EPA/600/R/13/033
Technical Manual for Basic Version of the
Markov Chain Nest Productivity Model (MCnest)
Richard S. Bennett and Matthew Etterson
Mid-Continent Ecology Division
National Health and Environmental Effects Research Laboratory
Office of Research and Development
U. S. Environmental Protection Agency
Duluth, MN 55804
February 2013
-------
Technical Manual for Basic Version of MCnest - February 2013
Blank Page
-------
Technical Manual for Basic Version of MCnest - February 2013
Table of Contents
Blank Page 1
Table of Contents 2
Table of Tables 3
Table of Figures 4
A. Brief overview of the conceptual approach used in MCnest 6
B. Selection of surrogate endpoints 11
1. Attributes of surrogate endpoints 14
2. Default surrogate endpoints from existing laboratory studies 15
a. Adultpre-laying body weight 17
b. 1/10 of the LD50 18
c. Eggshell thickness and number of eggs laid 18
d. Proportion of viable eggs per eggs set 19
e. Proportion of hatchlings per viable egg 20
/ Proportion of 14-day-old chicks per hatchling 21
g. Fraction of juvenile dietary LCso 21
h. 1/10 of the LDso (as it pertains to juveniles-not a separate input) 22
/'. Other integrative endpoints 22
3. Selecting toxicity threshold values for each surrogate endpoint from toxicity tests 23
4. Addressing interspecies variability in selecting toxicity thresholds 24
5. Modifying or adding surrogate endpoints 25
C. Parameterizing the pesticide-use scenario 26
1. Consequences of expressing exposure as dose instead of dietary concentration 29
D. Using default species life-history profiles 30
1. Species life-history parameters 32
a. Initiation probability 32
b. Daily background nest failure rate during laying and incubation (m\) 32
c. Daily background nest failure rate during nestling rearing (m^) 33
d. Date of first egg laid in first nest of season (Tj) 33
e. Date of first egg laid in last nest of season (Tiast) 33
/ Length of rapid follicle growth period in days for each egg (rfg) 33
g. Length of eggshell formation period in days for each egg (ef) 34
h. Mean clutch size (clutch) 34
/'. Mean inter-egg laying interval in days (eli) 34
j. Egg on which female typically begins incubation-penultimate vs ultimate (penult) 34
k. Duration in days from start of incubation to hatch (I) 34
/. Duration in days from hatch to fledging of nestlings (N) 35
m. Duration in days since nest failure due to non-pesticide reasons until female initiates
new nest (We) 35
n. Duration in days since nest failure due to pesticides until female initiates new nest (Wp)
35
o. Duration in days since successful fledging until female initiates new nest (Wf) 36
p. Mean female body weight and diet composition during breeding season 36
q. Mean number of fledglings per successful nest 36
2. Representativeness of species profiles 36
-------
Technical Manual for Basic Version of MCnest - February 2013
E. Modifying and creating species life-history profiles 38
F. Running MCnest: Understanding the phase-specific decision points 39
G. Model assumptions and uncertainties 42
1. Assumptions reflecting limitations of toxicity testing 42
2. Assumptions related to parameterizing the exposure profile 43
3. Assumptions related to modeling avian breeding seasons 45
H. Interpreting MCnest outputs 46
1. Interpreting outputs from individual MCnest simulations 46
2. Interpreting MCnest outputs in ecological risk assessment 48
I. Quality Assurance: Verifying that the model performs as intended 49
J. Analysis of model sensitivity to variation in input parameters 51
1. Approach for a sensitivity analysis of life-history parameters 51
2. Results of sensitivity analysis of life history parameters 52
3. Approach for examining model sensitivity to changes in pesticide-related parameters.... 57
K. Example of how MCnest could be used in a tiered risk assessment process 62
L. References 70
Appendix A. Estimating the length of the rapid follicle growth (RFG) period 76
Appendix B. Bar graphs representing sensitivity analysis of life history parameters 80
Table of Tables
Table 1. Types of effects possible during each avian breeding phase by major categories of
exposure 13
Table 2. Types of effects and corresponding surrogate endpoints used in the Basic Version of
MCnest for each avian breeding phase 16
Table 3. Maximum and mean (plus standard deviation) residue concentrations used in 28
Table 4. Calculated annual reproductive success (ARS) for each species compared to estimates
derived from the literature 39
Table 5. Changes in the number of successful broods per female due to variation in life-history
parameters from 27 baseline species profiles 55
Table 6. Proportional change from baseline value due to variation in life-history parameters for
27 baseline species profiles, with proportions > 0.15 highlighted 56
-------
Technical Manual for Basic Version of MCnest - February 2013
Table of Figures
Figure 1. Four phases of avian breeding cycle with phase-specific toxicity endpoints and
associated exposure estimates (i.e., estimated daily dose [EDD] or time-weighted average
dose [TWA]) used in risk quotients (RQs) at each decision point 7
Figure 2. MCnest model inputs and outputs 11
Figure 3. Example of phase diagram for eastern meadowlark 47
Figure 4. The effect of a single pesticide application on June 15, using three estimates of its
residue half-life (i.e., 3.5, 10, and 35 d), on the percent reduction in seasonal productivity
for 12 species profiles based on three daily nest mortality rates (i.e., 0.015, 0.03, and
0.045 d"1) and four egg-laying durations (i.e., 30, 60, 90, and 120 d) with a midpoint date
of June 15 59
Figure 5. The effect of a single pesticide application on May 15, using three estimates of its
residue half-life (i.e., 3.5, 10, and 35 d), on the percent reduction in seasonal productivity
for 12 species profiles based on three daily nest mortality rates (i.e., 0.015, 0.03, and
0.045 d"1) and four egg-laying durations (i.e., 30, 60, 90, and 120 d) with a midpoint date
of June 15 60
Figure 6. The effect of a single pesticide application on each of a series of application dates on
the percent reduction in seasonal productivity for 4 species profiles based on four egg-
laying durations (i.e., 30, 60, 90, and 120 d) with a midpoint date of June 15 62
Figure 7. Phase diagrams for American robins A) without pesticide exposure and B) with
exposure to a pesticide application on May 15 assuming maximum nomogram residue
values for diet and a 35-d half-life for residue degradation 64
Figure 8. Phase diagram for American robins with exposure to a pesticide application on May 15
assuming a lognormal distribution of initial residue values for diet and a 35-d half-life for
residue degradation 65
Figure 9. Phase diagram for American robins with exposure to a pesticide application on May 15
assuming a lognormal distribution of initial residue values for diet and a 7-d half-life for
residue degradation 65
Figure 10. The percent reduction in number of successful broods per female per year for a single
application if it occurred on a specific day in May 66
Figure 11. Phase diagrams for American robins with exposure to a pesticide application on May
1 or May 31 assuming a lognormal distribution of initial residue values for diet and a 7-d
half-life for residue degradation 67
Figure 12. Phase diagrams for dickcissels A) without pesticide exposure and B) with exposure to
a pesticide application on May 15 assuming maximum nomogram residue values for diet
and a 35-d half-life for residue degradation 68
-------
Technical Manual for Basic Version of MCnest - February 2013
Figure 13. Phase diagram for dickcissels with exposure to a pesticide application on May 15
assuming a lognormal distribution of initial residue values for diet and a 35-d half-life for
residue degradation 69
Figure 14. Phase diagram for dickcissels with exposure to a pesticide application on May 15
assuming a lognormal distribution of initial residue values for diet and a 7-d half-life for
residue degradation 69
Figure 15. The percent reduction in number of successful broods per female per year for a single
application if it occurred on a specific day in May 70
This technical support manual has been reviewed in accordance with U.S.
Environmental Protection Agency policy and approved for publication. Mention
of trade names or commercial products does not constitute endorsement or
recommendation for use.
-------
Technical Manual for Basic Version of MCnest - February 2013
A. Brief overview of the conceptual approach used in MCnest
In the United States Environmental Protection Agency's (USEPA) current pesticide risk
assessment process, a pair of laboratory avian reproduction tests with mallards (Anas
platyrhynchos) and northern bobwhites (Colinus virginianus) is conducted to evaluate how
dietary pesticide exposure affects a standard suite of reproduction endpoints (USEPA 1996).
The results of these tests are used in calculating risk quotients (RQ) by comparing the reported
no-observed-adverse-effect concentration (NOAEC) for the most sensitive measured endpoint(s)
with estimates of the maximum dietary exposure expected for a given application rate. As a
screening tool, RQs are compared to an established regulatory level-of-concern to categorize the
potential for unacceptable risk. Because of the high degree of uncertainty in these simple tools
for characterizing risk, RQs typically incorporate conservative or worst-case assumptions about
exposure and toxicity to reduce the chances of concluding a chemical has an acceptable level of
risk when in fact it does not (i.e., false negative conclusion). Consequently, risk quotients can be
used to identify the environmental concentration above which adverse effects to avian
reproduction may occur, but they cannot determine the probability or magnitude of potential
reproductive effects.
An alternative conceptual framework for interpreting the results of avian reproduction
tests was proposed by Bennett et al. (2005). Briefly, it involves linking the types of effects that
may occur during each phase of a bird's reproductive cycle (e.g., pair formation, egg laying,
incubation, nestling rearing) to selected surrogate endpoints from all three standard avian toxicity
tests and relates those effects to the estimated exposure during each phase under a given
pesticide-use scenario (Bennett et al. 2005). Because the great majority of avian reproduction
tests do not provide quantitative dose-response information for surrogate endpoints, by necessity
the alternative approach is based on a series of phase-specific deterministic decision points-
essentially RQs for specific surrogate endpoints at each breeding phase-for determining whether
the nest attempt fails or continues (Figure 1). Also, estimated exposure and effects endpoints are
expressed as ingested doses (e.g., mg/kg/day) rather than as concentrations on food (e.g., ppm in
diet). In the framework proposed by Bennett et al. (2005), if the estimated exposure during the
critical exposure period is less than the established toxicity threshold (e.g., the no-observed-
adverse-effect level or NOAEL) for surrogate endpoints at each phase, the nest continues without
disruption. However, if exposure exceeds the toxicity threshold for a surrogate endpoint, the
nest attempt is assumed to have failed and the female may be able to renest if conditions permit
and sufficient time remains in the breeding season. Also, for those species that can produce
multiple broods in a single breeding season, females may renest after successful nesting attempts
if conditions permit. The simulated performance of a population of females in relation to the
timing of pesticide applications is modeled over the course of a full breeding season (Bennett et
al. 2005). Consequently, using this framework, the effects of a pesticide on annual reproductive
success are not only a function of the results of avian toxicity tests, but also are quite sensitive to
the timing of pesticide applications relative to a species' breeding season and to differences in
life history characteristics among species.
-------
Technical Manual for Basic Version of MCnest - February 2013
Phase-specific
endpoints
NOAEL adult
body wt prelaying
1/10 of LD50
NOAEL eggs laid/
hen
NOAEL eggshell
thickness/hen
NOAEL adult
body wt prelaying
1/10ofLD50
NOAEL % viable/
eggs set/hen
NOAEL % hatch/
viable eggs/hen
NOAEL adult
body wt prelaying
1/10ofLD50
Fraction of LC50
NOAEL %14-d
chicks/hatchlings/
hen
Exposure
measure
1-dEDD
1-dEDD
1-dEDD
1-dEDD
1-dEDD
1-dEDD
1-dEDD
Follicle
TWA
2-dTWA
2-dTWA
5-dTWA
Follicle
TWA
PHASE OF
BREEDING CYCLE
PAIR FORM/
BREEDING SITE
SELECTION *>->_
^
^/
FOLLICLE NO
GROWTH/ iS*
EGG ^
PRODUCTION ^\
^^
^
NO
*r
INCUBATION TO
HATCHING ---^
^--^
S
/
NESTLING
REARING TO
FLEDGING ^~\.
^--«*
,/
N0<
Decision points Timepermitting.
restart breeding cycle
if EDD < adult <
behavioral endpoints
^ AnyRQ>1?_ ^YES * de^ed »
nest
attempt
failed
"* Any RQ > 1 ? > YES > nftst >
attempt
failed
/AnyRQ>1? ^YES » nest ^
attempt
~* AnvRO^I7 » YES * chicks
lost
^
Successful breedina atteirmt
(i.e..fledglings produced)
Figure 1. Four phases of avian breeding cycle with phase-specific toxicity endpoints and
associated exposure estimates (i.e., estimated daily dose [EDD] or time-weighted average dose
[TWA]) used in risk quotients (RQs) at each decision point.
The framework described above identifies three categories of effects resulting from direct
exposures that may occur: 1) effects on adult behavior and reproductive performance from
external exposure (e.g., dietary), 2) effects on nestling growth and survival from external
exposure, and 3) effects on nestling growth and survival from in ovo exposure. Some potential
effects have direct corollary measurements from the reproduction test (e.g., percent hatchability
related to in ovo exposure), while other effects have more indirect surrogate measures (e.g., using
change in adult body weight during the pre-laying period as an indicator of overall parental well-
being and behavioral effects). Some effects, such as nestling toxicity from external exposures,
have no surrogate directly from the reproduction test because chicks are not exposed to treated
diets. However, data from other toxicity tests are used as surrogates. While indirect effects of
chemicals also may reduce reproductive success, laboratory toxicity tests cannot provide
adequate surrogate endpoints for these effects, and they are not addressed currently in the
framework. However, on a case-by-case basis they could be included if sufficient information is
available for defining appropriate decision points in the framework.
-------
Technical Manual for Basic Version of MCnest - February 2013
A flexible mathematical model, known as the Markov chain nest productivity model or
MCnest, has been developed for implementing the conceptual framework of Bennett et al.
(2005). It projects estimates of pesticide effects on reproductive success for a broad range of
species and can be modified to incorporate either sparse or abundant life-history data. MCnest
builds on over 40 years of avian nest-survival modeling in the ornithological literature. Etterson
and Bennett (2005) showed that a simple Markov chain model is equivalent to the well-known
Mayfield (1961, 1975) nest-survival model when similar assumptions are imposed and unifies
many current formulations of nest survival estimation models (e.g., Johnson 1979, Hensler and
Nichols 1981, Bart and Robson 1982, Dinsmore et al. 2002, Shaffer 2004). Etterson et al. (2009)
demonstrated how alternative approaches for defining the length of the breeding season affected
productivity estimates. Bennett and Etterson (2007) presented a demonstration of an earlier
version of the model applied to a pesticide-use scenario. MCnest also can be applied to
contaminant effects questions in other USEPA Program Offices, though at present the model is
not designed to adequately estimate the effects of bioaccumulative chemicals where effects on
hatchability and hatchling survival may result from chemical residues accumulated prior to the
egg formation period.
The decision to develop MCnest as a Markov process was based in part on the iterative
nature of avian breeding (propensity to renest after either failure or success of a nest attempt),
which is naturally captured in the cyclic nature of Markov processes and easily represented in the
associated transition matrices. Another important consideration is the way in which important
nest survival parameters (m\ and mi} are estimated equivalent to the Mayfield estimator. Thus,
the choice to simulate nest survival and productivity as a Markov process is consistent with way
in which important model parameters are generated. We have considered other mathematical
methods (specifically individual-based models and differential equations) for simulating avian
reproduction (see Etterson et al. 2011) and found that the Markov approach was the most suitable
for development of MCnest.
Most of the data used in MCnest are in the form of input parameters provided by the
model user and represent three categories of input parameters: toxicity threshold values for
surrogate endpoints, pesticide application scenarios, and species life-history parameters (Figure
2). MCnest uses information for parameterizing toxicity threshold values and application
scenarios that is currently available in the risk assessment process. The model user may use
default life-history parameters from a library of avian species available to MCnest or create new
or modified species parameter profiles.
In the Basic Version of MCnest, the estimated exposure to a species immediately after a
pesticide application (i.e., initial dietary dose) is based largely on the approach used in the
Terrestrial Residue Exposure model (a.k.a. T-REX) developed by the USEPA Office of
Pesticide Programs. In addition to estimating the mean and maximum dose as in T-REX,
MCnest also gives the model user the option of using a distribution of initial doses based on the
mean and standard deviation of residues expected on various food types in the diets of each
species (discussed further in Section C below).
Once all input parameters are set, a model simulation in MCnest follows the breeding
activities of a population of females each day throughout a breeding season. The temporal
pattern of breeding activity of each female (i.e., transitions among breeding phases) varies due to
-------
Technical Manual for Basic Version of MCnest - February 2013
differences in the initiation date of the first nest attempt and due to a specific probability each
day that the nest attempt could fail from ecological causes such as predation or weather. When a
nest attempt fails, each female can make a new attempt if there is time remaining in the breeding
season, and for many species, females make a new attempt after completing a successful brood.
If the simulation incorporates one or more pesticide applications, the pesticide exposure may
represent an additional cause of nest failures depending on the types of pesticide effects observed
in tests and the timing of the application relative to the phase of the nesting attempt for each
female (Figure 1). When a nest attempt fails due to pesticide exposure, each female may make a
new attempt if there is time remaining in the breeding season and pesticide residues decline to
levels that would not affect parental well-being. As MCnest follows each female of the
population through the breeding season, it tabulates the number of nest attempts and successful
broods (i.e., broods surviving to fledging).
The primary output of each MCnest simulation is an estimate of the number of successful
broods per female per year, which can be multiplied by the number of fledglings per successful
nest to estimate the number of fledglings per female per year (i.e., annual reproductive success).
A "successful brood" is defined as a nest attempt producing one or more fledglings. However, to
put the output from a single MCnest simulation involving a pesticide exposure into perspective,
it is compared to the output of a simulation(s) without pesticides. Calculating the relative
difference between scenarios with and without pesticide exposure provides an estimate of the
potential magnitude of pesticide effects to annual reproductive success under a specific pesticide-
use scenario. MCnest also can provide information on which species are at greatest risk under a
specific pesticide-use scenario or which application dates have the greatest impact throughout a
breeding season. This quantitative estimate of pesticide effects on annual reproductive success is
needed for use in population modeling or probabilistic risk assessments.
While MCnest provides an approach for quantifying the effects of pesticide exposure on
avian reproductive success, caution is needed in interpreting the results, especially when
conservative assumptions are made concerning input parameters. Because of data gaps and
uncertainty in information on species life-history parameters and for quantifying toxicity and
exposure, it is inevitable that assumptions will be required when using available data to
determine input parameters. The ecological risk assessment process traditionally has used
conservative assumptions when addressing uncertainty to reduce the probability of concluding
that there is not an unacceptable risk of adverse effects when, in fact, there is an unacceptable
risk. Due to limitations of the toxicity data, the model structure of MCnest has incorporated a
conservative assumption by assuming that exposures that exceed a specified toxicity threshold of
a surrogate endpoint will lead to a nest failure. We recognize that not every exposure
exceedance of the toxicity threshold value in the field will result in a nest failure, but given the
lack of information from the reproduction test for quantitatively describing the dose-response
relationship, failure is assumed the worst outcome for the nest. Consequently, estimates of the
reduction in reproductive success from MCnest represent a conservative estimate of potential
pesticide effects that may occur in the field. Future versions of MCnest could incorporate dose-
response information into the decision points when it is available. Model users also may
incorporate conservative assumptions in MCnest simulations through selection of parameter
inputs, such as using the maximum nomogram values instead of the estimated distribution of
residue values or by selecting the 35-d default value for the residue half-life value instead of
chemical-specific information of the rate of residue degradation. However, when using
-------
Technical Manual for Basic Version of MCnest - February 2013
conservative assumptions, model users need to be aware that not only will the model results be
more conservative (i.e., estimate a greater reduction in annual reproductive success than may
occur in the field), but the relative responses among species and across applications dates will
change in ways that may make these comparisons less informative. This is discussed in greater
detail in Section H (Model Assumptions and Uncertainties) below.
While the primary goal of MCnest is translating available toxicity data into a currency
useful for population-level assessments by estimating the magnitude of change in annual
reproductive success for a species exposed to a specific pesticide-use scenario, the ultimate
challenge in ecological risk assessment is to estimate the effects of pesticide use on population
growth rates in complex landscapes where pesticide exposure often is patchily distributed.
MCnest can be an important tool in that challenge, but isn't the only tool needed. The model
user must recognize that MCnest intentionally simplifies aspects of the exposure scenario, but
these simplifications can be addressed by the model user within a population-modeling
framework. First, in each MCnest simulation a pesticide is applied on a single day for the entire
population, when in reality pesticide applications are asynchronous across a landscape in
response to pest levels and weather conditions. The MCnest user can run simulations for a series
of possible application dates to estimate the overall effect as a cumulative function of the relative
effect on each date and the probability of an application made on that date. This gives the model
user both an overall estimate of the effect of the pesticide for any distribution of possible
application dates and information on which specific application dates pose the least or greatest
risk. Second, all females in a simulation are exposed to the pesticide. MCnest currently does not
include a spatial component that allows the user to expose only a portion of the population to the
pesticide as would occur across a heterogeneous landscape, but the MCnest results can be used in
spatially-implicit or spatially-explicit population models to simulate patchy exposures. Third, all
females in a simulation are active breeders even though in field populations it is possible for a
certain portion of the population to remain in non-breeding status (i.e., floaters). Again, this can
be addressed best in a population model, especially since the proportion of active breeders each
year may be a density-dependent function. The need to address any of these issues depends
largely on the specific risk management question being posed.
10
-------
Technical Manual for Basic Version of MCnest - February 2013
Toxicity thresholds
1/10 of LD50
NOAEL for pre-laying body wts
NOAEL. f or # eggs laid per pair
NOAEL for eggshell thickness
NOAEL forviable embryos/set
NOAEL for hatchlings/viable embryos
NOAEL for 14 d old chicks/hatchlings
Fraction of LC50
Application
scenario
Applicationdate(s)
Application rate(s)
Residue half-life on food
Species life history parameters
Typical start and end dates of egg laying
Daily nest failure rates during laying/incubation & brood rearing
Mean clutch size and intra-egg laying interval
Incubation starts with penultimate or last egg
Duration of rapid follicle growth, incubation, & brood rearing periods
Duration of waiting periods after nest success & failure
Mean female body weight
Dietary composition of adults and nestlings
Phase-specific
endpoints
NOAEL adult
body wt prelaying
1/10ofLD50
NOAEL eggs laid/
hen
NOAEL eggshell
thickness/hen
NOAEL adult
body wt prelaying
1/10ofLD50
NOAEL % viable/
eggs set/hen
NOAEL % hatch/
viable eggs/hen
NOAEL adult
body wt prelaying
1/10 of LD50
Fraction of LC50
NOAEL %14-d
chicks/hatchlings/
hen
Exposure
measure
1-dEDD
1-dEDD
1-dEDD
1-dEDD
1-dEDD
1-dEDD
1-dEDD
Follicle
TWA
2-dTWA
2-dTWA
5-dTWA
Follicle
TWA
MCnest
PHASE OF
BREEDING CYCLE Decision points
PAIR FORM/
BREEDING SITE
SELECTION \^^
J^ Any RQ > 1 ? +
FOLLICLE NO
GROWTH/ *-""
EGG ..
PRODUCTION ^\^
^""*- AnyRQ>1? >
NO
if
INCUBATION TO
HATCHING -~-^^^
~~~~~~^. Any RQ > 1 ? ».
NO
/
NESTLING
REARING TO
FLEDGING "^^-^_
~"~> AnyRQ>1? »
NO*T
Successful breed!
Time permitting.
if EDD < aduit «
behavioral endpoints
YES ». de|ayed >
nest
attempt
failed
attempt
failed
YES -.. nest
attempt
wpg chicks
~*" lost
ft attamnt
(i.e. .fledglings produced)
^
MCnest outputs
# successful broods/female
# nest attempts/female
% nest success rate
Temporal phase diagram
Brood histogram
Temporal exposure profile
Figure 2. MCnest model inputs and outputs.
B. Selection of surrogate endpoints
The conceptual approach described in Bennett et al. (2005) recognizes that the current
avian reproduction test is not sufficient on its own to directly estimate effects on annual
reproductive success, but it provides data for measured endpoints that may represent several of
the specific types of effects that occur in the field. They define these test endpoints as surrogate
endpoints and briefly describe the rationale for their use as a surrogate for effects in the field.
However, there are several types of potential effects in the field that are not represented in the
reproduction test (Mineau et al. 1994a, Mineau 2005, Bennett and Etterson 2006), though there
may be endpoints in the avian reproduction test that can act indirectly as surrogate endpoints in
these cases. For other types of field effects there are no suitable surrogate endpoints available
from the reproduction test. In such cases, suitable surrogate endpoints may be available from
other laboratory tests or pen and field studies. For example, Bennett et al. (2005) proposed using
data from the 5-day LC50 test as a surrogate for mortality of chicks from direct pesticide
exposure, since chicks are not fed pesticide-treated diets in the reproduction test.
11
-------
Technical Manual for Basic Version of MCnest - February 2013
Pesticides may cause a variety of proximate effects (e.g., effects on parental behavior,
eggshell thickness, or hatchability of eggs) through all the breeding phases that can result
ultimately in a change in annual reproductive success. There are several major categories of
effects resulting from different pathways of exposure (Table 1). Bennett et al. (2005) described
three of these categories for effects resulting from direct exposures: 1) effects on adult behavior
and reproductive performance, including egg production and eggshell quality, from external
exposure (adult extrinsic); 2) effects on juvenile growth and survival from external exposure
(juvenile extrinsic); and 3) effects on juvenile growth and survival from in ovo exposure
(juvenile intrinsic). There are two additional categories involving indirect effects, namely
reduced food resources to adult and juveniles, which were not previously discussed, but could be
important to the assessment of overall effects on avian reproductive success.
During each of the breeding phases, effects may result via one or more of these exposure
pathways (Table 1). To fully assess the potential risks of pesticide exposure on overall
reproductive success, all potential effects of a pesticide should be identified and, where data
exist, surrogate endpoints should be defined for use in a model of reproductive effects. For many
types of effects, surrogate endpoints would need to be derived from sources other than the
laboratory avian reproduction test because, as Mineau et al. (1994) describe, the test measures "a
very unnatural and truncated reproductive performance." While the test simulates an extended
period of egg laying, there are many types of effects related to changes in adult behaviors that
cannot be observed, including behaviors affecting nest construction, clutch completion,
incubation, or rearing of nestlings that are important in determining overall reproductive success.
12
-------
Technical Manual for Basic Version of MCnest - February 2013
Table 1. Types of effects possible during each avian breeding phase by major categories of exposure.
Breeding phase
Pair formation/
Breeding site
selection
Follicle growth/
Egg production
Incubation to
hatching
Nestling rearing to
fledging
Adult direct
Territory loss or nest
abandonment due to
sublethal effects or
death
Reduced clutch size
Nest abandonment due
to sublethal effects or
death of adults or
eggshell failures
Nest abandonment due
to sublethal effects or
death of adults
Reduced hatch due to
infertility
Brood abandonment
due to sublethal effects
or death
Reduced juvenile
growth and survival
due to reduced parental
care and defense
Juvenile in ovo
Juvenile direct
Not applicable
Embryotoxicity due
to in ovo exposure
Reduced juvenile
growth and survival
due to in ovo
exposure
Embryotoxicity due
to external eggshell
exposure
Reduced juvenile
growth and survival
due to direct post-
hatch exposure
Adult indirect
Territory loss or
abandonment due to
reduced food
availability/ habitat
Reduced clutch size
Nest abandonment
due to reduced food
availability/ habitat
Nest abandonment
due to reduced food
availability/ habitat
Brood abandonment
due to reduced food
availability/ habitat
Reduced juvenile
growth/survival due
to reduced parental
foraging success
Juvenile indirect
Not applicable
Reduced juvenile
growth and survival
due to reduced food
availability/ habitat
13
-------
Technical Manual for Basic Version of MCnest - February 2013
The Basic Version of MCnest focuses on surrogate endpoints for direct effects though it
is capable of including decision points for all types of potential effects, assuming that a
meaningful surrogate endpoint and exposure period can be defined. There are many issues still
to resolve on how best to integrate information on indirect effects into this model. Also, because
the Basic Version of MCnest focuses on the female as the subject of the model, rather than
individual eggs or chicks, it does not follow the survival of juveniles after fledging or their
sexual maturation as adults. Post-fledging performance is better addressed using a separate
modeling approach.
1. Attributes of surrogate endpoints
Surrogate endpoints must have certain attributes to be useful in MCnest or in other
models of reproductive effects such as those presented in Roelofs et al. (2005) and Topping et al.
(2005). Not all measured endpoints of effects possess the attributes to be useful as surrogate
endpoints in estimating pesticide effects on overall avian reproductive success.
First, surrogate endpoints must be measurements of effect that can be linked to an
exposure concentration or dose. While this is relatively straightforward in controlled-dose
laboratory studies, linking effects to an estimated exposure can be more difficult in pen and field
studies, not only because exposure may be more difficult to measure, but also because exposure
can be very dynamic over time. To be most effective for pesticide risk assessments, the
estimated exposure concentration or dose should also be relatable to an application rate (i.e.,
application rate x > exposure dose y > effect z).
Second, surrogate endpoints must be measurements of effect that can be related directly
or indirectly to field effects that ultimately may lead to changes in reproductive success (See
Table 2). For example, an observed reduction in hatching rate from in ovo exposure in a
laboratory test may relate quite closely to an observed reduction in hatching rate measured in the
field from comparable pesticide concentrations in eggs. Reduced hatching success can result
directly in changes in reproductive success. However, some measurement endpoints (e.g.,
change in a biochemical concentration or in a behavioral measurement) may require additional
information to demonstrate their relationship to changes in reproductive success. Measurement
endpoints that are proposed as indirect surrogates of effects on reproductive success need to be
examined on a chemical-by-chemical basis to ensure the plausibility of these relationships. A
measured endpoint should not be used as a surrogate endpoint when there is no plausible linkage
between the endpoint and effects on overall reproductive success. Establishing plausibility may
require reliance on completely separate-and unrelated-models; e.g., an efficacy model relating a
given application rate to the proportion of invertebrate kill in the case of indirect effects.
Third, for use in deterministic phase-specific decisions, surrogate endpoints of direct
effects must be expressed as a daily dose (i.e., jig active ingredient [AI]/g body weight/day)
below which the risk of adverse effects is considered acceptable. This dose is referred to as the
toxicity threshold value. In Bennett et al. (2005) these toxicity threshold values were primarily
represented by the NOAEL for the surrogate endpoints from the avian reproduction test.
Finally, although not strictly an attribute of the surrogate endpoint itself, the toxicity
threshold value for each surrogate endpoint needs to be compared to an estimate of
14
-------
Technical Manual for Basic Version of MCnest - February 2013
environmental exposure during an appropriate exposure period. A key issue is determining what
constitutes an "appropriate" exposure period. For example, if effects can occur very rapidly,
such as changes in adult behavior leading to nest abandonment, the appropriate exposure period
may be as short as a single day immediately preceding the effect. On the other hand, for effects
such as reduced hatchability or chick survival due to in ovo exposure, the appropriate exposure
period for rapidly-metabolized pesticides is during the time of egg formation, which occurs days
or weeks before the effect can be observed, and may be defined as a time-weighted average
(TWA) dose based on the length of the rapid follicle growth period for each egg. The selection
of appropriate exposure periods is discussed in greater detail below.
2. Default surrogate endpoints from existing laboratory studies
The Basic Version of MCnest is programmed with a series of eight default surrogate
endpoints from standard laboratory toxicity tests done primarily on northern bobwhite and
mallards (Table 2), each of which can be used or not used at the discretion of the model user.
Many of the surrogate endpoints were proposed in Bennett et al. (2005). A few additional
surrogate endpoints were included in MCnest based on the reasoning presented in a recent
revision of the European Union (EU) Guidance Document on Risk Assessment for Birds and
Mammals (EFSA 2008). This section provides additional guidance for the use of default
surrogate endpoints from existing laboratory tests in MCnest. Depending on the nature of a
particular chemical, additional surrogate endpoints may be appropriate for representing the same
or other potential direct effects. The Basic Version of MCnest has the capability to modify
surrogate endpoints during each breeding phase when they can be justified; however, it is very
important to assure that new surrogate endpoints can be properly linked to an appropriate
exposure period. This is particularly critical where there is a lag time between exposure and
expression of the effect (e.g., in ovo exposure-related effects).
In MCnest the surrogate endpoints are expressed as toxicity threshold doses (i.e.,
mg/kg/day), which represent the dose below which unacceptable adverse effects are not
expected. For surrogate endpoints from the avian reproduction test, the toxicity thresholds may
be based on the NOAEC for that endpoint. However, test endpoints from the avian reproduction
test and the LC50 test are usually reported as the dietary concentration (mg/kg in diet), rather
than as a daily dose. There are multiple possible approaches for making this conversion, but no
standardized approach, so at this point MCnest does not attempt to standardize the method for
the conversion. In the Basic Version of MCnest, it is the responsibility of the model user to
convert dietary concentration for test endpoints into an estimate of the comparable dose using the
information on body weights and food ingestion rates provided in the test reports.
15
-------
Technical Manual for Basic Version of MCnest - February 2013
Table 2. Types of effects and corresponding surrogate endpoints used in the Basic Version of
MCnest for each avian breeding phase.
Breeding phase
Pair formation/
breeding site
selection
Follicle
development
and egg laying
Incubation and
hatching
Nestling rearing
until fledging
Phase-specific effect of
concern
Adult behavioral effects
leading to territory
abandonment or delayed
breeding
Adult behavioral effects
leading to abandonment
of nest attempt
Reduced eggshell quality
leading to abandonment
of nest attempt
Reduced clutch size
Adult behavioral effects
leading to abandonment
of nesting attempt or
reduced nest attentiveness
Reduced fertility
Embryotoxicity from in
ovo exposure leading to
reduced hatchability
Embryotoxicity from
external eggshell
exposure leading to
reduced hatchability
Adult behavioural effects
leading to brood
abandonment or abnormal
parental care
Reduced nestling survival
from direct exposure
Reduced nestling survival
and growth from in ovo
exposure
Type of
effect
Extrinsic
adult
Extrinsic
adult
Extrinsic
adult
Extrinsic
adult
Extrinsic
adult
Extrinsic
adult
Extrinsic
adult
Intrinsic
juvenile
Extrinsic
juvenile
Extrinsic
adult
Extrinsic
juvenile
Intrinsic
juvenile
Test endpoint used as
surrogate
l/lOofLDso
NO AEL for adult body wt
pre-laying
NO AEL for the number of
eggs laid per hen
NO AEL for mean eggshell
thickness
NONE
l/10ofLD50
NO AEL for adult body wt
pre-laying
NO AEL for proportion of
viable eggs per eggs set per
hen
NO AEL for proportion of
hatchlings per viable eggs
per hen :
NONE
1/lOofLDso
NO AEL for adult body wt
pre-laying
1/lOofLDso
Fraction of 5-d LC50
NO AEL for proportion of
14-day-old juveniles per
number of hatchlings per
hen
Comparable
exposure period for
phase-specific RQ
1 -day maximum
estimated daily dose
(EDO)
1 -day EDO
1 -day EDO
1 -day EDO
NONE
1 -day EDO
1 -day EDO
1 -day EDD during
follicle development
and egg laying
Follicle development
time-weighted
average (TWA)
NONE
2-day TWA
2-day TWA
1 -day EDD (juvenile
diet)
5-day TWA (juvenile
diet)
Follicle development
TWA
1 Alternatively, if the NO AEL for proportion of hatchlings per number of viable eggs is not available, use the lower
of the NO AEL for proportion of 3 -week live embryos per number of viable eggs or the NO AEL for proportion of
hatchlings per number of 3 -week live embryos.
16
-------
Technical Manual for Basic Version of MCnest - February 2013
The surrogate endpoints used in the Basic Version of MCnest are as follows:
a. Adultpre-laying body weight
Bennett et al. (2005) proposed using a change in adult body weight during the pre-laying
period of the avian reproduction test as a surrogate endpoint for parental well-being during all
breeding phases from territory establishment to fledging, except the egg-laying phase. This
proposal now has been refined to focus on changes in body weight observed in the first two
weeks after the onset of treatment because it is a surrogate for physiological or behavioral
responses resulting in nest and territory abandonment or reduced nest attentiveness that may
occur rapidly after an initial pesticide exposure. Adult body weight is the surrogate endpoint
with the most indirect connection between the measured effect and the potential responses of
birds in the field. The avian reproduction test is designed so that many of the pesticide-related
responses potentially expressed by breeding adults in the field cannot be observed in the
laboratory. While we cannot observe many of these adult behavioral changes that threaten the
success of a nest, we do measure pesticide-related changes in parental food consumption and
body weight. Food consumption measurements can be highly variable and biased by
unmeasured spillage, whereas changes in body weight can be more accurately measured.
Consequently, a rapid change in body weight during the first two weeks of the pre-laying period
in the avian reproduction test is considered to be an appropriate surrogate of possible effects on
the overall well-being of the adult females in the field that could ultimately lead to nest failure.
Because parental responses may occur rapidly following an initial pesticide exposure, Bennett et
al. (2005) proposed that the NOAEL of the body weight surrogate endpoint be compared to the
expected dietary dose on each day during a breeding phase (i.e., 1-day estimated daily dose or
EDD). For some pesticides, this may lead to a very conservative decision point if the NOAEL
for change in body weight is considerably lower than a dose causing behavioral effects leading to
a nest failure, though in most cases information on this field effect would be unknown.
However, other pesticides may cause effects on adult behavior without impacting body weight in
the laboratory test, leading to decisions that underestimate risks. The adequacy of using a
change in adult body weight as a surrogate endpoint should be evaluated on a pesticide-by-
pesticide basis.
During the nestling-rearing phase, Bennett et al. (2005) argued that the change in pre-
laying body weight should be compared to the 2-day TWA for exposures throughout the phase
because nestlings can withstand reduced parental care for approximately one day, but may not be
able to withstand longer periods of reduced attentiveness.
The comparison of changes in adult body weight during the pre-laying period to single
day (or 2-day TWA) exposure doses has been criticized because the measured effect (i.e., change
in body weight) takes more than a 1-day exposure to be observed. In determining the
appropriate period of exposure to compare with a surrogate endpoint, it is important to focus on
how rapidly the field effect of concern is expressed after an initial exposure, rather than the time
course of the effect measured in the laboratory. In this case, the concern is over sublethal
behavioral and/or physiological effects that can cause nest failures soon after initial exposures,
such as the nest or brood abandonments observed shortly after application by Busby et al. (1990)
and Brewer et al. (1988). Consequently, despite using a change in adult body weight as the
surrogate endpoint, the concern is not over how quickly birds lose weight following a pesticide
17
-------
Technical Manual for Basic Version of MCnest - February 2013
application, because weight loss may not be relevant to the response of wild birds if they
abandon the area or change feeding sites. However, if the laboratory birds show a significant
change in body weight in the first two weeks of exposure, then it is likely that sublethal effects,
such as reduced food consumption or reduced metabolic efficiency, began shortly after the initial
exposure, and that sublethal effects such as these may be indicative of other effects that threaten
the success of the nest attempt.
One issue to consider is that while original avian reproduction study reports present
analyses on treatment-related differences in body weights measured bi-weekly prior to the onset
of egg-laying, the data and analyses on pre-laying body weights are not currently included as part
of the OPP Data Evaluation Records (DER). While the body weight data at the end of the test is
included in the DERs, this endpoint IS NOT an adequate substitute for pre-laying weight as a
surrogate endpoint. Also, there are at least two methods for analyzing the data on changes in
pre-laying body weights. Many avian reproduction study reports simply perform an analysis of
variance on the actual body weights at each pre-laying measurement period. A more statistically
sensitive analysis for treatment effects on body weights would be to use a repeated measures
analysis of variance or to conduct the analysis of variance on the change in weights since test
initiation (i.e., difference in weight between weeks 0 and 2 for each individual) among
treatments.
b. 1/10 of the LD50
One concern raised about the original proposal for using pre-laying body weight as a
surrogate endpoint for adult well-being is that if changes in body weight are not observed until
several weeks after the onset of treatment, this period may be much longer than a realistic
exposure period with most current pesticides. Largely for that reason, the recent revisions to the
EU Guidance Document on Risk Assessment for Birds and Mammals (EFSA 2008) proposed
that an alternate surrogate endpoint of adult well-being be derived from results of the avian acute
toxicity (LD50) test. Ideally, the test could provide information on the single oral dose below
which mortality and/or overt signs of poisoning are not observed. However, many LDso tests
produce mortality and other signs of toxicity at each dose tested and do not adequately document
presence and severity of sublethal signs of poisoning. A review of LDso studies showed that
severe signs of toxicity likely to interfere with a bird's normal activities tend to be recorded at
dosing levels greater than 1/10 of the LDso (Callaghan and Mineau 2000). On the basis of this
work, it is proposed that, as a default, 1/10 of the LDso be used as a surrogate endpoint for effects
on adult behavior leading to disruption on nesting success and that it be compared to the
expected dietary dose on each day during a breeding phase (i.e., 1-day EDD), except during the
nestling rearing phase where it is compared to the 2-day TWA exposure dose. However, if a
model user has chemical-specific data on which to derive a refined estimate of the dose below
which mortality and/or overt signs of poisoning are not observed, this value can be used as an
alternative to the 1/10 of the LD50 surrogate endpoint. The model user is responsible for
providing the rationale for alternative values.
c. Eggshell thickness and number of eggs laid
Mean eggshell thickness per hen and the number of eggs laid per hen are surrogate
endpoints reflecting effects to adults from direct pesticide exposure during the egg-laying phase.
18
-------
Technical Manual for Basic Version of MCnest - February 2013
A reduction in eggshell thickness is a surrogate for nest failures related to cracked and broken
eggs with reduced eggshell quality. Although the effects on eggshell thickness observed in the
avian reproduction test occur after an extended period of pre-laying exposure of the parents,
other studies have shown that pesticides may affect eggshell quality rapidly after the initial
exposure (Bennett et al. 1990, Bennett and Bennett 1990). Adverse effects of reduced eggshell
thickness, such as egg breaking or punctures, may be expressed in the field during either the egg-laying or
incubation phases, but it is used in MCnest as a surrogate endpoint during the earlier breeding phase
where effects may occur and affect the outcome of the nest. A reduction in the number of eggs laid is
a surrogate for effects on adult well-being that can lead to nest abandonment or reduced nest
attentiveness. The cluster analysis conducted by Mineau et al. (1994a) showed that these two
endpoints segregated into different categories of responses observed in avian reproduction tests,
and both are needed as surrogate endpoints to represent the range of parental effects possible
during egg laying. However, reduced egg production in the laboratory test is not an appropriate
indicator of reduced clutch size in the field, because it is unclear if reduced production in the
laboratory translates into a proportional reduction in clutch size, complete abandonment of the
nest, or a longer period of time to complete a normal-size clutch (Mineau 2005). The
determinants of clutch size in the field involve hormonal and sensory cues that are not present in
a laboratory test where eggs are removed daily for artificial incubation. For this reason, reduced
egg production should be seen as a broader indicator of adult well-being during the egg-laying
phase that could ultimately affect reproductive success and that may be expressed in several
ways in the field. Because some pesticides can affect egg production and eggshell thickness
rapidly after initial exposures, Bennett et al. (2005) proposed that both endpoints be compared to
the estimated dietary dose (i.e., 1-day EDD) on each day during the egg-laying phase. If
evidence exists for a pesticide indicating that a longer period of exposure is necessary to produce
effects on these endpoints, an exposure estimate based on a longer TWA may be appropriate, but
the Basic Version of MCnest does not include this option. The existing avian reproduction test
itself does not provide information on the rapidity of onset of effects for the reproductive
endpoints because of the extended period of pre-laying exposure.
d. Proportion of viable eggs per eggs set
Bennett et al. (2005) also proposed that the percentage of viable eggs per number of eggs
set per hen be used in decisions during the egg-laying phase as a surrogate for direct parental
effects leading to reduced egg viability. While this endpoint is intended to be a surrogate
measure for pesticide effects on fertility (i.e., production of infertile eggs), egg viability is
determined by candling eggs at approximately 11 days of incubation, and it is very difficult using
this approach to separate infertility due to parental exposure from early embryo mortality due to
in ovo exposure. Consequently, without detailed analysis of failed eggs, this endpoint potentially
represents a combination of infertility and early embryo death. An additional factor is that
infertility due to parental exposure may be due to the direct chemosterilant activity of the
chemical to males or to sublethal intoxication that reduces or stops mounting behavior by the
male. Without additional testing or knowledge of the chemical, it is difficult to determine from
current practices in the avian reproduction test why eggs are not viable. Also, adverse effects on
egg viability usually would not be detected by the parent until late in incubation, making it more
appropriately a surrogate endpoint for effects observed at the end of the incubation phase (Table
2), rather than during egg laying, as proposed in Bennett et al. (2005).
19
-------
Technical Manual for Basic Version of MCnest - February 2013
Bennett et al. (2005) proposed that this surrogate endpoint be compared to the expected
dietary dose on each day (i.e., 1-day EDD) during the follicle growth and egg production phase.
Jones and Jackson (1972) and Jones et al. (1972) demonstrated that male fertility in Japanese
quail can be decreased within days after single-dose exposures to certain chemicals.
Consequently, using single-day exposure doses during the egg-laying period (from start of the
rapid follicle growth period through egg laying) as the default exposure period is intended to
represent chemicals that can rapidly affect fertility. However, some chemicals can depress
fertility rates for many weeks or even permanently (Jones and Jackson 1972, Jones et al. 1972,
Schafer et al. 1976), in which case, chemical exposures occurring well before the start of egg
laying could lead to higher rates of male infertility. The default exposure period in the Basic
Version of MCnest does not address this situation, and because the effects on the pattern of
fertility vary among chemicals, it would be difficult to establish a single default exposure period
that was suitable for all chemicals. When information exists on the specific effects of a chemical
on the proportion of viable eggs, it is possible that MCnest could be modified to incorporate that
information by modifying the exposure period for comparison with the toxicity threshold value.
The use of the 1-day EDD in the Basic Version of MCnest may represent a conservative estimate
of exposure for comparing with toxicity threshold of this surrogate endpoint.
e. Proportion of hatchlings per viable egg
During the incubation phase, the proportion of hatchlings per number of viable eggs per
hen is proposed as a surrogate for effects on hatchability in the field. This endpoint has been
modified from the proposal in Bennett et al. (2005). In currently accepted protocols, viability is
assessed by candling at 11 days (bobwhite) or 14 days (mallard) of incubation (USEPA 1996).
Originally, this endpoint was proposed as the percentage of hatchlings per number of eggs set per
hen. However, the proportion is more appropriately based on the number of viable eggs to
separate the effects of in ovo exposure on late embryotoxicity from the combined effects of
reduced fertility and early embryo mortality. This surrogate endpoint should be compared to the
time-weighted average of exposures occurring during the period of rapid follicle growth of each
egg prior to laying.
Many studies do not specifically report the proportion of hatchlings per number of viable
eggs per hen, but rather report the proportion of 3-week live embryos per number of viable eggs
and the proportion of hatchlings per number of 3-week live embryos. In such cases, the more
sensitive of the two endpoints should be used as the surrogate endpoint to represent the effects of
in ovo exposure on hatchability. Regardless of which of these endpoints is used as the surrogate
endpoint, it should be compared to the TWA of exposures occurring during the period of rapid
follicle growth prior to laying. Consequently, the duration for the TWA will vary among species
based on the length of their rapid follicle growth period (See additional background in Appendix
A). Because the Basic Version of MCnest estimates the exposure only during the period of rapid
follicle growth, it may not be suitable for bioaccumulative pesticides where the deposition of the
pesticide into yolk may be a function of a longer period of dietary exposure occurring prior to
egg formation.
20
-------
Technical Manual for Basic Version of MCnest - February 2013
/ Proportion of 14-day-old chicks per hatchling
During the juvenile rearing phase, the proportion of 14-day-old chicks per number of
hatchlings per hen is an indicator of effects to chick growth and survival from in ovo exposure.
This endpoint also should be compared to the TWA of exposures occurring during the period of
rapid follicle growth of each egg prior to laying.
g. Fraction of juvenile dietary LCso
Since chicks are raised on untreated diets in the avian reproduction test, the test does not
provide information on their sensitivity to direct pesticide exposures after hatching. As a
surrogate endpoint for direct pesticide exposure to hatchlings and fledglings, Bennett et al.
(2005) proposed using a dietary exposure dose derived from the 5-day dietary toxicity (LCso) test
with juveniles that does not result in adverse effectsessentially an effects threshold.
There are important issues to be addressed when using an endpoint derived from the 5-
day toxicity test in a reproductive success model. First, the 5-day toxicity test is not designed
specifically to determine a toxicity threshold value below which adverse effects on juvenile
survival are not expected, because the emphasis is on selecting treatment concentrations that
would produce some level of mortality between 0 and 100%. In the Basic Version of MCnest we
are proposing that a fraction of the LCso be used to represent a toxicity threshold based on the
levels of concern (LOCs) as defined by USEPA's Office of Pesticide Programs for classifying
risk to birds from short-term dietary exposure. The three LOCs related to the 5-day toxicity test
are 0.5 of the LCso for acute risk, 0.2 for acute restricted-use risk, and 0.1 for acute endangered
species risk (See additional discussion on LOCs at
http://www.epa.gov/oppefed 1 /ecorisk ders/toera risk.htm Last accessed 11 January 2013).
The model user is responsible for using the fraction of the LCso that is appropriate to the specific
pesticide-use scenario. Second, many concerns have been raised about the adequacy of the avian
5-day toxicity test as a quantitative measure of toxicity for use in risk assessment (Hill 1995;
Mineau et al. 1994b). It is considered to be a test of vulnerability instead of toxicity, where
vulnerability is the product of the willingness to consume treated feed, feeding rate, sensitivity to
the pesticide, and temporal pattern of pesticide availability (Hill 1995). Two studies designed to
directly compare the results of the laboratory 5-day toxicity test with same age birds in the field
observed that not only was the mortality rate higher in the field than in the laboratory at
comparable exposure levels, but the timing and nature of mortality was very different (Matz et
al. 1998, Vyas et al. 2006). Consequently, the adequacy and use of a surrogate endpoint derived
from the 5-day toxicity test should be evaluated and decided on a pesticide-by-pesticide basis.
Bennett et al. (2005) argued that the surrogate endpoint from the LCso should be
compared to a 5-day TWA of dietary exposure to juveniles. Consequently, comparisons between
the toxicity threshold and the 5-day TWA begin once nestlings have had 5 days of dietary
exposure (i.e., at beginning of 6 day since hatching) and continue each day until fledging. It
could be argued that some pesticides act very quickly (i.e., most mortalities occur in the first day
or two) so the toxicity threshold does not require a full 5 days of exposure before effects are
observed. However, LCso test reports may not adequately document the time course of
mortalities, and the analysis of time to death is not a primary purpose of the test. Given the other
21
-------
Technical Manual for Basic Version of MCnest - February 2013
limitations of the test discussed above, we have not tried to tailor the duration of the TWA to the
temporal pattern of observed mortalities.
h. 1/10 of the LDso (as it pertains to juveniles-not a separate input)
The EU Guidance Document on Risk Assessment for Birds and Mammals (EFSA 2008)
alternatively proposed to use 1/10 of the adult LDso to assess the ability of juveniles to grow and
develop. This is based on the assumption that for precocial young, at least, there is no systematic
difference between the relative sensitivity of juveniles and adults (Hudson et al. 1972). There
may be differences on a substance-by-substance basis, but no systematic correction factor is
available. It should be noted that this may not be the case for altricial young (i.e., species where
the young hatch blind and are tended by their parents, such as passerines). For example, altricial
juveniles have been shown to be more sensitive to cholinesterase-inhibiting chemicals than
adults (Wolfe and Kendall, 1998). However, it is not known whether this difference applies to
pesticides with other modes of action. In the absence of any further information, it is proposed
that 1/10 of the LD50 be used as the surrogate endpoint for direct toxicity to juveniles, and it
should be compared to the expected dietary dose to juveniles on each day during the nestling
rearing phase (i.e., 1-day EDD).
/'. Other integrative endpoints
There are other endpoints measured in the avian reproduction test that have not been
mentioned as possible surrogate endpoints; however, there are situations where they may be
adequate substitutes for the default surrogate endpoints. These endpoints are primarily expressed
as counts (i.e., the number of eggs set, viable eggs, 3-week live embryos, hatchlings, or 14-day-
old chicks per hen) and integrate information from a combination of factors including different
types of possible effects and husbandry issues. For example, the number of eggs set per hen
represents the number of eggs laid minus the number of eggs removed for eggshell quality
measurement and the number of eggs cracked for any number of reasons. The number of 14-
day-old chicks per hen integrates many of the types of effects on parents via direct exposure and
on embryos/chicks via in ovo exposure into a single metric. Surrogate endpoints are most
effective in models of reproductive success when they represent a specific type of field effect.
Usually a response observed in integrative endpoints should also be observed in the more effect-
specific endpoints that are proposed as surrogate endpoints. However, in some avian
reproduction tests the data analysis may indicate that an integrative endpoint is more sensitive
(i.e., lower NOAEL) than each of the proposed surrogate endpoints that form the basis for the
integrative endpoint. An example of this would be if the NOAEL for the number of hatchlings
per hen was lower than the NOAELs determined for the number of eggs laid per hen, the
proportion of viable embryos per egg set, and the proportion of hatchlings per viable egg. In
some cases, this may result from differences in statistical power among dissimilar endpoints. In
other cases, it may be the cumulative impact of integrating several types of specific effects.
When this occurs, it is critical to determine if the lower NOAEL in the integrative endpoint
simply reflects significant parental effects, such as a treatment-related decrease in the number of
eggs laid, or a true measure of in ovo effects independent of parental effects. Unless there is a
clear indication of in ovo effects independent of possible parental effects, an integrative endpoint
could be a very misleading substitute for surrogate endpoints already proposed.
22
-------
Technical Manual for Basic Version of MCnest - February 2013
3. Selecting toxicity threshold values for each surrogate endpoint from toxicity tests
For each of the default surrogate endpoints, a toxicity threshold value is determined that
represents the daily dose below which the risk of adverse effects is considered acceptable. In
most avian reproduction tests with experimental designs based on hypothesis testing, the toxicity
threshold value may be defined as the no-observed-adverse-effect level (NOAEL). Where
quantitative dose-response relationships have been defined, the toxicity threshold level may be
expressed as an ECXX. In the conceptual approach presented in Bennett et al. (2005), the toxicity
threshold values for each surrogate endpoint derived from the avian reproduction test were
defined as the NOAEL determined in the test. This reflects the common practice of using the
NOAEL of avian reproduction endpoints for characterization of risks via risk quotients. When
avian reproduction tests were conducted for two species (e.g., northern bobwhite and mallard),
Shore et al. (2005) used the lower of the two NOAELs for each surrogate endpoint. However,
the selection of the toxicity threshold value to use in decisions is a combination of science (e.g.,
what is a biologically-meaningful description of level of effect?) and policy (e.g., what is the
intended level of protection?). A European Union opinion paper on pesticide risks to mammals
presents an argument for why a higher value than the NOAEL may be appropriate for some
surrogate endpoints (EPSA 2006). Specifically, they argue that for sensitive endpoints a
statistically-significant difference may not equate to a biologically-significant effect and that the
acceptable toxicity threshold may be at the LOAEL or higher. In other cases, it may be argued
that due to low statistical power in a test, the NOAEL for a surrogate endpoint may not be
considered suitably protective, so a lower value may be more appropriate. Ultimately, risk
assessors need to evaluate the toxicity threshold value associated with each surrogate endpoint to
insure it is suitable for the intended level of protection in the assessment.
The EFSA (2006) document states that the use of "acceptable levels" in place of
NOAELs would require:
1) quantification of the dose-response relationship and its uncertainty;
2) knowledge of the functional relationship between the measured parameter and
individual reproductive success and survival;
3) knowledge of the relationship between individual reproductive success and impacts on
population dynamics;
4) knowledge of how these functional relationships in 2) and 3) vary between mammal
species with different life-history and developmental traits, and vary between
captive (often inbred) and wild mammals; and
5) a risk management judgment about what types and magnitude of effects are
acceptable.
Although these requirements were written regarding mammals, they are equally relevant for
birds. Adoption of such an approach for selecting "acceptable levels" is currently hampered by
poor definition of the dose-response relationship for most pesticides and by lack of knowledge in
areas 2), 3) and 4) above.
Although toxicity threshold values are expressed as daily dose (jig Al/g body wt/day), the
surrogate endpoints from the avian reproduction test and the dietary LC50 test are typically
reported as dietary concentrations (jig Al/g food). Dietary concentrations need to be converted
to daily doses by the model user for use in MCnest. The conversion from concentration to dose
23
-------
Technical Manual for Basic Version of MCnest - February 2013
can be accomplished using information on body weights and daily food ingestion rates (FIR)
from the toxicity test:
Daily dose (mg Al/kg /day) = Dietary concentration (mg Al/kg food) x FIR (g food/day)
Body weight (g).
This conversion is an approximation because body weight and food ingestion rates are changing
during the course of both the reproduction test and the LC50 test. In the reproduction test, both
body weights and food ingestion rates increase as birds move into egg production. The LC50
uses juvenile birds that grow rapidly during the test. Also, laboratory studies vary in the degree
to which they quantify food spillage during the tests. One option for converting dietary
concentrations to daily doses is to calculate the average daily food ingestion rate per bird and the
average body weight at the beginning and end of the test period for each bird. Because of
differences among tests in exactly how body weights and food ingestion rates are reported, it
may be difficult to standardize a specific algorithm for making this conversion. Consequently,
the Basic Version of MCnest does not convert dietary concentrations into daily doses. This is
the responsibility of the model user.
Another issue that complicates the selection of toxicity threshold values for some
surrogate endpoints from the avian reproduction test is not having "bounded NOAELs," defined
as the next lower tested dietary treatment group below the lowest-observed-adverse-effect level,
or LOAEL, determined by analysis of variance. Even in well-designed avian reproduction
studies, some endpoints may not be affected by the pesticide within the range of dietary
treatments tested (i.e., the mean responses for the endpoints at all treatment levels are not
statistically different from controls). In such cases, the NOAEL is reported as being the highest
tested treatment level (i.e., an "unbounded NOAEL"), but it is not known if a statistically-
significant effect would occur at levels just above those tested or not at all because the endpoint
is insensitive to the chemical. This highest treatment level could be selected as a conservative
estimate of the toxicity threshold value for a surrogate endpoint, but this becomes problematic if
a proposed application rate results in an estimated exposure dose that is higher than the highest
treatment level used in the test. When this occurs at a MCnest decision point, the nests would be
considered to have failed even though there may be no evidence from the test to indicate that
such failures might occur or be possible. Consequently, the model user should evaluate each
surrogate endpoint on a case-by-case basis. An alternative approach when a surrogate endpoint
has an unbounded NOAEL is to not designate a toxicity threshold value (e.g., leave it at the
default value of 9999) that would lead to a nest failure, unless there is additional evidence on
which to base an appropriate toxicity threshold value.
4. Addressing interspecies variability in selecting toxicity thresholds
There is little guidance for addressing interspecies variability for effects on reproductive
success. Development of species sensitivity distributions has focused primarily on acute toxicity
data since avian reproduction tests are conducted on so few species-typically northern bobwhite
and mallards. Mineau et al. (2001) considered that variation among species in reproductive tests
would be at least as great as that observed in acute tests and recommended that extrapolation
factors derived from acute toxicity data could be applied to reproductive endpoints. Luttik et al.
(2005) summarize possible approaches for addressing interspecies variability in effects on avian
24
-------
Technical Manual for Basic Version of MCnest - February 2013
reproductive endpoints and recommend a method for estimating extrapolation factors proposed
by Luttik and Aldenberg (1995, 1997).
The Basic Version of MCnest does not explicitly address modifications of toxicity
thresholds based on differences in toxicological sensitivity among species. It is the responsibility
of the model user to determine if interspecies extrapolation factors (or other means of addressing
interspecies variability) are warranted. If they are, the model user may modify the input values
for the toxicity thresholds accordingly. Shore et al. (2005) present an example of this approach.
5. Modifying or adding surrogate endpoints
The list of default surrogate endpoints in the Basic Version of MCnest was developed to
address most types of direct effects to reproduction. However, for some pesticides it may be
appropriate to modify existing surrogate endpoints or add new surrogate endpoints based on
knowledge of the chemical. Users should clearly articulate the rationale for changes to the
default list of surrogate endpoints and realize that changes appropriate for a specific pesticide
may not be appropriate for others.
Some modifications to existing surrogate endpoints have already been discussed above,
including:
1. Using a toxicity threshold value different from a NOAEL from the avian reproduction
test for surrogate endpoints where there is information to justify using a higher or lower
value.
2. Using related endpoints that make up one of the existing surrogate endpoints (e.g.,
using the proportion of 3-week live embryos per number of viable eggs and/or the
proportion of hatchlings per number of 3-week live embryos in place of the proportion of
hatchlings per number of viable eggs).
3. Using related integrative endpoints (i.e., counts rather than proportions) to replace an
existing surrogate endpoint when an argument can be made that the integrative endpoint
is a better reflection of the response to the pesticide.
For many pesticides, there may be additional information available from non-standard
laboratory studies and field studies for forming the basis of a new surrogate endpoint, especially
if it can be argued that an endpoint based on this information is a better, more direct indicator of
effects in the field. For example, if a field study provided data on the rate of nest abandonment
during incubation caused by a pesticide application of known application rate, this information
may provide a more direct surrogate endpoint than using changes in body weight or a fraction of
the LD50. Similarly, pen or field studies may provide more useful information on juvenile
mortality from direct exposure (i.e., not in ovo exposure) than can be provided from the LC50 or
LD50 test.
A critical element in creating a new surrogate endpoint is making sure the endpoint is
matched up with the appropriate exposure period. This is relatively straightforward if the
endpoint represents an effect that can occur very rapidly after exposure. In these cases the
surrogate endpoint might be matched with single-day exposures occurring during each day of
25
-------
Technical Manual for Basic Version of MCnest - February 2013
that particular breeding phase. However, if a surrogate endpoint reflects a delayed response after
a period of exposure (i.e., effects on hatchability occurring from in ovo exposure during the time
of egg formation), the user must make sure that the endpoint is matched with the appropriate
time-weighted average during egg formation.
The Basic Version of MCnest is not designed to insert completely new surrogate
endpoints in addition to the current default list. If a user had data sufficient to create a new
surrogate endpoint that is essentially a replacement for an existing one and using the same
exposure period, then the user can simply insert the toxicity threshold value for the new
surrogate endpoint in place of the existing endpoint. For example, if data from a field study
indicated that the pesticide had a greater impact on hatchability via in ovo effects compared to
using the NOAEL for hatchability from the reproduction test, the toxicity threshold value derived
from the field could be used to parameterize the hatchability endpoint in the model. However, if
a new surrogate endpoint requires a different exposure period than exists for the default surrogate
endpoints, the Basic Version of MCnest does not have this capability, though future versions of
MCnest could be modified to include additional surrogate endpoints as needed.
C. Parameterizing the pesticide-use scenario
In addition to selecting pesticide toxicity information for use in phase-specific decisions
in MCnest, the pesticide-use scenario needs to be specified for each simulation run, including the
timing of application(s), the application rate(s) expressed in pounds active ingredient per acre,
and the half-life of residue degradation on foods relevant to the species of interest. The Basic
Version of MCnest can simulate single or multiple applications of a pesticide by specifying one
to five dates of applications and the application rate for each date. In each model simulation, all
individuals in a population of breeding birds are exposed to the pesticide application(s) on the
specified date(s).
It is recognized that application dates are not synchronized within a geographical unit,
and that the individual birds within a free-ranging population may experience applications
occurring throughout a range of dates. MCnest does not try to simulate a distribution of
application dates, which can be quite variable over time and space due to factors such as weather
and pest levels. Also, running simulations based on a distribution of application dates can
obscure information about the potential effects on specific dates when an avian species may be at
greatest risk to a particular chemical. Instead it is recommended that simulations be run for a
series of specific dates (using the batch function) covering the range of possible application
dates. By examining the model results for a series of possible application dates, the user has
more information from which to determine which specific dates can lead to the greatest impacts
on each species. The user also gains sufficient information to estimate the overall effect to a
population exposed to any assumed distribution of application dates.
To account for the dissipation of pesticide residues over time, the model user also must
specify the half-life of residues on avian food items for a particular pesticide. Residue half-life
values on plant foods should be obtained from the literature or from registrant submitted studies.
If no foliar residue half-life value is available for a pesticide, USEPA Office of Pesticide
Programs uses a default value of 35 days based on the work by Willis and McDowell (1987) that
reported foliar residue half-lives for approximately 80 pesticides, with a maximum value of 36.9
26
-------
Technical Manual for Basic Version of MCnest - February 2013
days. The average half-life (± standard deviation) across all formulations and extraction methods
for organochlorine, pyrethroid, organophosphorus, and carbamate insecticides was 5.0 ± 4.6 d,
5.3 ± 3.6 d, 3.0 ± 2.7 d, and 2.4 ± 2.0 d, respectively.
In the Basic Version of MCnest, pesticide exposure to a population of birds is simulated
by estimating the initial daily dose for each species on the day of application and assuming that
exposure decreases over time based on the half-life of residue degradation on foods relevant to
the species of interest. The model user has four options for determining the initial daily dose for
each species. Three of the options are based on the approach used in OPP's Terrestrial Residue
EXposure model or T-REX (USEPA 2012) for translating an application rate into the dietary
concentrations on various food types and, finally, into an initial daily dietary dose (|ig/g body
weight/day) for each species based on its body weight, food ingestion rate, and diet composition.
The dietary concentrations on plant food types used in T-REX (Table 3) are based on the reviews
by Hoerger and Kenaga (1972) and Fletcher et al. (1994) that estimated the mean and maximum
residue concentrations on various categories of plant food types after chemical application. OPP
has developed a nomogram showing the relationship of initial residue concentrations among
various plant categories as a function of application rate. In MCnest, the model user can base the
initial exposure doses on: 1) the maximum residue values for each food type, 2) the mean residue
values, or 3) values drawn from a log-normal distribution of residue values for each bird based
on the mean and standard deviation of each food type in Fletcher et al. (1994). For the third
option, once an initial value is drawn from the distribution for each female, the individually-
calculated exposure dose decreases based on the residue half-life. This assumes that the
distribution of initial daily dose values reflects the variation in the initial pesticide concentrations
possible within and among different fields, rather than individual food items, and that the
average initial pesticide concentration in some bird feeding territories is higher than in other
bird's territories. Consequently, instead of having each female in the simulation receive the
same mean or maximum initial dose, the third option results in each female being randomly
assigned an initial dose from the distribution. It should be noted that the lognormal distribution
results in values at the upper end of the distribution that exceed the maximum nomogram value.
The fourth option allows the model user to directly set a specific initial dose for both adults and
juveniles independent of an application rate.
In the Basic Version of MCnest, birds are assumed to get 100% of the daily food intake
from treated fields. This is consistent with the approach used in T-REX. Future versions of the
model could integrate species-specific information of the proportion of the diet derived from
treated fields if a suitable approach is available.
Each species life-history profile includes the estimated proportion of each of six food
type categories in the total diet. The six categories are: 1) short grass, 2) tall grass, 3) broadleaf
forage plants, 4) fruits, 5) seeds and pods, and 6) insects and other invertebrates. Where
possible, the dietary composition information is based on breeding season diets.
27
-------
Technical Manual for Basic Version of MCnest - February 2013
Table 3. Maximum and mean (plus standard deviation) residue concentrations used in
T-REX for the initial dietary concentrations expected on various avian food types immediately
after a pesticide application of 1 Ib active ingredient/acre.
Dietary-based EECs (|ig/g food)
Short Grass
Tall Grass
Broadleaf Forage Plants
Fruits
Seeds & pods
Insects & Other Invertebrates
Initial Residue Concentrations
Maximum
240
110
135
15
15
94
Mean
85
36
45
7
7
65
Standard deviation1
60.3
40.6
56.7
12.4
12.4
48
1 Standard deviations for plant food types derived from Fletcher et al 1994.
Initial residue values for insects derived from Appendix B of the User's Guide for T-REX
Version 1.5 (USEPA2012)
(http://www.epa.sov/oppefedl/models/terrestrial/trex/t rex user suide.htm#app a).
Just as in T-REX, MCnest converts an application rate (expressed as pounds active
ingredient/acre) into an initial daily dose for adults (expressed as |ig/g body weight/day) using
the information on average body weight, food ingestion rate (FIR), and diet composition for each
species of interest. The FIRdry (g dry weight/day) is estimated from an allometric equation for
all birds (FIRdry = 0.648 * Body weight (g wet weight) A 0.651) from Nagy (1987). To convert
the FIR from dry to wet weight, the diet composition must be considered because the moisture
content for seeds is assumed to be lower than for other food types. In T-REX, the mass fraction
of water in food types is 0.1 for seeds and 0.8 for all other food types (e.g., insects, fruits,
leaves). To calculate the FIR as wet weight (FIRWet):
FIRwet = FIRdry / (S * 0.9 + (1 - S) * 0.2),
where S = proportion of seeds in the diet. The amount of food ingested as a proportion of body
weight (FIR/BW) is calculated by dividing FZRwet by the average body weight. The initial daily
dose (IDD) for each species is calculated as the sum across food types of the initial residue
concentration for each food type (Q) multiplied by FIR/BW and the proportion of that food type
in the diet of a species (Pi):
IDD = X Q * FIR/BW * Pi.
Juvenile body weights and FIRs change rapidly from hatching to fledging, and empirical
information on the changes in weights and food consumption of nestlings is lacking for most
avian species. Appendix R of the European Union (EU) Guidance Document on Risk
Assessment for Birds and Mammals (EFSA 2009) calculated the FIR/BW ratios for each day of
the nestling period based on the work of Williams and Prints (1987) on savannah sparrows and
Kendeigh et al. (1977) on house sparrows and found a peak ratio for FIR/BW of 1.08 at about
day 3 after hatch. They proposed that the juvenile IDD be calculated by multiplying the residue
concentrations of various food types in the diet by the 1.08 FIR/BW ratio. This may be
appropriate for species that feed nestlings invertebrates and other high moisture-content foods,
28
-------
Technical Manual for Basic Version of MCnest - February 2013
but for species that feed seeds to nestlings (e.g., mourning doves, American goldfinches), this
will overestimate juvenile food ingestion rates. In the EFSA (2009) analysis, the 1.08 FIR/BW
rate was based on a 4.24 g nestling consuming 4.58 g of insects (wet weight) per day, but not all
species feed nestlings an all invertebrate diet. Using the moisture-content assumptions in T-
REX, the 4.58 g of insects (wet weight) is equivalent to 0.916 g (dry weight) per day. Assuming
the moisture content of seeds is 10%, the equivalent wet weight daily consumption rate would be
1.02 g seeds, resulting in a FIR/BW ratio of 0.24 for a seed diet. Until better information is
available, the Basic Version of MCnest uses a modification of the approach recommended for
the EU to calculate the IDD for juveniles where the seed portion of the juvenile diet is multiplied
by a FIR/BW of 0.24 and the non-seed portion is multiplied by 1.08.
MCnest also has the capability to use exposure profiles generated by EFED's Terrestrial
Investigation Model (TIM) instead of the simple T-REX exposure profile described above.
Details for implementing the integration with TIM exposure profiles are still being developed
and are beyond the scope of the current draft of the technical manual.
1. Consequences of expressing exposure as dose instead of dietary concentration
Historically, risk quotients for avian reproductive effects have been calculated on the
basis of dietary concentration of the pesticide. The lowest NOAEC from acceptable avian
reproduction tests, expressed as the concentration of test substance in the diet, is compared to the
estimated environmental concentration on various types of avian foods just after an application.
By basing reproduction risk quotients on dietary concentrations, T-REX considers the
differences in expected concentrations among food type categories listed in Table 3, but it does
not address differences in moisture content among food categories or differences in the amount
of food consumed as a function of body weight (i.e., FIR/BW). In MCnest, by incorporating the
approach used in T-REX to calculate acute risk quotients, risks to reproductive success are based
on the daily ingested dose by considering not only the residue concentrations on food types, but
also moisture content and FIR/BW.
Although the allometric equation in Nagy (1987) estimates the daily ingestion rate
(expressed on a dry weight basis) for each species as a function of body weight, non-seed food
types have a much higher moisture content than seeds, so species that consume non-seed food
types ingest a greater amount of food on a wet weight basis than granivores. Based on the
moisture content values used in T-REX, a non-granivore would consume 4.5 times more food on
a wet weight basis than a granivore of the same body weight. Consequently, even though the
residue concentrations on fruits and seeds are the same after application based on values in Table
3, the IDD for a frugivore would be 4.5 times higher than the IDD for a same-sized granivore.
Dose also varies as a function of body weight due to the allometric equation in Nagy
(1987), resulting in the FIR/BW ratio increasing as body weight decreases. Consequently,
regardless of the food type eaten, a 20-g or 100-g bird ingests a dose 3.9 or 2.2 times higher than
a 1000-g bird, respectively, based on the calculations used in T-REX. Within a food type
category, the specific food items selected by a 20-g bird vs a 1000-g bird likely differ, but there
are insufficient data to determine how this might affect the relative difference in dose ingested
between these two birds.
29
-------
Technical Manual for Basic Version of MCnest - February 2013
The consequences of expressing exposure as a dose instead of as a dietary concentration
are that when the expected environmental exposure is near the toxicity threshold for sensitive
surrogate endpoints, MCnest simulations may indicate a reduction in seasonal productivity for
some species, even though the traditional risk quotient approach that considers only the dietary
concentration does not indicate a risk of adverse effects to those species. This is primarily
possible for smaller species and non-granivores. However, there are several points for model
users to keep in mind when evaluating the output from MCnest in relation to assessments based
on the traditional risk quotient approach for reproductive effects. At the screening level, risk
quotients for reproductive effects (i.e., often referred to as "chronic risk quotients") are used to
determine if there are values that exceed the established level of concern (LOG) so that the
pesticide can be classified as to whether there is or is not a presumption of unacceptable risk.
Although T-REX calculates reproductive risk quotients based on diets containing short or tall
grass, broadleaf plants, fruits, seeds, or invertebrates, the risk conclusions are based on the
highest quotients calculated, and the risk quotients for consumers of short grass are always 16
times higher than for granivores and 2.5 times higher than for insectivores (based on the
maximum values in Table 3). Consequently, risk quotients are primarily designed to determine
which pesticides do not need further assessment because even when using worst-case
assumptions their quotients do not exceed LOCs, but risk quotients do not provide information
on the probability or magnitude of risk and tell us little about which species or life-history
strategies may be at risk.
MCnest is intended to quantify the effects to reproductive success for a range of avian
species to provide more information on which species or life-history strategies are at greatest risk
or which exposure scenarios produce the greatest effects. It does that, in part, by refining the
exposure profile for each species by considering the diet composition during the breeding season
and by refining the estimated daily dietary exposure by integrating information on the moisture
content of foods in the diet and on the food ingestion rate for the body weight of each species.
MCnest is based on the assumption that daily dose is a more ecologically-relevant expression of
exposure than simply using residue concentrations on food types as the only factor. We also
assume that MCnest is best suited to be used in higher tier risk assessments for pesticides where
there is already a presumption of unacceptable risks based on screening-level assessments that
requires further refinement and analysis. However, refining the estimates of daily dose for a
series of species focuses attention on the empirical basis for the default assumptions used in
those estimates. For example, how much variation is there in moisture content within food types
and are the default moisture content values similarly representative of the moisture content of
diets among species? Are the maximum and mean nomogram values (along with the standard
deviations) for each food type presented in Table 3 representative of the distribution of residue
values occurring on the foods consumed by each species? These default values may be the best
information we have currently, but given their importance in the estimation of daily doses for
avian species, additional research may be needed to improve the quality of dose estimates.
While the basic version of MCnest uses default values from T-REX for moisture content and
nomogram values, all of these input parameters are editable by the model user.
D. Using default species life-history profiles
To run simulations in MCnest for specific avian species, a suite of life-history parameters
is required to describe the typical series of events during a breeding season. In developing
30
-------
Technical Manual for Basic Version of MCnest - February 2013
MCnest, we focused on a generalized model requiring a limited number of life-history
parameters that can be applied to a broad range of species with limited life-history data. Also,
the model treats avian nesting in a manner consistent with the way in which nesting parameters
are typically estimated in the field, thus reducing the potential for currency mismatch between
the model and available ecological data for the species of interest. Because MCnest runs on
daily time steps, all life-history parameters expressed as durations (i.e., number of days) must be
set as integer values.
The Basic Version of MCnest accesses a series of default life-history profiles for avian
species associated with agroecosystems. These species profiles are draft versions for a range of
life-history strategies and at this point are available for demonstration purposes. Once the Basic
Version of MCnest is finalized for distribution and the exact structure for life-history profiles has
been approved, the current series of life-history profiles will be finalized for peer review and
work will begin on expanding the library of species, including variations in species profiles
where important differences exist among geographical regions within a species range. The
current species profiles include one example of this for the tree swallow, where two profiles
represent northern and southern parts of their range.
Typical values for each of the following life-history parameters were harvested from the
published literature. Each of the parameters is subject to variation among locations, within and
among breeding seasons, and among techniques and study designs used to collect data. For most
species, there is insufficient information to understand the extent to which these factors truly
affect parameter estimates. For species with several published studies with consistent results, a
mean value may be selected, while for other species the selected values may come from a single
high quality study. Some life-history parameters are relatively easy to quantify and have a long
history of being reported in the published literature (e.g., clutch size), while other parameters
may be difficult to quantify or their quantification varies depending on the field methods
employed (e.g., waiting periods from success or failure until first egg is laid in new nest).
Consequently, when using the word "typical" in selecting values for each life-history parameter,
the goal is to create a species profile that provides a reasonable representation of a species'
breeding season across its entire range. It is recognized that the life-history parameters for many
species with extensive ranges may vary among regions. This is especially true of the timing and
duration of the egg-laying period. Future versions of the default species profiles may include
region-specific species profiles for species with adequate data. Currently, model users can use
the default life-history parameters or modify values to reflect regional variability.
Model users also may want to modify a species profile to represent a "reasonable worst-
case" profile for a species in order to assure that model projections are adequately protective of
all regional populations and subpopulations. While this may seem simple in concept, it must be
approached with great caution in MCnest where temporal issues are extremely important. For
example, because the start and end of egg laying and the length of the egg-laying period may
vary from site to site, a model user may want to know what start and end egg-laying dates
produce the greatest relative impact on reproductive success. However, the reality is that
changing these dates may increase the apparent impact under some pesticide-use scenarios and
decrease it under others. The bottom line is that no modifications to species life-history
parameters will result in a relative impact that is universally a worst-case scenario. This is an
issue that we will continue to explore as the species library expands and regional variations in
31
-------
Technical Manual for Basic Version of MCnest - February 2013
species profiles are considered. However, at this point model users are advised to use caution
and explore changes in species profiles on a case by case basis.
1. Species life-history parameters
Each species life-history profile is defined by the following suite of parameters:
a. Initiation probability
Although species vary in the degree to which nest initiation is synchronized, there is
variation in nest initiation dates among females in a population. For most species there is
insufficient information to empirically define a distribution of first nest initiations, so the Basic
Version of MCnest uses a geometric distribution defined by the value of the initiation probability
to create a distribution of nest initiation dates for each simulation. Starting on the date of the
first egg laid in the first nest of a season (Ji), the initiation probability defines the probability that
a female that has not yet started laying will initiate the first nesting attempt. The default value
for the initiation probability in each species life-history profile is 0.25 (i.e., each day 25% of the
remaining females in the population initiate their first nest attempt). The initiation probability
must be > 0 and < 1. If the initiation probability is set equal to 1, all females initiate nests on the
same date (i.e., the first day of egg laying, T\). The lower the initiation probability, the broader
the distribution of nest initiation dates.
Research will continue into methods for improving the definition for the start and end of
the egg-laying period for females of each species, but those methods are not currently available
for the Basic Version of MCnest. In the meantime, if a model user has data for a specific
distribution of first nest start dates for a particular species, there is a way (albeit more
cumbersome) to use that distribution instead of the default initiation probability. The model user
will need to run a series of simulations by resetting the initiation probability to 1 and using a
series of dates for T] that represent the range of first nest start dates. Next calculate the sum of
number of successful broods per female on each date multiplied by the probability of nests
starting on that date. This will provide an overall estimate of the number of successful nests per
female based on the user-provided distribution of start dates.
b. Daily background nest failure rate during laying and incubation
Nests fail due to a variety of non-pesticide causes, such as predation and adverse weather
events. Many nesting studies report data on nest survival during laying and incubation as a daily
nest survival rate using methods, such as those described by Mayfield (1961, 1975), to account
for biases existing when the fate of each nest is not known throughout the entire period. The
daily nest mortality rate (mi) is 1 minus the daily nest survival rate. Other studies may report
only the apparent nest survival or failure rate during the egg-laying and incubation periods (i.e., #
successful or failed nests/total # nests). The apparent rates can be converted to the daily rate
over the number of days for the egg-laying and incubation periods, as follows:
where,
32
-------
Technical Manual for Basic Version of MCnest - February 2013
si is daily nest survival rate,
Si is overall survival rate for egg-laying and incubation periods, and
a\ is the age, in days since the first egg was laid, at which nests typically hatch. Then:
ml = 1 - s1.
However, this approach introduces bias by assuming knowledge of the fate of nests over the
entire period when this is not the case. Apparent nest success rates overestimate the daily nest
survival rates because some nests may fail before detection; however the degree of
overestimation varies considerably due to the specific methods used in studies.
Some studies will report only an overall apparent nest survival rate for the entire nesting
period (i.e., egg and nestling phases). As in other situations, overall nest survival rates can be
converted to the daily rate over the number of days for the entire nest period (i.e., egg-laying,
incubation, and nestling-rearing periods), with the same daily rate assigned to mi and mi.
c. Daily background nest failure rate during nestling rearing (m^)
Many nesting studies report data on daily nest survival rates during the nestling rearing
phase. When only an apparent nest survival rate for the brood rearing phase is reported, it can be
converted to a daily nest survival rate over the number of days in the brood rearing phase.
d. Date of first egg laid in first nest of season (Ti)
The length of the clutch initiation period is defined by the difference between the first
egg in the first and last nests of the season (i.e., Tiast - TI). In the Basic Version of MCnest, new
nests can only be initiated within this period. However, even though some literature sources
report extreme egg-laying dates, the TI parameter is intended to represent when egg laying
typically begins for the species of concern. Using extreme laying dates for TI and Tiast may
overestimate the length of the clutch initiation period for a species, resulting in more nest
attempts than are commonly observed. The value for Tj could represent the mean of several
"first dates" reported from multiple studies or from a high quality study over multiple years.
e. Date of first egg laid in last nest of season (T^
Similar to TI, this represents the typical date for the first egg in the last nest of the season.
Again, this is not intended to represent extreme egg-laying dates as that would serve to extend
the simulated length of the breeding season and bias the model output.
/ Length of rapid follicle growth period in days for each egg (rfg)
Unlike fish, amphibians, and reptiles, birds do not lay their eggs in masses. Instead, most
birds lay an egg each day until they complete a clutch, while some birds may lay an egg every
other day or some other period longer than one day. The follicles that develop into egg yolks
also start growing on a staggered schedule over a several day period, known as the rapid follicle
growth (rfg) period. During the rfg period yolk material is deposited to the growing follicle until
it reaches the size of a fully formed yolk just prior to ovulation.
33
-------
Technical Manual for Basic Version of MCnest - February 2013
Although estimates for the duration of the rfg period are available only for a subset of
species, there is sufficient empirical information for estimating the rfg period for most species.
Many song birds (i.e., passerines) have an rfg period of 3 to 4 days while in doves and pigeons
the period is approximately 6 days. See Appendix A (Estimating the length of the rapid follicle
growth period) for additional detail. The value selected for the duration of the rfg period must be
an integer.
g. Length of eggshell formation period in days for each egg (ef)
After the yolky follicle is ovulated, it enters the oviduct for deposition of the albumin
followed by formation of the egg membrane and shell. This process usually takes approximately
24 hours for most species. Consequently, the eggshell formation period is fixed at 1 day in the
model, and thus does not show up in MCnest as one of the input parameters.
h. Mean clutch size (clutch)
Clutch size is one of the most commonly reported avian life-history parameters in the
literature. Here again, the intent is to select a value representing the typical clutch size of a
species rather than extreme values. The value selected for mean clutch size must be an integer.
/'. Mean inter-egg laying interval in days (eli)
As mentioned above, most birds lay one egg each day, while other species may have a
longer mean inter-egg laying interval. The value for eli can be any value > 1 day and can be
expressed as a decimal value.
/'. Egg on which female typically begins incubation-penultimate vs ultimate (penult)
Those species beginning incubation after the last egg is laid are assigned a value for
penult of 0, while those beginning with the penultimate egg are assigned a value of 1. For those
species where both options are possible, a judgment is made as to which option is more typical
for the species.
k. Duration in days from start of incubation to hatch (1)
The duration of the incubation period (I) also is a commonly reported life-history
parameter in the literature. A typical value for the duration of the incubation period should be
selected and it must be expressed as an integer. For pesticides that affect egg hatchability
because of embryotoxicity due to in ovo exposure or infertility, there is a related parameter
known as "doomed incubation" or Id. When pesticide exposure is high enough to cause
embryotoxicity or reduced fertility, it is assumed the female does not become aware that these
effects have occurred until the time at which eggs are expected to hatch. Consequently, the
female continues to incubate the clutch of eggs that has failed a decision point and is considered
to be "doomed," but the nest attempt does not actually fail until the end of incubation period
when the eggs fail to hatch on schedule. In the Basic Version of MCnest the duration of the Id
period is set to the same value as /.
34
-------
Technical Manual for Basic Version of MCnest - February 2013
/. Duration in days from hatch to fledging of nestlings (N)
The duration of the brood rearing period also is a commonly reported life-history
parameter in the literature. A typical value for the duration of brood rearing should be selected
and it must be expressed as an integer. Nestlings of some species can leave the nest early when
stressed by predators or weather events such as floods. While many studies report a wide range
of fledging durations that reflect that some nests fledge early under stress, the intent is to select a
typical value for the brood rearing period reflective of non-stressed conditions (i.e., how long do
nestlings typically remain in the nest if not stressed?).
m. Duration in days since nest failure due to non-pesticide reasons until female initiates
new nest (WJ
After a nest failure due to environmental causes such as predation or weather, females
may attempt to renest after a period of recovery and reinitiation of the egg formation process.
The value for We represents the duration from nest failure until the first egg is laid in a new nest
and must be expressed as an integer. Many high quality studies have data on the duration of this
period.
n. Duration in days since nest failure due to pesticides until female initiates new nest
After a nest failure due to pesticide exposure, females also may attempt to renest after a
period of recovery and reinitiation of the egg formation process. Wp represents the duration from
nest failure until the first egg is laid in a new nest and must be expressed as an integer. Of all the
life-history parameters, Wp may have the poorest amount of information for selecting a value.
Pesticide field studies typically do not provide information on the probability or timing of
renesting after a pesticide-related nest failure. Occasionally, laboratory reproduction studies are
designed to include a period of untreated food at the end of the treatment period. These studies
can provide information on the potential for egg production to increase or restart after treatment
ends, though it is not clear if this is indicative of the potential for free-ranging birds to renest.
Depending on the nature of the pesticide, the model user might assume that Wp equals We if birds
recover quickly from an exposure. However, for chemicals with prolonged or delayed effects
after exposure, a longer duration may be appropriate for Wp. A conservative assumption would
be that females do not renest after pesticide failure (i.e., set Wptoa value larger than the length
of the breeding season). However, suitable field examples of renesting periods after a pesticide-
induced failure have not been found that provide a basis for additional guidance. The Wp cannot
be shorter than We.
As a default in the Basic Version of MCnest, Wp is set equal to We. If a model user
decides to change the value for Wp, it is unlikely that there would be sufficient information to set
species-specific values, so the value for Wp is not located on the "Life History" window of each
species. If the model user is running simulations on a single species, the value for Wp can be
changed on the "Set Pesticide" window; however, if multiple species are being simulated using
the "Batch mode," there is a toggle switch for overriding the Wp value used for all selected
species. If a model user inserts a value that overrides the default Wp values, this value will be
35
-------
Technical Manual for Basic Version of MCnest - February 2013
used as the Wp for all species, except for species where the We value is larger, in which case the
value of We is substituted for Wp.
o. Duration in days since successful fledging until female initiates new nest (Wf)
After successfully fledging an initial brood, some species will attempt one or more
additional broods. Wf represents the duration from successful fledging until the first egg is laid
in a new nest and must be expressed as an integer. In some species, fledglings become the
responsibility of the male while the female immediately initiates a new clutch of eggs. In other
species, both males and females continue to feed and care for fledglings until they become
independent-a period of up to several weeks-after which the female may become available to
start a new nest if time remains in the breeding season. When there is a period of female
involvement in post-fledging care, the estimates of the period of time until the female renests
found in the literature can be quite variable. It is not always clear if the shorter estimates reflect
that some females renest relatively rapidly even if they still are assisting with fledgling care or
that some females have lost their entire broods prior to becoming independent.
p. Mean female body weight and diet composition during breeding season
MCnest simulations involving a pesticide exposure require information on the mean
female body weight (in grams) and diet composition, ideally representing weights and diet
during the breeding season, as well as diet composition of juveniles prior to fledging. The body
weight and diet information is used in converting application rates into the estimated daily
dietary dose (in mg/kg body weight/day) for each species based on the algorithm used in OPP's
T-REX model (http://www.epa.gov/oppefedl/models/terrestrial/trex/t_rex_user_guide.htm). The
diet composition in MCnest species' profiles is expressed as the proportion of the diet in the six
food categories presented in Table 3. It is intended that the proportions are based on the mass
(wet weight) of each type, but in some species literature information on diet composition may
only be expressed as volume or counts of food items.
q. Mean number of fledglings per successful nest
Each species profile contains an estimate of the mean number of fledglings per successful
nest, which is multiplied by the number of successful nests per female per season (i.e., the
primary output from MCnest simulations) to estimate the number of fledglings per female per
season.
2. Representativeness of species profiles
In pesticide risk assessments, the number of successful broods produced under a scenario
without pesticides is compared with results under a specific pesticide-use scenario to calculate
the percent reduction in annual reproductive success due to pesticide exposure. The estimate of
the number of successful broods also is used as a check of how well the default life-history
profile of a species represents its breeding season output compared to available information in
the literature. A comparison of MCnest output with literature-derived estimates of annual
reproductive success is limited by the small number of studies designed specifically to monitor
the cumulative production of juveniles throughout the breeding season in most species. Even
though there are some studies that report estimates of annual reproductive success at a specific
36
-------
Technical Manual for Basic Version of MCnest - February 2013
place and time, it is often difficult to determine how representative those estimates are of the
species in general.
To estimate annual reproductive success, the MCnest estimate of the number of
successful broods per female was multiplied by an estimate of the number of fledglings per
successful brood from the literature. The baseline estimates of the number of successful broods
per female were based on MCnest simulations without a pesticide exposure using a population
size of 100,000. Information on the mean number of fledglings per successful brood and sample
sizes was gathered from literature sources. Although some sources report the number of
fledglings produced per nesting attempt, we only used data based on the number of fledglings
from successful nest attempts (i.e., one or more fledglings). Where there are multiple studies
reporting mean numbers of fledglings, we based the estimate used in the species profile on the
weighted mean number of fledglings per successful nest where sample sizes are reported or a
simple arithmetic mean among studies where sample size information is lacking or incomplete.
Similarly, estimates of annual reproductive success were gleaned from the literature for
16 of the draft default species currently in the MCnest Species Library. This is limited by the
small number of studies designed specifically to monitor the cumulative production of juveniles
throughout the breeding season in most species. The methods for estimating reproductive
success varied greatly among species-from models to empirical field estimates from marked
populations. For eastern meadowlarks (Kershner et al. 2004) and dickcissels (Walk et al. 2004),
field studies with radio-marked females documented annual reproductive success for marked
populations. For many other species information about the number of fledglings per nest attempt
or successful nest was integrated with information on the estimated number of broods per female
to estimate annual reproductive success, though specific approaches varied by study. Also,
Ricklefs and Bloom (1977) developed a simple model using length of breeding season, clutch
size, nest success rate, and waiting periods following success or failure to estimate annual
reproductive success for six species in the MCnest Species Library: horned lark, black-capped
chickadee, eastern bluebird, American robin, northern mockingbird, and red-winged blackbird.
Their model estimate was used for the chickadee and horned larks, but empirically-based
estimates were used for the other species.
Based on the discussion in Etterson et al. (2009) on the use of a fixed breeding season
length, we expected that MCnest-derived estimates of annual reproductive success may exceed
estimates derived from field data. The comparison in Table 4 indicates that this is true in
general. In two species where the literature-derived estimate was higher (i.e., horned lark and
black-capped chickadee), these estimates came from the Ricklefs and Bloom (1977) model
which also was based on a fixed-length breeding season. For many species, using a fixed-length
breeding season will overestimate the number of nesting attempts because each female in the
simulation continues nesting as long as there is still time in the breeding season, whereas in the
field some individuals may quit earlier after a successful or failed nest attempt. This seems to
explain the disparity observed for northern mockingbirds where the reported egg-laying dates
span approximately 4 months, but in MCnest this results in a mean of over 5 nest attempts per
female per year, which is much higher than the 2.7 nest attempts per female reported in
Derrickson and Breitwisch (1992). The probability of quitting after a successful nest attempt
may be relatively high in mockingbirds. In general, while there is anecdotal evidence that some
females quit breeding early, there is little empirical evidence for quantifying quitting
37
-------
Technical Manual for Basic Version of MCnest - February 2013
probabilities. Additional research is needed on methods for improving the definition of breeding
season for modeling annual reproductive success.
[NOTE: The comparisons of MCnest and literature-based estimates of annual
reproductive success presented in Table 4 are preliminary because they are based on draft species
profiles. Once the basic version of MCnest is finalized and the species profile database has been
peer-reviewed and finalized, a more complete comparison of MCnest and literature estimates
will be conducted. However, the preliminary comparisons are provided to help model users
assess the representativeness of the MCnest species profiles.]
E. Modifying and creating species life-history profiles
Users may wish to create new species life-history profiles or modify existing default
species profiles to fit a specific risk assessment application. For species with large geographic
ranges, users may want to modify species profiles to reflect regional variability in key
parameters, especially the starting and ending dates for egg laying.
To create a new species profile, species-specific estimates of each of the life-history
parameters (described above) are needed. While there may be several studies published for a
species that provide data on breeding season parameters, the challenge is in synthesis of
available data because studies differ in their experimental designs, size and duration, techniques
to collect data, methods for analyzing data, and formats for reporting data. Because of these
differences, it is often difficult to determine realistic mean estimates of parameters across
studies. As emphasized in the description of the life-history parameters, the goal is to select
parameter estimates that are typical for the species that collectively provide a reasonable
representation of the breeding season of the species.
38
-------
Technical Manual for Basic Version of MCnest - February 2013
Table 4. Calculated annual reproductive success (ARS) for each species compared to estimates
derived from the literature.
Species
Mallard
Mourning dove
Eastern phoebe
Horned lark
Barn swallow
Black-capped chickadee
Blue-gray gnatcatcher
Eastern bluebird
American robin
Northern mockingbird
Vesper sparrow
Dickcissel
Red-winged blackbird
Eastern meadowlark
American goldfinch
House sparrow
Successful
broods/
female
0.36
2.79
1.58
2.70
1.45
0.94
0.7
1.91
2.02
2.07
0.73
0.62
0.72
1.08
1.66
2.95
Fledglings/
successful
brood
5.0
1.85
4.0
2.46
3.67
5.3
3.0
3.63
2.8
2.8
3.0
2.9
1.86
3.46
2.9
2.68
Calculated
ARS1
1.8
5.2
6.3
6.6
5.3
5.0
2.1
6.9
5.7
5.8
2.2
1.8
1.3
3.7
4.8
7.9
Literature
reported
ARS
1.2
3.6
5.8
6.82
5.5
6.22
1.8
5.0
5.0
2.5
4.2
1.2
1.3
2.6
3.7
7.2
1 Calculated ARS was derived by multiplying the MCnest estimate of the
number of successful broods per female by the mean number of fledglings
per successful nest from the literature.
ARS based on model estimate in Ricklefs and Bloom 1977.
To modify an existing species profile, users can review the data and studies used in
generating the default species profile and determine if modifying the selected parameter
estimates is relevant and appropriate for their specific risk assessment application. For example,
if a risk assessment is focused on a specific state or region, and there is a high quality breeding
season study conducted in that area, the user may want to use that study to modify default
parameters to be more reflective of that area. The timing of the breeding season, reflected in the
starting and ending dates for egg laying, may be the parameters that are most variable among
regions.
F. Running MCnest: Understanding the phase-specific decision points
When running a simulation on MCnest, each female will have one or more nesting
attempts during a breeding season. During each breeding phase of the nesting attempt (see Table
2) there is a series of decision points related to possible responses to chemical exposures. This
section discusses the decisions to be made during each of the breeding phases and the response
39
-------
Technical Manual for Basic Version of MCnest - February 2013
of the female if the estimated exposure is higher than the toxicity threshold values for the
surrogate endpoints. This decision framework is diagrammed in Figure 1.
During the pair formation phase, there are two surrogate endpoints (i.e., the NOAEL of the
pre-laying adult body weight and 1/10 of the LD50) for effects on adult well-being.
Both surrogate endpoints are compared each day of the pair formation phase to the
estimated daily dose for that day. On any day that the estimated exposure dose is greater
than either of the toxicity thresholds for the two surrogate endpoints, the initiation of egg
laying is delayed until the estimated exposure dose decreases below the toxicity threshold
for both surrogate endpoints, at which point the female would move into rapid follicle
growth for the first egg.
During the follicle growth and egg production phase, there are two surrogate endpoints (i.e.,
the NOAELs for the number of eggs laid per hen and mean eggshell thickness per hen) for
effects on adult well-being.
The toxicity threshold for the number of eggs laid is compared to the estimated daily
doses during each day of the follicle growth and egg production phase. On any day
during the phase that the estimated exposure dose is greater than the toxicity threshold for
the number of eggs laid, the nest attempt is considered failed and the female enters the
waiting period after pesticide failure.
The toxicity threshold for eggshell thickness is compared to the estimated daily doses
during each day of shell formation. On the day of shell formation for each egg laid if the
estimated exposure dose is greater than the toxicity threshold for mean eggshell
thickness, the nest attempt is considered failed and the female enters the waiting period
after pesticide failure.
During the incubation phase, there are three surrogate endpoints (i.e., the NOAEL of the pre-
laying adult body weight, 1/10 of the LD50, and the NOAEL for proportion of eggs set that
contain viable embryos) for effects on adult well-being and one surrogate endpoint (i.e. the
NOAEL for the percentage of viable embryos producing hatchlings) for effects on egg
hatchability from in ovo exposure.
Two of the surrogate endpoints for adult well-being (i.e., the NOAEL of the pre-laying
adult body weight and 1/10 of the LD50) are compared each day to the estimated daily
dose for that day. On any day that the estimated exposure dose is greater than either of
the toxicity thresholds for these surrogate endpoints the nest attempt is considered failed,
and the female enters the waiting period after pesticide failure.
The third surrogate endpoint for adult well-being (i.e., the NOAEL for proportion of eggs
set that contain viable embryos) is compared to the estimated daily dose on each day of
the follicle growth and egg production phase. If the estimated daily dose during any day
of the egg-laying period exceeds the toxicity threshold, the clutch is completed and enters
a doomed incubation state, which means the female incubates the clutch, but the nest
40
-------
Technical Manual for Basic Version of MCnest - February 2013
attempt is considered failed on the date the eggs should have hatched, and the female
enters the waiting period after pesticide failure.
The toxicity threshold for egg hatchability endpoint is compared with a time-weighted
average (TWA) for exposure during the rapid follicle growth and shell formation period
of each egg in a clutch. When the TWA exposure for one or more eggs exceeds the
toxicity threshold for hatchability, the clutch is completed and enters a doomed
incubation state, the nest attempt is considered failed on the date the eggs should have
hatched, and the female enters the waiting period after pesticide failure.
During the nestling rearing phase, there are two surrogate endpoints (i.e., the NOAEL of the
pre-laying adult body weight and 1/10 of the LD50) for effects on adult well-being and three
surrogate endpoints for nestling survival-one from in ovo exposure (i.e., the NOAEL for the
percentage of hatchlings surviving to 14 days of age) and two from dietary exposure (i.e.,
fraction of the LC50 and 1/10 of the LD50).
Both surrogate endpoints for adult well-being are compared each day to the 2-day TWA
exposure (i.e., on each day of the phase the average exposure dose on that day and the
preceding day is calculated). On any day that the 2-day TWA exposure is greater than
either toxicity threshold the nest attempt is considered failed, and the female enters the
waiting period after pesticide failure.
The nestling survival endpoint from in ovo exposure is compared with time-weighted
average (TWA) for exposure during the rapid follicle growth and shell formation period
of each egg in a clutch. When the TWA exposure for one or more eggs exceeds the
toxicity threshold for nestling survival, the clutch is completed and enters a "doomed"
incubation state even though it is assumed that at least some of the nestlings will hatch.
Although study reports give the proportion of hatchlings that survive to 14 days of age,
they may not report the distribution of nestling deaths during the 14-day period.
However, since the in ovo exposure may cause mortality in hatchlings soon after they
hatch, the nest attempt then is considered failed on the date the nestlings hatched, and the
female enters the waiting period after pesticide failure.
The two surrogate endpoints for nestling survival from dietary exposure are compared to
different measures of exposure. The fraction of the LC50 endpoint is compared with the
5-day TWA for dietary exposure to nestlings (i.e., average of nestling exposure doses on
each day and the preceding 4 days). Consequently, comparisons between the toxicity
threshold and the 5-day TWA begin once nestlings have had 5 days of dietary exposure
(i.e., at beginning of 6th day since hatching) and continue each day until fledging. With
this surrogate endpoint, nest failure does not occur until nestlings are older than 5 days.
The 1/10 of the LD50 endpoint is compared each day of the nestling rearing phase to the
estimated daily dose for that day based on the nestling diet. If either one of the measures
of nestling exposure exceeds the corresponding toxicity threshold on one or more days,
the nest attempt is considered failed immediately, and the female enters the waiting
period after pesticide failure.
41
-------
Technical Manual for Basic Version of MCnest - February 2013
There is also a probability of nest failure due to ecological causes (e.g., nest predation,
adverse weather, etc.) during each day of the egg-laying, incubation and nestling-rearing phases.
If a nest fails due to ecological causes, the female enters the waiting period after ecological
failures. If a nest attempt proceeds through each breeding stage without a failure due either to an
ecological cause or a pesticide exposure that exceeds a toxicity threshold, the nest attempt is
considered to have been successful, and the female enters the waiting period after fledging
success. This waiting period may be very brief for species where the female is not involved with
post-fledging juvenile care or may be an extended period for species where the female cares for
juveniles until they become more independent.
At the end of the waiting periods after success or failure from either pesticides or ecological
causes, the female would return to egg production in a new nest attempt if there is time
remaining in the breeding season and the estimated daily exposure dose is below the toxicity
thresholds for both the NOAEL of the pre-laying adult body weight and 1/10 of the LD50. If the
estimated daily exposure dose is higher than either of these toxicity thresholds, the female
remains in that waiting period until the exposure dose falls below both toxicity thresholds, at
which time a new nest attempt begins if time remains in the breeding season. If, on the other
hand, the end of a waiting period comes after the date set for the first egg in the last nest of the
season (i.e., 7/arf), the female quits breeding for the season.
G. Model assumptions and uncertainties
Many of the assumptions and uncertainties associated with using the MCnest model have
been mentioned in the previous sections. This section is intended to present a more thorough
discussion of model assumptions and uncertainties and their possible consequences on the model
outcomes. Also, where appropriate, suggestions will be provided for reducing uncertainties and
lessening the reliance on assumptions.
1. Assumptions reflecting limitations of toxicity testing
By necessity, laboratory toxicity tests simplify the way test organisms are exposed to test
substances and limit the range of possible responses to them. Several types of potential field
effects either cannot be observed in the laboratory test (e.g., parental behavioral effects during
incubation or nestling rearing) or, if observable, the response may be restricted by the limitations
of laboratory test systems (e.g., may see a decrease in the rate of egg production with no
information about how this could relate to changes in clutch size in the field). Also, most
reproductive toxicity tests, including those for birds, were not originally designed to quantify the
magnitude of effects on reproductive success, so there is limited information for quantitatively
describing dose-response relationships. To address these limitations, a key assumption in
MCnest is that the surrogate endpoints selected from available toxicity tests are suitable
indicators for the possible field effects due to pesticide exposure. As mentioned above, surrogate
endpoints vary in the degree to which they directly represent the field effect of concern. Even
for relatively indirect surrogate endpoints such as a change in body weight during the pre-laying
period, it is assumed that a pesticide-related reduction in body weight in the laboratory is
indicative of other adverse effects on adult well-being in the field that could lead to nest failure.
42
-------
Technical Manual for Basic Version of MCnest - February 2013
A related assumption is that the NOAEL for surrogate endpoints from the avian reproduction
test, as well as the fractions of the LD50 and LC50, are adequate approximations of the exposure
dose below which unreasonable adverse effects are not expected. Conversely, when exposure
exceeds these toxicity thresholds it is assumed that adverse effects are possible, though because
of the lack of quantitative dose-response information, it often is not possible to estimate the
probability or magnitude of adverse effects. Consequently, when estimated exposure doses
exceed the toxicity thresholds for surrogate endpoints, it is assumed that the most severe effect is
that the nest attempt would fail, with the female having the opportunity to renest //the exposure
dose drops below the toxicity threshold for effects on adult behavior and there is time remaining
in the breeding season. In reality, it is unlikely that all exceedances of the toxicity threshold in
the field would lead to a nest failure, but without additional information it is not possible to
determine the probability of nest failure. By assuming that all exceedances of toxicity thresholds
lead to nest failure, MCnest outputs may represent a conservative estimate of potential pesticide
effects on reproductive success (i.e., may overestimate the effects of the pesticide on overall
reproductive success).
The estimation of pesticide effects on overall reproductive success in MCnest could be
improved if avian reproduction tests were designed to quantitatively describe dose-response
relationships for sensitive endpoints using regression analysis. This has been discussed in
several research papers (Bennett et al. 1990, Stromborg 1986a, b) and review papers (Bennett
and Etterson 2006, Bennett and Ganio 1991, OECD 1996). However, this alone does not address
the fact that many of the measured test endpoints used as surrogates are indirect measures of the
field effect of concern. We also would need to understand the functional relationship between
the surrogate endpoint and the response of the field effect. For example, if the avian
reproduction test were designed to quantify the dose-response relationship for changes in pre-
laying body weight, we would need to understand how that relates to changes in adult well-being
that could lead to nest failure in order to improve the estimation of the percent change in
reproductive success from pesticide exposure. Alternatively, development of an avian
reproduction test using natural incubation has been discussed (Bennett and Ganio 1991, Mineau
2005), which, if available, would reduce the need for indirect surrogate endpoints by more
directly measuring the types of effects possibly occurring in the field. While these issues have
been in discussion for many years, the current avian reproduction test protocol will be the
standard used for the foreseeable future. Consequently, the Basic Version of MCnest focuses on
how to make the greatest use of the existing testing data in improving the quality of risk
assessments, while helping to illuminate testing deficiencies that could be improved upon in the
future.
2. Assumptions related to parameterizing the exposure profile
When data are lacking or highly variable, assumptions are made about exposure
parameter values such as the half-life of pesticide residues on various food types, initial residue
concentrations on food types, dietary composition of each species, and the proportion of the diet
derived from treated fields. In screening-level risk assessments, conservative default
assumptions are usually used for residue half-life estimates or initial residue concentrations to
examine a worst-case exposure scenario. However, beyond screening-level assessments, if the
goal is to more realistically estimate the magnitude of effect on reproductive success in a
population, the use of conservative exposure assumptions may not accomplish this goal because
43
-------
Technical Manual for Basic Version of MCnest - February 2013
not only can they overestimate a pesticide's effect on avian reproductive success under a specific
use scenario, but they also can distort what is learned from comparisons of the relative effects
among species or application dates. To more realistically estimate the magnitude of effects,
values for exposure parameters should be as representative as possible of empirical observations
from the field. Admittedly, this is difficult because often there are limited empirical data on
which to base a parameter estimate. For example, there may be data on the degradation half-life
of a pesticide only on plant leaves, so can one assume the same half-life is appropriate for seeds
and insects? Even when empirical data exist, it is often not possible to determine how
representative the data are of field scenarios. For example, while Fletcher et al (1994)
summarized pesticide residue concentrations on various plant food types from a large number of
existing studies, these studies were not conducted from the perspective of describing the
distribution of residues on foods of wildlife species, so it is unclear how well the reported mean
and standard deviation of pesticide residues on each food type represent the distribution of
residues in the diets of a population a particular species. These uncertainties due to data
limitations need to be considered in light of the management questions being addressed when the
model user is setting exposure parameters.
Dietary composition also varies among individuals with populations and among seasons.
The goal in establishing the default values for diet composition in each species' profile was to
estimate the typical proportion consumed in each of the six diet categories from Table 3 by
females during the breeding season based on available diet information. For species that are
purely granivorous or insectivorous, this is straightforward even though there may be
tremendous variability among individuals or regions in the species of seeds or invertebrates
consumed. For omnivores consuming foods from multiple categories it is more difficult to
estimate proportions consumed from each not only because of the variability among individuals
and regions, but also because of the paucity of quantitative analysis available for some species.
The values selected for each diet category affect the calculation of daily ingested dose because of
differences in the nomogram values and estimates of percent moisture for each diet category.
The Basic Version of MCnest uses nomogram values and percent moisture estimates from OPP's
T-REX model. However, because a model user may want to modify these values with scenario-
specific information, all nomogram values, dietary composition estimates and percent moisture
estimates are editable by the model user. It is the responsibility of the model user to justify
changes to these default values.
Finally, while the food ingestion rate per body weight (FIR/BW) ratio used in calculating
the daily dietary dose is based on species-specific data for adults, this calculation is much more
difficult in nestlings for a couple of reasons. First, nestling body weights and their daily
ingestion rates are changing rapidly from hatching to fledging. Second, while daily growth of
nestlings has been documented in some species, daily estimates of the food ingestion rate in
nestlings are uncommon. The Basic Version of MCnest uses an estimate of nestling FIR/BW
ratio from an analysis discussed in Section C above where the FIR/BW was calculated each day
of the nestling period for an insect diet (EFSA 2009). By choosing the day with the highest
FIR/BW ratio, this approach is intended to represent a conservative estimate of the FIR/BW ratio
throughout the nestling rearing period. This is an exposure parameter endpoint that could be
improved by additional research, but until better information is available the default values for
the FIR/BW ratio in the Basic Version of MCnest are 1.08 for the non-seed portion of the
44
-------
Technical Manual for Basic Version of MCnest - February 2013
nestling diet and 0.24 for the seed portion. The model user can use these default values or
modify the values if suitable data are available.
3. Assumptions related to modeling avian breeding seasons
In developing the MCnest model we tried to minimize the number of life-history
parameters required and to rely on parameters that are typically estimated in field studies and
reported in the literature. Our goal is a model that is applicable to a broad range of species with
limited life-history data. In general, we assume that the species profiles developed are adequate
representations of breeding seasons for those species. It is difficult to verify MCnest estimates of
reproductive success against field estimates because very few field studies are designed
specifically to measure the annual reproductive success of a population or the number of
successful broods per female per year. Even when some field studies do estimate these
parameters, they may reflect the results at a specific place and time, but it is unclear how
representative those results are for the species in general. We develop the most realistic
representation we can of the breeding season of each species, but in risk assessment applications
we are not relying on the absolute estimate of a chemical's effect on reproductive success
because the MCnest model results will be used primarily to estimate the relative difference
between chemical and no-chemical scenarios. However, there still are assumptions made in
modeling a species' breeding season that could bias the relative effects. Most notable among
these assumptions is that a fixed length breeding season is an appropriate description for the start
and end of breeding activity in a population of birds. Under this assumption, all females renest
after both nest successes and failures as long as there is still time remaining in the breeding
season, but all re-nesting ends after a fixed end date. In reality, for many species there is
evidence that the probability that a female will quit breeding for the season increases over time
until finally all females have quit, but there is little data to quantitatively describe these time-
dependent quitting probabilities (Etterson et al. 2009). In the Basic Version of MCnest when a
fixed end to the breeding season is assumed, there may be more females in the simulation nesting
late in the breeding season than occurs in the field. Consequently, if a pesticide is applied late in
a species' breeding season, more nesting attempts may be at risk in the model simulation than
might occur in the field at that time. Research continues on methods for improving the definition
of the length of the breeding season. Similarly, other life-history parameters are assumed to be
constant throughout the breeding season in the Basic Version of MCnest, such as the daily rate of
nest mortality and clutch size. However, for many species these parameters also may vary over
time though there is often not sufficient data available to quantify these temporal relationships.
In MCnest simulations involving a pesticide application(s), every female in the
population is assumed to be exposed to the pesticide. However, given the spatially
heterogeneous use of a pesticide in the field, it is unlikely that all females of a population would
be exposed to a specific pesticide except for possibly locally-defined populations. The results of
MCnest can be used in conjunction with spatially-explicit or spatially-implicit modeling
approaches to simulate the proportion of a population expected to be exposed given the use
patterns of a pesticide, but the Basic Version of MCnest does not incorporate this capability.
Similarly, MCnest simulations assume that each female in the population is an active breeder,
whereas populations of many species may include a proportion of males and females that are not
reproductively active (i.e., non-breeding floaters). To the extent that information exists for
quantifying the proportion of non-breeders in a population, it would be best to use results of
45
-------
Technical Manual for Basic Version of MCnest - February 2013
MCnest in a population modeling framework that can address the issue of floaters, especially
since the proportion of non-breeders each year may be a density-dependent function.
H. Interpreting MCnest outputs
The interpretation of MCnest outputs occurs at two levels. First, the model user needs to
understand the information presented in the series of outputs from each model simulation.
Second, the model user needs to understand how the outputs from a series of MCnest simulations
can be used in ecological risk assessments, including understanding how uncertainties in the
model affect the interpretation of results.
1. Interpreting outputs from individual MCnest simulations
At the completion of each model simulation, several outputs are available. The most
basic output is displayed on the MCnest main window and includes the mean number of
successful broods per female in the population, the 95% confidence interval (CI), and the number
of pesticide applications. The full table of results also can be displayed and includes the mean
number of nest attempts (plus 95% CI), the overall nesting success (plus 95% CI), the date(s) and
application rate(s) for each application, and the initial exposure dose for adults and nestlings
after each application.
The confidence intervals for each of the main output parameters reflect the variability
among the replicate populations used in the simulation. For example, if a simulation is setup to
run 10 replicate populations of 100 females, for a total population size of 1000, the confidence
intervals represent the variability observed among the 10 replicates. The variability among
populations is due to the fact that the temporal pattern of nesting activity for each female varies
due to different dates for the start of the first nest attempt (determined by the initiation
probability) and differences in the timing of nest failures from ecological causes (determined by
mi and m^). Variability among populations also results when using the lognormal distribution
option for creating the exposure profile. Consequently, for simulations that hold pesticide
exposure constant for all females (i.e., using maximum or mean nomogram values), if the
initiation probability for a species is set to 1 and m\ and m^ are set to 0, the responses of all
females and all replicates are identical. Since there are other important sources of uncertainty in
the model that are not quantified (e.g., inability to adequately quantify dose-response
relationships, field effects that are not observed in laboratory test, simple assumptions about the
length of the breeding season), the confidence intervals should not be interpreted as reflection of
the variability around the estimate of the true response of the population, but rather as a measure
of how well the model has converged upon the projected average number of successful broods
per female, given the number of females simulated.
46
-------
Technical Manual for Basic Version of MCnest - February 2013
The model user also can display the "Phase diagram" for each simulation (see example in
Figure 3) which shows the proportion of females in each of 11 states (see Section D for
definitions) throughout the entire breeding season, including:
1. Pair formation phase (PF)
2. Rapid follicle growth period (rfg)
3. Overlap of rapid follicle growth & eggshell formation periods (rfg/ef)
4. Eggshell formation period (ef)
5. Incubation phase (7)
6. Nestling rearing phase (N)
1. Doomed incubation phase (Id)
8. Waiting period after pesticide failure (Wp)
9. Waiting period after ecological failure (We)
10. Waiting period after success (Wf)
11. Quit breeding (0
PF
I rfg | | rfg/ef [
Apr-17 Apr-30 May-13 May-26 Jun-08 Jun-21 Jul-04 Jul-17
Figure 3. Example of phase diagram for eastern meadowlark.
The "Phase diagram" is used to visualize the response of the population to a pesticide
exposure scenario over time, and Section F above discusses several of the transitions between
breeding phases resulting from pesticide-related nest failures. The next few paragraphs discuss
some additional aspects about phase transitions in MCnest that help in the interpretation of the
"Phase diagrams."
At the beginning of the breeding season, the "Phase diagram" shows all females starting
in the pair formation phase before transitioning into egg laying based on a geometric distribution
where 25% of the remaining females each day initiate their first clutch (i.e., initiation probability
of p = 0.25). Three of the diagram states (i.e., 2, 3, and 4) represent the egg-laying period. Each
egg goes through a several day period of rapid follicle growth until the egg yolk is ready for
ovulation and a period of approximately one day from ovulation to laying where the egg albumin
is deposited and the eggshell is formed (See Figure 3 above.). Because birds lay eggs one at a
47
-------
Technical Manual for Basic Version of MCnest - February 2013
time, during the formation of a clutch of eggs there is an initial period where only rapid follicle
growth (RFG) is occurring for the first egg (i.e., state 2), followed by a period where both RFG
for subsequent eggs and eggshell formation (EF) is occurring (i.e., state 3). Finally, there is a
one-day period of EF for the final egg laid (i.e., state 4).
All of the waiting periods are defined as the duration from a nest failure or success until
laying the first egg in a new nest. Because there is also a period of rapid follicle growth for the
first egg of the new nest, the waiting periods, as visualized in the "Phase diagram," may seem
shorter than they are defined for each species because on the days where the waiting period and
rapid follicle growth period overlap, the birds are shown as being in the rapid follicle growth
period (i.e., state 2) in the "Phase diagram." For example, if the waiting period after success is 7
d and the rapid follicle growth period is 4 d, a female will only be in state 10 (i.e., Waiting period
after success) for 3 d before transitioning to state 2. This is especially noticeable in species
where the duration of the waiting periods and rapid follicle growth period are similar (e.g.,
mourning dove). The exception is when pesticide residue concentrations on food remain higher
than the toxicity threshold values for adult behavioral surrogate endpoints, because when this
occurs the birds will remain in the waiting period until residue concentrations fall below these
thresholds.
The MCnest model also produces a "Brood histogram" that plots a histogram based on
the frequency distribution of the number of successful broods per female and an "Exposure
series" that plots the daily exposure dose over time for both adults and juveniles. When the
model user chooses to draw calculated doses from a distribution, the "Exposure series" also plots
the 25th and 75th percentile from that distribution for both adults and juveniles. Finally, the
model produces a "Log file" that documents the results of a simulation run as well as all of the
input parameters. This provides the model user with all the information required to exactly
recreate a model simulation if need be.
After each model simulation, the model user can save a specific simulation or, if multiple
simulations have been run, the entire table of simulations can be saved. When a single
simulation is saved, MCnest automatically names the file with the following format: species
name, date of the run (yyyy.mmm.dd), and time of the run (hh.mm.ss AM/PM). An example file
name is "dickcissel.2011Jun.20.12.30.45 PM." By using the exact date and time in the file
name, there is no possibility of overwriting previously saved files. By saving a model
simulation, the model user has access to all of the model output discussed above. When an entire
table is saved, MCnest prompts the user to supply a file name.
2. Interpreting MCnest outputs in ecological risk assessment
As stated in the introductory overview, to put the output from a single MCnest simulation
involving a pesticide exposure into perspective, it is compared to the output of a simulation(s)
without pesticides. Calculating the relative difference between scenarios with and without
pesticide exposure gives the model user an estimate of the proportional reduction in annual
reproductive success due to the specific pesticide-use scenario simulated. In ecological risk
assessments, this estimate should stimulate at least two additional questions. First, how
reflective is the estimate of the response expected in the field? Second, what does the estimated
reduction in annual reproductive success mean to the sustainability of the population?
48
-------
Technical Manual for Basic Version of MCnest - February 2013
While there is no way of knowing exactly how well the model estimate of reduced annual
reproductive success reflects the potential field response, the model user can examine how
assumptions made in valuation of parameter inputs affect the model outcomes. As is stated in
earlier sections of the Technical Manual, using conservative assumptions about the value of input
parameters, instead of more empirically-based values reflecting observed field data, can lead to
model outcomes that overestimate the proportional reduction in reproductive success. Often it is
necessary to use conservative assumptions in risk assessment when the needed data are absent or
of poor quality. However, when there are data for a specific parameter, even if not considered
adequate for use in risk assessment, the model user can run simulations to examine to what
extent a conservative assumption for a specific parameter estimate affects the overall results
compared to the empirically-based data. This provides some insight into the degree to which
conservative assumptions overestimate the proportional reduction in reproductive success.
There are a couple of conservative assumptions (discussed in previous sections) that are
built into the Basic Version of the MCnest model, and model users cannot examine alternatives.
First, when the estimated exposure exceeds the toxicity threshold for each decision point, the
nest attempt is considered to have failed. If avian reproduction tests were designed to quantify
dose-response relationships for surrogate endpoints AND if we knew the quantitative
relationship between the laboratory-measured surrogate endpoints and the field effects they
represent, decision points could be based on proportional responses rather than the current
success/failure dichotomy, and model users would have the ability to explore alternative
assumptions about these quantitative relationships. There is much about these relationships that
is poorly understood, but the architecture of MCnest already can incorporate dose-response
relationships. The options available in MCnest can expand as our knowledge of these
relationships grows and can be formalized. Second, the length of the egg-laying period for each
species is defined by a start and end date (i.e., Tiast - TI) and all females are assumed to continue
making nest attempts until reaching the end date. Research continues on alternative methods for
defining the length of the egg-laying period for each female. These methods incorporate
knowledge that after each nest success or failure, a portion of the population will quit breeding
for the year, but those alternatives currently are not available in the Basic Version of MCnest.
The second question relevant to risk assessments addresses our understanding of how
changes in one demographic parameter-in this case the fecundity rate-affect the population
level. This question is often simplified to ask how much is too much? The answer to this
question is outside the scope of MCnest and is better addressed through population modeling.
However, MCnest plays a role by providing population modelers with better quantitative
estimates of the change in annual reproductive success from a specific pesticide-use scenario.
I. Quality Assurance: Verifying that the model performs as intended
Beyond the model outputs discussed in the previous section, each simulation produces
several additional outputs that are not available currently in the compiled version of MCnest for
use on computers without Matlab software. However, these output files are used when running
the model in Matlab primarily for diagnostic purposes to verify that the model is performing as
intended. All of the transitions from one state to another (discussed in Sections F and G) can be
verified for each female in the simulation by using a series of output files. The "StateMatrix" is
the primary diagnostic tool for determining if the transitions among states are occurring as
49
-------
Technical Manual for Basic Version of MCnest - February 2013
intended in the model, especially at each of the decision points. For each female in a simulation,
it reports which of the 11 states a female is in on each day of the breeding season. To verify
model performance a series of simulations are run to isolate specific decision points or specific
types of effects, and the "StateMatrix" is reviewed to evaluate if transitions are occurring as
expected. For example, a series of simulations can be set up where only a single surrogate
endpoint is triggered by a brief period of exposure exceedance to isolate specific decision points.
The "StateMatrix" also can be compared with a file called "Endpoints" that calculates the
exposure dose used for each of the MCnest model decision points on each day of the breeding
season. This is the exposure dose that is compared with the toxicity threshold for surrogate
endpoints to determine whether or not there is an exceedance that would lead to a nest failure or
the female can return to egg laying from a waiting period. A related file called "Exceedances"
compares the daily exposure value from the "Endpoints" file with the toxicity threshold values
for each decision point and displays a "1" if exposure exceeds the toxicity threshold or "0" if not.
These three files can be used to verify that all decision points in the simulation are functioning as
intended. Whenever changes are made to the code of a version of MCnest, assessments are
conducted to verify that all transitions continue to function as intended.
We also need to verify that the model is correctly calculating the mean response for
endpoints from all the individual bird responses, as well as the algorithms for calculating the
95% confidence intervals. During each simulation in MCnest, a file called "Broods" is created
that tabulates the number of successful broods, the number of nest attempts, and the proportion
of nest attempts that are successful for each female in the population. For example, if the
simulation used a population size of 1000, the "Broods" file is a matrix of three columns and
1000 rows. To verify that MCnest has correctly calculated the mean number of successful nests
and nest attempts, this matrix can be copied and pasted into Excel and the mean function can be
used for columns 1 (number of successful broods) and 2 (number of nest attempts). To verify the
proportion of nest attempts that are successful for the population, we divide the mean number of
successful nests by the mean number of nest attempts. Simply taking the mean of column 3 (i.e.,
ratio of successful nest/total nests for each female) will not correctly estimate the proportion of
nest attempts that are successful for the population because individual birds differ in the number
of nests attempted. The 95% confidence intervals can be verified using another file created with
each MCnest simulation called "BroodReps." When setting up a simulation in MCnest, the user
chooses the number of population replications and the number of individual females in each
population replicate. For example, the default when MCnest opens is set to 10 replicates of 100
females for a total population size of 1000 females. MCnest captures the mean number of
successful broods, the mean number of nest attempts, and the mean proportion of nest attempts
that are successful for each replicate in the "BroodReps" file as a matrix with three columns and
one row for each replicate. MCnest calculates the 95% confidence interval for these three output
parameters as the mean ± (1.96 * std dev). The calculations are verified by copying the
"BroodReps" matrix into Excel to calculate the mean, standard deviation, and upper and lower
95% confidence limit using Excel functions.
There also is a need to verify that the calculations in MCnest of the exposure doses at application
are functioning as intended. Currently, MCnest uses the same approach as OPP EFED's T-REX
model for converting an application rate, expressed as pounds per acre, into an initial daily
dietary dose, expressed as mg/kg body weight/day, for each species. MCnest uses the same
formulas for integrating information on diet composition and body weight to perform this
50
-------
Technical Manual for Basic Version of MCnest - February 2013
conversion, although MCnest is estimating the exposure for specific species rather than
generalized species used in T-REX. For quality assurance purposes, an Excel spreadsheet has
been developed that calculates the initial dietary dose for the specific body weights and diet
compositions of the species currently used in the species life-history database at a specified
application rate using the T-REX formulas. These calculations are compared to the initial dose
calculations used in MCnest simulations to verify that MCnest is calculating initial daily
exposure doses exactly the same as the approach used in T-REX. This spreadsheet currently
exists only as an internal quality assurance tool. Once the species library is expanded and
finalized, the spreadsheet may be available as part of the species library documentation.
J. Analysis of model sensitivity to variation in input parameters
Sensitivity analysis is the study of how variation in the model input parameters affects
model outputs. In MCnest, sensitivity analysis is used to determine how changes in input
parameters affect the estimated number of successful broods per female in the simulated
populations-the primary output parameter from MCnest. The first step is to determine which of
the input parameters to consider in the sensitivity analysis. As mentioned in Section A
(Overview of the conceptual approach used in MCnest), there are three categories of input
parameters: 1) species life-history parameters, 2) pesticide application scenarios, and 3) toxicity
threshold values for surrogate endpoints.
1. Approach for a sensitivity analysis of life-history parameters
Species life-history profiles used in MCnest are based on a series of parameters gleaned
from a variety of literature sources including journal articles, books, and reports. Many of the
life-history parameters reported in the literature may vary considerably within and among
studies. The variability evident within studies may be from year-to-year differences or
differences among sites with different habitat quality or weather-related parameters. Variability
among studies may be evident because studies were conducted at different times and/or different
locations or because of differences in experimental methods. Selection of typical values for life-
history parameters is made more complicated when information from the literature is variable,
but the factors responsible for that variability are poorly understood or described. A sensitivity
analysis of the effects of variation in life-history parameters on MCnest outputs would identify
which parameters have the greatest effect on changing MCnest outputs and would be useful in
the development of species life-history profiles.
To determine which parameters have the greatest impact on MCnest output, input
parameters can be varied by the same relative amount (e.g., each input parameter varied by plus
or minus a specific percentage of the estimate to examine the relative change in model output
relative to a baseline simulation). First a baseline simulation was run for a specific suite of life-
history parameters. Next, a series of simulations was run modifying one parameter at a time by
either increasing or decreasing the baseline value by 20%. The difference in response due to
variation of each life-history parameter was calculated as the proportional change in the number
of successful broods per female relative to the baseline value, i.e., (modified - baseline)/baseline.
The life-history parameters considered were: 1) nest initiation probability, 2) daily nest
mortality rates during egg laying/incubation (m\) and during nestling rearing (mi}, 3) duration of
51
-------
Technical Manual for Basic Version of MCnest - February 2013
the breeding season (7), 4) length of the rapid follicle growth period for each egg (rfg), 5) mean
clutch size (clutch)., 6) duration of incubation period (7), 7) duration of the nestling rearing period
(TV), 8) duration of the waiting period after ecological failure (We), and 9) duration of the waiting
period after success (W/). Because daily nest mortality rates for the two periods (i.e., egg-
laying/incubation and nestling-rearing) are often similar, simulations examined variation in these
rates individually and combined (i.e., varied both m\ and m^ by same amount). The length of the
breeding season was defined here as the difference between the dates of the first egg in the first
nest (Ti) and the first egg in the last nest (Tiast). All durations were expressed in days.
Some of the life history parameters are similar among species, whereas others are quite
different-reflecting the diversity of breeding season strategies among species. Consequently,
there is no single baseline simulation that can be used for exploring variation in input parameters
that would be representative of all species. A series of baseline simulations was used that
reflected the variation observed in three of the most dynamic parameters, while holding the other
parameters constant. The first is the duration of the breeding season (7) to represent short,
medium, and long breeding seasons. Durations of 60, 90, and 120 days were used for Tin
baseline simulations. The second in the combined daily nest mortality rates (mi & m^) to
represent low (0.015 d"1), medium (0.03 d"1), and high (0.045 d"1) daily nest mortality rates. The
third is the waiting period after fledging (WJ) to represent short (10 d), medium (20 d), and long
(40 d) periods of post-fledging juvenile care by the female prior to initiating a new clutch of
eggs. The three levels of three parameters result in 27 combinations of input parameters to
define the suite of baseline simulations. All other input parameters were held constant for the 27
baseline simulations, including: 1) length of the rapid follicle growth period (5 d), 2) mean
clutch size (5 eggs), 3) duration of incubation period (10 d), 4) duration of the nestling rearing
period (10 d), and 5) duration of the waiting period after ecological failure (10 d). Also, all
simulations set the nest initiation probability at 0.25, used an egg-laying interval of 1 day, and set
incubation to start with the laying of the last egg.
For each of the 27 baseline simulations, 20 additional simulations were run by either
increasing or decreasing the baseline value for each life history parameter by 20%. For mi and
mi, simulations were run on each parameter separately and combined (i.e., a 20% increase or
decrease during both m\ and m2). Each simulation was run using 10 replicates of 1000 females
for a total population size of 10,000 females.
2. Results of sensitivity analysis of life history parameters
An immediate pattern that emerged is that variation in the initiation probability and the
duration of the rapid follicle growth period had no effect on MCnest outputs. Changing the
initiation probability by 20% resulted in only small changes to the distribution of first nest
attempts and had no observable impact on overall MCnest outputs. The lack of effect of changes
in the rapid follicle growth period is because the first egg overlaps entirely with the waiting
period after either success or failure of the previous nest attempt. Waiting periods are defined as
the duration between a nest's success or failure (from any cause) until the first egg is laid in the
subsequent nest and are almost always longer than the rapid follicle growth period.
Consequently, varying the rapid follicle growth period has virtually no influence on the number
of nest attempts possible within the breeding season. The results for simulations varying the
length of the follicle growth period or initiation probability are not discussed further.
52
-------
Technical Manual for Basic Version of MCnest - February 2013
Among the 27 baseline simulations the number of successful broods per female ranged
from 0.71 (short season, high nest mortality rates, long Wf) to 2.96 (long season, low nest
mortality rates, short Wf) (Table 5). Not surprisingly, as the length of the breeding season T
increases, the number of successful broods also increases. Also, the number of successful broods
increases as the daily nest mortality rate decreases and the duration of the waiting period after
success Wf decreases. For each simulation, the number of successful broods per female is
reported in Table 5 and the proportional changes from the baseline value are reported in Table 6.
Bar graphs for visualizing the relative sensitivity of variation in life history parameters for each
of the 27 baseline simulation are presented in Appendix A.
Across all baseline scenarios, the largest proportional changes in the number of
successful broods per female were due to variation in T. A 20% decrease in T resulted in a 3% to
19% decrease in the number of successful broods, while a 20% increase in T resulted in an 11%
to 45% increase (Table 6). For most of the baseline scenarios, the proportional increase and
decrease for each baseline scenario are relatively similar, but one set of conditions stand out as
different. For baseline scenarios where the duration of a complete successful nesting attempt
(including the duration of the waiting period after success Wf) is slightly longer than T, a 20%
increase in T provides sufficient time for many of the females to attempt an additional nest. The
most extreme example of this is for the baseline scenario defined as T= 60 d, m\=mi = 0.015 d"
l, and Wf= 40 d (Table 6), where a 20% increase in T resulted in a 45% increase in the number of
successful nests, while a 20% decrease resulted in only a 3% decrease in the number of
successful broods. A similar result was observed for the baseline scenario defined as T= 60 d,
mi=m2 = 0.03 d"1, and Wf= 40 d. There are additional baseline scenarios where a 20% increase
in T resulted in a noticeable increase in the number of successful broods per female because the
additional time allowed for one more nesting attempt than was observed in the baseline
simulation (e.g., T= 60 d, m\ = m2 = 0.045 d"1, and Wf= 40 d and T= 120 d, mi=m2 = 0.015 d"1,
and Wf= 40 d).
Variation in combined daily nest mortality rates (i.e., m\ and m^) also resulted in large
proportional changes in the number of successful broods per females when the baseline level of
daily nest mortality was high (i.e., 0.045 d"1). The largest impact is for the baseline scenario
defined as T= 60 d, m\ = m^ = 0.045 d"1, and Wf= 10 d (Table 6), where a 20% increase in m\
and m^ resulted in an 18% decrease in the number of successful nests, while a 20% decrease
resulted in 22% increase in the number of successful broods. The significance of variation in m\
and m^ combined decreased as the baseline level for m\ & m^. decreased. When considering
variation in m\ and m^ separately, the vast majority of simulations resulted in a less than 10%
change in the number of successful broods per female.
Variation in the waiting period after success Wf resulted in a pattern of responses across
all 27 baseline scenarios that was similar to the pattern observed for T. In general variation in Wf
resulted in less than a 10% change in the number of successful broods (Table 6). However, for
baseline scenarios where the duration of a complete successful nesting attempt (including the
duration of Wf) is slightly longer than T, a 20% decrease in ff/provides sufficient time for many
of the females to attempt an additional nest. Consequently, the largest increases (up to 32%) in
the number of successful broods due to decreases in Wf occurred for the same baseline scenarios
as responded to increases in T (Table 6). This can be visualized in Appendix A where the
MCnest output is skewed to the right for T and W/for those simulations where sufficient time
53
-------
Technical Manual for Basic Version of MCnest - February 2013
exists for an additional nesting attempt relative to the baseline simulation (e.g., T= 60, m\ = mi =
0.015, Wf=40).
Variation in other life-history parameters, such as clutch size (clutch), duration of the
incubation (7) and nestling rearing (TV) periods, and the waiting period after environmental failure
(We), resulted in smaller impacts on the number of successful broods per female, and impacts
were more consistent across the 27 baseline scenarios. Varying clutch size by 20% resulted in
changes in the number of successful broods from 0% to 9% (Table 6), while variation in / or N
resulted in 0% to 13% change in number of broods. Variation in We had a small (1% to 7%)
impact on the number of successful broods per female.
Unfortunately, the life-history parameters that had the greatest impact on the MCnest
outputs (i.e., T, m\, mi, and Wf) are also among the most variable parameters in the literature.
The length of the breeding season, as well as the typical starting and ending dates for egg laying,
vary considerably among geographical regions and in the way they are reported in the literature.
As a life-history profile is developed for a species, considerable thought needs to be given during
evaluation of available literature for selecting dates to represent the start and end of egg laying
that produce an overall species profile that is reflective of that species. Also, for risk
assessments focused on a particular region, it may be appropriate to select dates for the start and
end of egg laying that are representative of that region. Variation in m\ and mi in the literature
reflects that nest success varies due to factors such as habitat quality, weather patterns, and the
abundance of nest predators and parasites. This analysis indicates that it is important to
document the rationale for selecting typical values of m\ and mi in the development of a species
profile because of the impact these values can have on MCnest output. Although the duration of
the waiting period after fledging (W/) differs greatly among species because of differences in the
role of females in post-fledging juvenile care, there also is considerable variation within some
species probably reflecting differences among individual females in how rapidly they can
liberate themselves from fledgling care for renesting. This parameter also suffers because few
studies follow females throughout the breeding season to actually document Wf. Consequently,
much of the information in the literature relevant to W/is anecdotal or based on small sample
sizes. Also, some estimates may reflect females that renest after losing fledged juveniles that
had not completely reached independence. The waiting period after ecological failure (We~) is
similar to Wf in that few studies follow females throughout the breeding season to actually
document We. However, the degree of independence of juveniles is not a factor in We, but the
length of time for rapid follicle growth and/or to build a new nest does affect We. The manner in
which a female loses a nest attempt may also influence We, but few studies provide insight into
these factors.
The avian literature contains much more information on clutch size and the duration of
incubation and nestling rearing phases, so establishing typical values for these parameters is
relatively easy for most species. Given the results of this analysis, variation in selecting typical
values for these parameters will have minor impacts on MCnest outputs.
54
-------
Technical Manual for Basic Version of MCnest - February 2013
Table 5. Changes in the number of successful broods per female due to variation in life history parameters from 27 baseline species
profiles.
Baseline
T
60
90
120
ml&m2
0.015
0.03
0.045
0.015
0.03
0.045
0.015
0.03
0.045
Wf
10
20
40
10
20
40
10
20
40
10
20
40
10
20
40
10
20
40
10
20
40
10
20
40
10
20
40
# broods
1.59
1.47
0.97
1.21
1.11
0.86
0.89
0.83
0.71
2.30
1.97
1.63
1.75
1.54
1.31
1.28
1.18
1.03
2.96
2.52
1.88
2.26
1.98
1.61
1.67
1.51
1.30
Mean number of successful broods per female for life history parameters increased ( A ) or decreased ( T ) by 20%
ml
1.62
1.51
0.97
1.27
1.18
0.89
0.98
0.92
0.75
2.36
2.02
1.65
1.84
1.63
1.36
1.41
1.29
1.11
3.05
2.46
1.90
2.38
1.88
1.67
1.82
1.65
1.39
1.57
1.42
0.96
1.14
1.04
0.83
0.81
0.76
0.67
2.25
1.92
1.60
1.64
1.45
1.24
1.17
1.08
0.95
2.88
2.56
1.86
2.11
2.08
1.55
1.51
1.37
1.19
m2
1.63
1.50
0.97
1.29
1.16
0.89
0.98
0.90
0.76
2.38
2.02
1.68
1.84
1.63
1.38
1.40
1.29
1.11
3.06
2.45
1.91
2.38
1.89
1.66
1.83
1.64
1.39
1.55
1.43
0.96
1.14
1.05
0.83
0.82
0.76
0.66
2.24
1.93
1.59
1.67
1.46
1.24
1.19
1.09
0.96
2.89
2.58
1.86
2.12
2.10
1.54
1.53
1.39
1.21
ml&2
1.69
1.55
0.98
1.37
1.24
0.91
1.09
0.98
0.81
2.43
2.06
1.70
1.95
1.70
1.43
1.54
1.40
1.18
3.15
2.40
1.92
2.53
1.79
1.73
1.99
1.77
1.48
1.52
1.39
0.95
1.07
0.99
0.80
0.73
0.70
0.62
2.18
1.88
1.56
1.55
1.38
1.19
1.07
1.01
0.88
2.81
2.63
1.84
1.98
2.18
1.48
1.38
1.29
1.12
T
1.42
1.26
0.94
1.03
0.94
0.80
0.75
0.71
0.63
1.98
1.64
1.41
1.47
1.28
1.10
1.07
0.97
0.87
2.41
2.15
1.71
1.85
1.65
1.38
1.35
1.25
1.09
1.94
1.63
1.41
1.43
1.27
1.10
1.07
0.97
0.85
2.78
2.32
1.82
2.07
1.81
1.50
1.51
1.38
1.19
3.54
2.99
2.31
2.66
2.32
1.88
1.95
1.79
1.52
clutch
1.62
1.51
0.97
1.27
1.15
0.88
0.93
0.87
0.73
2.38
2.09
1.67
1.82
1.63
1.35
1.35
1.24
1.08
3.07
2.58
1.90
2.35
2.04
1.65
1.76
1.59
1.36
1.55
1.44
0.96
1.16
1.07
0.84
0.85
0.80
0.69
2.25
1.80
1.59
1.67
1.45
1.27
1.21
1.12
0.99
2.87
2.45
1.86
2.16
1.90
1.57
1.57
1.44
1.25
I
1.67
1.53
0.98
1.30
1.19
0.88
0.98
0.93
0.76
2.42
2.16
1.70
1.90
1.68
1.39
1.44
1.32
1.12
3.16
2.63
1.92
2.47
2.13
1.69
1.86
1.67
1.40
1.52
1.40
0.95
1.12
1.03
0.83
0.80
0.76
0.66
2.20
1.78
1.57
1.60
1.41
1.23
1.15
1.06
0.94
2.80
2.38
1.84
2.08
1.82
1.53
1.47
1.35
1.20
N
1.66
1.53
0.97
1.32
1.19
0.88
0.99
0.92
0.76
2.44
2.16
1.71
1.90
1.70
1.39
1.45
1.31
1.13
3.17
2.64
1.92
2.45
2.13
1.69
1.87
1.67
1.40
1.52
1.41
0.95
1.12
1.02
0.82
0.80
0.76
0.67
2.18
1.78
1.57
1.60
1.40
1.22
1.15
1.06
0.94
2.78
2.40
1.84
2.06
1.84
1.51
1.48
1.36
1.20
We
1.62
1.49
0.98
1.27
1.16
0.88
0.94
0.89
0.74
2.35
1.99
1.67
1.81
1.59
1.35
1.35
1.23
1.06
3.04
2.56
1.91
2.33
2.04
1.65
1.76
1.58
1.36
1.56
1.44
0.96
1.18
1.07
0.84
0.85
0.80
0.68
2.26
1.93
1.60
1.67
1.51
1.27
1.22
1.14
0.98
2.90
2.46
1.86
2.17
1.91
1.57
1.59
1.46
1.25
Wf
1.62
1.51
1.28
1.23
1.14
1.02
0.91
0.86
0.78
2.37
2.14
1.74
1.79
1.61
1.40
1.32
1.22
1.08
3.07
2.61
2.20
2.31
2.08
1.76
1.72
1.56
1.38
1.57
1.42
0.96
1.19
1.07
0.86
0.88
0.81
0.71
2.26
1.81
1.54
1.70
1.46
1.23
1.26
1.14
0.97
2.89
2.39
1.84
2.20
1.90
1.55
1.62
1.45
1.25
55
-------
Technical Manual for Basic Version of MCnest - February 2013
Table 6. Proportional change from baseline value due to variation in life history parameters for 27 baseline species profiles, with
proportions > 0.15 highlighted.
Baseline
T
60
90
120
ml&m2
0.015
0.03
0.045
0.015
0.03
0.045
0.015
0.03
0.045
Wf
10
20
40
10
20
40
10
20
40
10
20
40
10
20
40
10
20
40
10
20
40
10
20
40
10
20
40
# broods
1.59
1.47
0.97
1.21
1.11
0.86
0.89
0.83
0.71
2.30
1.97
1.63
1.75
1.54
1.31
1.28
1.18
1.03
2.96
2.52
1.88
2.26
1.98
1.61
1.67
1.51
1.30
Proportional change from baseline for life history parameters increased ( A ) or decreased ( T ) by 20%
ml
0.02
0.03
0.00
0.05
0.06
0.03
0.10
0.11
0.06
0.03
0.03
0.01
0.05
0.06
0.04
0.10
0.09
0.08
0.03
0.02
0.01
0.05
0.05
0.04
0.09
0.09
0.07
-0.01
-0.03
-0.01
-0.06
-0.06
-0.03
-0.09
-0.08
-0.06
-0.02
-0.03
-0.02
-0.06
-0.06
-0.05
-0.09
-0.08
-0.08
-0.03
-0.02
-0.01
-0.07
-0.05
-0.04
-0.10
-0.09
-0.08
m2
0.03
0.02
0.00
0.07
0.05
0.03
0.10
0.08
0.07
0.03
0.03
0.03
0.05
0.06
0.05
0.09
0.09
0.08
0.03
0.02
0.02
0.05
0.06
0.03
0.10
0.09
0.07
-0.03
-0.03
-0.01
-0.06
-0.05
-0.03
-0.08
-0.08
-0.07
-0.03
-0.02
-0.02
-0.05
-0.05
-0.05
-0.07
-0.08
-0.07
-0.02
-0.03
-0.01
-0.06
-0.05
-0.04
-0.08
-0.08
-0.07
ml&2
0.06
0.05
0.01
0.13
0.12
0.06
0.22
0.18
0.14
0.06
0.05
0.04
0.11
0.10
0.09
0.20
0.19
0.15
0.06
0.04
0.02
0.12
0.10
0.07
0.19
0.17
0.14
-0.04
-0.05
-0.02
-0.12
-0.11
-0.07
-0.18
-0.16
-0.13
-0.05
-0.05
-0.04
-0.11
-0.10
-0.09
-0.16
-0.14
-0.15
-0.05
-0.05
-0.02
-0.12
-0.10
-0.08
-0.17
-0.15
-0.14
T
-0.11
-0.14
-0.03
-0.15
-0.15
-0.07
-0.16
-0.14
-0.11
-0.14
-0.17
-0.13
-0.16
-0.17
-0.16
-0.16
-0.18
-0.16
-0.19
-0.15
-0.09
-0.18
-0.17
-0.14
-0.19
-0.17
-0.16
0.22
0.11
0.45
0.18
0.14
0.28
0.20
0.17
0.20
0.21
0.18
0.12
0.18
0.18
0.15
0.18
0.17
0.16
0.20
0.19
0.23
0.18
0.17
0.17
0.17
0.19
0.17
clutch
0.02
0.03
0.00
0.05
0.04
0.02
0.04
0.05
0.03
0.03
0.06
0.02
0.04
0.06
0.03
0.05
0.05
0.05
0.04
0.02
0.01
0.04
0.03
0.02
0.05
0.05
0.05
-0.03
-0.02
-0.01
-0.04
-0.04
-0.02
-0.04
-0.04
-0.03
-0.02
-0.09
-0.02
-0.05
-0.06
-0.03
-0.05
-0.05
-0.04
-0.03
-0.03
-0.01
-0.04
-0.04
-0.02
-0.06
-0.05
-0.04
I
0.05
0.04
0.01
0.07
0.07
0.02
0.10
0.12
0.07
0.05
0.10
0.04
0.09
0.09
0.06
0.13
0.12
0.09
0.07
0.04
0.02
0.09
0.08
0.05
0.11
0.11
0.08
-0.04
-0.05
-0.02
-0.07
-0.07
-0.03
-0.10
-0.08
-0.07
-0.04
-0.10
-0.04
-0.09
-0.08
-0.06
-0.10
-0.10
-0.09
-0.05
-0.06
-0.02
-0.08
-0.08
-0.05
-0.12
-0.11
-0.08
N
0.04
0.04
0.00
0.09
0.07
0.02
0.11
0.11
0.07
0.06
0.10
0.05
0.09
0.10
0.06
0.13
0.11
0.10
0.07
0.05
0.02
0.08
0.08
0.05
0.12
0.11
0.08
-0.04
-0.04
-0.02
-0.07
-0.08
-0.05
-0.10
-0.08
-0.06
-0.05
-0.10
-0.04
-0.09
-0.09
-0.07
-0.10
-0.10
-0.09
-0.06
-0.05
-0.02
-0.09
-0.07
-0.06
-0.11
-0.10
-0.08
We
0.02
0.01
0.01
0.05
0.05
0.02
0.06
0.07
0.04
0.02
0.01
0.02
0.03
0.03
0.03
0.05
0.04
0.03
0.03
0.02
0.02
0.03
0.03
0.02
0.05
0.05
0.05
-0.02
-0.02
-0.01
-0.02
-0.04
-0.02
-0.04
-0.04
-0.04
-0.02
-0.02
-0.02
-0.05
-0.02
-0.03
-0.05
-0.03
-0.05
-0.02
-0.02
-0.01
-0.04
-0.04
-0.02
-0.05
-0.03
-0.04
Wf
0.02
0.03
0.32
0.02
0.03
0.19
0.02
0.04
0.10
0.03
0.09
0.07
0.02
0.05
0.07
0.03
0.03
0.05
0.04
0.04
0.17
0.02
0.05
0.09
0.03
0.03
0.06
-0.01
-0.03
-0.01
-0.02
-0.04
0.00
-0.01
-0.02
0.00
-0.02
-0.08
-0.06
-0.03
-0.05
-0.06
-0.02
-0.03
-0.06
-0.02
-0.05
-0.02
-0.03
-0.04
-0.04
-0.03
-0.04
-0.04
56
-------
Technical Manual for Basic Version of MCnest - February 2013
3. Approach for examining model sensitivity to changes in pesticide-related
parameters
The other two categories of input parameters are specific to the pesticide and pesticide-
use scenario being evaluated. Toxicity threshold values are determined from avian toxicity tests.
The pesticide-use scenario parameters such as the application rate are specified on the pesticide
label, while the residue half-life parameter often is determined from registrant-submitted fate
studies. The range of application dates may not be specified on the label because applications
are tied to the conditions (i.e., weather, pest levels) in each region, but information about typical
application dates may be available from registrants or extension agents.
A specific pesticide-use scenario for a particular pesticide is defined by this suite of
parameters including toxicity threshold values, applications dates(s) and rate(s), and residue half-
life estimates. Variation in any of the toxicity threshold values or pesticide-use scenario
parameters may or may not affect the MCnest output, depending on the specific scenario being
considered. For example, the timing of a pesticide application relative to the timing of the
breeding season of a species is critically important to the magnitude of effect observed. In fact,
because there is virtually an infinite number of combinations of toxicity threshold values and
pesticide application scenarios that could be considered, it is very difficult to generalize how
variation in any particular parameter will affect the MCnest output. For example, increasing or
decreasing a toxicity threshold value by a specific amount will significantly change the MCnest
output results under some pesticide application scenarios (especially when the toxicity threshold
value is close to the estimated dietary exposure at application), but have virtually no effect under
other scenarios (e.g., if a pesticide application causes nest attempts to fail near the end of the
egg-laying window).
Because the response of the MCnest output parameter is so dependent on the unique
toxicity characteristics of each pesticide and on the specific pesticide application scenario for the
simulation (especially timing of the application), a complete sensitivity analysis based on all
combinations of toxicity threshold values and pesticide-use scenario parameters is not possible.
However, once a specific pesticide is identified and the toxicity thresholds and application
scenario parameters are determined, the model user may want to examine variation in certain
parameters, especially the application date. As mentioned above, MCnest runs simulations with
each pesticide application occurring on a single date, but within a geographic area the pesticide
may be used over a range of dates. To examine the differences in model response to pesticide
applications on different dates, it is suggested that the model user run a series of simulations for
various possible application dates. This is simple to do using the batch options and provides
considerable insight into the effect of application timing on avian reproductive success.
Similarly, although a specific pesticide use may have a proposed application rate, the model user
may want to examine how risk to reproductive success changes when changing the proposed
application rate. There may be additional reasons why a model user may want to examine
variation in the toxicity threshold values, application dates, or the estimated residue half-life
values on a pesticide-by-pesticide basis, but this was not done as part of a formal sensitivity
analysis because of the difficulty of generalizing how variation in any particular pesticide
parameter will affect the MCnest output. However, the sensitivity of changes in two parameters
-residue half-life and application date-will be addressed in the remainder of this section to
illustrate how variable the response may be depending on the values chosen for other parameters.
57
-------
Technical Manual for Basic Version of MCnest - February 2013
The quality of information on the half-life of pesticide residues on avian food types varies
among pesticides; i.e., some pesticides may have specific measured degradation rate data for
food types such as seeds or fruits while others may have no degradation data specifically on
avian foods. The Office of Pesticide Programs uses a default degradation half-life on food types
of 35 days based on the work by Willis and McDowell (1987) that reported available foliar
residue half-lives for approximately 80 pesticides, with a maximum value of 36.9 days. They
reported an average half-life (± standard deviation) across all formulations and extraction
methods for organochlorine, pyrethroid, organophosphorus, and carbamate insecticides was 5.0 ±
4.6 d, 5.3 ± 3.6 d, 3.0 ± 2.7 d, and 2.4 ± 2.0 d, respectively. Since the MCnest outputs can be
very sensitive to changes in the residue degradation half-life, the default value may significantly
overestimate the magnitude of effects on reproduction if the actual pesticide half-life on foods is
much shorter. When information on the half-life of residues on foods is uncertain, the model
user can examine the effects of changes in the degradation half-life on MCnest outputs to
understand the implication of input parameter selections.
To evaluate the role of changes to the residue half-life value, a series of simulations was
conducted using a portion of the baseline profiles used above for evaluating life history
parameters. Twelve profiles were developed based on four durations for the egg-laying period
(i.e., 30, 60, 90, and 120 days) and three combined daily nest mortality rates (mi & m2) to
represent low (0.015 d"1), medium (0.03 d"1), and high (0.045 d"1) daily nest mortality rates. In
this series of simulations the waiting period after fledging (Wf) was held constant at 20 d. All
other life history input parameters were held constant for the 12 baseline simulations, including:
1) length of the rapid follicle growth period (5 d), 2) mean clutch size (5 eggs), 3) duration of
incubation period (10 d), 4) duration of the nestling rearing period (10 d), and 5) duration of the
waiting period after ecological failure (10 d). Also, all simulations used an egg-laying interval of
1 day and set incubation to start with the laying of the last egg.
Compared to the simulations above evaluating the sensitivity of life-history parameters,
additional input parameters must be defined. Because these simulations focus on the effect of a
pesticide on estimated annual reproductive success, the timing of the pesticide application
relative to the timing of the breeding season is critical and the body weight and diet composition
information are needed to calculate the initial daily dose. Consequently, in addition to defining
the duration of the egg-laying period (7), the dates for Tj and Tiast must be defined. All 12
baseline profiles used a midpoint date for T of June 15, resulting in dates for T] and 7/arfof May
31-June 30, May 16-July 15, May 1-July 30, and April 16-August 14 for the 30, 60, 90, and 120
d durations, respectively. The application date used for all simulations was June 15-midway
through the window for egg laying.
Also, all simulations were based on a 20 g insectivore (i.e., 100% of diet for adults and
juveniles was invertebrates). Using an application rate of 1 Ib/acre results in an initial daily dose
for adults of 107 mg/kg/d. The simulated pesticide was defined to primarily affect egg
hatchability via in ovo exposure, and the toxicity threshold used in all simulations was 10.7
mg/kg/d (i.e., equivalent to a risk quotient of 10). To examine the effect of changes in residue
half-life on the model results, each of the 12 baseline profiles was run using the default 35-d
half-live as well as half-lives of 10 d and 3.5 d. All simulations were run using 10 replicates of
1000 females for a total population size of 10,000 females.
58
-------
Technical Manual for Basic Version of MCnest - February 2013
Simulation results indicate that the residue half-life estimate had little impact on seasonal
productivity for birds with short egg-laying durations (i.e., 30 d), but at longer durations the
percent reduction in productivity is greater using a 35 d half-life than one of 3.5 d (Figure 4).
For birds with the longest egg-laying duration (i.e., 120 d), the percent reduction is greater using
a 35 d half-life than for both 3.5 d and 10 d. If a model user wanted to compare the effects of the
pesticide across a series of species defined by the characteristics of these 12 profiles, using a 35-
d residue half-life the conclusion would be that species with short egg-laying durations (i.e., 30
d) are at less risk than species with longer egg-laying durations (Figure 4). However, if the
actual residue half-life of the pesticide was 3.5 d, the conclusion would be that species with short
egg-laying durations are at greatest risk.
60%
o%
Figure 4. The effect of a single pesticide application on June 15, using three estimates of its
residue half-life (i.e., 3.5, 10, and 35 d), on the percent reduction in seasonal productivity for 12
species profiles based on three daily nest mortality rates (i.e., 0.015, 0.03, and 0.045 d"1) and four
egg-laying durations (i.e., 30, 60, 90, and 120 d) with a midpoint date of June 15.
If the series of simulations in Figure 4 is repeated with only a single change in the input
parameters-the application date is changed to May 15 instead of June 15-the results change
significantly, primarily for profiles with shorter egg-laying durations. Species profiles with the
59
-------
Technical Manual for Basic Version of MCnest - February 2013
shortest egg-laying periods (i.e., 30 d) experience virtually no impact if the residue half-life is set
at 3.5 d, but a 100% reduction in productivity when the half-life is set at 35 d. With a May 15
application date, the pesticide residues would have decreased below the toxicity threshold by
May 31 (i.e., the first day of egg laying for the profiles with 30-d egg-laying durations) if the
residue half-life was 3.5 d, but would have remained about the toxicity threshold throughout the
egg-laying period if the half-life was set at 35 d.
Figure 5. The effect of a single pesticide application on May 15, using three estimates of its
residue half-life (i.e., 3.5, 10, and 35 d), on the percent reduction in seasonal productivity for 12
species profiles based on three daily nest mortality rates (i.e., 0.015, 0.03, and 0.045 d"1) and four
egg-laying durations (i.e., 30, 60, 90, and 120 d) with a midpoint date of June 15.
When information on a pesticide's residue half-life on avian foods is lacking or
inadequate, the model user may need to use a conservative assumption about the half-life value,
such as OPP's default of 35 d. However, these two series of simulations illustrate that for some
species and pesticide-use scenarios the selection of an estimate for the residue half-life has little
or no influence on the results for that species, but for other species or scenarios the selection of a
half-life can significantly affect not only the absolute magnitude of the estimated effect, but the
relative response among species. Model users should be aware of this potential outcome when
interpreting model results.
The above series of simulations also illustrates the importance of the timing of the
pesticide application relative to the time of a species' breeding season. As mentioned above,
although each pesticide application in MCnest simulations occurs on a single date, pesticide
60
-------
Technical Manual for Basic Version of MCnest - February 2013
applications in a certain geographical region may occur over a range of possible dates depending
on factors such as the status of the crop and insect population levels. Since the model response
can vary among application dates, the model user is encouraged to examine pesticide effects on
seasonal productivity over a range of possible application dates. If the model user has
information on the distribution of application dates, the cumulative response could be estimated
as the weighted average of responses from a series of application dates.
To examine the reproductive response to changes in application dates, a series of model
simulations was conducted using four profiles based on four durations for the egg-laying period
(i.e., 30, 60, 90, and 120 d). In this series of simulations, the waiting period after fledging (Wf)
was held constant at 20 d and the daily nest mortality rate was held constant at 0.03 d" . All other
life-history input parameters were held constant for the 4 baseline simulations, including: 1)
length of the rapid follicle growth period (5 d), 2) mean clutch size (5 eggs), 3) duration of
incubation period (10 d), 4) duration of the nestling rearing period (10 d), and 5) duration of the
waiting period after ecological failure (10 d). Also, all simulations used an egg-laying interval of
1 day and set incubation to start with the laying of the last egg.
The dates for T] and 7/arf remain the same as in the previous simulations, i.e., May 31-
June 30, May 16-July 15, May 1-July 30, and April 16-August 14 for the 30, 60, 60, and 120d
durations, respectively. The application dates occurred on March 1 or on dates at 15-d intervals
thereafter through the growing season.
Like above, all simulations were based on a 20 g insectivore. Using an application rate of
1 Ib/acre results in an initial daily dose for adults of 107 mg/kg/d. The simulated pesticide was
defined to primarily affect egg hatchability via in ovo exposure, and the toxicity threshold used
in all simulations was 10.7 mg/kg/d. In this series of simulation the pesticide residue half-life
was set at 10 d. All simulations were run using 10 replicates of 1000 females for a total
population size of 10,000 females.
Simulation results indicate that depending on the application date, the effect on seasonal
productivity can range from no impact (i.e., 0% reduction) to complete failure (i.e., 100%
reduction). The greatest impacts are observed for species with short breeding seasons (i.e., 30 d)
where the pesticide is applied just prior to the onset of egg laying, but applications before and
after egg laying had little impact. For long breeding seasons the magnitude of effect was lower
(i.e., < 33%), but effects in the range of 20 to 30% were observed following applications ranging
over a four-month period. Consequently, determining the species at greatest risk is dependent on
the application date or range of dates used in the simulations.
61
-------
Technical Manual for Basic Version of MCnest - February 2013
100%
3
o
o
Q.
_£
£
O
B
3
o
0)
£
8
QJ
Q.
Figure 6. The effect of a single pesticide application on each of a series of application dates on
the percent reduction in seasonal productivity for 4 species profiles based on 4 egg-laying
durations (i.e., 30, 60, 90, and 120 d) with a midpoint date of June 15.
K. Example of how MCnest could be used in a tiered risk assessment process
Throughout the Technical Manual we have identified uncertainties that exist in the
toxicity data and knowledge of species life history and have discussed how uncertainties in
ecological risk assessments are often addressed by making conservative assumptions about
parameter values. This is consistent with a tiered risk assessment process that starts with a
simple screening level based on conservative, worst-case assumptions about exposure and
toxicity values. If the screening level assessment indicates that unacceptable adverse effects are
possible, higher tier assessments are performed that refine these assumptions with more realistic
values when acceptable data are available while retaining conservative values where
uncertainties remain that cannot be resolved due to lack of information. The use of conservative
62
-------
Technical Manual for Basic Version of MCnest - February 2013
assumptions at any tier in the risk assessment process is intended to avoid wrongly concluding
that there is no unacceptable risk when, in fact, there is.
MCnest can be used in a tiered risk assessment to explore the change in model outcomes
as input parameters are refined from conservative default values to values more reflective of field
conditions. While it is possible to use MCnest as a screening tool, it is more informative when
used in higher tier assessments with the best data available for the input parameters, especially
when the risk management questions involve comparing the magnitude of effects among species
or among application dates. This section presents an example of how MCnest might be used in a
tiered process using a hypothetical pesticide that primarily affects the hatchability of eggs and its
effects on American robins.
Based on the default species profile, robins begin laying eggs on April 12 and the first
egg in the last nest of the season is laid on July 22. Robins weigh approximately 77 g and during
the breeding season consume approximately 72% invertebrates and 28% fruit. Without
pesticides, MCnest estimates that robins make an average of 4.1 nest attempts per female and
produce 2.02 successful broods per female per year for an overall nest success rate of 49%. The
mean number of fledglings per successful nest is 2.8, so the annual reproductive success for
robins is 5.7 fledglings per female per year without pesticide exposure.
In this example, we assume the pesticide is applied at 1 pound active ingredient per acre
and that the typical date of application is May 15. Based on the avian reproduction test, we
assume that the most sensitive endpoint is the proportion of hatchlings per viable eggs and that
the NOAEL, when transformed from a dietary concentration to a daily dose, is 10 mg/kg body
wt/day. Initially, the model uses the default 35-d half-life for residue degradation and uses the
maximum nomogram values for calculating initial daily dose. Under this scenario, an
application on May 15 would not affect the first nest attempt of robins, but would cause all
subsequent nest attempts during the breeding season to fail (Figure 7), resulting in 0.71
successful broods per female (i.e., a 65% reduction in productivity).
63
-------
Technical Manual for Basic Version of MCnest - February 2013
A
| PF _ rig I rfg/ef L ef I I I I N | | Id | Wp | | We | Wf _| Q
Apr-02 Apr-16 Apr-30 May-14 May-28 Jun-11 Jun-25 Jul-09 Jul-23 Aug-06 Aug-20
Wp | | We | Wf | Q
Apr-02 Apr-16 Apr-30 May-14 May-28 Jun-11 Jun-25 Jul-09 Jul-23 Aug-06 Aug-20
Figure 7. Phase diagrams for American robins A) without pesticide exposure and B) with
exposure to a pesticide application on May 15 assuming maximum nomogram residue values for
diet and a 35-d half-life for residue degradation.
It is unlikely that all of the females in a population would be exposed to diets with the
maximum nomogram residue concentrations. If we rerun the simulation using a lognormal
distribution of initial residue concentrations based the mean and standard deviation for residues
on each food type in the diet, the robin population produces a mean of 1.29 successful broods per
female per year (i.e., 36% reduction in productivity compared to control). Again, the pesticide
application does not affect the first nest attempt (Figure 8), but in subsequent nest attempts only
those females randomly assigned to the higher concentrations from the lognormal distribution
experience lost nest attempts.
64
-------
Technical Manual for Basic Version of MCnest - February 2013
Apr-02 Apr-16
Apr-30 May-14 May-28 Jun-11
Date
Jun-25 Jul-09 Jul-23 Aug-06 Aug-20
Figure 8. Phase diagram for American robins with exposure to a pesticide application on May 15
assuming a lognormal distribution of initial residue values for diet and a 35-d half-life for residue
degradation.
If additional information was available that demonstrated that this pesticide's residue
half-life was 7 d, instead of the default 35 d, the rerun simulation indicates that the robin
population produces 1.77 successful broods per female per year (i.e., 12% reduction in
productivity from control). With the shorter residue half-life, the period of time where exposure
doses exceed the toxicity thresholds is greatly reduced, resulting in fewer nest failures near the
end of the breeding season (Figure 9).
Apr-16 Apr-30 May-14 May-28 Jun-11 Jun-25 Jul-09 Jul-23 Aug-06 Aug-20
Figure 9. Phase diagram for American robins with exposure to a pesticide application on May 15
assuming a lognormal distribution of initial residue values for diet and a 7-d half-life for residue
degradation.
65
-------
Technical Manual for Basic Version of MCnest - February 2013
Although May 15 was identified as the typical date for application of this pesticide, we
may have information that it could be applied any time in May. For simplicity if we assume that
there is an equal probability that the pesticide is applied on each day in May, we can use the
batch function in MCnest to rerun the simulation for the 31 days in May and average the results
to get an estimate of the overall effect for the application of this pesticide during a typical season.
If empirical data exist, any other distribution of application days could be assumed in deriving an
overall estimate. Taking the mean of simulations over the month of May indicates the robin
population produces 1.81 successful broods per female per year (i.e., a 10% reduction in
productivity compared to control), which is only slightly different than the response on May 15
because there was relatively little change in response over this one-month period (Figure 10).
However, the timing of pesticide applications, for example May 1 vs May 31, produced failures
of nest attempts at different portions of the breeding season (Figure 11).
14%
'5 12%
i
-g 10%
o
8%
Q.
C
C
o
'^
(J
c
(U
+J
01
u
i_
(U
a.
6%
4%
2%
0%
11 13 15 17 19 21 23 25 27 29
May
31
Figure 10. The percent reduction in number of successful broods per female per year for a single
application if it occurred on a specific day in May.
66
-------
Technical Manual for Basic Version of MCnest - February 2013
Mayl
Apr-02 Apr-16 Apr-30 May-14 May-28 Jun-11 Jun-25 Jul-09 Jul-23 Aug-06 Au
May31
Apr-02 Apr-16 Apr-30 May-14 May-28 Jun-11 Jun-25 Jul-09
Figure 11. Phase diagrams for American robins with exposure to a pesticide application on May
1 or May 31 assuming a lognormal distribution of initial residue values for diet and a 7-d half-
life for residue degradation.
We can run this same series of pesticide-use scenarios with dickcissels that typically
produce only one brood per year. Based on the default species profile, dickcissels begin laying
eggs on May 24 and the first egg in the last nest of the season is laid on July 21. Dickcissels
weigh approximately 25 g and during the breeding season consume approximately 70%
invertebrates and 30% seeds. Without pesticides, MCnest estimates that dickcissels make an
average of 2.8 nest attempts per female and produce 0.62 successful broods per female per year
for an overall nest success rate of 22%. The mean number of fledglings per successful nest is
2.9, so the annual reproductive success for dickcissels is 1.8 fledglings per female per year
without pesticide exposure.
Again, we assume the pesticide is applied at 1 pound active ingredient per acre and that
the typical date of application is May 15. Based on the avian reproduction test, we assume that
the most sensitive endpoint is the proportion of hatchlings per viable eggs and that the NOAEL,
when transformed from a dietary concentration to a daily dose, is 10 mg/kg body wt/day.
Initially, the model uses the default 35-d half-life for residue degradation and uses the maximum
nomogram values for calculating initial daily dose. Under this scenario, an application on May
15 would result in the failure of all nest attempts until the very end of the breeding season
67
-------
Technical Manual for Basic Version of MCnest - February 2013
(Figure 12), resulting in 0.02 successful broods per female (i.e., a 97% reduction in productivity
compared to control).
May-14 May-23 Jun-01 Jun-10 Jun-19 Jun-28 Jul-07 Jul-16 Jul-25 Aug-03 Aug-12
B
ef ^^ I ^^ N ^^ Id ^^ Wp ^^ We [ZM Wf ^^| Q
May-14 May-23 Jun-01 Jun-10 Jun-19 Jun-28 Jul-07
Date
Jul-16 Jul-25 Aug-03 Aug-12
Figure 12. Phase diagrams for dickcissels A) without pesticide exposure and B) with exposure to
a pesticide application on May 15 assuming maximum nomogram residue values for diet and a
35-d half-life for residue degradation.
If we rerun the simulation using a lognormal distribution of initial residue concentrations
based the mean and standard deviation for residues on each food type in the diet, the dickcissel
population produces a mean of 0.32 successful broods per female per year (i.e., a 48% reduction
in productivity compared to control). Under this scenario, the pesticide application does affect a
portion of the nest attempts throughout the breeding season (Figure 13), but a portion of the nest
attempts by females randomly assigned to the lower concentrations from the lognormal
distribution are successful.
68
-------
Technical Manual for Basic Version of MCnest - February 2013
PF
I rfg/ef I I ef I I I
] Id ^^ Wp ^^ We Wf |
May-14 May-23 Jun-01 Jun-10 Jun-19 Jun-28 Jul-07
Date
Jul-16 Jul-25 Aug-03 Aug-12
Figure 13. Phase diagram for dickcissels with exposure to a pesticide application on May 15
assuming a lognormal distribution of initial residue values for diet and a 35-d half-life for residue
degradation.
Again, if additional information is used to decrease the pesticide's residue half-life to 7 d,
the rerun simulation indicates that the dickcissel population produces 0.54 successful broods per
female per year (i.e., 13% reduction in productivity from control). With the shorter residue half-
life, the period of time where exposure doses exceed the toxicity thresholds is greatly reduced,
resulting in fewer nest failures near the end of the breeding season (Figure 14).
I rig I I rfg/ef [
May-14 May-23 Jun-01 Jun-10 Jun-19 Jun-28 Jul-07 Jul-16
Date
Jul-25 Aug-03 Aug-12
Figure 14. Phase diagram for dickcissels with exposure to a pesticide application on May 15
assuming a lognormal distribution of initial residue values for diet and a 7-d half-life for residue
degradation.
Again, if we assume that the pesticide could be applied anytime in May and that there is
an equal probability that the pesticide is applied on each day in May, the batch function can be
used to run a series of simulations resulting in a mean of 0.54 successful broods per female per
year (i.e., a 13% reduction in productivity compared to control). While this is the same overall
69
-------
Technical Manual for Basic Version of MCnest - February 2013
result as was observed on May 15, there was considerable change in the response over time
during the month (Figure 15), ranging from a 5% reduction or less at the beginning of the month
to 24% reduction on May 24, before dropping again by the end of the month.
> 25%
7 9 11 13 15 17 19 21 23 25 27 29 31
Figure 15. The percent reduction in number of successful broods per female per year for a single
application if it occurred on a specific day in May.
These examples illustrate some ways that MCnest could be used in a tiered risk
assessment process to refine the description of the risk posed by a specific pesticide-use scenario.
They also illustrate the degree to which refinement of input parameters can change the outcome
of the model and help identify where additional data may provide the greatest improvement in
model performance. The model user is free to use additional data from specific locations or from
other sources of information of pesticide toxicity or use characteristics to refine to refine the
model inputs further. Also, as mentioned above, research continues on alternative methods for
defining avian breeding seasons or improving the exposure profile that will offer users more
options for refining the risk description in future versions of MCnest.
L. References
Alisauskas RT, Ankney CD. 1992. The cost of egg laying and its relationship to nutrient reserves
in waterfowl. Pages 30-61 in: Batt BDJ, Afton AD, Anderson MG, Ankney CD, Johnson
DH, Kadlec JA, Krapu GL, (eds.), Ecology and Management of Breeding Waterfowl.
University of Minnesota Press, Minneapolis, MN.
Bart J, Robson DS. 1982. Estimating survivorship when the subjects are visited periodically.
Ecology 63:1078-1090.
Bennett JK, Dominguez SE, Griffis WL. 1990. Effects of dicofol on mallard eggshell quality.
Archives of Environmental Contamination and Toxicology 19:907-12.
70
-------
Technical Manual for Basic Version of MCnest - February 2013
Bennett RS, Bentley R, Shiroyama T, Bennett JK. 1990. Effects of the duration and timing of
dietary methyl parathion exposure on bobwhite reproduction. Environmental Toxicology and
Chemistry 9:1473-1480.
Bennett RS, Dewhurst 1C, Fairbrother A, Hart ADM, Hooper MJ, Leopold A, Mineau P,
Mortensen SR, Shore RF, Springer TA. 2005. A new interpretation of avian and
mammalian reproduction toxicity test data in ecological risk assessment. Ecotoxicology
14: 801-815.
Bennett RS, Etterson MA. 2006. Estimating pesticide effects on fecundity rates of wild birds
using current laboratory reproduction tests. Human and Ecological Risk Assessment 12:762-
781.
Bennett RS, Etterson MA. 2007. Incorporating results of avian toxicity tests into a model of
annual reproductive success. Integrated Environmental Assessment and Management
3:498-507.
Bennett RS, Ganio LM. 1991. Overview of Methods for Evaluating Effects of Pesticides on
Reproduction in Birds. EPA/600/3-91/048. USEPA, Office of Research and Development,
Corvallis, OR.
Brewer LW, Driver CJ, Kendall RJ, Zenier C, Lacher TE Jr. 1988. Effects of methyl parathion
in ducks and duck broods. Environmental Toxicology and Chemistry 7:375-379.
Busby DG, White LM, Pearce PA. 1990. Effects of aerial spraying of fenitrothion on breeding
white-throated sparrows. Journal of Applied Ecology 27:743-755.
Callaghan J, Mineau P. 2000. An Evaluation of Clinical Sign Data from Avian Acute Oral
Toxicity Studies. Appendix 11 to The EFSA Journal (2008) 734:1-181. Risk Assessment for
Birds and Mammals. Scientific Opinion of the Panel on Plant Protection Products and their
Residues (PPR) on the Science behind the Guidance Document on Risk Assessment for Birds
and Mammals, (http://www.efsa.europa.eu/en/scdocs/scdoc/734.htm Last accessed 11
January 2013).
Dinsmore SJ, White GC, Knopf FL. 2002. Advanced techniques for modeling avian nest
survival. Ecology 83:3476-3488
Etterson MA, Bennett RS. 2005. Including transition probabilities in nest survival estimation: A
Mayfield Markov Chain. Ecology 86:1414-1421.
Etterson MA, Bennett RS, Kershner EL, Walk JW. 2009. Markov chain estimation of avian
seasonal fecundity. Ecological Applications 19(3):622-630.
Etterson MA, Ellis-Felege SN, Evers D, Gauthier G, Grzybowski JA, Mattsson BJ, Nagy LR,
Olsen BJ, Pease CM, van der Burg MP, Potvien A. 2011. Modeling fecundity in birds:
conceptual overview, current models, and considerations for future developments.
Ecological Modelling 222:2178- 2190.
71
-------
Technical Manual for Basic Version of MCnest - February 2013
[EFSA] European Food Safety Authority. 2006. Opinion of the Scientific Panel on Plant health,
Plant protection products and their Residues (PPR) on a request from the EFSA related to the
choice of endpoints to assess the long-term risk to mammals. 06 April 2006. The EFSA
Journal (2006) 344, 1-22. (http://www.efsa.europa.eu/en/efsajournal/pub/344.htm Last
accessed 11 January 2013).
[EFSA] European Food Safety Authority. 2008. Background information for reproductive risk
assessment. Appendix 16 to The EFSA Journal (2008) 734: 1-181. Risk Assessment for
Birds and Mammals. Scientific Opinion of the Panel on Plant Protection Products and their
Residues (PPR) on the Science behind the Guidance Document on Risk Assessment for Birds
and Mammals, (http://www.efsa.europa.eu/en/scdocs/scdoc/734.htm Last accessed 11
January 2013).
[EFSA] European Food Safety Authority. 2009. Nestling scenarios for long-term assessments.
Appendix R to The EFSA Journal (2009) 7(12): 1438. Guidance Document on Risk
Assessment for Birds and Mammals, (http://www.efsa.europa.eu/en/scdocs/scdoc/1438.htm
Last accessed 11 January 2013).
Fletcher JS, Nellessen JE, Pfleeger TG. 1994. Literature review and evaluation of the EPA food-
chain (Kenaga) nomogram, an instrument for estimating pesticide residues on plants.
Environmental Toxicology and Chemistry 13:1383-1391.
Hensler GL, Nichols JD. 1981. The Mayfield method of estimating nesting success: a model,
estimators and simulation results. Wilson Bulletin 93:42-53.
Hill EE. 1995. Organophosphorus and carbamate pesticides. Pages 243-273 in Handbook of
Ecotoxicology. Hoffman, DJ, Rattner BA, Burton GA Jr, Cairns J Jr. (eds). Lewis
Publishers. Boca Raton, FL.
Hoerger F, Kenaga EE. 1972. Pesticide residues on plants: Correlation of representative data as a
basis for estimation of their magnitude in the environment. Pages 9-28. in Coulston F, Korte
F (eds.), Environmental Quality and Safety: Chemistry, Toxicology and Technology.
Stuttgart: Georg Thieme Publishers.
Hudson RK, Tucker RK, Haegele MA. 1972. Effects of age on sensitivity: acute oral toxicity
of 14 pesticides to mallard ducks of several ages. Toxicology and Applied Pharmacology 22:
556-561.
Johnson DH. 1979. Estimating nest success: the Mayfield method and an alternative. The Auk
96:651-661.
Jones P, Jackson H. 1972. Estimation of the duration of spermatogenesis in Japanese quail,
Coturnix coturnixjaponica, using antispematogonial chemicals. Journal of Reproduction and
Fertility 31:3\9-322.
Jones P, Kominkova E, Jackson H. 1972. Effects of antifertility substances on male Japanese
quail. Journal of Reproduction and Fertility 29:71-78.
72
-------
Technical Manual for Basic Version of MCnest - February 2013
Kemeny JL, Snell JL. 1983. Finite Markov Chains. Springer-Verlag, New York, NY.
Kendeigh SC, Dol'nik VR, Gavrilov VM. 1977. Avian energetics. Pages 127-204 in Pinowski J,
Kendeigh SC (eds.), Omnivorous birds in ecosystems: Their evolution, energetics,
adaptations, impacts and control. Cambridge, UK: Cambridge University Press.
King JR. 1973. Energetics of reproduction in birds. Pages 78-107 in Farner, DS (ed.), Breeding
Biology of Birds. National Academy of Sciences, Washington, DC.
Luttik R, Aldenberg T. 1995. Extrapolation factors to be used in case of small samples of
toxicity data (with a special focus on LD50 values for birds and mammals). RIVM Report
679102029. National Institute of Public Health and Environment, Bilthoven, The
Netherlands.
Luttik R, Aldenberg T. 1997. Extrapolation factors for small samples of pesticide toxicity data:
Special focus on LD50 values for birds and mammals. Environmental Toxicology and
Chemistry 16:1785-88.
Luttik R, Mineau P, Roelofs W. 2005. A review of interspecies toxicity extrapolation in birds
and mammals and a proposal for long-term toxicity data. Ecotoxicology 14:817-832.
Mayfield HF. 1961. Nesting success calculated from exposure. Wilson Bulletin 73:255-261.
Mayfield HF. 1975. Suggestions for calculating nest success. Wilson Bulletin 87:456-466.
Matz AC, Bennett RS, Landis WG. 1998. Effects of azinphos-methyl on northern bobwhite: A
comparison of laboratory and field effects. Environmental Toxicology and Chemistry 17:
1364-1370.
Mayfield HF. 1961. Nesting success calculated from exposure. Wilson Bulletin 73:255-261.
Mayfield HF. 1975. Suggestions for calculating nest success. Wilson Bulletin 87:456-466.
Mineau P. 2005. A review and analysis of study endpoints relevant to the assessment of "long
term" pesticide toxicity in avian and mammalian wildlife. Ecotoxicology 14:775-799.
Mineau P, Baril A, Collins BT, Duffe J, Joerman G, Luttik R. 2001. Pesticide acute toxicity
reference values for birds. Reviews of Environmental Contamination and Toxicology
170:13-74.
Mineau P, Boersma DC, Collins B. 1994a. An analysis of avian reproduction studies submitted
for pesticide registration. Ecotoxicology and Environmental Safety 29:304-329.
Mineau P, Jobin B, Baril A. 1994b. A Critique of the Avian 5-day Dietary Test (LC50) as the
Basis of Avian Risk Assessment. Technical Report Series No. 215, Canadian Wildlife
Service, Environment Canada.
73
-------
Technical Manual for Basic Version of MCnest - February 2013
Nagy KA. 1987. Field metabolic rate and food requirement scaling in mammals and birds.
Ecological Monographs 57:111-128.
[OECD] Organization for Economic Co-operation and Development. 1996. Report of the
SETAC/OECD Workshop on Avian Toxicity Testing. OCDE/GD(96)166. Series on Testing
and Assessment No. 5. Organization for Economic Co-operation and Development, Paris.
Roelofs W, Crocker DR, Shore RF, Moore DRJ, Smith GC, Akcakaya HR, Bennett RS,
Chapman PF, Clook M, Crane M, Dewhurst 1C, Edwards PJ, Fairbrother A, Person S,
Fischer D, Hart ADM, Holmes M, Hooper MJ, Lavine M, Leopold A, Luttik R, Mineau P,
Mortenson SR, Noble DG, O'Connor RJ, Sibly RM, Spendiff M, Springer TA, Thompson
HM, Topping C. 2005. Case study Part 2: Probabilistic modeling of long-term effects of
pesticides on individual breeding success in birds and mammals. Ecotoxicology 14:895-923.
Schafer EW Jr, Brunton RB, LockyerNF. 1976. Evaluation of 45 chemicals as chemosterilants
in adult male quail (Coturnix coturnix). Journal of Reproduction and Fertility 48:371-375.
Schaffer TL. 2004. A unified approach to analyzing nest success. The Auk 121:526-540.
Shore RF, Crocker DR, Akcakaya HR, Bennett RS, Chapman PF, Clook M, Crane M, Dewhurst
1C, Edwards PJ, Fairbrother A, Person S, Fischer D, Hart ADM, Holmes M, Hooper MJ,
Lavine M, Leopold A, Luttik R, Mineau P, Moore DRJ, Mortenson SR, Noble DG,
O'Connor RJ, Roelofs W, Sibly RM, Smith GC, Spendiff M, Springer TA, Thompson HM,
Topping C. 2005. Case study 1. How to calculate appropriate deterministic long-term
toxicity to exposure ratios (TERs) for birds and mammals. Ecotoxicology 14:877-893.
Stromborg KL. 1981. Reproduction tests of diazinon on bobwhite quail. Pages 19-30 in Lamb
DW, Kenaga EE (eds), Avian and Mammalian Wildlife Toxicology: Second Conference.
American Society of Testing and Materials. ASTM STP 757, Philadelphia, PA.
Stromborg KL. 1986a. Reproduction of bobwhites fed different dietary concentrations of an
organophosphate insecticide, methamidophos. Archives of Environmental Contamination and
Toxicology 15:143-47.
Stromborg KL. 1986b. Reproductive toxicity of monocrotophos to bobwhite quail. Poultry
Science 65:51-57.
Topping CJ, Sibly RM, Ak9akaya HR, Smith GC, Crocker DR. 2005. Risk assessment of UK
skylark populations using life-history and individual-based landscape models. Ecotoxicology
14: 925-936.
[USEPA] U.S. Environmental Protection Agency. 2012. Avian Reproduction Test. Ecological
Effects Test Guidelines OCSPP 850.2300. Washington, DC. EPA/712/C-023.
(http://www.epa.gov/ocspp/pubs/frs/publications/Test Guidelines/series850.htm Last
accessed 11 January 2013).
[USEPA] U.S. Environmental Protection Agency. 2012. User's Guide T-REX Version 1.5
(Terrestrial Residue Exposure model). Office of Pesticide Programs. Washington, DC.
74
-------
Technical Manual for Basic Version of MCnest - February 2013
(http://www.epa.gov/oppefedl/tnodels/terrestrial/trex/t_rex_user_guide.htm#app_a Last
accessed 11 January 2013).
Vyas NB, Spann JW, Hulse CS, Borges SL, Bennett RS, Torrez M, Williams BI, Leffel R. 2006.
Field evaluation of an avian risk assessment model. Environmental Toxicology and
Chemistry 25:1762-1771.
Williams JB, Prints A. 1986. Energetics of growth in nestling savannah sparrows: A comparison
of doubly labeled water and laboratory estimates. The Condor 88:74-83.
Willis GH, McDowell LL. 1987. Pesticide persistence on foliage. Reviews of Environmental
Contamination and Toxicology 100:23-73.
Wolfe MF, Kendall RJ. 1998. Age-dependent toxicity of diazinon and terbufos in European
starlings (Sturnus vulgaris) and red-winged blackbirds (Agelaiusphoeniceus).
Environmental Toxicology and Chemistry 17:1300-1312.
75
-------
Technical Manual for Basic Version of MCnest - February 2013
Appendix A. Estimating the length of the rapid follicle growth (RFG) period
Estimating the length of the RFG period for each species is important because it is used
in defining the exposure period for comparing with the surrogate endpoints for hatchability and
nestling survival based on in ovo exposure. The time-weighted average for exposure during the
RFG period for each egg is used as the exposure measure in these decision points. Because the
Basic Version of MCnest focuses on exposures that occur during rapid follicle growth, it is best
suited for rapidly metabolized chemicals where deposition to egg yolks is primarily from
recently consumed foods. It may not be well suited for bioaccumulative chemicals where the
chemical deposition to the egg comes primarily from tissue residue stores accumulated over a
longer period of time.
As mentioned above, since birds lay one egg at a time, the ovarian follicles that develop
into egg yolks start growing on a staggered schedule about 1 d apart or longer. Each follicle
grows over a several day period, known as the RFG period. During the RFG period yolk
material in deposited to the growing follicle until it reaches the size of a fully formed yolk just
prior to ovulation. The rate of follicle growth is approximately sigmoidal, and the energy
requirement for each developing follicle over time approximates a bell-shaped curve (King
1973). After ovulation, the yolky follicle enters the oviduct for about 1 d where albumin is
deposited and the eggshell is formed. In many species there is an overlap in the RFG and
eggshell formation (EF) periods where eggs are being laid at the same time that follicles of
subsequent eggs are growing (Figure Al).
Species vary in the reported length of their RFG period (Table Al). Alisauskas and
Ankney (1992) show that the duration of the RFG period is related to the egg mass. The general
relationship for all birds between RFG (days) and egg mass (E), measured in grams, is:
RFG = 2.852*E °31 (r2=0.62).
Follicle growth rates for waterfowl are faster relative to egg mass, so the waterfowl-specific
relationship is:
RFG= 1.273*E°'43 (r2=0.71).
76
-------
Technical Manual for Basic Version of MCnest - February 2013
c
o
13
E
£
O)
O)
o
M-
o
0
D)
a
CO
r
Rapid follicle growth
(RFC)
Overlap of RFC & EF
A.
Eggs
Yolky
follicles
Eggshell
Formation (EF)
o
o o
*
A
Days
Figure Al. Example of the rapid follicle growth (RFG) and eggshell formation (EF) periods for
a clutch of five eggs (adapted from an illustration in Haywood 1993)^
77
-------
Technical Manual for Basic Version of MCnest - February 2013
Table Al. Estimated duration of rapid follicle growth (RFG) period for avian species.
Order
Anseriformes
Falconiformes
Galliformes
Gruiformes
Charadriiformes
Species
Mute swan (Cygnus olor)
Giant Canada goose (Branta canadensis maxima)
Cackling goose (Branta canadensis minima)
Domestic duck
Mallard (Anas platyrhychos)
Northern pintail (Anas acuta)
American wigeon (Anas americana)
Northern shoveler (Anas clypeata)
Wood duck (Aix sponsa)
Canvasback (Aythya valisimria)
Redhead (Aythya americana)
Lesser scaup (Aythya affmis)
Lesser scaup (Aythya affmis)
Greater scaup (Aythya marila)
Ring-necked duck (Aythya collaris)
Common goldeneye (Bucephala clangula)
Common eider (Somateria mollissima dresseri)
White-winged scoter (Melanitta fusca)
Ruddy duck (Oxyurajamaicensis)
Common merganser (Mergus merganser)
Eurasian kestrels (Falco tinnunculus)
Domestic chicken
Chukar (Alectoris graeca)
California quail (Lophortyx californicus)
Japanese quail (Coturnix coturnix)
American coots (Fulica americana)
Bar-tailed godwit (Limosa lapponica)
Ruddy turnstone (Arenaria interpres)
Western sandpiper (Calidris mauri)
Red phalarope (Phalaropusfulicarius)
Northern phalarope (Phalaropus lobatus)
Herring gull (Larus argentatus)
Herring gull (Larus argentatus)
Glaucous gull (Larus hyperboreus)
Great black-backed gull (Larus marinus)
Glaucous-winged gull (Larus glaucescens)
Western gull (Larus occidentalis)
Ring-billed gull (Larus delawarensis)
Mew gull (Larus canus)
Black-legged kittiwake (Rissa tridactyla)
Sabine's gull (Xema sabini)
Arctic tern (Sterna paradisaea)
Common murre (Una aalge)
Black guillemot (Cepphus grylle)
RFG
period
(days)
12
13
12
6-7
6
4.2
5.1
5
7
7
7
6
5
5.2
6
8
6
6
5-6
9
7
7-8
~6
6-7
5
7
8-12
5-6
5-8
4-5
6-7
9-10
11-13
12
13
12
10-11
12
5-8
9
7-8
6
12-18
8
References
Alisauskas & Ankney 1 992
Alisauskas & Ankney 1 992
Alisauskas & Ankney 1 992
King 1973
Alisauskas & Ankney 1 992
Eslerl994
Eslerl994
Alisauskas & Ankney 1 992
Drobney 1980
Barzen&Serie 1990
Alisauskas & Ankney 1 992
Alisauskas & Ankney 1 992
Esler 1994
Gorman et al. 2007
Alisauskas & Ankney 1 992
Alisauskas & Ankney 1 992
Alisauskas & Ankney 1 992
Alisauskas & Ankney 1 992
Alisauskas & Ankney 1994
Alisauskas & Ankney 1 992
Meijeretal. 1989
King 1973
King 1973
King 1973
Sonodaetal. 1996
Alisauskas & Ankney 1 985
Roudybush et al. 1979
Roudybush et al. 1979
Roudybush et al. 1979
Roudybush et al. 1979
Roudybush et al. 1979
King 1973
Roudybush et al. 1979
Roudybush et al. 1979
Roudybush et al. 1979
Roudybush et al. 1979
Roudybush et al. 1979
Roudybush et al. 1979
Roudybush et al. 1979
Roudybush et al. 1979
Roudybush et al. 1979
Roudybush et al. 1979
Roudybush et al. 1979
Roudybush et al. 1979
78
-------
Technical Manual for Basic Version of MCnest - February 2013
Columbiformes
Passeriformes
Pigeon guillemot (Cepphus columba)
Cassin's auklet (Ptychoramphus aleuticus)
Tufted puffin (Lunda cirrhata)
Domestic pigeon
Ring dove (Streptopelia sp.)
Jackdaw (Corvus monedula)
Starling (Sturnus vulgaris)
Great tit (Parus major)
Hermit warbler (Dendroica occidentalis)
Townsend's warbler (Dendroica townsendi)
House sparrow (Passer domesticus)
Eastern meadowlark (Sturnella magna)
Western meadowlark (Sturnella neglecta)
Tricolored blackbird (Agelaius tricolor)
Brown-headed cowbird (Molothrus ater)
White-crowned sparrow (Zonotrichia leucophrys)
Song sparrow (Melospiza melodia)
Zebra finch (Taeniopygia guttata)
10
8
12-13
5-8
5-7
~S
3
3-4
3
3
4
4
4
3-4
~3
~4
~4
4
Roudybush et al. 1979
Roudybush et al. 1979
Roudybush et al. 1979
King 1973
King 1973
King 1973
Ricklefs 1976
King 1973
Pearson & Rohwer 1998
Pearson & Rohwer 1998
Krementz & Ankney 1986
Pearson & Rohwer 1998
Pearson & Rohwer 1998
King 1973
Scott 1978
King 1973
King 1973
Haywood 1993
Alisauskas, R. T., and C. D. Ankney. 1992. The cost of egg laying and its relationship to nutrient reserves in
waterfowl. Pp. 30-61 in: Ecology and Management of Breeding Waterfowl. (B.D. J. Batt, A. D. Afton, M.
G. Anderson, C. D. Ankney, D. H. Johnson, J. A. Kadlec, and G. L. Krapu, eds.),. University of Minnesota
Press, Minneapolis. MN.
Alisauskas, R. T., and C. D. Ankney. 1994. Costs and rates of egg formation in ruddy ducks. Condor 96: 11-18.
Alisauskas, R. T., and C. D. Ankney. 1985. Nutrient reserves and the energetics of reproduction in American
coots. The Auk 102:133-44.
Barzen, J. A. , and J. R. Serie. 1990. Nutrient reserve dynamics of breeding canvasbacks. The Auk 107: 75-85.
Drobney, R. D. 1980. Reproduction bioenergetics of wood ducks. The Auk 97: 480-490.
Esler, D. 1994. Dynamics of ovarian follicles in breeding ducks. Wilson Bulletin 106: 679-88.
Haywood, S. 1993. Sensory control of clutch size in the zebra finch (Taeniopygia guttata). TheAukllQ'. 778-86.
Gorman, K. B., P. L. Flint, D. Esler, and T. D. Williams. 2007. Ovarian follicle dynamics of female greater
scaup during egg production. Journal of Field Ornithology 78:64-73.
King, J. R. 1973. Energetics of reproduction in birds. Breeding Biology of Birds. Pp. 78-107. Washington, DC:
National Academy of Sciences.
Krementz, D. G., and C. D. Ankney. 1986. Bioenergetics of egg production by female house sparrows. The Auk
103: 299-305.
Meijer, T., D. Masman, and S. Daan. 1989. Energetics of reproduction in female kestrels. The Auk 106: 549-59.
Pearson, S. F., and S. Rohwer. 1998. Determining clutch size and laying dates using ovarian follicles. Journal of
Field Ornithology 69: 587-94.
Ricklefs, R. E. 1976. The chemical composition of the ovary, oviduct, and follicles of the starling. The Auk 93:
184-87.
Roudybush, T. E., C. R. Grau, M. R. Petersen, D. G. Ainley, K. V. Hirsch, A. P. Oilman, and S. M. Patten. 1979.
Yolk formation in some charadriiform birds. The Condor 81: 293-98.
Scott, D. M. 1978. Using sizes of unovulated follicles to estimate the laying rate of the brown-headed cowbird.
Canadian Journal of Zoology 56: 2230-2234.
Sonoda, Y., T. Sato, and K. Imai. 1996. Rapid growth of the ovarian follicle in relation to age and laying
performance in Japanese quail. Japanese Poultry Science 33, no. 3: 170-177.
79
-------
Technical Manual for Basic Version of MCnest - February 2013
Appendix B. Bar graphs representing sensitivity analysis of life history parameters
For each of the 27 baseline scenarios, a bar graph is presented to help visualize how
increasing or decreasing each life history parameter by 20% affects the MCnest output (as
summarized in Table 6). The MCnest output is represented on the x-axis of each graph and is
expressed as the proportional change in the number of successful broods per female relative to
the baseline scenario (i.e., (modified-baseline)/baseline). The blue bar represents the
proportional change in broods due to a 20% increase in the specific parameter value, while the
red bar represents a 20% reduction in parameter value.
At the top of each figure is the identifier for each baseline scenario, such as T= 60, m\ =
mi = 0.015, Wf= 10, which indicates that for this particular baseline simulation the length of the
breeding season (7) was set at 60 d, the daily nest mortality rates during egg laying, incubation,
and nestling rearing (m\ and m^ was 0.015 d"1, and the waiting period after nest success (Wj) was
10 d. All other input parameters were held constant for the 27 baseline simulations, including:
1) length of the rapid follicle growth period (rfg; 5 d), 2) mean clutch size (clutch; 5 eggs), 3)
duration of incubation period (/; 10 d), 4) duration of the nestling rearing period (N; 10 d), and 5)
duration of the waiting period after ecological failure (We; 10 d). Also, all simulations used an
egg-laying interval (elf) of 1 d and set incubation to start with the laying of the last egg.
80
-------
Technical Manual for Basic Version of MCnest - February 2013
T= 60, ml=m2 = 0.015, Wf= 10
Wf
We
N
I
clutch
T
ml &m2
m2
ml
baseline
-0.25 -0.2 -0.15 -0.1 -0.05 0 0.05 0.1 0.15 0.2 0.25
1 1 1
1=1
-
1
-
1
ZH
i
1
T= 60, ml=m2 = 0.015, Wf= 20
Wf
We
N
1
clutch
T
ml &m2
m2
ml
baseline
i
c
i
I
i
-
i
i
-
-
i
^
i
i
i
i
I
1
-0.15
-0.1
-0.05
0.05
0.1
0.15
T= 60, ml=m2 = 0.015, Wf= 40
\A/f
We
N
1
clutch
T
ml &m2
m2
ml
baseline
I
|
*
I
i i
[
[
-0.45 -0.35 -0.25 -0.15 -0.05 0.05 0.15 0.25 0.35 0.4
81
-------
Technical Manual for Basic Version of MCnest - February 2013
T= 60, ml=m2 = 0.03, Wf= 10
Wf
We
N
1
clutch
T
ml &m2
m2
ml
baseline
l
II
I
i
C
*
1
-
-
i
-
-
-
-
i
i
i
i
i
-0.2 -0.15 -0.1 -0.05 0
0.05 0.1 0.15 0.2
T= 60, ml=m2 = 0.03, Wf= 20
Wf
We
N
I
clutch
T
ml &m2
m2
ml
baseline
r
c
I
I
[
I
^^
1 I 1
-
-
1
-
-
-
=l
I
I
I
=
1
-0.2 -0.15 -0.1 -0.05 0
0.05 0.1 0.15 0.2
T= 60, mi=m2 = 0.03, Wf= 40
Wf
We
N
I
clutch
T
ml &m2
m2
ml
baseline
l=
i
l^=
1=
i
-
i
i
zn
i
i
i
i
-0.3
-0.2
-0.1
0.1
0.2
0.3
82
-------
Technical Manual for Basic Version of MCnest - February 2013
T= 60, ml=m2 = 0.045, Wf= 10
Wf
We
N
I
clutch
T
ml &m2
m2
ml
baseline
-0.25 -0.2 -0.15 -0.1 -0.05
T= 60, ml=m2 = 0.045, Wf= 20
Wf
We
N
I
clutch
T
ml &m2
m2
ml
baseline
-0.2 -0.15 -0.1 -0.05 0
T= 60, ml=m2 = 0.045, Wf= 40
Wf
We
N
I
clutch
T
ml &m2
m2
ml
baseline
0.05 0.1 0.15 0.2 0.25
0.05 0.1 0.15 0.2
-0.2 -0.15 -0.1
-0.05
0.05 0.1
0.15 0.2
83
-------
Technical Manual for Basic Version of MCnest - February 2013
T= 90, ml=m2 = 0.015, Wf= 10
Wf
We
N
1
clutch
T
ml &m2
m2
ml
baseline
-0.
25 -0
.2 -0.
i
15 -0
.1 -0
1
-
i
i
-
-
i
i
=
i
i
i
1
1
05 0 0.05 0
1 0.15 0
HI
2 0.25
T= 90, ml=m2 = 0.015, Wf= 20
Wf
We
N
1
clutch
T
ml &m2
m2
ml
baseline
i
i
i
i
^
-
-
-
-
i
i
i
i
i
i
i
i
i
i
-0.2 -0.15 -0.1 -0.05 0 0.05 0.1 0.15 0.2
T= 90, ml=m2 = 0.015, Wf= 40
Wf
We
N
1
clutch
T
ml &m2
m2
ml
baseline
i
.
1
i
i
i
-
i
i
i
i
i
=-
1
-0.15
-0.1
-0.05
0.05
0.1
0.15
84
-------
Technical Manual for Basic Version of MCnest - February 2013
7=90, mi
Wf
We
N
1
clutch
T
ml &m2
m2
ml
baseline
-0
T=90,mi
Wf
We
N
I
clutch
T
ml &m2
m2
ml
baseline
-0
7=90, mi
Wf
We
N
I
clutch
T
ml &m2
m2
ml
baseline
-0
= m2 = 0.0
I
.2 -0.
= m2 = 0.0
i
.2 -0.
= w2 = 0.0
.2 -0.
3, 09= 10
i
15 -0
3, fFf= 20
C
15 -0
3, Wf= 40
15 -0
i
i
1
.1 -0.
I
i
i
r^
i
r~
.1 -0.
I
1
1
1
c
1
.1 -0.
=-
1
-
-
1
-
-
1
05 G
i
05 1
i
i
05 1
^m
i
i
o.c
i
i
) 0.
1
1
) 0.
1
1
15 0.
i
in
in
in
05 0
1
I
I
i
n
05 0
|
1 0.1
T
n
.1 0.
.1 0.
I
5 0.2
i
15 0.
15 0.
85
-------
Technical Manual for Basic Version of MCnest - February 2013
T= 90, ml=m2 = 0.045, Wf= 10
Wf
We
N
I
clutch
T
ml &m2
m2
ml
baseline
-0.25 -0.2 -0.15 -0.1 -0.05 0 0.05 0.1 0.15 0.2 0.25
T= 90, ml=m2 = 0.045, Wf= 20
Wf
We
N
I
clutch
T
ml &m2
m2
ml
baseline
-0.2 -0.15 -0.1 -0.05
0.05 0.1 0.15 0.2
T= 90, ml=m2 = 0.045, Wf= 40
Wf
We
N
I
clutch
T
ml &m2
m2
ml
baseline
-0.2 -0.15 -0.1 -0.05
0.05 0.1
0.15 0.2
86
-------
Technical Manual for Basic Version of MCnest - February 2013
T= 120, ml=m2 = 0.015, Wf= 10
Wf
We
N
1
clutch
T
ml &m2
m2
ml
baseline
i
l
c
i
C^
1
=
-
1
-
-
1
1
1
1
^^^^
1
1
-0.2 -0.15 -0.1 -0.05
0.05 0.1 0.15 0.2
T= 120, ml=m2 = 0.015, Wf= 20
Wf
We
N
I
clutch
T
ml &m2
m2
ml
baseline
-0.2 -0.15 -0.1 -0.05
T= 120, ml=m2 = 0.015, Wf= 40
Wf
We
N
I
clutch
T
ml &m2
m2
ml
baseline
0.05 0.1 0.15 0.2
-0.25 -0.2 -0.15 -0.1 -0.05 0 0.05 0.1 0.15 0.2 0.25
87
-------
Technical Manual for Basic Version of MCnest - February 2013
T= 120, m
Wf
We
N
1
clutch
T
ml &m2
m2
ml
baseline
-0
7=120, m
Wf
We
N
I
clutch
T
ml &m2
m2
ml
baseline
-0
7=120, m
Wf
We
N
I
clutch
T
ml & m2
m2
ml
baseline
-0
777 2 0.
I
.2 -0.
L = 7772 = 0.
i
.2 -0.
= mi = 0.
.2 -0.
03, Wf= 1C
i
15 -0
03, fFf= 2(
15 -0
03, Wf= 4(
|
15 -0
)
i
i
I
i
.1 -0.
)
i
i
i
.1 -0.
)
1
.1 -0.
n
-
-
-
I
-
-
-
"
05 (
i
i
-
-
i
-
-
i
"
05 (
i
i
-
-
i
-
-
i
i
05 (
^m
i
i
) O.I
i
i
) O.I
1
1
1
1
) O.I
1
1
]
]
35 0
i
i
HI
35 0
i
35 0
i
1 0.
1 0.
1 0.
I
15 0.
i
15 0.
i
15 0.
88
-------
Technical Manual for Basic Version of MCnest - February 2013
T= 120, ml=m2 = 0.045, Wf= 10
Wf
We
N
1
clutch
T
ml &m2
m2
ml
baseline
-0
7=120, m
Wf
We
N
I
clutch
T
ml &m2
m2
ml
baseline
-0
7=120, m
Wf
We
N
I
clutch
T
ml &m2
m2
ml
baseline
-0
i
i
I
i
^^
1
-
-
1
-
-
1
-
1
1
1
1
1
1
1
.2 -0.15 -0.1 -0.05 0 0.05 0.1 0.15 0.
L=y»2 = 0.045, Wf=20
c
1
1
-
-
1
-
1
1
-
-
"
1
1
1
1
1
1
1
.2 -0.15 -0.1 -0.05 0 0.05 0.1 0.15 0.
L=y»2 = 0.045, Wf=4Q
1
-
1
-
-
1
i
1
-
-
1
1
-
"
1
1
1
1
1
1
1
.2 -0.15 -0.1 -0.05 0 0.05 0.1 0.15 0.
89
------- |