to vary across the field. Anytime within 48 hours of the application, a
bird may encounter a puddle. At each hour, the depth of the puddle (dw) has a different random
water depth, selected from a uniform distribution, ranging 1.3-15 cm (0.5 and 6 inches). The soil
depth (dsoii) that contains pesticide at equilibrium with the puddle is set to 2.6 cm (1 inch).
Default parameter values for soil properties, including bulk density (pb) and fraction of organic
carbon (fOC(soii)), are based on EFED scenarios for the Pesticide Root Zone Model (PRZM). The
default values of 1.5 kg/L for pb and 0.015 for fOC(soii) are based on the mean values from the field
crop and orchard scenarios. The user may select alternative values to represent specific scenarios
or fields. Porosity (9Soii) and bulk density are related (Equation 7.3), where pp is the density of
soil particles (kg/L). A typical value of 2.65 (Smettem 2006) is used for soil particle density.
Over time, degradation of the pesticide is assessed in the puddle and on the field after the puddle
has dried up. Degradation of the pesticide at every time step following a pesticide application is
based on the aerobic soil metabolism half-life. The degradation rate constant (r) used in
Equation 7.2 is calculated using Equation 4.2 and the soil metabolism half-life. At each time
step, pesticide mass from all previous applications is added. It is assumed that after a pesticide
application, the pesticide degrades while in the puddle, or in the soil if the puddle is dried up.
The mass that remains from the first application is added to the mass from all other relevant
applications made prior to the time step.
7.2. Pesticide Concentration in Dew from Contaminated Forage
Pesticide concentrations in dew are estimated through the use of a simple equilibrium
partitioning model. This model assumes two compartments, water and leaf cuticle, into which the
pesticide may associate. Equation 7.4 is used to estimate the pesticide concentration in dew
(CW(dew)(t)). Partitioning between the two compartments is based on the octanol-water partition
coefficient of the chemical (K0W), where octanol is a surrogate for the waxy, external
(epicuticular) layer of the leaf cuticle. Cpiant(t) is the total concentration of pesticide in broadleaf
forage leaves (mg/kg-ww) at time t after application. (See Section 4 above for discussion of how
this value is calculated; note that this also considers the fraction of contaminated foliage or
FCbroadieaf). Fdfris used to account for the amount of pesticide that is present on the surface of the
leaf, and thus may partition between the waxy layer of the leaf cuticle and dew. This approach
establishes a distribution of pesticide concentrations in dew that is correlated with random
selection of pesticide concentrations on broadleaf forage. The pesticide partitions into the
epicuticular layer of the cuticle, which is influenced by the mass of wax (mWax). Available data
indicate that the mass of wax in the epicuticular layer varies by species, with ranges of 5-30
|ig/cm2 (Buschhaus and Jetter, 2011). Therefore, a default value of 0.012 kg/m2is selected for
mWax to represent the central tendency of this parameter. The density of water is used to generate
an estimate of the pesticide concentration in water. It is assumed that the density of water is 1
Equation 7.3.
a 1 Pb_
^soil ^
Pp
kg/L.
Page 56 of 77
-------
Equation 7.4.
C * F * o
plant(t) dfr r water
7.3. Drinking Water Intake Rate (DWIR)
The total daily water flux rate (FluxWater; mL/day) for birds is derived from work carried out by
Nagy and Peterson (1988) that involved the development of allometric relationships between
avian BW and daily water flux rate. According to Nagy and Peterson, water flux represents
".. .the amount of water moving into an animal each day..." (p. 3). Thus, this includes water
intake from all sources, including water from food and from drinking. The daily water flux rate
for passerines in the field is estimated according to Equation 7.5. Nagy and Peterson (1988)
noted that passerines take in 3.7 times more water compared to other birds. The authors attribute
this difference to a higher metabolic rate in passerines, as well as differences in behavior related
to diet and drinking water. Equation 7.6 is used for non-passerine birds in order to account for a
lower flux.
The original Nagy and Peterson (1988) drinking water ingestion equation is modified using a
water adjustment scale factor (Sw). The water adjustment scale factor is a random variable that is
selected from a beta distribution that is established assuming that the mean is 1.0 and the
minimum and maximum values are 0.9 and 1.1, respectively. This factor is intended to allow for
variability in the FluxWater of an individual bird from one day to the next.
The daily water flux rate is assumed for a bird in water equilibrium, such that water balance is
maintained each day (i.e., incoming water = outgoing water from all pathways). It is assumed
that a proportion of this daily water flux is fulfilled by water obtained through the consumption
of each day's dietary items, with the remainder satisfied through daily drinking water intake. The
calculation of water from dietary items (waterf00d) is made by multiplying the daily fresh mass of
each food item consumed by the bird by the corresponding fractional water content of that food
item (Equation 7.7). As indicated in Section 4, TDIR is the total daily intake rate for food items
(g/day), DFk is the constant fraction of TDIR attributed to the M1 food item, and FWk is the unit-
less and constant fraction of water in a fresh food item as cited in Table 7.2. The daily drinking
water intake rate (DWIR) is calculated by subtracting the food water intake rate from the total
daily water flux rate (Equation 7.8). This value assumes a standard density of water of 1 kg/L.
Parameters used to calculate DWIR are summarized in Table 7.1.
Equation 7.5.
FluxMr = {l.m*BWm)*Sw
Equation 7.6.
Equation 7.7. waterfood = ^ TDIR * DF *FWk
Equation 7.8.
DWIR = Flux water-waterfood
Page 57 of 77
-------
Table 7.2. Fraction of Water in Fresh Food Items (FWk) of Birds (USEPA, 1993).
Food Item
FWk
Insects
0.69
Seeds
0.093
Fruit
0.77
Grasses
0.79
Broadleaf forage
0.85
8. Establishing Sensitivity and Mortality of Individuals
8.1. Determining Survival and Mortality
At a given time step, an individual bird is considered either alive or dead. This status is
determined by considering the internal dose of the bird at a time step (t) due to all intake sources
(Equation 1.1). These sources include diet (Ddiet(t)), drinking water (Ddrinking(t)), inhalation
(Dinhaiation(t)), and dermal contact (Ddermai(t)). The dose at a given time step also includes the
pesticide dose carried over from the previous time period (Dtotai(t-i)), with consideration of the
fraction of the pesticide retained after elimination of the pesticide {i.e., Fretained, discussed in
Section 8.5). All doses are converted to an oral equivalent.
Equation 1.1. ^total(t) ^diet(t) ^drinking ^ ^inhalation (t) ^dermal (t) ^'total (7-1) ^retained
The internal dose of the pesticide at a given time step is compared to the individual threshold for
mortality (Tmortaiity) of that bird. The threshold is randomly assigned to each bird based on the log
probit dose/response distribution derived from the avian acute oral toxicity data. More
information on how the threshold is determined is provided below. If the internal dose is below
the threshold, the bird is considered to be alive at that time step, and the bird survives to the next
hour, where the process is repeated {i.e., if Dtotai(t) < Tmortaiity, the bird survives to t+1). If the dose
exceeds the threshold, the bird is considered dead, and is no longer included in the simulation
{i.e., if Dtotai(t) > Tmortaiity, the bird dies during t).
It is assumed that doses from all routes of exposure are comparable to the acute oral dose
because the different doses are converted to an oral equivalent using available toxicity data
comparing the dermal and inhalation routes to the oral route. It is also assumed that the
elimination rate constant may be applied equally to any dose, regardless of exposure route.
8.2. Establishing an Individual Threshold for Mortality (Tmortaiity)
The individual threshold of a bird is calculated according to Equation 8.1. The ZSCore is a random
number selected from a normal distribution. The slope in this equation is based on the user input
defining the slope of the dose-response curve of the acute oral toxicity data available for the
assessed pesticide. The intercept of the dose/response curve is calculated according to Equation
8.2. In this approach, approximately 50% of birds receiving a dose equivalent to the LD50 would
die.
Page 58 of 77
-------
Zscore-int ercept
Equation 8.1. Tmortality =10 shpe
Equation 8.2. intercept = —slope * log10 (LD50)
Separate thresholds are calculated for adult birds simulated by TIM and their juvenile offspring
that are incorporated into MCnest. This approach assumes that the sensitivity of the juvenile is
not the same as its mother since the parental sensitivity is likely different from the mother. If
acute oral toxicity data are available to observe a difference in sensitivity of juveniles and adults,
the model user may account for this difference by entering an appropriate value to represent the
ratio of juvenile to adult toxicity. If these data are not available, the model uses the LD50 and
slope data for adult birds to determine the individual sensitivity values for juveniles.
8.3. Avian Acute Oral LD50
The avian acute oral toxicity test (OPPTS 850.2100) provides a measurement of acute toxicity to
the test population from a single oral dose administered at geometrically spaced doses after a
fasting period to groups of individuals. The number of mortalities at each dosage level is
recorded over time (usually fourteen days). Probit analysis or another appropriate statistical
method is used to estimate the dose response curve and other descriptive statistics, including the
LD50, the slope and confidence limits around the estimates. Typically, LD50 values are available
for only two test species exposed to a pesticide {i.e., mallard and bobwhite quail). Recently, the
data requirements for pesticides were altered to include an acute oral study with a passerine
species {e.g., zebra finch, {Taeniopygia guttata), canary {Serinus canaria)). Data for other test
species may be available through the scientific literature.
The avian acute oral LD50 is one of the most important parameters for determining the risk of a
chemical to birds. Therefore, the model user should select an input value with care. If only two
or three values are available, the model user may choose to run the model with the high and low
LD50 values. If the model user is simulating a specific species and toxicity data are available for
that species or for one that is closely related (taxonomically), then the single endpoint may be
sufficient. The model user may still wish to explore uncertainty associated with this endpoint by
simulating LD50 values that represent the confidence bounds of the available LD50. Non-
definitive LD50 values should not be used, as they do not represent the dose-response relationship
for the chemical of interest.
If several LD50 values are available for different species, the model user may choose to develop a
species sensitivity distribution (SSD) to represent the distribution of species responses to the
pesticide of interest. If the sensitivity of the species of interest is unknown, the model user can
explore the influence of uncertainty associated with the LD50 by selecting LD50 values that
represent sensitive {e.g., 5th percentile of SSD), average {e.g., median of SSD), and tolerant {e.g.,
95th percentile of SSD) test species. Guidance on developing SSDs is included in Appendix J.
Before LD50 values from available studies are used as input parameters, they must be scaled to
account for the BW of the assessed species. For SSDs, each endpoint should be expressed in
Page 59 of 77
-------
units of mg a.i./kg-bw for the mean BW of the assessed species (AW) and should be normalized
according to Equation 8.3. The LD50 value on the right side of this equation is the endpoint
reported from the study (units expressed as mg a.i./kg-bw). TW represents the BW (in g) of the
species tested. Generally, acute oral tests involve adult animals. If BW data are not available in
the study report, the literature can be cited for species specific BWs. Default BWs for the
bobwhite quail and mallard duck are 178 and 1580 g, respectively. BWs for additional bird
species can be found in Dunning (1984). The Mineau scaling factor (s) is used to adjust bird
BWs. When chemical specific values are available in Mineau (1996), they should be used. If not,
the default value of 1.15 should be used.
Equation 8.3. Normalized LD50 = LD50 [ — )
8.4. Slope
The slope provides an estimate of the variation of the individual response in the tested sample.
Steep slopes (e.g., 9) indicate a low variance among individuals, and shallow slopes (e.g., 2)
indicate a greater variance among individuals. If possible, the LD50 and slope values should be
from the same study. If the LD50 is based on a single study, the corresponding slope should be
used. If no slope was established, a default of 4.5 (with confidence bounds of 2-9) should be
used. If an SSD is used, the slope representing species at the percentile where the LD50 is
selected may be used. Alternatively, the geometric mean of all available slope values may be
used.
8.5. Metabolism (Fretained)
Elimination is included in TIM using a chemical-specific fraction of the pesticide that is retained
in the bird from one hour to the next (Fretained). This value is generally obtained from empirical
data, e.g., residue chemistry studies with chickens (OCSPP guideline 860.1480). Alternatively,
data representing recovery of an organism from a pesticide can be used, such as decrease in
acetylcholinesterase inhibition as a surrogate for elimination of a carbamate.
9. Model Results
As discussed in Section 1.5, the TIM executable generates several output files (Table 1.4). This
section describes the model's outputs that are intended for the user. Output files for the QC of the
code and MCnest are not discussed in this section.
9.1. Model_results.txt Output File
The Model_results.txt output file includes the input values contained in the TIM_inputs.txt file
as well as some inputs generated by the model (e.g., species parameters for user selected generic
species). This information can be accessed though the GUI by selecting Output -> File -> Model
results.
Page 60 of 77
-------
This output file includes summary statistics for the number of dead birds and the percent of total
simulated birds that died. This output also includes two tables of values that can be used to
characterize the risks of the simulated pesticide use to birds: including information to identify
dominant routes of exposure contributing to mortality (Table 9.1) and probabilities of mortality
to individuals within a flock (Table 9.2).
For each bird that dies during a simulation, the relative contributions of each exposure route to
the total pesticide dose are calculated. Table 9.1 provides an example output with the minimum,
maximum, mean and standard deviations of the fractions of total pesticide dose by exposure
route for those birds. Figure 9.1 depicts this information in a whisker-box plot. As demonstrated
by the figure for this example simulation, pesticide doses received by diet and drinking dew were
the major routes of exposure leading to mortality.
Table 9.1. For Dead Birds: Median, Mean, Standard Deviation, Minimum and Maximum of
Fractions of Total Pesticide Dose
)y Exposure Route.
Exposure route
Median
Mean
SD
Minimum
Maximum
Food Ingestion
0.774
0.789
0.136
0.221
1
Drinking: Puddle
0.02
0.022
0.013
0
0.105
Drinking: Dew
0.205
0.187
0.138
0
0.777
Inhalation: Vapor
0.003
0.003
0.001
0
0.005
Inhalation: Spray
0
0
0
0
0
Dermal Contact
0
0
0
0
0
Dermal Spray
0
0
0
0
0
o
T3
1
0.9
0.8
0.7
0,6
tc
2 0.5
0.4
0,3
0,2
0,1
0
Ingestion
Vapor
spray
Figure 9.1. Relative Contributions of Different Exposure Pathways to Lethal Doses in
Simulated Birds. Box plots represent mean and standard deviations of fractions, with
minimum and maximum represented by lines.
Page 61 of 77
-------
Table 9.2 includes the Probability Density Function (PDF), Cumulative Distribution Function
(CDF) and Complementary Cumulative Distribution Function (CCDF) associated with all of the
simulated birds sorted into flocks of 25, which is based on user input for flock size. Figure 9.2
depicts the PDF, CDF and CCDFs for the example data provided in Table 9.2. The PDF is used
to determine the probability associated with killing exactly x birds. The CDF describes the
probability of killing x or fewer birds. The CCDF provides the probability of killing greater than
x birds.
Table 9.2. Probabilities of Mortality to x Birds out of the Flock. Values generated for PDF,
DCF and CCDF. Note that values depicted as 0 or 1 are rounded.
Dead (x)
PDF
(probability of killing x birds)
CDF
(probability of killing x birds)
0
0.329882
0.329882
0.670119
1
0.374082
0.703963
0.296037
2
0.203618
0.907581
0.092419
3
0.07081
0.978391
0.021609
4
0.017665
0.996056
0.003944
5
0.003365
0.999422
0.000579
6
0.000509
0.99993
6.96E-05
7
6.26E-05
0.999993
0.000007
8
6.4E-06
0.999999
6E-07
9
5E-07
1
0
10
0
1
0
11
0
1
0
12
0
1
0
13
0
1
0
14
0
1
0
15
0
1
0
16
0
1
0
17
0
1
0
18
0
1
0
19
0
1
0
20
0
1
0
21
0
1
0
22
0
1
0
Page 62 of 77
-------
0.8
/
•CDF
0.7
CCDF
0
0
5
10
15
20
25
Number of dead birds out of 25
Figure 9.2. PDF, CDF and CCDF Values Generated by TIM for Risk to All Simulated Birds.
The PDF is useful in telling the risk assessor the most likely magnitude of mortality to birds
based on the parameters of the simulation. For example, based on the model results provided in
Table 9.2, it is most likely that 1 bird out of a flock of 25 will die (probability = 0.37). The
results of the PDF may be used to interpret population level effects of avian mortality. For
example, the most likely decreases in survival of adult birds could be input to a population model
to determine impacts to a species. This information could be useful in interpreting whether a risk
that is likely to adversely affect a listed bird species may also result in jeopardy to that
population. The results of the CCDF, may be the most relevant when considering risks to listed
species of birds. For pesticide effects determinations, EPA determines the potential that a
pesticide may affect one individual. The CCDF can be used to describe the likelihood of killing
one or more individuals. In the example data provided in Table 9.2, the likelihood of killing one
individual of more individuals of a flock of 25 birds near and on a treated field is approximately
The equation used to determine the probability of mortality to x birds (P(x)) out of the user
selected flock size is based on the PDF for a binomial distribution (Equation 9.1). In this
equation, p is the fraction of the total number of simulated birds that died, n is the flock size and
x is the number of dead birds out of the flock for which the probability is being generated. For
the CDF, the probability of killing x or fewer birds is calculated by summing the values of P(x)
that correspond to the value of x and less. For the CCDF, the probability of killing greater than x
birds is calculated by subtracting the probabilities of killing all birds less than or equal to x, in
other words, by subtracting the CDF probability from 1.
0.3.
Equation 9.1. P(x) = (^£^7) * px * (1 - p)n x
Page 63 of 77
-------
9.2 Dead_per_hour.txt
This output file contains two columns that are separated by a space. This information can be
accessed though the GUI by selecting Output -> File -> Dead per hour. The first column
indicates the simulation hour and the second is the number of birds that died during the
simulation. This information can be used to understand the timing of the mortalities relative to
the applications. Figure 9.3 below depicts an example figure generated using the mortalities per
hour.
25
20
I I I
ooLntocnor->.^r
MLnCOOfOlDOIH^
HrlrlrlfMfMfMfMfflfflffl
oo lo cm to
r*s (T> CM LD P-. o
I I I 11 I I I I
roor-v^r^HCOLnrMcntDro
rotDoo^H'^-tD(j>rM^rp>.o
^^ti^LnLnLniDioiDN
Simulation hour
Figure 9.3. Example of Output Depicting the Number of Simulated Birds That Died Per
Hour.
10. Uncertainties
This section discusses the uncertainties associated with the assumptions of TIM v.3.0. It is
possible that in future versions of the model, refinements can be made to address these
uncertainties.
10.1. Exposure Routes Not Considered
At this time, TIM does not consider pesticide uptake through the following routes:
- Dietary consumption of granular formulations or treated seeds;
- Dietary consumption of contaminated small vertebrates (e.g., mammals, birds), carrion,
worms or aquatic organisms;
Incidental ingestion of soil;
Inhalation of particulate-associated pesticide (fugitive dust emissions associated with soil
or seed treatments);
- Dermal contact with soil (e.g., dust baths, foot contact);
- Dermal contact with contaminated water;
- Drinking contaminated guttation fluid from water;
Page 64 of 77
-------
- Drinking from larger water bodies receiving spray drift and runoff from the field (e.g.,
ponds;
Oral uptake through preening; and
Oral uptake through nest building (i.e., collection and manipulation of nest materials that
may be contaminated with pesticides).
The model also does not account for soil or seed treatments or tree trunk injections. Therefore,
exposures from systemic pesticides that translocate into plant tissues and are associated with
consumption of plants are not considered.
In TIM v.3.0, the model user can simulate up to 5 separate pesticide applications. Although the
ability to simulate 5 applications represents an upgrade in the capability of TIM, this fact may
represent a limitation for model users in cases where pesticide labels allow for more than 5
applications per season.
For ground applications, it is assumed that birds on the field will flush, thus, preventing exposure
to direct spray via inhalation and dermal routes (Section 1.3). Although this is a reasonable
assumption for birds on the treated field, it is possible that birds in the edge habitat will not flush.
Therefore, birds in the edge could be exposed to pesticide spray that is transported to the edge
habitat via spray drift.
10.2. Avian species
10.2.1. Diet and Feeding
The model assumes a constant diet composition for all individuals of the simulated species over
the course of the simulation. It is expected that for many species, diet composition will vary over
time based on the availability of food. In addition, it is expected that there will be some
variability in diet composition among individuals within a species.
As indicated by Appendix D, many species that visit agricultural fields have diets represented
predominantly by multiple food items (i.e., omnivores). The generic omnivore species is
parameterized so that its diet is equal parts arthropods, seeds, grass, broadleaf and fruit. In
reality, the proportions of these items in an omnivore's diet are not equal.
The model does not account for changes in feeding that may be associated with the growth of the
crop and plants in adjacent habitat. Also, differences in feeding strategies among birds and
potential impacts on pesticide exposure are not considered. For example, aerial feeders are
assumed to have the same exposure as ground feeders. It is possible that aerial feeders may have
lower exposures if they feed above the canopy, thus not receiving dermal or inhalation
exposures. This possibility may be accounted for by the model user by turning off pathways that
may not be relevant to a specific modeled species.
Page 65 of 77
-------
10.2.2. Body Weight (BW)
The generic species do not simulate individual birds weighing 30-66 g and 152-660 g. If a bird
species represented by these BWs is of interest to the user, the custom species can be used to
account for the potential risks to birds within these BW ranges.
The BW of an individual does not vary over time. Therefore, seasonal changes in BW (e.g., in
preparation for migration, due to reproduction) are not accounted for. In addition, juvenile BW is
set to 0.5 of the value for its parent (see Appendix F). Thus, changes in BW due to growth of
juveniles is not accounted for.
10.2.3. Frequency on Field and Residency
One notable uncertainty associated with the empirical FOF data used to derive the default mean
FOF values for the generic species (and summarized in Appendix D) is that the majority of the
census studies were conducted during the spring and summer months. Therefore, FOF values
representative of fall and winter months are unknown.
As indicated by available avian census studies, use of fields by individuals within the same
species varies in time and location. There is also uncertainty associated with the use of avian
census studies to represent frequency on field. As noted by the SAP, radio telemetry data
tracking movements of individual birds on and off of treated fields would be ideal for
determining FOF; however, these data are not generally available. Observations of individuals of
a species on agricultural fields and orchards relative to the edge habitat is used as a surrogate to
estimate FOF for a species.
The model does not consider impacts of the pesticide on prey availability and alterations to FOF.
For example, decreases in availability of insect prey due to application of a pesticide does not
result in decreases in FOF of insectivores.
For specific species, residency status was assigned based on the nest location of a bird (e.g.,
ground nesters in grassland are assumed to be residents). Ideally, these assignments could be
confirmed using studies documenting nesting on agricultural fields and orchards; however, these
studies were not available.
10.3. Modeling Bird Behavior: Influence of the Fidelity Factor
Although the model incorporates bird behavior into exposure and risk estimates, there are some
uncertainties that are difficult to quantify, in particular the fidelity factor, which represents the
tendency of a bird to return to a specific area to feed. As noted in Section 2.6, no data are
available in the literature to support the parameterization of the fidelity factor (Q). Since TIM's
outputs are sensitive to the value of this parameter, this represents an uncertainty. Figure 10.1
illustrates how Q shifts the shape of the triangular distribution of Pn. Q influences the central
tendency of the distribution of Pnand thus the probabilities of birds to stay on the treated field
from one hour to the next.
Page 66 of 77
-------
When the FOF is kept constant, an increase in Q results in an increase in Pn (Figure 10.2),
indicating that as a bird's fidelity factor increases, the bird is more likely to remain on the treated
field from one time step to the next, resulting in higher pesticide exposure. Therefore, the
selection of the default fidelity factor value of 0.8 for field residents indicates that individual
birds that start out on the treated field at the time of the simulation will most likely stay on the
treated field throughout the simulation.
As Q increases, Poo also increases (Figure 10.3), indicating that as a bird's fidelity factor
increases, the bird is more likely to remain in the edge habitat from one time step to the next,
resulting in lower pesticide exposure. Therefore, individual field resident birds (Q = 0.8) located
in the edge habitat will be more likely to stay there, resulting in lower pesticide exposures to
those birds. Lower Q values generate lower Poo values, which increase the likelihood that the
bird will move from the edge at time t-1 into the field at time t (during feeding hours only).
Therefore, the selection of the default fidelity factor value of 0.6 for edge residents indicates that
individual edge resident birds will more likely move from the edge to the field than field resident
birds and vice versa.
In this example, FOF = 0.6
0.3 0.4 0.5 0.6 0,7 0.8 0.9 1.0
Figure 10.1. Effect of Fidelity Factor (Q) on the Shape of the Triangle Distribution of Pn
(Probability that a bird, now on the field, will be on the field in the next hour; depicted on x-axis).
Page 67 of 77
-------
Q
Figure 10.2. Effect of Q on Pn. Note that Pn values are equivalent for FOF values of 0.1, 0.25 and 0.5
(bottom line).
Q
Figure 10.3. Effect of Q on Poo. Note that Poo values are equivalent for FOF values of 0.5, 0.75 and 0.9
(bottom line).
10.4. Dietary Exposure
The method used in TIM to represent initial pesticide residues on avian food items assumes that
all fields exhibit a residue variability comparable to a mixed data estimate of variance (i.e.,
within field and among field data contributing to the variance estimate), which may represent a
somewhat conservative approach. An alternative assumption would be that all variance
associated with the underlying avian food item residue data are only attributable to among field
Page 68 of 77
-------
variance and that there is no residue variance within a field. EPA has reviewed a number of
pesticide residue datasets and has concluded that at best, variance within a field is lower than
variance among fields, but under some circumstances, variance within a field could approach
variance estimates among fields.
Initial pesticide residues on avian food items are assumed to be linearly related with the
application rate of the pesticide. This assumption was examined for residues on plants by
Fletcher et al. (1994), and the authors found this assumption was consistent with the pesticide
residue data from the uptake/accumulation, translocation, adhesion, and biotransformation
(UTAB) database that they evaluated. However, most of the data points included in their
analysis were for typical application rates between approximately 0.2 and 4 lb a.i./A. The extent
to which this relationship holds for pesticides that are applied at exceptionally low rates or
exceptionally high rates is unknown. Also, the extent to which the linear relationship holds for
insects is unknown.
Data compiled on studies of pesticide residues on insects were used to estimate peak exposure of
birds from consumption of terrestrial invertebrates. The extent to which the residues on
terrestrial insects represent those of other invertebrates that birds consume (e.g., arachnids and
annelids) is unknown. Also, while Day-0 residues were assumed to represent peak levels,
residues on some mobile invertebrates may actually peak after Day 0 (Brewer et al., 2003).
The model assumes that dissipation of the pesticide from each food item is a constant value. It is
likely that there is variability in dissipation across a field and among different fields due to
varying weather conditions and other factors. This variability is not accounted for in TIM v.3.0.
The dietary intake model used to estimate daily food intake for birds is based on allometric
equations for the average daily field metabolic rate. This model predicts the average intake
needed to achieve balance with daily caloric requirements. In addition, the model also does not
consider the impact of egg laying on foraging behavior (e.g., duration, intensity) on female birds.
10.5. Inhalation Exposure
The exposure assumptions are based upon a vapor concentration at saturation and at a
temperature of 25°C. Temperatures at the time of pesticide applications could differ from 25°C,
with higher temperatures resulting in higher vapor pressures. The value of 25°C is advantageous,
however, because vapor pressure data are generally available at this temperature. In addition, it
does not seem to be an unreasonable estimate of an environmentally relevant temperature at the
time of pesticide application. This estimate does, however, add uncertainty to the calculations.
The model does not consider volatilization from the soil. It also does not consider exposures in
edge habitat from volatilization and redeposition.
The respiration rate (Rrate) is calculated using an allometric relationship from USEPA (1993) that
relates avian resting respiration rate to BW (Equation 5.5). This value is multiplied by 3 in order
to translate this laboratory based allometric equation into one that is representative of the field.
Equation 5.5 was derived from non-passerine birds with a range of BWs and is associated with
Page 69 of 77
-------
standard metabolism (post-digestive, at rest). Equation 5.5 may underestimate inhalation rates
for passerine species because passerines have somewhat higher metabolic rates than non-
passerines (USEPA, 1993); however, allometric relationships were not available to allow for
estimates of inhalation rates for passerines in USEPA (1993). Although birds may have
decreased respiration rates in the field when they are not feeding and are thus less active, the
model does not account for a decrease in respiration.
Initial limitations of the air model include the assumption that equilibrium conditions exist.
Consequently, the rate of change in exposure as a function of changing meteorological
conditions on air concentrations cannot be determined. Also, the model does not alter the height
and mass of the crop over time.
The model cannot be applied to situations where pesticides are applied to soils with little or no
ground cover; an important limitation because many volatile pesticides, such as soil fumigants,
are applied to non-vegetated soils. Finally, the model is limited in the ability to address
exposures at varying heights within the canopy.
In TIM v.3.0, it is assumed that birds can inhale particles of 100 [j,m in diameter or less of the
direct spray droplet distribution immediately after application of the pesticide. The review of
available literature by the 2004 SAP (USEPA, 2004a) identified limitations with the data and
suggested that larger particle sizes may be able to enter the respiratory system of a bird.
Therefore, the larger particle size of 100 [j,m was chosen in order to determine the respirable
fraction of spray droplets.
In order to convert the inhalation dose to an oral-equivalent, avian oral and inhalation data are
used. If avian inhalation data are not available, the relationship between mammalian oral and
inhalation toxicity is used, with use of equivalency factors to account for differences between
avian and mammalian lungs. Physiological and biochemical differences in avian and mammalian
lungs can lead to uncertainty in equivalency factors (e.g., differences in vascularization influence
diffusion rates, enzymatic rates impacting chemical transformation). OPP has been calling in
avian inhalation toxicity data as part of registration review. As more avian inhalation toxicity
data become available, EFED may be able to derive relationships between avian oral and
inhalation toxicity data that can be used to predict acute inhalation toxicity endpoints when they
are not available for a specific chemical.
10.6. Dermal Exposure
Contact exposure with contaminated foliage is estimated using exposure values from human
hands (workers). The model also assumes that only the bird foot is exposed to these residues.
The relationship between exposure values of human hands and bird feet is unknown. In addition,
there is no consideration for bird foot morphology.
In calculating the dermal spray dose at the time of an application, the model does not account for
a decrease in exposure that may occur due to foliar interception of the pesticide spray. This is a
conservative assumption.
Page 70 of 77
-------
There is uncertainty in the approach used to relate external dermal dose response to acute oral
dose response. The relationships are based on a limited data set, including only 3 test species
and 6 chemicals, all of which are organophosphate pesticides. In addition, a poor correlation was
established between available measurements of acute oral and acute dermal toxicity. The
correlations were not improved when other physical/chemical properties were considered.
However, the simple correlation models used to test for physical/chemical property influences
were not mechanistic. It is possible that improved predictive models may be developed in the
future that relate pesticide physical/chemical properties to rates of absorption across, and
metabolism within, avian skin tissue. A more complete understanding of such mechanisms
affecting bioavailability may aid in the establishment of more robust predictive models of avian
dermal toxicity. These issues remain topics for further research and future model development.
10.7. Drinking Water Exposure
The relative importance of different sources of drinking water is expected to vary based on
environmental factors, weather, geography, climate and species. The model does not account for
these potential influences on drinking water source selection. In cases where a bird species of
interest does not ingest drinking water, the user may choose to "turn off' the drinking water
switches.
Contaminated water sources are not addressed for banded and in-furrow applications. Banded
and in-furrow applications present specific modeling challenges for estimating drinking water
contamination, and applicable models are not presently available. It is possible that
concentrations of water in puddles forming in treated furrow and banded areas could be higher
than modeled for aerial sprays. This potential for increased exposure magnitude warrants future
field investigation and model refinements to account for these application methods.
The model does not consider pesticide exposures through larger bodies of water, such as ponds
and streams. It is expected that exposures through consumption of these drinking water sources
will be lower compared to puddles and dew. Since dew exposures are higher than puddles, the
lack of consideration of other sources of water is expected to be conservative. The FIFRA SAP
(SAP, 2001) indicated that birds are less likely to consume dew if standing water is available
(e.g., ponds).
Both the puddle and dew models rely upon equilibrium based partitioning models that are
dependent upon a chemical's properties (i.e., Koc and Kow). There are several uncertainties and
assumptions associated with these models, including:
surface characteristics of the soil and vegetation are not accounted for in determining the
potential for modifying solubilization of vegetation/associated pesticide residues;
equilibrium is established quickly between the two compartments of each model; and
degradation is the only route of dissipation assumed to occur.
The puddle model assumes that puddles are on the field at the time of the application and are
present for 48 hours afterword. There are uncertainties associated with these assumptions.
Page 71 of 77
-------
First, pesticide applicators may not be likely to apply pesticides if 6 inch puddles are present
on the field. Especially if the puddles limit the ability of the application equipment to
maneuver the field.
Second, puddles may persist on a field for longer than 2 days, potentially prolonging this
exposure route.
Third, this approach does not account for evaporation of water from puddles that could result
in higher concentrations.
Fourth, in reality, puddles would be expected to be on the field at different times outside of
the pesticide application.
In addition, the following assumptions apply to the dew model:
Pesticide concentrations in dew are based on concentrations of the pesticide applied to the
surfaces of broadleaf plants. This approach does not consider concentrations of pesticides in
guttation water from plant sap that may be representative of systemic pesticides.
Relative compartment volumes are assumed unimportant, such that the mass of pesticide
initially on the leaf compartment is sufficient to reach maximum equilibrium concentrations
in water.
10.8. Determining Mortality
10.8.1. Toxicity Data
One of the largest sources of uncertainty associated with predicting effects of pesticides to
nontarget species comes from the large variability in the sensitivity of species to toxic chemicals.
A review of toxicity studies for 53 carbamate and organophosphate insecticides showed that the
range between LDso's among birds is from 5 to more than 100 (ECOFRAM, 1999). For 70% of
the products, this range extends between 10 and 100. If the species of the assessment is the same
as the species tested in a toxicity study, the effects profile may be the same as the dose-response
relationship derived from the study. More often the assessment is focused on species that have
not been tested. Therefore, the effects profile needs to account for the uncertainty introduced by
the high variability in sensitivity among species. In the absence of toxicity data on specific
species with unknown sensitivities, uncertainty is introduced into the assessment of risk to
individual species.
The SSD approach generates a distribution of species sensitivity from the results of available
acute oral toxicity tests. This approach is based on the concept that the sensitivity of species is a
stochastic variable that can be characterized by fitting a probability density function to the results
of the toxicity tests. This assumes that the distribution of wild species sensitivity closely
approximates the estimated distribution from laboratory tests, and the sensitivity of species used
in laboratory tests is an unbiased measure of the variance and the mean of the distribution of
sensitivity of wild species.
The slope of the dose-response curve is an estimate of the population's variability in individual
sensitivities and therefore has inherent statistical uncertainty. Also, the slope of the dose-
response curve is thought to differ among species due to the differences in morphology and
biochemical and physiological processes, which interact with the inherent pharmacokinetic
Page 72 of 77
-------
characteristics of the compound. Information on the extent of the variability of the slope between
species is lacking and limits, at this point, predictions about the slope based on taxonomic
relationships. Therefore, few species, other than the standard species used for laboratory tests,
are tested in such a way that slopes can be determined, which prevents a more thorough
evaluation of the species differences in slopes at this time (ECOFRAM, 1999).
Uncertainty is introduced into the model results from the major variables that influence the acute
response of individual animals. These include intra- and inter-species variability, age, sex,
nutritional status, breeding status, environmental conditions, formulation, routes of exposure and
duration and extent of exposure. For the majority of these variables, while data has been
developed that indicate they contribute to the variability of the response of an individual to
exposure to a toxicant, limited information is available to quantify their influence on the
numerous wildlife species exposures under the countless environmental conditions that occur
under field conditions.
Acute oral toxicity studies have limitations for estimating the risk to wild avian species exposed
to pesticides in the environment. One limitation is that the study includes a fixed exposure
period, which does not allow for the differences in response of individuals to different duration
of exposure. In the model, exposure occurs over several days or weeks. The study involves a
single dose of the pesticide, which does not mimic wild birds' exposure. In addition, for
exposure through different environmental matrices, the acute oral LD50 does not account for the
effect of the matrices on the absorption rate of the chemical into the animal.
The assumption is made that there is no cumulative effect of repeated doses that reduce the
sensitivity of an individual to successive doses, and that the peak cumulative dose per day, taking
into account the elimination rate of the chemical per day, is equivalent to the single bolus
exposure in the acute oral toxicity test. In essence, the foundation of the approach is the toxic
response of the individual, which is a function of the body burden of the compound. Likewise,
the body burden is a function of the ingestion rate plus the residual from previous exposure
periods, using a defined time step.
The construct of the risk assessment model relies on the peak exposure over the course of a
series of time steps. That is, the assessment of individual bird survivorship within a cohort of
birds is based on the interpolation of a mortality risk for the highest exposure time step that is
modeled. In this way, risk of mortality for the highest time step is evaluated independently from
previous exposure history. Therefore, the model cannot account for any potential increase or
decrease in susceptibility to intoxication that occurs at lower dosages from earlier time steps.
By relying on toxicity data derived under laboratory conditions, mortality is only considered
based on exposure under controlled environmental conditions. As a consequence, the potential
for additional reduced survivorship as a result of sublethal effects is not considered. These types
of sublethal effects may include increased susceptibility to temperature stress, reduced ability to
obtain food, reduced ability to care for offspring, and impaired ability to avoid predation.
Page 73 of 77
-------
10.8.2. Elimination
The exposure assessment model assigns a fixed estimate of pesticide clearance rate to every
individual bird in the simulation. This rate is based on chemical-specific metabolism data from
domesticated chickens. It is possible that other birds will exhibit different metabolic clearance
rates for the same pesticide. For instance, smaller birds are likely to have overall higher
metabolic rates than chickens. If this is the case, higher clearance rates would mean less
carryover from one time step to another, and peak exposures for most individual birds that are
modeled would be lower, corresponding to lower risks. It is also uncertain that chickens will be
representative of all bird species. It is expected that there will be differences in metabolism and
clearance among species.
It is assumed that uptake and elimination kinetics are equivalent for all exposure routes. There is
uncertainty in lumping all of the exposure routes into one dose. For instance, this approach does
not account for potential differences in elimination via respiratory and dermal routes.
10.9. Other Considerations
The model relies upon data relevant to a pesticide active ingredient. Impacts of components of a
formulation on exposure and toxicity are not considered directly. For example, potential
increases in dermal uptake due to carriers present in a formulation are not considered. There is
some consideration of active ingredients inherent in the food exposure calculations because the
initial residue distributions are based on data from field studies that involved formulated
products. The model user may choose to account for the toxicity of a formulation of interest
using acute avian toxicity data for that formulation.
This model does not consider impacts of indirect effects to birds. For example, reduced
availability of invertebrate food items, a variety of food items and suitable drinking water
(through conditioned response to avoid chemical contamination) are not considered.
The model does not consider sublethal effects to birds, including decreased feeding or
movement. Regurgitation and avoidance of food are not considered.
Changes in weather, including rainfall and temperature, are not considered. Changes in
temperature in particular may impact the fate of the chemical (e.g., alter degradation rates, alter
partitioning), metabolism of birds and the toxicity of the chemical.
11. References
Best, L.B., R.C. Whitmore, G.M. Booth. 1990. Use of cornfields by birds during the breeding
season: The importance of edge habitat. Am. Midi. Nat. 123:84-99.
Boutin, C., Freemark, K.E., Weseloh, D.V., Donaldson, G.M., Csizy, M., Martin, P.A.,
Wormington, A., McCracken, J., Shepherd, D., 1996. Bird use of crops in southern Ontario:
Page 74 of 77
-------
implications for assessment of pesticide use. Can. Wildl. Serv. Tech. Rep. Ser., No. 264,
Environment Canada, Hull, Quebec.
Brewer, L.W., H.L. McQuillen, Jr., M.A. Mayers, J.M. Stafford, and S.L. Tank. 2003.
Chlorpyrifos residue levels in avian food items following applications of a commercial EC
formulation to alfalfa and citrus. Pest Management Science. 59:1179-1190.
Buschhaus, C. and R. Jetter. 2011. Composition differences between epicuticular and
intracuticular wax substructures: How do plants seal their epidermal surfaces? J. Exp. Bot.
(2011)62 (3): 841-853.
Driver, C.J., M.W. Ligotke, P. Van Voris, B.D. McVeety, D.B. Brown. 1991. Routes of uptake
and their relative contribution to the toxicologic response of northern bobwhite (Colinus
virginianus) to an organophosphate pesticide. Environ Toxicol Chem 10:21-33.
ECOFRAM, Terrestrial Workgroup. 1999. ECOFRAM Terrestrial Draft Report. May 10,
1999. www.epa.gov/oppefedl/ecorisk/index.htm
FIFRA SAP. 2001. SAP Minutes No. 2001-06, A Set of Scientific Issues being Considered by
the Environmental Protection Agency regarding: Probabilistic Models and Methodologies:
Advancing the Ecological Risk Assessment Process in the EPA Office of Pesticide Programs.
March 13-16. http://www.epa.gov/scipoly/sap/meetings/2001/march/marchl32001.pdf
FIFRA SAP. 2004. SAP Minutes No. 2004-01, A Set of Scientific Issues Being Considered by
the Environmental Protection Agency Regarding: Refined (Level II) Terrestrial And Aquatic
Models Probabilistic Ecological Assessments For Pesticides: Terrestrial. March 30-31.
http://www.epa.gov/scipolv/sap/meetings/20Q4/march/meetingminutesmarch.pdf
Fletcher, J.S., J.E. Nellessen, and T.G. Pfleeger.1994. Literature review and evaluation of the
EPA food-chain (Kenaga) nomogram, an instrument for estimating pesticide residues on plants.
Environmental Toxicology and Chemistry. 13(9): 1383-1391.
Mace, G.M. and P.H. Harvey. 1983. Energetic constraints on home-range size. The American
Naturalist, v 121(1): 120-132.
Mineau, P., B.T. Collins, A. Baril. 1996. On the use of scaling factors to improve interspecies
extrapolation to acute toxicity in birds. Reg. Toxicol. Pharmacol. 24:24-29.
Nagy, K.A. 1987. Field metabolic rate and food requirement scaling in mammals and birds. Ecol.
Monogr. 57:111-128.
Nagy, K.A. and C.C. Peterson. 1988. Scaling of Water Flux Rate in Animals. University of
California Press. 172pp.
Page 75 of 77
-------
Smith, P.N., Cobb, G.P., Godard-codding, C., Hoff, D., McMurry, S.T., Rainwater, T.R., and
K.D. Reynolds. 2007. Contaminant exposure in terrestrial vertebrates. Environmental Pollution
150: 41-64.
Schoener, T.W. 1968. Sizes of feeding territories among birds. Ecology, 49(1): 123-141.
Smettem, 2006. Particle density. In: Encyclopedia of soil science, edited by R. Lai. CRC Press.
Pp. 1243-1244.
Teske ME, SL Bird, DM Esterly, SL Ray, and SG Perry. 2001. A User's Guide for AgDRIFT
2.01: A Tiered Approach for the Assessment of Spray Drift of Pesticides, Regulatory Version.
Continuum Dynamics Report No 01-02.
U.S. Environmental Protection Agency (USEPA). 1993. Wildlife Exposure Factors Handbook.
EPA/600/R-13/187a, Office of Research and Development, Washington, DC.
USEPA. 1999. Farm Food Chain Module: Background and Implementation for the Multimedia,
Multipathway, and Multiple Receptor Risk Assessment (3MRA) Model for HWIR 99. Office of
Solid Waste, Washington, DC.
USEPA. 2001. A Probabilistic Model and Process to Assess Acute Lethal Risks to Birds.
Prepared for the March 13 - 16, 2001 Scientific Advisory Panel Meeting. February 16, 2001.
Prepared by Edward Fite, M.S., EFED/OPP; Ed Odenkirchen, Ph.D., EFED/OPP; and Timothy
Barry, Sc.D., Office of the Administrator/Office of Policy, Economics and Innovation, US EPA.
USEPA. 2004. A Discussion with the FIFRA Scientific Advisory Panel Regarding the Terrestrial
and Aquatic Level II Refined Risk Assessment Models (Version 2.0). USEPA/OPP/EFED.
March 4, 2004. http://www.epa.gov/scipoly/sap/meetings/2004/033004 mtg.htm#materials
USEPA 2007a. Thesaurus of Terms Used in Microbial Risk Assessment. Chapter 5, section 9.
United States Environmental Protection Agency. Office of Water. Available online at:
http://water.epa.gov/scitech/swguidance/waterquality/standards/criteria/health/microbial/thesauru
sindex.cfm.
USEPA. 2007b. Tier I Rice Model for Estimating Pesticide concentrations in Rice Paddies.
United States Environmental Protection Agency, Office of Pesticide Programs, Environmental
Fate and Effects Division. Available online at:
http://www.epa.gOv/oppefedl/models/water/index.htm#rice.
USEPA. 2012a. User's Guide: T-REX Version 1.5 (Terrestrial Residue Exposure Model).
Office of Pesticide Programs, Environmental Fate and Effects Division. Available online at:
http://www.epa.gOv/oppefedl/models/terrestrial/trex/t rex user guide.htm
USEPA. 2012b. Standard operating procedures for residential pesticide exposure assessment.
Office of Pesticide Programs, Health Effects Division. Available online at:
http://www.epa.gov/pesticides/science/residential-exposure-sop.html
Page 76 of 77
-------
Vose, D. 1996. Quantitative risk analysis: A guide to Monte Carlo simulation modeling. New
York, NY: John Wiley and Sons Publishers.
Willis and McDowell. 1987. Pesticide persistence on foliage. Environ. Contam. Toxicol.
100:23-73.
Page 77 of 77
------- |