ORNL/TM-13261
MODELING ECOLOGICAL RISKS OF PESTICIDE APPLICATION:
A REVIEW OF AVAILABLE APPROACHES
Lawrence W. Barnthouse
Environmental Sciences Division
Oak Ridge National Laboratory1
January 1996
Prepared for
Office of Pesticide Programs
U.S. Environmental Protection Agency
under Interagency Agreement No. 1824-D073-A1
with the U. S. Department of Energy
1 Oak Ridge National Laboratory, managed by Lockheed Martin
Energy Research Corp. for the U.S. Department of Energy under
contract number DE-AC05-96OR22464.

-------
ORNL/TM-13261
MODELING ECOLOGICAL RISKS OF PESTICIDE APPLICATION:
A REVIEW OF AVAILABLE APPROACHES
Lawrence W. Barnthouse
Environmental Sciences Division
Oak Ridge National Laboratory1
January 1996
Prepared for
Office of Pesticide Programs
U.S. Environmental Protection Agency
under Interagency Agreement No. 1824-D073-A1
with the U. S. Department of Energy
1 Oak Ridge National Laboratory, managed by Lockheed Martin
Energy Research Corp. for the U.S. Department of Energy under
contract number DE-AC05-96OR22464.

-------
I
EXECUTIVE SUMMARY
In 1992 an Ecological Fate and Effects Task Force within the EPA Office of Pesticide
Programs issued a report that substantially revised the agency's approach to pesticide risk
assessment. Among the most important recommendations from the Task Force was a
recommendation to eliminate the field test as a routine component of the pesticide risk
assessment process. As an alternative, the Task force recommended that initial decisions be
based solely on laboratory test data. Where initial test data showed a potential for long-term
effects, a refined assessment would be performed. Mitigating measures such reduced
application rates, mandatory tillage practices, or changes in formulation could be
implemented depending on the outcome of the refined risk assessment.
Modeling of ecological effects was specifically identified by the Task Force as a
research need, to "predict and/or characterize adverse effects in non-target organisms, their
populations and communities." This report provides a review of available modeling
approaches that could meet this need, with an emphasis on new approaches to population
modeling published in the scientific literature within the past five years. Four model types
were reviewed: age/stage-structured models, individual-based models, metapopulation
models, and spatially explicit models. Four criteria were used to evaluate the modeling
approaches:
	ability to characterize ecologically relevant effects (e.g., the abundance and/or persistence
of populations),
	ability to characterize the spatial and temporal distributions of exposures and effects.
	applicability to a variety of types of biota and exposure situations, and
	current degree of acceptance within the scientific community.
All four modeling approaches were found to be applicable within one or more phases of
the New Paradigm. Age/stage-structured models appear most likely to be useful for initial
registration decisions. For refined assessments (e.g., special reviews regional assessments)

-------
ii
the other three modeling approaches, because of their greater flexibility and ability to
simulate realistic pesticide application regimes, may be more appropriate.
The review found that age/stage-structured models and metapopulation models are
already extensively used in natural resource management and no further scientific
development is needed to support use by EPA. Individual-based models and spatially-
explicit models have appeared in the refereed scientific literature only within the past five
years and have had few management applications. Although in the long run these approaches
may well be the most useful, substantial research and development are still needed to support
regulatory use by the Agency.
Four steps were identified that EPA could take to increase the use of population models
in pesticide risk assessment:
	Broaden the management basis for decisions concerning pesticides to include measured
and effects on populations
	Develop reference population data sets for representative species and local environments
	Demonstrate new models through application in actual assessements
	Fund research in individual-based and spatially-explicit modeling, including the support
of graduate and postgraduate fellowships
	Train agency staff in the theory and application of population models

-------
iii
TABLE OF CONTENTS
1.0 INTRODUCTION 	1
1.1	Endpoints	2
1.2	Model Evaluation Criteria	3
1.3	Past Reviews 	4
2.0 DESCRIPTION OF MODELING APPROACHES	7
2.1	Age/Stage-structured models	7
2.2	Individual-Based Models	10
2.3	Metapopulation models	14
2.4	Spatially-Explicit Models 	21
3	0 INTEGRATION OF ECOLOGICAL MODELS INTO THE NEW PARADIGM 	25
3.1	Types of Model Applications	27
3.2	State of Development of Modeling Approaches	30
4	0 CONCLUSIONS AND RECOMMENDATIONS	32
5.0 REFERENCES 	35
APPENDIX
41

-------
LIST OF TABLES AND FIGURES
iv
TABLES
1.	Modeling Approaches Suitable for Three Types of Pesticide Risk Assessment
Applications 	29
2.	Comparative Evaluation of Modeling Approaches	30
FIGURE
1. Implementation Process for the New Paradigm Perform Preliminary
Risk Assessment 	26

-------
1
1.0 INTRODUCTION
In 1992 an Ecological Fate and Effects Task Force within the EPA Office of Pesticide
Programs issued a report that substantially revised the agency's approach to pesticide risk
assessment. Prior to 1986, the agency had operated under a tiered approach that included
mesocosms and field tests (Urban and Cook 1986). The first tiers involved simple acute
toxicity tests and exposure criteria using the "quotient approach." Estimates of standard
toxicity test endpoints such as LC50s and LD50s were compared directly to Estimated
Environmental Concentrations (EECs) derived from standardized surface-water runoff
models. If the quotients exceeded specified Levels of Concern (LOCs) then more extensive
testing (more species, full life-cycle tests, reproduction tests) would be required. If these
tests showed the pesticide to be safe, then no further testing was required. If there was still
uncertainty concerning the effects of the pesticide, then mesocosm or field tests could be
required prior to decision making.
In theory, mesocosm and field tests were supposed to provide information on long-term,
chronic, population and ecosystem-level effects of pesticide use under realistic application
conditions. In practice, data obtained from these tests proved to be difficult to use in
regulatory decision making. The systems were difficult to standardize and replicate, and
were expensive to use. The Task Force found that mesocosm/field data were adding little to
the information available for regulatory decision making, and were adding greatly to the time
and money required to make decisions. Consequently, the Task Force recommended
eliminating the field test as a routine component of the pesticide risk assessment process. As
an alternative, the Task force recommended that initial decisions be based solely on
laboratory test data. Where initial test data showed a potential for long-term effects,
mitigating measures such as reduced application rates, mandatory tillage practices, or changes
in formulation would be implemented.
The Task Force's recommendations were formalized in an "Implementation Paper for the
New Paradigm" issued in 1993. This document defined the paradigm to consist of the
following components (SETAC 1994, Appendix B):

-------
2
	Risk assessment is the scientific phase of the overall process and consists of hazard
identification, and exposure assessment, ultimately integrating hazard and exposure to
characterize risk.
	Risk mitigation involves mitigation measures to reduce or eliminate source
contamination and adverse environmental impact.
	Risk management is a policy-based activity that defines risk assessment questions and
endpoints to protect human health and ecological systems. It takes the scientific risk
assessment and incorporates social, economic, political, and legal factors, which impinge
or influence the final decision and selects regulatory actions.
The New Paradigm emphasizes early decision making and mitigation. For new
registrations, the paradigm employs essentially the same tier-1 LOCs that were used in the
old assessment scheme. If these levels are not exceeded, then registration may proceed. If
they are exceeded, then a "refined risk assessment" must be performed, and potential
mitigation measures must be identified. Costs and benefits of the mitigation measures must
be evaluated.
Specific procedures for performing refined risk assessments and evaluating benefits
of mitigation were not specified in the implementation paper and are still to be defined.
Modeling of ecological effects was specifically identified as a research need, to "predict
and/or characterize adverse effects in non-target organisms, their populations and
communities." This report provides a review of available modeling approaches that could
meet this need, with an emphasis on new approaches published in the scientific literature
within the past five years.
1.1 Endpoints
FIFRA requires decisions to incorporate analysis of risks and benefits of regulatory
decisions. If a pesticide is to be restricted, it must be shown that reductions in risks to
ecosystems will outweigh economic benefits of unrestricted use. Ecological endpoints in the

-------
3
past have emphasized (1) predicted effects on individual exposed organisms, and (2)
observed fish and bird kill incidents. However, neither predicted effects on individuals nor
enumerations of kill incidents can provide scientifically rigorous or defensible estimates of
the benefits of regulation. All organisms die. Except in the case of threatened or endangered
species, the abundance and persistence of populations are a more relevant endpoint.
Ecological risk assessments for pesticides would be more useful and scientifically credible if
it were possible to base decisions on risks to populations rather than risks to individuals, and
to consider both spatial scale and temporal scale in the assessment. If only a small fraction
of a population is exposed, or if the population recovers rapidly after exposure events, risks
associated with pesticide use may be small even if lethal exposures occasionally occur. Some
pesticides are acutely toxic (e.g., neurotoxins in birds) but not persistent in the environment.
Such substances are arguably preferable to persistent pesticides that have long-term chronic
effects on avian reproduction that might only be detected after substantial environmental
damage has occurred. If it were possible to estimate the spatial variations in pesticide
exposure and to evaluate the rate of recovery of exposed populations, this information could
be used to design pesticide application regimes that would minimize ecological risks.
1.2 Model Evaluation Criteria
None of the above considerations are captured in the tier-1 assessment criteria.
However, they could be included in "refined risk assessments," if appropriate models could
be found that extrapolate effects on individual organisms (mortality, reproduction etc.) to
effects on populations and ecosystems. Are there models available that can, in principle, be
used for this purpose? What additional research or demonstration is required before they can
be used within the new paradigm? The following criteria were used in this report to evaluate
the potential utility of the modeling approaches reviewed:
Endpoints: ability to characterize ecologically relevant effects, i.e., the abundance and/or
persistence of populations.

-------
4
Spatiotemporal resolution: ability to characterize the spatial distribution of exposures and
effects; ability to account for variations in temporal exposure over a period of days, weeks, or
months.
Generality: applicability to a variety of types of biota and exposure situations relevant to
pesticide risk assessment.
Current degree of acceptance: degree of acceptance within the scientific community, as
evidenced by the number of successful applications (e.g., in resource management or
conservation biology), and the number of refereed publications that employ the approach.
"Data requirements" and "data availability" were not included as evaluation criteria,
because within each of the major categories of modeling approaches one can find models
possessing a wide a range of data requirements and availabilities. It is almost always possible
to adjust a model to fit the available data.
1.3 Past Reviews
There have been several recent reviews of models potentially useful in ecological risk
assessment for chemicals and pesticides. Barnthouse et al. (1986) reviewed the general
history of successes and failures of population and ecosystem theory in resource management
and environmental impact assessment. Emlen (1989) reviewed general types of theoretical
population models with regard to applicability to terrestrial ecological risk assessments. Both
of these reviews concluded that age-structured population models, i.e., models of populations
that categorize organisms in terms of age and reproductive status, have been widely used in
fish and wildlife management, have been validated under many circumstances, and are ready
for use in ecological risk assessment of chemicals and pesticides. The widely-used
population modeling program RAMAS (Ferson and Akcakaya 1989; Ferson 1990) is an
age-structured population model. Barnthouse (1993) provided a discussion of the historical
development and basic principles of age-structured population models, with a review of the
literature and specific examples of applications to toxic chemicals. Barnthouse (1992)

-------
5
highlighted several new developments, specifically in landscape modeling and
individual-based modeling that appeared potentially useful for ecological risk assessment.
There is no need to repeat these reviews. Most of the literature reviewed was
published prior to 1990. The age/stage-structured models emphasized by Emlen (1989) and
Barnthouse (1993) are refinements and applications of a theory that is more than 50 years old.
These reviews should be consulted for detailed accounts of the historical development and
underlying theory of population models. The following kinds of models are included in this
review:
Age/stage-structured population models : These models subdivide populations into
discrete classes based on age, size, sex, or reproductive status. The best-known model of this
type is the Leslie Matrix (Leslie 1945). In this model information on the age-specific
reproductive rates and probabilities of survival can be used to project the future growth or
decline of the population and to show how the future status of a population should change in
response to changes in survival and reproduction. Variants on age-structured models have
been principal tools in natural resource management, especially for fish, since the 1950s. The
matrix representation of population dynamics is highly flexible and has been modified
variously to accommodate stochastic environmental variation, density-dependent survival and
reproduction, and other biological or physical processes. The model can be formulated in
terms of the size rather than the age of organisms. Caswell (1989) provided a thorough
discussion of the theoretical development of these models. Applications through 1990 were
reviewed by Barnthouse (1993). The discussion in this report will be limited to a few
innovations that have appeared in the literature since 1990.
Individual-based models: Individual-based models are models that characterize the
dynamics of populations in terms of the physiological, behavioral, or other relevant properties
of the individual organisms. The "core" of an individual-based population model is a model
of the organism, including its physiology, behavior, reproduction, spatial location, or any
other relevant property. For some simple models, the population-level consequences of

-------
individual properties can be generated analytically (e.g., Kooijman and Metz 1984, Hallam et
al. 1990). For more complex organisms or realistic environmental scenarios, these properties
are calculated by numerical simulation: a fixed number of individuals are simulated
day-by-day or week-by-week and quantities such as abundance, spatial distribution, or
probability of extinction are generated by tabulating the numbers and distributions of
organisms. Most of the published examples of individual-based organisms involve forest
composition (Huston and Smith 1987, Shugart 1984, Dale and Gardner 1987), Cladocera
(McCauley et al. 1990, Gumey et al. 1990, Hallam et al. 1990) or fish (Beyer and Laurence
1980), DeAngelis et al. 1991, Madenjian and Carpenter 1991, Rose and Cowan 1993). More
recently, models that simulate the behavior and distribution of animals moving over a
complex landscape have been developed (Loza et al. 1992, Liu 1993).
Metapopulation models: As discussed by Hanski and Gilpin (1991), a metapopulation can
be defined as a "set of populations that interact via individuals moving among populations."
Levins (1969), performed the first quantitative analysis of conditions under which a species
consisting of many populations could remain extant even though individual populations were
frequently fluctuating and going extinct. Many metapopulation models, including complex
ones involving interacting species (e.g., hosts and parasitoids, predators and prey, plants and
herbivores) have been developed for use in biological pest control studies (Murdoch et al.
1985). Many recent applications are in conservation biology, most notably in studies of the
Northern spotted owl (Lande 1987, Lamberson et al., 1994) and other endangered species
with fragmented spatial distributions (Lindenmayer and Lacy 1995). Hanski and Gilpin
(1991) provided a good recent review.
Spatially-explicit models: Spatially-explicit models are models that incorporate realistic
features of landscape structure. These representations can range from idealized arrangements
of "patches" of suitable and unsuitable habitat (Lamberson et al. 1994) to vegetation maps
generated by Geographic Information Systems (GIS) (Pulliam et al. 1992, Liu 1993, Turner et

-------
al. 1993). These models can be thought of as extensions of the metapopulation and
individual-based modeling concepts to complex spatial environments.
7
2.0	DESCRIPTION OF MODELING APPROACHES
This section provides detailed descriptions of a few recent models of each type, drawn
from the recent peer-reviewed literature. The objective of these descriptions is to provide a
foundation for evaluating the consistency of each approach with the evaluation criteria and
for determining the potential applicability of each within the New Paradigm.
2.1	Age/Stage-structured Models
Both Emlen (1989) and Bamthouse (1993) noted that age/stage-structured population models,
i.e., models in which all organisms belonging to a population are classified into groups
according to age, size, and reproductive status, have a long history of application in natural
resource management. These models are the most readily available quantitative methods for
assessing risks of toxic chemicals to populations. The simplest such model assumes that (1)
all organisms of the same age are identical, (2) all rates of birth and death are constant and
independent of environmental variation, and (3) the rate of population growth is independent
of population size:
where
lx = fraction of organisms surviving from birth to age x
mx = fecundity of individuals at age x
X = finite rate of natural increase

-------
8
Provided that lx and mx remain constant, any population growing according to equation 1 will
assume a stable age distribution in which the fraction of organisms in each age class * will
remain the same from each generation to the next. Once the stable age distribution is
achieved, the population will either grow or exponentially according to the following
equation:
<2>
where
No = population size at time 0
Nt = population size at time t
The above model is biologically unrealistic in many ways, however, it has been successfully
applied to many population management problems and is still the most widely-applied
approach to assessment of the impacts of human activities on the abundance and persistence
of fish and wildlife populations.
Leslie (1945) developed a matrix form of equation (2) that permits detailed analysis
of the influence of age-specific survival and reproduction rates on the rate of population
growth:
N(t)=LN{t-\)	^
where N(t) and N(t-l) are vectors containing the numbers of organisms in each age class
(N0,... Nk and L is the matrix defined by

-------
9
Vi
s/z
s:
 st-/k
0
*0
0
0 .
. 0
0
0

0 .
. 0
0
0
0
S2
. 0
0
0
0
0

0
where
sk = age-specific probability of surviving from one time interval to the next and
fk = average fecundity of an organism of age k
As discussed by Bamthouse (1993), a wide variety of population models can be derived from
equation 4 by making the survival and reproduction parameters random variables or functions
of environmental parameters or population size. The model can be defined in terms of sizes
or life stages rather than ages. Caswell (1989) presents a detailed discussion of the
mathematical properties of all of these models. Barnthouse (1993) and Emlen (1989)
described the range of resource management and risk assessment applications to which
age/stage-structured models have been applied. Some of these applications involve
pesticides and toxic chemicals (e.g., Tipton et al. 1980, Samuels and Ladino 1983,
Barnthouse et al. 1990). The literature search performed for this report revealed no
qualitatively new types of age/stage-structured models.
Some new research has, however, been published concerning methods for comparing
the influence of different life-history characteristics on the rate of population growth (X). The
term elasticities (deKroon et al. 1986) has been applied to a measure of the proportional
sensitivity of X to each element of the population transition matrix (ay):

(5)

-------
10
Elasticities as defined by equation 5 measure the relative contribution of each life-cycle
element to the population growth rate. They have been found useful in theoretical studies of
the relative fitness of different life-history strategies, especially for organisms such as plants
that have extremely complex life histories compared to most vertebrate animals.
Refinements of the basic methodology have been described by van Groenendael et al. (1994)
and van Tienderen (1995). Meyer and Boyce (1994) used elasticity to compare the influence
of changes in fecundity and survival due to hypothetical pesticide exposures on bird
populations with different age and size-structures.
The elasticity methodology itself does not provide any new approaches for modeling
impacts of pesticides on populations. It may, however, be useful in the design of model-
based assessment schemes. Models representative of a range of life-history types (small,
fast-growing, short life-span vs. large, slow-growing, long life-span) would be developed.
Elasticity analyses would be used to identify the life-cycle stages most strongly influencing
the long-term population growth rate. These are the life stages at which pesticide exposure
would be likely to have the greatest impacts.
2.2 Individual-Based Models
As the term implies, individual-based models provide the opportunity to evaluate the
influence of characteristics of individual organisms on the abundance of whole populations.
Age- and stage-based models already account for the influence of age and (for some
stage-based models) size on population dynamics, but individual-based models can expand
the list of characteristics considered to any aspect of organismal biology believed to be
relevant. For the purposes of pesticide risk assessment, the most relevant of these appear to
be physiology and behavior. The general procedure is to develop a model of the individual
organism to whatever level of detail is required, and then to infer the properties of the
population as a whole either by analytical solution of equations or by numerical simulation of
the activities of hundreds or thousands of individual organisms.
Physiological characteristics included in individual-based models have emphasized
metabolism, growth and contaminant pharmacodynamics. Work on metabolism was

-------
11
pioneered by Kooijman and Metz (1984), who examined the influence of contaminants on
metabolism and population growth using Daphnia as a model organism. Hallam and Lassiter
(Hallam et al. 1990, Lassiter and Hallam 1990) extended this approach to include (1) a
thermodynamically-based model of the uptake of contaminants from aqueous media and (2) a
definition of death in tenns of the internal dissolved contaminant concentration within an
organism. McCauley et al. (1990) and Gurney et al. (1990) developed an energetics-based
model of Daphnia growth and reproduction and used the model to predict time-dependent
changes in the age and size-structure of Daphnia populations in response to changes in food
availability.
All of the above models were developed for aquatic organisms with relatively simple
life-cycles. The emphasis in model analysis was on evaluation of general properties of the
models through analytical investigation of the equations. DeAngelis et al. 1991 and Rose and
Cowan (1993) developed models of fish populations that include metabolism, growth,
foraging behavior, and prey selection as functions of the life stage and age of the fish.
DeAngelis et al. (1991) even included the nesting and nest defense behavior of male
smallmouth bass. The approach followed in developing both of these models was to use the
existing extensive theoretical literature on bioenergetics, reproduction, and foraging of
individual fish, coupled with exhaustive evaluation of the life history of specific fish species,
to develop detailed models of each life-stage from egg through reproductive adult.
Population-level consequences of changes in the physiology, behavior, or reproduction of
individual fish are inferred by brute-force simulation of the birth, growth, and death of
hundreds or thousands of individual fish. The models are calibrated to extensive data sets
collected for specific fish populations.
The model of DeAngelis et al. (1991), for example, simulates the spawning, growth,
and survival of a year-class of smallmouth bass (Micropterus dolomeui) in Lake Opeongo,
Ontario. It is structured as a set of discrete submodels that simulate the daily activities and
physiological condition of each individual fish in the model population. Reproductive
behavior in smallmouth bass is quite complex. Adult males excavate nests, and after eggs are
deposited in the nests the males defend the nests from predators for several weeks while the

-------
12
eggs hatch and develop through their early larval stages. Spawning behavior and spawning
success have been shown to be both temperature and size-dependent. Larger males spawn
earlier; larvae that are spawned early have a size advantage over larvae that are spawned late.
Success in rearing a brood has also been shown to be related to the size and condition of the
guarding males. The males do not feed during the brood period; smaller males and males in
poorer condition at spawning abandon their nests much more frequently than do larger,
healthier males. Environmental conditions also have an important influence on spawning
success because storm events that cause water temperatures to fall below a critical threshold
cause the males that have spawned to abandon their nests.
The model of DeAngelis et al. (1991) simulates all of these processes. Given an
initial size/condition distribution of males and a specified daily temperature regime, the
model simulates the nesting and brood-rearing success of each male over the course of the
reproductive season. The probability of successfully rearing a brood and the number of
young fish produced from each brood are defined as probabilistic functions of the size and
condition of the male at the time of nesting.
Growth and survival of the young-of-the-year fish is simulated on a daily time-step.
The model incorporates well-established models of fish bioenergetics and foraging. Prey
abundances and size distributions are specified in the model; the prey selectivities and
feeding successes of each fish are specified through established predation models that
account for the swim speed and visual acuity of the fish (both dependent on size) and on the
frequency of encounter with prey of appropriate sizes. All of these processes are stochastic.
Rather than producing single numbers for the abundance and average size of fish, the model
produces, each day, a size and age (in days) distribution of fish. It is possible to calculate the
relative probability of survival of larvae spawned on any given day or having any given initial
size.
The objective of DeAngelis et al. (1991) was to understand and explain the processes
responsible for recruitment success in smallmouth bass: why more fish are produced in some
years than in others and why large males spawn earlier than small males in spite of the
increased risk of low-temperature events. Because of the physiological detail included in the

-------
13
model, however, it would be relatively easy to modify it to include lethal and sublethal effects
of toxic chemicals.
Individual-based models have also been applied to terrestrial biota. In many cases the
emphasis has been on behavior rather than metabolism and physiology. Pulliam et al. (1992)
developed a model of the Bachmann's sparrow population on the Savannah River Site, South
Carolina, derived from the individual foraging behavior and habitat selection of the birds
(this model is discussed in detail in section 2.4).
Lacy (1993) described a generalized computer program (VORTEX) that simulates the
local population dynamics of terrestrial vertebrate populations. VORTEX was intended for
use in the management of small populations threatened by habitat loss, environmental
variability, and loss of genetic variation. The core of VORTEX is stochastic model of the
birth, growth, reproduction, and death of each individual animal. At the start of simulation,
an initial number of animals, age/sex structure, genetic composition, and carrying capacity
are specified. At each subsequent time-step (normally one year, but modifiable by the user),
mature animals mate and produce young. The percentage of mature females producing young
in a given year, and the number of young produced by each reproducing female, are drawn
from probability distributions. Several different options are available to specify mate
selection and, thus, inbreeding effects. Following reproduction, the individual animals are
randomly "killed" according to numbers drawn from age-specific probability distributions.
Both reproduction and survival are assumed to be density-dependent, with expected values
changing depending on the relationship between the current population size and the carrying
capacity. Reproduction and survival are subject to two sources of environmental variation:
"normal" annual variation specified by a binomial probability distribution, and "catastrophic"
variation that occurs randomly according to a uniform probability distribution. When
catastrophes occur, survival and reproduction rates of all animals are reduced by a constant
fraction specified by the user.
Each simulation "run" produces a single stochastically-determined time-series of
population sizes and age/sex/genetic compositions. The simulated population either persists
throughout the simulation or goes extinct at some point during the simulation. Estimates of

-------
the probability of persistence and expected time-to-extinction of the simulated populations
are obtained by performing multiple runs (hundreds or thousands); the results of these
simulations can be summarized as (I) probability of extinction as a function of simulation
time, (2) median and mean time to extinction of populations that go extinct, and (3) mean
size and genetic variation within populations that survive.
VORTEX, in the version described by Lacy (1993), can simulate up to 20
populations, between which immigration and emigration can occur. In this form it can be
viewed as a metapopulation model (see section 2.3). VORTEX has been applied to a variety
of endangered bird and mammal species (Lacy et al. 1989, Seal and Foose 1989, Seal and
Lacy 1989, Lacy and Clark 1989, Maguire et al. 1990, Foose et al. 1992, Lindenmayer et al.
1993). VORTEX, unlike the model of DeAngelis et al. (1991) does not explicitly simulate
ecological or physiological processes relevant to pesticide exposure and effects assessment.
However, information concerning(l) the distribution of doses within an exposed population
and (2) dose-response relationships for reproduction and mortality could be used to modify
the survival and reproduction functions used in VORTEX.
2.3 Metapopulation Models
Most species do not exist as continuous interbreeding populations. They consist of
subpopulations inhabiting patches of suitable habitat mixed in with patches or regions of
unsuitable habitat. All are subject to environmental variability that may be either large or
small. Small populations frequently go extinct, but habitat patches are then recolonized by
colonists arriving from other patches. This view of species as "metapopulations" was first
formalized by Andrewartha and Birch (1954), although they did not use the term. The first
quantitative studies or metapopulation biology were published in the 1960s by MacArthur
and Wilson (1967) and den Boer (1968). Levins (1969) is credited with developing the first
formal model to specifically address the central question in metapopulation biology: how are
(1) the fraction of occupied patches and (2) the expected time to extinction of the species as a
whole affected by the probabilities of extinction and recolonization of individual patches?

-------
15
Levins (1969) formulated a simple relationship between the fraction of habitat patches
occupied by a species at any given time (p(t))> the rate of extinction of occupied patches (e),
and the rate of production of propagules from each occupied patch (m). He reasoned that at
any time t, mp propagules would be produced. Assuming equal probability of dispersal to
occupied and unoccupied patches, a fraction equal to (1-/?) of these would colonize
unoccupied patches. At the same time, a total number of patches equal to ep would become
extinct. The rate of change in p at any time would be determined by the equation:
dpldt-mp{ \ -p) -ep	(g)
It follows from this equation that the equilibrium frequency of occupied patches (p*^ is
determined by the ratio of the extinction (e) and colonization (m) rates:
It is intuitively obvious even without Levins' model that if extinction is more likely than
dispersal (i.e., e is larger than m) the species must become extinct. It is not, obvious,
however, that the if these two parameters are similar in magnitude the fraction of occupied
patches can be expected to be very small, even if the rate of dispersal of propagules from
occupied patches is very high. Levins also investigated the influence of temporal variation in
extinction and colonization rates on the size and probability of persistence of species
subdivided into local populations.
The above model is clearly too simplistic to be of much value in the management of
real populations. Many subsequent authors (see review by Hanski 1991) have replaced
Levins' simple assumptions with more biologically realistic representations of both the
dispersal of organisms between habitat patches and of the local dynamics of populations
within patches. However, the fundamental processes and variables of interest, i.e., dispersal,

-------
16
extinction, percent occupancy of available habitat, and metapopulation persistence, have not
changed.
Most early work involving metapopulation models was concerned with insect
populations, either with understanding the reasons for persistence of insect species subject to
wide fluctuations in local population abundance (den Boer 1968) or with designing control
strategies to reduce the frequency of widespread pest outbreaks (Levins 1969). In the 1980s
conservation biologists turned to metapopulation theory as a means of designing preservation
strategies for vertebrate species that, although once widespread, were becoming restricted to
isolated subpopulations because of increasing habitat fragmentation. The early models were
extended to include influences of local population size (Hanski 1985), local population
structure (Lande 1987) and spatial dispersal patterns (Ray and Gilpin. 1991). The theory has
also been extended to include predator-prey and host-parasitoid dynamics (Murdoch et al.
1985, Sabelisetal. 1991).
The relevance of this work to pesticide risk assessment comes from the observations
that many wildlife species of management interest are, effectively, metapopulations. Their
distribution patterns have been changed by decades of habitat conversion as the original
forests and prairies of North America have been transformed into a mosaic of agricultural,
urban/suburban, and successional landscapes. Pesticides of equal toxicity will have
differential impacts on wildlife species depending on patterns of habitat utilization, degree of
population isolation, dispersal ability, and other aspects of population biology included in
metapopulation models.
Lande (1987) formulated a model of extinction and persistence in territorial
populations that is a direct descendent of Levins' (1969) original model. Subsequently, other
authors have developed much more detailed models of the Northern Spotted Owl, intended
for use in predicting the influence of specific habitat management regimes on the recovery of
this endangered subspecies. Lamberson et al. (1992) described a metapopulation model of
the Northern Spotted Owl (Strix occidentalis caurina). In their model habitat patches are
defined as nesting territories and local populations are defined as nesting pairs. A nesting
pair annually produces young according to either a fixed fecundity rate or a randomly varying

-------
17
fecundity rate. The juvenile birds disperse at the end of each breeding season, with juvenile
males seeking an unoccupied nesting territory and juvenile females seeking a site occupied by
a solitary male. The probability that a dispersing juvenile finds a suitable site before it dies is
determined by the fraction the total landscape that consists of suitable sites, the fraction of
those sites that is already occupied by nesting pairs, and the number of sites a juvenile can
search before it dies. Adult birds are subjected to annual mortality, and nesting sites are
subjected to disturbance through timber harvesting, adults nesting on a harvested site must
disperse and locate new, unoccupied sites. The outputs from the model, which are updated
on an annual time-step, include the total number of suitable nesting sites, the number of sites
that are occupied by nesting pairs, and the number of sites occupied by single males.
Lamberson et al. (1992) used the model to evaluate the influence of initial population
size, the proportion of the landscape suitable for occupancy by spotted owls, and the degree
of interannual variability in fecundity (reflecting variability in food supply). Several
interesting effects were found. First, because the sexes were modeled separately and
colonization of a suitable site by both a male and a female s required to establish a nesting
pair, the entire metapopulation invariably declined to extinction if the number of nesting pairs
fell below a threshold determined by the searching abilities of the dispersing juveniles and
the proportion of suitable habitat. This well-known phenomenon is termed by population
biologists the "AUee effect" after the biologist who first described it. Second, Lamberson et
al. (1992) investigated the influence of habitat availability on the probability of survival of
the metapopulation under conditions of no, low, and high environmental variability. They
found that, for the case of no environmental variability, extinction always occurred if less
than a fixed percentage of the landscape (determined by dispersal ability) was suitable, and
that extinction never occurred if the percentage of suitable habitat was greater than the
threshold. Environmental variability had the effect of smoothing the transition from
inevitable extinction to indefinite persistence, so that there was a small probability of
persistence for habitat suitabilities slightly below the deterministic threshold and a small
probability of extinction for suitabilities somewhat higher than the threshold.

-------
18
By examining a range of parameter values consistent with the current state of
knowledge of spotted owl population biology, the authors found that the effective persistence
threshold for the metapopulation lies somewhere between 10% and 25% suitability of
available habitat. Lamberson et al. (1992) also simulated a potential future timber harvesting
regime in which the availability of nesting habitat was reduced gradually for 20 years and
then stabilized at a level of 20%. They found that, although the simulated owl
metapopulation eventually stabilized, there was significant time lag during which the total
population of adult owls and the per cent occupancy of suitable sites was relatively constant.
The implication of this result, according to the authors, is that it would be relatively difficult
to use population monitoring data collected during the harvest period to determine the
ultimate response of the owl metapopulation to habitat reduction.
Lamberson et al. (1994) used a different metapopulation model to evaluate the
influence of patch size and spacing on the viability of the Northern Spotted Owl. The
landscape was portrayed as a rectangular array of identical circular clusters containing
potential owl habitat. Each cluster consisted of a collection of territories, some or all of
which were assumed to be suitable as nesting sites. All of the space between clusters was
assumed to be unsuitable habitat. This idealized landscape was intended to approximate the
real landscape inhabited by spotted owls, which consists of patches of old-growth forest of
with differing abilities to support spotted owls separated by areas of cut forest. Within each
site, owl reproduction, survival, and dispersal were modeled in the same way as Lamberson
et al. (1992), except that only females were considered. Dispersal was assumed to be
successful if a juvenile female found an unoccupied but suitable territory within a specified
number of searches.
Dispersal within clusters was simulated using a "random walk": starting from the
territory of birth, a dispersing juvenile female was assumed to have an equal probability of
searching any adjacent territory. If that territory was suitable and unoccupied, it became
occupied by the dispersing bird. Otherwise, the bird moved, again in a random direction, to
another adjacent territory. After a fixed number of unsuccessful searches, the dispersing
juvenile was assumed to exit the cluster in a random direction. Two sources of mortality

-------
19
were imposed on juvenile females dispersing between clusters: first, a juvenile was assumed
to die if a straight line in the selected direction did not intersect another cluster. If a line in
the selected direction did intersect a cluster, then the probability of survival during transit
between the clusters was assumed to decline exponentially with the distance between
clusters. A female successfully arriving at a new cluster searched it in the same way as her
natal cluster and continued to search until she either died during transit, found an unoccupied
suitable territory, or searched a total of 22 sites. All birds that searched 22 sites without
success were assumed to die.
Landscape parameters investigated by the authors included the percentage of the total
landscape included within habitat clusters, the number of sites within each cluster, the
percentage of sites within each cluster suitable for nesting, the fraction of sites within a
cluster searched prior to exiting, and the rate of mortality during dispersal. The authors
evaluated the influence of different reserve design patterns on the mean occupancy of nesting
sites, defined as the fraction of suitable sites occupied by nesting females.
According to the authors, field studies suggest that approximately 60% of the forested
area within the range of the Northern Spotted Owl provides suitable nesting habitat.
Assuming that 60% of the sites within each cluster are suitable and that a maximum of 22
sites can be searched by dispersing juveniles, simulated spotted owl metapopulations did not
achieve a stable mean occupancy unless the average cluster contained at least 15 sites. Using
a mean suitability of 80%, a stable population could be achieved at a smaller mean cluster
size but mean occupancy declined rapidly at cluster sizes below 10 sites per cluster. For any
fixed total reserve size and percent suitability, the equilibrium mean occupancy was found to
increase with the size of the clusters. Lamberson et al. (1994) also investigated the influence
of (1) the fraction of the total landscape occupied by clusters and (2) the distance between
clusters on mean occupancy. They found that for small cluster sizes (i.e., 10 or fewer sites
per cluster) increasing the fraction of the landscape in clusters and decreasing the distance
between clusters both significantly increased mean occupancy; for cluster sizes greater than
25 there was little effect.

-------
20
The authors used their results to evaluate the adequacy of the reserve design proposed
for the Northern Spotted Owl. They concluded that, in general, the proposed sizes and spatial
distributions of proposed "Habitat Conservation Areas" is adequate, provided that the
recovery of currently degraded habitat within the HCAs is rapid.
Lindenmayer and Lacy (1995a, b) used the multipopulation version of VORTEX
(Section 2.2) to evaluate the metapopulation stability (expressed as probability of persistence
in the metapopulation as a whole and the inter-annual variability in abundance of local
populations) of Leadbeater's Possum (Gymnobelidius leadbeateri) in fragmented Australian
old-growth forests. Effects of patch size and number on stability were simulated by varying
the carrying capacities and number of local populations; no attempt was made to simulate the
influence of inter-patch distance or spatial distribution. The authors found, like Lamberson et
al., that increasing the size of patches enhanced the stability of the metapopulation as a
whole. When all patch sizes were small, metapopulation extinction rates were invariably
high and emigration actually decreased metapopulation stability.
Lande's (1987) and Lindenmayer and Lacy's (1995a, b) models are more obviously
relevant to pesticide risk assessment problems than is the more species-specific model of
Lamberson et al. Neither, however, may provide sufficient biological realism to support
pesticide regulation. In particular, neither provides for explicit consideration of local habitat
requirements and distributions within agricultural landscapes. By following the example of
Lamberson et al. (1992,1994) population biologists could develop models specifically
tailored to species and exposure regimes of interest in pesticide regulation. Such models
could provide useful information for risk assessment if (1) the species of interest is, because
of its intrinsic biological requirements or because of habitat fragmentation caused by habitat
change, restricted to relatively isolated subpopulations between which dispersal and
recolonization occur, (2) pesticide applications have the potential to increase the risk of
extinction of local populations, and (3) it is the persistence of the species as a whole, not the
persistence of individual local populations, that is the regulatory endpoint of interest.

-------
21
2.4 Spatially-Explicit Models
Spatially-explicit models may be thought of as extensions of individual-based or
metapopulation models in which the organisms or subpopulations are distributed over a
realistic rather than an idealized landscape (Dunning et al. 1995). "Suitable" and
"unsuitable" habitat types can defined explicitly in terms of vegetation, topography, or soil
type. Temporal changes in habitat suitability can readily be simulated. For management
applications, spatially-explicit models can utilize landscape maps derived from aerial surveys
and remote sensing. For obvious reasons, growth of research and application of these models
did not really begin until the late 1980s and the majority of the research performed using this
approach has been published within the last five years.
The approach appears especially suited to the study of mobile animal populations that
forage and disperse over large, heterogeneous areas. The spatially-explicit approach permits
ecologists to integrate theory and observation on foraging behavior and reproduction in
individual animals, relate these to specific measurable habitat characteristics, and infer
influences of habitat change on populations. As noted by Pulliam (1994), information on
environmental contaminant distributions and effects can easily be integrated into the same
framework. Because spatially-explicit models often deal with individuals, the full array of
individual physiological characteristics can also be incorporated. Such models can be
thought of simply as individual-based models in which the location and directional
movement of the organism are included as additional characteristics.
The most thoroughly explored and tested models of this type have been developed for
populations of ungulates foraging in Yellowstone National Park (Turner et al. 1993,1994)
and for the population of Bachmann's Sparrow nesting on the U.S. Department of Energy
Savannah River Site (Pulliam et al. 1992). Turner et al. (1993) simulated the influence of
landscape heterogeneity on winter grazing in "generic" ungulates. A standard energetics
model was used to simulate the daily foraging intake and energy balance of an animal as a
function of body weight, forage availability, and activity level. The authors then investigated
the influence of different ungulate movement "rules" and patterns of forage availability on the
energy balance and survival of model populations. Landscapes in which resource patches

-------
22
(sagebrush-grassland communities) were randomly distributed across the landscape were
compared with landscapes derived directly from vegetation maps for Yellowstone. During
each time step of the simulation, an animal feeds on resources within the patch it occupies. It
may move to another patch; the probability of movement increases as it feeds and depletes
the forage at its current location. While an animal is moving between patches it cannot feed.
Results obtained from the model generally supported previous theoretical predictions
that (1) when resources are abundant, landscape pattern and movement rules should have no
influence on weight maintenance and survival, (2) when resources are scarce, aggregated
resources (i.e., the real Yellowstone landscape) should support more animals than randomly
dispersed resources, and (3) when resources are scarce, behavioral rules that allow the
animals to discern resource abundance at distant sites or to move over greater distances
should improve survival.
Turner et al. (1994) extended their original model and used it to explore the effects of
fire on free-ranging elk (Cervus elaphus) and bison (Bison bison) populations in northern
Yellowstone Park. In the new analysis, the authors derived a six-category habitat map from
GIS data maintained by the National Park Service and assigned to each category a winter
forage abundance derived from actual field measurements (available separately for unburned
sites and for sites burned during the 1988 fires). The foraging rule used assumed that each
animal visually searches within a circle around its current location and moves to the site with
the highest quality; it may continue searching and moving until it either obtains its maximum
daily intake or reaches its maximum daily movement distance. Because snow conditions are
an important determinant of winter ungulate survival, snow was simulated in the model. A
snow subroutine assigned monthly snow depth values to each grid cell based on observed
data and on known influences of topography on snow depth. Foraging behavior and energetic
costs were both assumed to be affected by snow depth.
The authors were able to calibrate and test their model using observed data collected
both before and after the 1988 fire. For all three years, data were available on winter
precipitation, fall elk/bison count, and overwintering elk/bison survival. After model
parameters were calibrated so that overwintering survival during these three years matched

-------
23
the available data, simulation experiments were performed to evaluate the influence of winter
severity, fire size, and fire pattern on ungulate survival. Observed snowfall during the most
severe and most mild winters recorded in this century at Yellowstone were used to evaluate
the influence of winter severity. Three levels of fire severity, expressed as the percentage of
the study area burned, were examined. A range of alternative fire patterns was evaluated: a
fragmented bum was simulated by distributing burned grid cells at random over the whole
map; a clumped burn was simulated by generating a single patch of burned cells centered on
an arbitrary location. Several intermediate patch distributions were also evaluated, including
the actual observed burn distribution of the 1988 fire. In all, 24 different scenarios were
evaluated.
The authors found that winter snow was the most important determinant of ungulate
survival. Fire severity and pattern influenced survival only during average and severe
winters. Provided winters were mild or average, large fires actually produced better
long-term survival than small fires due to their stimulating effect on forage availability during
post-fire winters. For small to moderate fires, ungulate survival was greater with clumped
than fragmented fire patterns. The authors concluded that fires and spatial fire patterns have
an important influence on ungulate population dynamics in Yellowstone only if severe winter
conditions occur in the post-fire winter.
Pulliam et al. (1992) described a generalized spatially-explicit population model for
bird dispersal, applied to the Bachmann's sparrow (Aimophila aestivalis). The objective of
the model was to describe influences of spatial variation in habitat suitability on the
abundance and persistence of sparrow populations in a managed pine plantation. The model
as described in the original paper is closely similar to the spotted owl model of Lamberson et
al. (1994), discussed above. Only female birds are included, and the only life-history
characteristics simulated is dispersal. The principal differences between the models are in
descriptions attached to the grid cells. In the model of Pulliam et al., grid cells are identified
with pine stands of different ages. Bachman's sparrows nest only in young (5 years old or
less) or mature (>80 years old) pine stands. The simulated plantation consists of a number of
tracts of different ages. As the simulation proceeds, newly-seeded tracts become suitable

-------
24
nesting sites and previously suitable tracts age and become unsuitable. Trees are harvested
on a 21-year rotation, except for a certain number of tracts of mature pine forest that provided
a stable source of dispersing birds. The authors evaluated the influence of different model
parameters on the abundance and persistence of populations simulated for 100 years (five
rotations). They found that parameters relating to mortality and reproduction were more
important than those relating to dispersal (site selectivity, dispersal mortality). Population
size increased linearly with the number of tracts left in mature forest, but mature forest was
not required to maintain viable sparrow populations.
Liu (1993) extended the BACHMAP model in two significant ways. First, he
modified it to accept landscape classification information from a GIS. Second, he developed
an economics subroutine that calculates growth, yield, income, cost, and net-present-value
estimates for each tract. The extended model is coded in an object-oriented programming
language, so that it is modular and can easily be adapted to different species or landscape
types. Results of actual management applications are not yet published, as of the date of
preparation of this report.
Other recently-published spatially-explicit models simulate physiology as well as
behavior. Loza et al. (1992) described a model of cattle grazing on open rangeland that
simulates the influence of physiological status (energy and water balance) on the grazing
behavior and land use of grazing animals, Jager et al. (1993) described a spatially-explicit
version of the smallmouth bass model of DeAngelis et al. (1991) that simulates reproduction,
foraging, and growth in a riverine population of smallmouth bass.
The principal advantages of spatially-explicit models include flexibility and realism,
especially realism with respect to spatial representation of the environment. Virtually any
physical or biological process can be included in such models, provided a model of that
process can be developed. Both short-term and long-term events can be simulated.
Extremely detailed representations of the landscape, including direct interfacing with GIS
systems, is possible. The principal disadvantages are complexity of some of the models (a
disadvantage if there are no data) and, especially, the relative immaturity of the applications.
The vast majority of the published models are very recent. Few applications utilizing GIS

-------
25
technology have yet been published, although several are currently being developed. Given
the flexibility of the approach, specific applications for pesticides should be straightforward:
specify the spatial scale (local or regional), the species and landscape types of interest, and
the pesticide application scenarios. The object-oriented programming approach described by
Liu (1993) appears to provide an important advance in modeling technique because it permits
a generalized model structure to be specifically tailored to a variety of risk assessment
scenarios.
3.0 INTEGRATION OF ECOLOGICAL MODELS INTO THE NEW PARADIGM
Figure 1 presents a schematic version of the implementation process for the New
Paradigm (Avian Effects Dialogue Group 1994). According to this scheme, if the initial
screening of a pesticide shows potential exceedence of Levels of Concern, then a "refined
risk assessment" may be performed. The Avian Effects Dialogue Group identified fourteen
specific kinds of information that could contribute to these refined risk assessments. The
modeling approaches discussed in this report could contribute to these refined assessments by
(1) allowing effects to be expressed in terms of population abundance and persistence rather
than as fractions of an LC50 or other laboratory test endpoint, (2) integrating information on
lethal and sublethal effects (including behavioral effects), and (3) accounting for
spatiotemporal variations in exposure.
The Avian Effects Dialogue Group discussed simulation modeling as a potential
approach to risk characterization, but provided no specific recommendations. This review
demonstrates that progress in some types of ecological modeling has been significant in
recent years. All four types of models discussed in this report could be used to implement the
New Paradigm.
Viewed within the Framework for Ecological Risk Assessment (U.S. Environmental
Protection Agency. 1992), all of these models are risk characterization techniques: they
express effects of pesticides or other stressors in terms that are understandable to decision
makers and are compatible with cost-benefit analyses or other management-related
evaluations.

-------
Figure 1. Implementation Process for the New
Paradigm
Perform Preliminary Risk Assessment
Revisit Decision
in the Future
I
Decision
Negotiation
Identify LOC
Exceedence\
Refine Risk Assessment
and Consider Risk
Reduction Measures
Notify Registrant
J
Conduct Preliminary Benefits
Assessment
fO
o\

-------
3.1 Types of Model Applications
There are at least three possible uses of models within the New Paradigm:
27
Initial registration
Initial pesticide registration is a predictive activity involving the estimation of
potential changes in population size or reproductive success from standard toxicity test data.
Applications of age-structured models to predictive assessment problems are described by
Barnthouse et al. (1990), Emlen (1990), and Barnthouse (1993). Inter-species and inter-life
stage extrapolation methods described by Barnthouse et al. (1990) can be used to quantify
prediction uncertainty and express risks in terms of ranges of potential effects associated with
a given exposure or ranges of exposure associated with a given effects levels. As suggested
by Emlen and Pikitch (1989), a few generic population models can be made to represent
major life-history types of organisms, especially vertebrates (mouse, moose, sparrow, eagle).
Such models could be used to illustrate the consequences of contaminant exposure for
populations of organisms having different life-history characteristics.
Because of the inherent limitations of age/stage-structured models, the above
approach cannot accommodate spatial or temporal variations in exposure and thus cannot be
used to evaluate the influence of application regimes or site-specific environmental
variability on the effects of pesticides. Such information may, however, not be needed for
initial decision making. If predicted effects using conservative exposure scenarios are shown
to be inconsequential, then no additional analyses may be necessary.
Another way to introduce population-level phenomena into initial registration
decisions would be to expand the concept of of reference environments, as it is already used
in exposure assessment, to include reference populations. Descriptions of a reference field
could include, in addition to estimates of soil type, slope, rainfall, and other physical
parameters used in pesticide fate models, estimates of the seasonal distribution and foraging
characteristics of a reference ground-feeding bird. An individual-based model would use this
information to assess the impact of pesticide application on a reference bird population.
Besides permitting detailed evaluation of alternative application patterns, this approach

-------
28
would facilitate assessments of pesticides with high potential toxicity but low environmental
persistence.
Substantial quantities of field data are often available for pesticides undergoing
special reviews. For these cases, site-specific population models may aid in interpreting the
results of field studies. In the case of granular carbofuran (Houseknecht 1993), for example,
information on acute mortality of birds due to carbofuran exposure was available both from
incidental observations and from field experiments. The value of these data for risk
assessment was limited, however, because pesticide effects could only be quantified in terms
of numbers of dead birds. An individual-based model of a foraging bird population,
developed using the approach described by Pulliam (1994), could have been used to estimate
the distributions of exposures within a local population of birds, the fraction of those birds
likely to have received a lethal dose of carbofuran, and the long-term consequences of
continued carbofuran application for the abundance and persistence of a reference population.
Such model applications are especially feasible for birds because of the well-developed state
of avian foraging theory and the relative ease of validating species-specific avian foraging
models (Pulliam 1994).
Regional assessment
There is increasing interest within the EPA in performing regional, or "place-based"
assessments that integrate a variety of kinds of environmental hazards and can be used to
prioritize regulatory activities for specific watersheds or regions. For example, pesticide use
patterns and ecosystem types present in the Midwest corn belt differ substantially from those
present in the Gulf Coast region. If the ecological importance of these differences could be
incorporated in risk assessments, then region-specific restrictions or labeling requirements
could be designed. Issues that could be addressed might include the cumulative impacts of
runoff from a region within which large quantities of a few pesticides are used, or impacts on
wide-ranging species that forage over a mosaic of landscape types. For these applications,

-------
29
depending on scale, spatially-explicit models or metapopulation models would be
appropriate. Such models can accommodate land classification data that are already available
from many sources. Turner et al. (1995) And Liu et al. (In press) have already shown that
such applications are technically feasible.
Table 1 summarizes the types of models appropriate for each of the above
applications. For initial registration decisions, little or no site or region-specific information
is likely to be available. Relatively little toxicity data may be available as well. Thus, the
models with the fewest information requirements, i.e., age/stage-based models, are the most
likely to be useful. Provided that (1) reference population descriptions have been developed,
and (2) some information about physiology or time-dependent pesticide fate patterns is
available, then individual-based models could enable consideration of more complex
phenomena. The same types of models may be used for special reviews. However, for a
special review more information is likely to be available, and spatial patterns of use may have
been established. If local or regional extinctions are identified as an ecological issue of
concern, then metapopulation models may be useful. For regional assessments, it is likely
that spatiotemporal pesticide use patterns and explicit characteristics of the landscape will be
relevant. Models that cannot accommodate space, i.e., age/stage-based models and
individual-based models, will not be able to address the relevant questions. Metapopulation
models, and, especially, spatially-explicit models, would appear to be the best choices in
principle.
Table 1. Modeling approaches suitable for three types of pesticide risk assessment
applications
Models
Applications
age/stage
based
individual-
based
meta-
population
spatially
explicit
Initial Registration
X
X


Special Review
X
X
X

Regional assessment


X
X

-------
30
3.2 State-of-Development of Modeling Approaches
Table 2 compares the four modeling approaches discussed in this report with respect
to the evaluation criteria discussed in section 1.2. All four approaches are highly flexible in
form and can represent a wide range of populations of interest. Less data would in general be
required to implement age/stage-structured models than to implement the other model types,
however, as noted in section 2, all four approaches can encompass a range of complexity so
that complexity per se is not a useful evaluation criterion.
Table 2. Comparative evaluation of modeling approaches.
Models
Criteria
age/stage-
based
individual-
based
meta-
population
spatially
explicit
Endpoints
H
H
H
H
Resolution
L
H
M
H
Generality
H
M
H
L
Acceptance
H
M
M
L
The four approaches differ significantly with respect to the other three evaluation
criteria. Age/stage-structured models are, according to the definitions used in this report,
spatially homogeneous (spatially-structured variants of this model type would be classified as
metapopulation models). As normally applied, age/structured models are used to characterize
the long-term or steady-state behavior of populations and cannot directly address phenomena
(e.g., transient pulses of contaminant exposure) that are short with respect to the lifetime of a
single organism. Individual-based models and spatially explicit models, in contrast, have
arbitrarily high degrees of resolution. The activities of individual organisms can be simulated
on any time-scale; the size of cells in spatially-distributed models can be made arbitrarily

-------
31
large or small. In both cases the resolution of the available data and the needs of the
assessment are the limiting factors. Metapopulation models are intermediate in resolution:
space can be at least implicitly represented in terms of immigration/emigration/extinction
processes. Like age/stage-structured models, however, metapopulation models are generally
best suited to addressing effects of long-term exposures.
As noted by Levins (1966), a tradeoff can usually be expected between generality and
spatiotemporal resolution in models. Age/stage-structured models have been developed for
virtually every type of living organism. The versatility of the metapopulation approach, at
least when applied to vertebrates, is demonstrated by the number of species for which the
multipopulation version of the VORTEX model has been implemented. In contrast,
physiologically and behaviorally-oriented individual-based models such as those of
DeAngelis et al. (1991) and Pulliam et al. (1992) are highly specific. The underlying theories
of foraging, bioenergetics, and reproduction are quite general, however, the number of
species-specific parameters needed to implement an individual-based model can be quite
large. Spatially-explicit models require, in addition, site-specific data on landcover, weather,
and other environmental influences on the activities of the organisms being modeled.
With respect to degree of acceptance by the scientific community, age/stage-based
models are by far the best-developed type discussed in this report. They are the type most
people immediately think of when they hear the term 'population model". The basic Leslie
matrix and its variants are the backbone of quantitative fisheries assessment, with literally
hundreds of applications over 50 years. User-friendly modeling software is widely available.
The more general stage-based models have been less-widely used in management, although
they are common in plant demography and applied entomology.
Metapopulation models have a much shorter history, but have become very widely
used in conservation biology over the last decade. The more complex models, such as the
multipopulation version of VORTEX, provide a useful extension of age/stage-based models
for situations in which differential exposures to isolated subpopulations are important. This
will often be the case for rare or endangered species, which are restricted to specific habitat
types. Like age/stage-structured models, they can be relatively general and applicable to a

-------
32
variety of different life-history types, as has been shown by the ease with which the
VORTEX model has been adapted to a variety of mammalian species.
Despite its formulation in terms of individual organisms, VORTEX is in principle
equivalent to a classical age/stage-structured model. Caswell and John (1992) showed that
all matrix-type population models can be derived from models of the birth, reproduction, and
death processes of individual organisms; for large population sizes VORTEX and its matrix
equivalent would provide identical results. Models that incorporate physiological or
behavioral influences on individuals, or that involve explicit simulation of interactions
between organisms and their surrounding landscape, are fundamentally distinct from any
previous approach to population modeling. Such represent the future rather than the present
of population biology. The majority of published accounts of these kinds of models have
appeared in the peer-reviewed literature only within the past three years. No standardized
modeling software is available to support their development. Only a few experts, primarily
associated with the authors of the papers cited in this report, have had any significant
experience in developing and applying these models. The object-oriented software described
by Liu (1993) provides a general framework that would, if widely adopted, significantly
simplify the programming aspect of model development. However, developing a sound
biological content for the models will still be a major undertaking.
4.0 CONCLUSIONS AND RECOMMENDATIONS
1.	The state-of-the-science in population biology is sufficient to support the development and
use by EPA of models that express risks to aquatic and terrestrial biota from pesticide
exposure in terms of population-level rather than individual-level endpoints.
2.	Age/stage-structured models and metapopulation models can be used to qualitatively
describe the influences of specific levels of mortality and reproduction on organisms with
different life-history types and distributional patterns. These approaches to modeling are
extensively documented in the scientific literature and are widely used in resource
management. No further scientific development is needed to support use by EPA.

-------
33
3.	Individual-based models and spatially-explicit models can quantitatively describe effects
of lethal and sublethal exposures to pesticides for specific target species in specific
environments. These approaches are new and there have been few management
applications. Implementation of these models requires substantial information
concerning (1) the modes of action and environmental fate of the pesticides being
assessed, and (2) the behavior, spatial distribution, and population dynamics of the
species of interest.
4.	Steps the Agency can take to implement these models include:
Broaden the management basis for decision making concerning pesticides. As
documented by Troyer and Brody (1994), there are no consistent ecological assessment
endpoints within EPA. Visible kills of birds and evidence of toxicity in the laboratory, have
been used as a basis for decision making but predicted or observed effects on populations
have not been. Specific population-level assessment endpoints and characteristic assessment
scales, consistent with the types of decisions made at each stage of the pesticide assessment
cycle (Figure 1), should be established.
Develop reference population data sets - Data sets should be developed for representative
species and local environments that can be used in the same way as the "reference
environments" used in pesticide fate modeling. These can then be used to parameterize
age/stage based and individual-based models. The wildlife exposure handbook recently
developed by EPA wildlife can serve as a good model.
Demonstrate models through actual applications. Application to a high-profile
assessment problem is the best way to demonstrate the value of any assessment methodology.
The ideal candidate studies would be special reviews or other assessments expected to
involve sensitive ecological resources and high-value pesticides. Such assessments might be
expected to take several years to complete and to justify significant expenditures of funds for
laboratory and field data. The availability of time and data would, in turn, allow the
development of credible models.

-------
34
Fund research - Individual-based models and spatially-explicit models in particular lack a
solid scientific foundation and, perhaps even more important, lack a corps of experienced
practitioners. The best way to increase both the quality of the science and the number of
practitioners is to fund research. Support of graduate and postgraduate fellowships in
particular would be a highly cost-effective approach.
Train agency staff - Agency technical staff will not use a new methodology if they are
unfamiliar with it and are not confident that it will improve their work. Training classes, and
hands-on experience will be necessary. If the agency funds extramural scientists to develop
any of the model types described in this report, then the contracts should require the model
developers to work with EPA staff during the initial phases of implementation by the agency.

-------
35
5.0 REFERENCES
Andrewartha, H.G., and L. C. Birch. 1954. The Distribution and Abundance of Animals.
University of Chicago Press, Chicago, Illinois.
Avian Effects Dialogue Group. 1994. Assessing Pesticide Impacts on Birds. RESOLVE,
Inc., 1250 24th St. N.W. Washington, D.C. 20037.
Bamthouse, L.W., R. V. O'Neill, S. M. Bartell, and G. W. Suter II. 1986. Population and
ecosystem theory in ecological risk assessment, pp. 82-96 IN: T.M. Poston and R. Purdy
(eds.), Aquatic Toxicology and Environmental Fate: Ninth Volume, ASTM STP 921,
American Society for Testing and Materials, Philadelphia, Pennsylvania.
Bamthouse, L. W. 1992. The Role of Models in Ecological Risk Assessment: A 1990's
Perspective. Environmental Toxicology and Chemistry 11:1751-1760.
Bamthouse, L.W. 1993. Population-level effects. Pp. 247-274 IN: G. W. Suter II (ed.)
Ecological Risk Assessment. Lewis Publishers, Chelsea, Michigan.
Bamthouse, L.W., G. W. Suter, II, and A. E. Rosen. 1990. Risks of Toxic Contaminants to
Exploited Fish Population: Influence of Life History, Data Uncertainty, and Exploitation
Intensity. Environmental Toxicology and Chemistry 9:297-311.
Beyer, J. E., and G. C. Laurence. 1980. A stochastic model of larval fish growth. Ecological
Modelling 8:109-132
Caswell, H. 1989. Matrix Population Models: Construction, Analysis, and Interpretation.
Sinauer, Inc., Sunderland, Massachusetts. 328 pp.
Caswell, H., and A. M. John. 1992. From the individual to the population in demographic
models. Pp. 36-61 IN: D. L. DeAngelis and L. J. Gross (eds.) Individual'Based Models
and Approaches in Ecology. Chapman and Hall, New York.
Dale., V. I., and R. H. Gardner 1987. Assessing regional impacts of growth declines using a
forest succession model. Journal of Environmental Management 24:83-93.
DeAngelis, D. L. L. Godbout, and B. J. Shuter. 1991. An individual-based approach to
predicting density-dependent compensation in smallmouth bass populations. Ecological
Modelling 57: 91-115.
denBoer, P. J., 1968. Spreading of risk and stabilization of animal numbers. Acta
Biotheoretica 18:165-194.

-------
36
deKroon, H., A. Plaisier, J. Van Groenendael, and H. Caswell. 1986. Elasticity: the relative
contribution of demographic parameters to population growth rate. Ecology 67:1427-
1431.
Dunning, J. B., D. J. Stewart, B. J. Danielson, B. R. Noon, T. L. Root, R. H. Lamberson, and
E. E. Stevens. 1995. Spatially explicit population models: current forms and future uses.
Ecological Applications 5: 3-11.
Emlen, J. M. 1989. Terrestrial population models for ecological risk assessment: a state-of-
the-art review. Environmental Toxicology and Chemistry 8:831-842.
Emlen, J. M., and E. K. Pikitch. 1989. Animal Population Dynamics: Identification of
Critical Components. Ecological Modeling. 44:253-273.
Ferson, S. 1990. RAMAS/Stage. Generalized stage-based modeling for population dynamics.
Applied Biomathematics, Setauket, New York.
Ferson, S., and H. R. Akcakaya. 1989. RAMAS/Age user manual. Modeling fluctuations in
age-structured populations. Applied Biomathematics, Setauket, New York.
Foose, T. J., R. C. Lacy, R. Brett, and U. S. Seal. 1992. Kenya black rhinoceros population
and habitat viability assessment. International Union for the Conservation of Nature and
Natural Resources, Captive Breeding Specialist Group, Apple Valley, Minnesota.
Gurney, W. S. C., E. McCauley, R. M. Nisbet, and W. W. Murdoch. 1990. The physiological
ecology of Daphnia: a dynamic model of growth and reproduction. Ecology 71:703-715
Hallam, T. G.,R. R. Lassiter, J. Li, and W. McKinney. 1990. Toxicant-induced mortality in
models of Daphnia populations. Environmental Toxicology and Chemistry 9: 597-621.
Hanski, I. 1985. Single-species spatial dynamics may contribute to long-term rarity and
commonness. Ecology 66:335-343.
Hanski, I. 1991. Single-species metapopulation dynamics: concepts, models and
observations. Biological Journal of the Linnaean Society 42:17-38.
Hanski, I. and M. Gilpin. 1991. Metapopulation dynamics: a brief history and conceptual
domain. Biological Journal of the Linnaean Society 42:3-16,

-------
37
Houseknecht, C. R. 1993. Ecological risk assessment case study: Special review of the
granular formulations of carbofuran based on adverse effects on birds. Section 3 IN U.S.
Environmental Protection Agency, A review of ecological assessment case studies from a
risk assessment perspective. EPA/630/R-92/005, U.S. Environmental Protection Agency,
Washington, D.C.
Huston, M. A., and T. M. Smith. 1987. Plant succession: life history and competition.
American Naturalist 130:168-198.
Jager, H. I., D. L. DeAngelis, M. J. Sale, W. Van Winkle, D. D. Schmoyer, M. J. Sabo, D. J.
Orth, and J. A. Lukas. 1993. An individual-based model for smallmouth bass
reproduction and young-of-year dynamics in streams. Rivers 4:91-113.
Kooijman, S. A. L. M., and J. A. J. Metz. 1984. On thy dynamics of chemically stressed
populations: the deduction of population consequences from effects on individuals.
Ecotoxicology and Environmental Safety 8:254-274.
Lacy, R. C., and T. W. Clark. 1990. Population viability assessment of the eastern barred
bandicoot in Victoria. Pp. 131-146 IN: T. W. Clark and J. H. Seebeck, (eds.) The
management and conservation of small populations. Chicago Zoological Society,
Brookfield, Illinois.
Lacy, R. C. 1993. VORTEX: a computer simulation model for use in population viability
analysis. Wildlife Research 20:45-65.
Lacy, R. C., N. R. Flesness, and U. S. Seal. 1989. Puerto Rican parrot population viability
analysis. International Union for the Conservation of Nature and Natural Resources,
Species Survival Commission, Captive Breeding Specialist Group, Apple Valley,
Minnesota.
Lamberson, R. H., R. McKelvey, B. R. Noon, and C. Voss. 1992. The effects of varying
dispersal capabilities on the population dynamics of the northern spotted owl.
Conservation Biology. 6:505-512.
Lamberson, R. H., R. McKelvey, B. R. Noon, and C. Voss, ,1994. Reserve design for
territorial species: the effects of patch size and spacing on the viability of the northern
spotted owl. Conservation Biology. 8:185-195.
Lande, R. 1987. Extinction thresholds in demographic models of terrestrial populations.
American Naturalist 130:624-635.

-------
38
Lassiter, R. R. and T.G. Hallam. 1990. Survival of the fattest: implications for acute effects
of lipophilic chemicals on aquatic populations. Environmental Toxicology and Chemistry.
9:585-596.
Leslie, P. H. 1945. On the use of matrices in certain population mathematics. Biometrika
33:183-212.
Levins, R. 1966. The strategy of model building in population biology. American Scientist
54:421-431.
Levins, R. 1969. Some demographic and genetic consequences of environmental
heterogeneity for biological control. Bulletin of the Entomological Society of America
15:237-240.
Lindenmayer, D. B., T. W. Clark, R. C. Lacy, and V. C., Thomas. 1993. Population Viability
Analysis as a tool in wildlife management: a review with reference to Australia.
Environmental Management 17:745-758.
Lindenmayer, D. B., and R. C. Lacy. 1995a. Metapopulation viability of Leadbeater's
possum, Gymnobelideus leadbeateri, in fragmented old-growth forests. Ecological
Applications 5:164-182.
Lindenmayer, D. B. and R. C. Lacy. 1995b. Metapopulation viability of arboreal marsupials
in fragmented old-growth forests: comparison among species. Ecological Applications.
5:183-199.
Liu, J. 1993. An introduction to ECOLECON: a spatially-explicit model for ECOLogical
ECONomics of species conservation in complex forest landscapes. Ecological Modelling
70:63-87.
Liu, J., J. B. Dunning, and H. R. Pulliam. Assessing alternative management strategies: an
example coupling GIS with spatially-explicit models. Conservation Biology (in press)
Loza, H. J., W. E. Grant ,J. W. Stuth, and T. D. A. Forbes. 1992. Physiologically based
landscape use model for large herbivores. Ecological Modeling 61:227-252
Mac Arthur, R. H., and E. O. Wilson. 1967. The Theory of Island Biogeography. Princeton
University Press, Princeton, New Jersey.
Madenjian, C. P., and S. R. Carpenter. 1991. Individual-based model for growth of young-of-
the-year walleye: a piece of the recruitment puzzle. Ecological Applications 1:268-278.

-------
39
Maguire, L. A., R. C. Lacy, R. J. Begg, and T. W. Clark. 1990. An analysis of alternative
strategies for recovering the eastern barred bandicoot. Pp. 147-164 IN: T. W. Clark and
J. H. Seebeck (eds.) The Management and Conservation of Small Populations. Chicago
Zoological Society, Brookfield, Illinois.
McCauley, E., W. W. Murdoch, R. M. Nisbet, and W. S. C. Gurney. 1990. The physiological
ecology of Daphnia: a dynamic model of growth and reproduction. Ecology 71:716-732.
Meyer, J. S. and M. S. Boyce. Life historical consequences of pesticides and other insults to
vital rates, pp. 349-363 IN: Kendall, R. J. and T. E. Lacher, Jr. (eds.) Wildlife
Toxicology and Population Modeling, Lewis Publishers, Boca Raton, Florida.
Murdoch, W. W., J. Chessson, and P. Chesson. 1985. Biological control in theory and
practice. American Naturalist 125:344-366.
Pulliam, H. R., J. B. Dunning, Jr., and J. Liu. 1992. Population dynamics in complex
landscapes: a case study. Ecological Applications 2:165-167.
Pulliam, H. R. 1994. Incorporating concepts from population and behavioral ecology into
models of exposure to toxins and risk assessment, pp. 13-26 IN: Kendall, R. J. and T. E.
Lacher, Jr. (eds.) Wildlife Toxicology and Population Modeling. Lewis Publishers, Boca
Raton, Florida.
Ray, C., and M. E. Gilpin. 1991. The effect of conspecific attraction on metapopulation
dynamics. Biological Journal of the Linnaean Society 42:123-134.
Rose, K. P., and J. H. Cowan Jr. 1993. Individual-based model of young-of-the-year striped
bass population dynamics. I. Model description and baseline simulations. Transactions of
the American Fisheries Society 122:415-438
Sabelis, M. W., O. Diekman, and V. A. A. Jansen. 1991. Metapopulation persistence despite
local extinction: predator-prey patch models of the Lotka-Volterra type. Biological
Journal of the Linnaean Society 42:267-283.
Samuels, W. B. and A. Ladino. 1983. Calculations of Seabird Population Recovery from
Potential Oil spills in the Mid-Atlantic Region of the United States. Ecological Modeling
21:63-84.
Seal, U.S., and T. J. Foose. 1989. Javan rhinoceros population viability analysis and
recommendations. International Union for the Conservation of Nature and Natural
Resources, Captive Breeding Specialist Group, Apple Valley, Minnesota.

-------
40
Seal, U. S., and R. C. Lacy. 1989, Florida panther population viability analysis. International
Union for the Conservation of Nature and Natural Resources, Captive Breeding Specialist
Group, Apple Valley, Minnesota.
Society of Environmental Toxicology and Chemistry (SETAC). 1994. Aquatic Dialogue
Group: Pesticide risk assessment and mitigation. SETAC Press, 1010 North 12th Ave.,
Pensacola, Florida 32501.
Shugart, H. H. 1984. A Theory of Forest Dynamics. Springer-Verlag, New York.
Tipton, A. R., R. J. Kendall, J. F. Coyle, and P. F. Scanlon. 1980. A Model of the Impact of
Methyl Parathion Spraying on a Quail Population. Bulletin of Environmental
Contamination and Toxicology 25:586-593.
Troyer, M. E., and M. S. Brody. 1994. Managing ecological risks at EPA: Issues and
recommendations for progress. EPA/600/R-94/183. U.S. Environmental Protection
Agency, Washington, D.C.
Turner, M. G., Y. Wu, W. H. Romme, and L. L. Wallace. 1993. A landscape simulation
model of winter foraging by large ungulates. Ecological Modelling 69:163-184.
Turner, M. G., Y. Wu, W. H. Romme, L. L. Wallace ,and A. Brenkert. Simulating winter
interactions among ungulates, vegetation, and fire in northern Yellowstone Park.
Ecological Applications 4:472-496.
Turner, M. G., G. J. Arthaud, R. T. Engstrom, S. J. Hejl, J. Liu, S. Loeb, and K. McKelvey.
1995. Usefulness of spatially explicit population models in land management. Ecological
Applications 5:12-16.
Urban, D. L., and N. J. Cook. 1986. Hazard Evaluation, Standard Evaluation Procedure,
Ecological Risk Assessment. EPA-540/9-85-001. U.S. Environmental Protection Agency,
Washington, D.C.
U.S. Environmental Protection Agency. 1992. Framework for Ecological Risk Assessment.
EPA/630/R-92/001. U.S. Environmental Protection Agency, Washington, D.C.
van Groenendael, J., H. DeKroon, S. Kalisz, and S. Tuljapurkar.. 1994. Loop analysis:
Evaluating life history pathways in population projection matrices. Ecology 75:2410-
2415.
van Tienderen. P. H. 1995. Life cycle trade-offs in matrix population models. Ecology
76:2482-2489.

-------
41
APPENDIX: ANNOTATED BIBLIOGRAPHY OF ECOLOGICAL MODELS
1.	Barnthouse, L. W. Population-level effects. IN: G. W. Suter, (ed.). Ecological Risk
Assessment. Chelsea, Michigan: Lewis Publishers; 1993; pp. 247-274.
Note: Overview of population modeling theory and applications, including
case studies involving toxic chemicals.
2.	Barnthouse, L. W. The Role of Models in Ecological Risk Assessment: A 1990's
Perspective. Environmental Toxicology and Chemistry, 1992; 11:1751-1760.
Note: A review emphasizing (a) modeling approaches developed within the
last decade; (b) applications to a broad array of environmental problems on
local, regional, and global scales; and (c) the relevance of different types of
models to different components of the risk assessment process.
3.	Barnthouse, L. W.; Suter, G. W.II, and Rosen, A. E. Risks of Toxic Contaminants to
Exploited Fish Population: Influence of Life History, Data Uncertainty, and
Exploitation Intensity. Environmental Toxicology and Chemistry. 1990;
9:297-311.
Note: Analysis of risks of contaminants to fish populations performed by
coupling standard toxicity test data to matrix-type population models derived
from fisheries data. Risks related to chemical exposure compared to risks
related to fishing and to environmental variability.
4.	Bartell, S. M.; O'Neill, R. V., and Gardner, R. H. Aquatic ecosystem models for risk
assessment. Analysis of Ecological Systems: State-of-the-Art in Ecological
Modelling. Lauenroth, W.K.; Skogerboe, G.V.; Flug, M. ed. Elsevier
Scientific Publishing Company; 1983; pp. 123-127. (Developments in
Environmental Modelling; v. 5).
Note: Overview of Fate of Aromatics Model (Foam), originally developed by
Bartell, and Standard Water-Column Model (SWACOM) originally developed
by O'Neill. Both models derive from early aquatic ecosystem models
developed for the International Biological Program in the early 1970s.
Programming techniques and computer hardware have improved significantly
since these models were first described, but the underlying ecosystem theory
has changed very little.
5.	Baveco, J. M. and A. M. de Roos. Assessing the impact of pesticides on lumbricid
populations: an individual-based modelling approach. Jounral of Applied
Ecology. (submitted).
Note: Physiologically-based model of earthworm population exposed to
pesticides. Includes analytical results for steady-state and Monte Carlo
analysis of stochastic responses.

-------
42
6.	Baveco, J. M. and R. Lingeman. An object-oriented tool for individual-oriented
simulation: host-parasitoid system application. Ecological Modelling. 1992;
61:267-286.
Note: Classical model of host-parasitoid interactions in a patchy environment,
but illustrating use of object-oriented programming in a PC environment.
7.	Belovsky, G. E. Modeling avian foraging: implications for assessing the ecological
effects of pesticides. Wildlife Toxicology and Population Modeling, R. J.
Kendall and T. E. Lacher, Jr. Boca Raton, FL: Lewis Publishers; 1994.
Note: Presents a linear programming approach to optimal foraging (and
consequent pesticide ingestion). Applied to cowbirds foraging on seeds and
grasshoppers. Processes/factors modeled include foraging, ingestion,
digestive constraints, energy requirements. Constraints consist of minimum
daily energy intake, maximum digestive capacity, maximum foraging time.
Addresses direct effects of pesticide on birds and indirect effects related to
reduction in grasshopper abundance. Scenario analyses evaluated impact of
reduced cowbird predation on effectiveness of pesticides at controlling
grasshoppers.
8.	Beyer, J. E. and G. C. Laurence. A stochastic model of larval fish growth. Ecological
Modelling. 1980; 8:109-132.
Note: Individual-based model of growth and survival of larval winter
flounder.
9.	Bird, S. L. Fate and exposure modeling in terrestrial ecosystems: a process approach.
Wildlife Toxicology and Population Modeling, Kendall, R. J. and T. E.
Lacher, Jr. Boca Raton, FL: Lewis Publishers; 1994; pp. 149-159.
Note: Overview of PIRANHA modeling system developed by EPA Athens
Lab. Consists of multimedia environmental fate models and supporting
databases. Includes a drift model (FSCBG), a soil model (PRZM), a model of
uptake in plants (no acronym), an of Craig Barber's FGETS model to uptake
of contaminants by worms, and a pharmacokinetic model of contaminant
uptake (ingestion pathway) by birds. Database information includes
geographical distributions of soil thpe, crop production, and rainfall. No
applications presented.
10. Bowers, M. A. Use of space and habitats by individuals and populations: dynamics
and risk assessment. Wildlife Toxicology and Population Modeling, Kendall
R. J. and T. E. Lacher Jr. Boca Raton, FL: Lewis Publishers; 1994.
Note: Discusses importance of habitat selection as determinant of
demographic "success." Source/sink populations, etc. no models, but some
good references.

-------
43
11.	Brisbin, I. L. Jr. Application of a modified Richards sigmoid model to assess the
uptake and effects of environmental contaminants upon birds. Wildlife
Toxicology and Population Modeling, R. J. Kendall and T. E. Lacher, Jr. Boca
Raton, FL: Lewis Publishers; 1994; pp. 161-170.
Note: Application of "Richards" 3-parameter model (an asymptotic sigmoid
growth model that includes the von Bertalanffy, logistic, and Gompertz
models as special cases) to contaminant uptake. Used to quantify change in
growth rate of coots and wood ducks as functions of contaminant exposure.
12.	Brown, G. M. Jr.; Hammack, J., and Tillman, M. Mallard population dynamics and
management models. Journal of Wildlife Management. 1976; 40(3):542-555.
Note: Application of classical fisheries methods to mallard populations.
Beverton-Holt model, with carrying capacity a function of habitat quantity.
Calculates MSY for humnting. Comparison to previously used "empirical"
techniques.
13.	Brown, J. H. and A. Kodric-Brown. Turnover rates in insular biogeography: effect of
immigration on extinction. Ecology. 1977; 58:445-449.
Note: metapopulation model.
14.	Cacho, 0. J. Protein and fat dynamics in fish: a bioenergetic model applied to
aquaculture. Ecological Modelling. 1990; :33-56.
Note: Individual-based bioenergetic model of catfish - separate fat & protein
compartments.
15.	Cantwell, M. D. and R. T. T. Forman. Landscape graphs: ecological modeling with
graph theory to detect configurations common to diverse landscapes.
Landscape Ecology. 1993; 8(4):239-255.
Note: Network analysis of landscape patterns. Application of graph theory to
"pattern recognition" problem in landscape characterization.
16.	Caswell, H. Matrix Population Models: Construction, Analysis, and Interpretation.
Sunderland, MA: Sinauer; 1989; p. 328 pp.
Note: Classic textbook/review of matrix population models. Includes
thorough literature review up through late 1980s.

-------
44
17.	Chesser, R. K.; K. B. Williams, and N. E. Mathews. Impacts of toxicants on
population dynamics and gene diversity in avian species. Wildlife Toxicology
and Population Modeling, Kendall, R. J. and T. E. Lacher, Jr. Boca Raton,
Florida: Lewis Publishers; 1994; pp. 171-188.
Note: Includes three types of models and analyses: (1) an individual-based
bird population model with sex, genotype, toxicant body burden, and
reproductive rate as individual characteristics. Includes stochastic variation in
mortality and reproduction. Model "experiments" used to investigate influence
of toxicant body burden, toxicant critical level, and population carrying
capacity on average population size and loss of genetic variablility. (2)
Logistic model used to estimate time-of-recovery as a function of mortality
level and intrinsic population growth rate. (3) Population genetics theory used
to estimate influence of contaminant-induced mortality on effective
population size, and consequently on the rate of loss of genetic variation.
18.	Cobb, G. P. and M. J. Hooper. Nonlethal wildlife monitoring to determine exposure
to xenobiotics and resulting impacts. Wildlife Toxicology and Population
Modeling, Kendall, R. J. and T. E. Lacher Jr. Boca Raton, FL: Lewis
Publishers; 1994; pp. 35-46.
Note: Describes empirically-based models of uptake of diazinon from the
gastrointestinal tract; empirical relationships between GI concentration of
diazinon and bird mortality; biochemical indicators of pesticide exposure
(alkylphosphate excretion, plasma AChE).
19.	Conroy, M. J. Y. Cohen F. C. James Y. G. Matsinos and B. A. Maurer. Parameter
estimation, reliability, and model improvement for spatially explicit models of
animal populations. Ecological Applications. 1995; 5(1): 17-19.
Note: Reveiw of parameter estimation methods for Spatially explicit
population models.
20.	Cowan, J. H. Jr. and K. A. Rose. Individual-based model of young-of-the-year striped
bass population dynamics. II. Factors affecting recruitment in the Potomac
River, Maryland. Transactions of the American Fisheries Society. 1993;
122:439-458.
Note: Analysis of individual-based striped bass model, emphasizing
environmental influences on year-class strength.
21.	Cowan, J. H. K. A. Rose E. S. Rutherford and E. D. Houde. Individual-based model
of young-of-the-year striped bass population dynamics. II. Factors affecting
recruitment in the Potomac River, Maryland. Transactions of the American
Fisheries Society. 1993; 122:459-466.

-------
45
22.	Cullen, Murray C. and Connell, Des W. Pesticide Bioaccumulation in Cattle.
Ecotoxicology and Environmental Safety. 1994; 28:221 -231.
Note: Pharmacokinetic model of pesticide bioaccumulation in grazing food
chain. Suitable for inclusion in individual-based models.
23.	Dale, V. H. and R. H. Gardner. Assessing regional impacts of growth declines using a
forest succession model. Journal of Environmental Management. 1987; 24:83-
93.
Note: Individual-based forest stand model used to simulate forest die-back in
New England.
24.	Danielson, B. J. Communities in a landscape: the influence of habitat heterogeneity
on the interaction between species. American Naturalist. 1991; 138:1105-
1120.
25.	DeAngelis, D. L. What food web analysis can contribute to wildlife toxicology.
Wildlife Toxicology and Population Modeling, Kendall, R. J. and T. E.
Lacher, Jr. Boca Raton, Florida: Lewis Publishers; 1994; pp. 365-382.
Note: General overview of food web theory and analysis. Discusses uses,
information requirements of different approaches.
26.	DeAngelis, D. L. and L. J. Gross eds. Individual-Based Models and Approaches in
Ecology. New York: Chapman and Hall; 1992.
Note: Edited volume containing 23 papers on various aspects of individual-
based modeling. Includes overviews of concepts, mathematical techniques,
specific applications.
27.	DeAngelis, D. L. K. A. Rose L. Crowder E. Marschall and D. Lika. Fish cohort
dynamics: Application of complementary modeling approaches. American
Naturalist. 1993; 68:273-292.
28.	DeAngelis, D. L. L. L. Godbout and B. J. Shuter. An individual-based approach to
predicting density-dependent compensation in smallmouth bass populations.
Ecological Modelling, 1991; 57:91-115.
Note: Model of a smallmouth bass population formulated in terms of the
reproduction, development, growth, foraging, and survival of individual fish.
Calibrated to data on the smallmouth bass population in Lake Opeongo,
Ontario. Simulates production of a single year-class from spawning through
overwintering survival. Includes influence of environmental conditions and
size/health of nest-guarding males.

-------
46
29.	deKroon, H. A. Plaisier J. Van Groenendael and H. Caswell. Elasticity: the relative
contribution of demographic parameters to population growth rate. Ecology.
1986; 67:1427-1431.
Note: Provides definition and rationale for "elasticity" index, a measure of the
influence of different life-cycle parameters on the rate of growth of a
population.
> 
30.	den Boer, P. J. The survival of populations in a heterogeneous and variable
environment. Oecologia. 1981; 50:39-53.
Note: Simulation of beetle population fluctuations: simple autocorrelation
model based on 20+ years of empirical data on abundance of subpopulations
of beetles on a heath. Demonstration that species composed of mutliple
independent subpopulations can persist for long time periods in spite of high
variability within subpopulations.
31.	DiGiulio, R. T.; Washburn, P. C.; Wenning, R. J.; Winston, G. W., and Jewell, C. S.
Biochemical responses in aquatic animals: a review of determinants of
oxidative stress. Environmental Toxicology and Chemistry. 1989; 8:1103-
1123.
Note: Review of biomarker studies: markers of exposure to free-radical-
producing chemicals. Discusses methodological problems; limitations on use
in regulation. Possibly relevant to development if individual-based models of
contaminant effects.
32.	Doak, D.; Kareiva, P., and Klepetka, B. Modeling population viability for the desert
tortoise in the western Mojave desert. Ecological Applications, 1994; 4:446-
460.
33.	Dobson, A. and P. Hudson. Assessing the impact of toxic chemicals: temporal and
spatial variation in avian survival rates. Wildlife Toxicology and Population
Modeling, Kendall, R. J. and T. E. Lacher, Jr. Boca Raton, FL: Lewis
Publishers; 1994; pp. 85-98.
Note: Describes sources of variation in bird survival: interspecific variation,
geographic variation, age-dependent variation, and seasonal variation. No
models. Developed from long-term banding studies of common British bird
species.
34. Dunning, J. B. B. J. Danielson and H. R. Pulliam. Ecological processes that affect
populations in complex landscapes. Oikos. 1992; 65169-175.

-------
47
35.	Dunning, J. B. D. J. Stewart B. J. Danielson B. R. Noon T. L, Root R, H. Lamberson
and E. E, Stevens. Spatially explicit population models: current forms and
future uses. Ecological Applications. 1995; 5:3-11.
Note: Overview of spatially-explicit populations models. Lots of references.
36.	Ebenhard, T. Colonization in metapopulations: a review of theory and observations.
Biological Journal of the Linnaean Society. 1991; 42:105-121.
Note: metapopulation models.
37.	Ellison, A. M. and B. L. Bedford. Response of a wetland vascular plant community to
disturbance: a simulation study. Ecological Applications. 1995; 5:109-123.
Note: Wetland ecosystem simulation model, based on cellular automata.
Integrates plant life history and population biology with hydrology.
38.	Emlen, J. M. Terrestrial population models for ecological risk assessment: a state-of-
the-art review. Environmental Toxicology and Chemistry. 1989; 8:831-842.
Note: Good review of models of terrestrial animal populations. Emphasizes
age-structured models.
39.	Emlen, John M. Pikitch Ellen K. Animal Population Dynamics: Identification of
Critical Components. Ecological Modeling. 1989; 44:253-273.
Note: Stage-based models of terrestrial vertebrate populations. Attempts to
identify "critical life stages" for large, long-lived vs. small, short-lived
organisms.
40.	Fahrig, L. and G. Merriam. Habitat patch connectivity and population survival.
Ecology. 1985;66:1762-1768.
Note: metapopulation models
41.	Graham, R. L.; Hunsaker, C. T.; O'Neill, R. V., and Jackson, B. L. Ecological risk
assessment at the regional scale. Ecological Applications. 1991; 1:196-206.
Note: Application of risk analysis concepts to landscape-level problems.
Includes analysis (hypothetical) of impacts of air pollution and bark beetle
infestations on forest landcover.
42.	Gurney, W. S. C. E. McCauley R. M. Nisbet and W. W. Murdoch. The physiological
ecology of Daphnia: a dynamic model of growth and reproduction. Ecology.
1990; 71:716-732.
Note: Individual-based model of Daphnia population growth derived from a
physiological model of the individual organisms. The individual model is
described in a companion paper by McCauley et al. (1990) in the same issue.

-------
48
43.	Gustafson, E. J. and T. R. Crow. Modeling the effect of forest harvesting on
landscape structure and the spatial distribution of cowbird brood parasitism.
Landscape Ecology. 1994; 9:237-248.
Note: Influence of timber harvesting patterns and resulting landscape structure
on the vulnerability of neotropical migrant nests to cowbird parasitism.
44.	Hallam, T. G.; Clark, C. E., and Lassiter, R. R. Effects of Toxicants on Populations:A
Qualitative Approach I. Equilibrium Environmental Exposure. Ecological
Modeling. 1983; 18:291-304.
Note: Individual-based model of Daphnia population exposed to
contaminants.
45.	Hallam, T. G. and J. T. DeLuna. Effects of Toxicants on Populations:a Qualitative
Approach III. Environmental and Food Chain Pathways. Journal of
Theoretical Biology. 1984; 109:411-429.
Note: Individual-based Daphnia population model based on three-
compartment model of contaminant uptake and toxicity.
46.	Hallam, T. G., and J. T. DeLuna.. Extinction and Persistance in Models of
Population-Toxicant Interactions. Ecological Modeling, 1983; 22:13-20.
47.	Hallam, T. G.; Lassiter, R. R.; Li, J., and McKinney, W. Toxicant-induced mortality
in models of Daphnia populations. Environmental Toxicology and Chemistry.
1990;9:597-621.
Note: Incorporation of contaminant pharmacokinetics/pharmacodynamics in
Hallam's individual-based Daphnia model.
48.	Hallam, T. G.; Lassiter, R. R.; Li, J., and Suarez, L. A. Modelling individuals
employing an integrated energy response: application to Daphnia. Ecology.
1990; 71:938-954.
Note: Bioenergetic model of Daphnia life cycle. Basis for Hallam's
individual-based Daphnia population model.
49.	Hanski, I. and M. Gilpin. Metapopulation dynamics: a brief history and conceptual
domain. Biological Journal of the Linnaean Society. 1991; 423-16.
Note: Nice review of metapopulation theory, from Andrewartha and Birch
through the date of publication.
50. Hansson, L. Dispersal and connectivity in metapopulations. Biological Journal of the
Linnaean Society. 1991; 42:89-103.

-------
49
51.	Holt, R. D. S. W. Pacala T. W, Smith and J. Lu. Linking contemporary vegetation
models with spatially explicit animal population models. Ecological
Applications. 1995;5:20-27.
Note: Discussion of issues involved in linking vegetation dynamics to
population dynamics.
52.	Huston, M. A. and T. M. Smith. Plant succession: life history and competition.
American Naturalist. 1987; 130:168-198.
Note: Uses forest-stand model to explain alternative tree life-history strategies
in terms of competition for space and light.
53.	Jager, H. I. D. L. DeAngelis M. J. Sale W. Van Winkle D. D. Schmoyer M. J. Sabo
D. J. Orth and J. A. Lukas. An individual-based model for smallmouth bass
reproduction and young-of-the-year dynamics in streams. Rivers. 1993; 4:491-
113.
Note: Modification of original DeAngelis et al. smallmouth bass model.
Major innovation consists of incorporation of spatial location and movement
as individual descriptors. Based on data from North Anna River, Virginia.
54.	Kareiva, P. Experimental and mathematical analyses of herbivore movement:
quantifying the influence of plant spacing and quality on foraging
discrimination. Ecological Monographs. 1982; 52:261-282.
55.	Kooijman, S. A. L. M. and Metz, J. A. J. On the Dynamics of Chemically Stressed
Populations.The Deduction of Population Consequences from Effects on
Individuals. Ecotoxicology and Environmental Safety. 1984; 8:254-274.
Note: A general and explicit model for the age dependent growth and
reproduction of individuals as a function of food supply.
56.	Lacy, R. C. Loss of genetic diversity from managed populations: interacting effects of
drift, mutation, immigration, selection and population subdivision.
Conservation Biology. 1987; 1:143-158.
Note: Generalized metapopulation model.
57.	Lacy, R.C.. VORTEX: a computer simulation model for use in population viability
analysis. Wildlife Research. 1993; 20:45-65.
Note: An individual-based model of the dynamics of small populations - has a
metapopulation version that can simulate up to 20 interacting subpopulations.
58.	Lamberson, R. H. B. R. Noon C. Voss and K. McKelvey. Reserve design for
terrestrial species: the effects of patch size and spacing on the viability of the
Northern Spotted Owl. Conservation Biology. 1994; 8:185-195.
Note: Metapopulation model of Northern spotted owl.

-------
50
59.	Lamberson, R. H. R. McKelvey B. R. Noon and C. Voss. A dynamic analysis of
Northern Spotted Owl viability in a fragmented forest landscape.
Conservation Biology. 1992; 6:505-512.
Note: Analysis of influence of landscape structure on Northern Spotted Owl
metapopulation.
60.	Lande, R. Demographic models of the northern spotted owl (Strix occidentalis
caurina). Oecologia. 1988; 75:601-607.
Note: Application of (1) Mertz's life table approach, and (2) Levins' patch
dynamics approach to survival of spotted owl. Argues that the population is
currently approximately stable based on life table statistics. Argues that
USDA management plan will not preserve sufficient habitat to prevent
extinction in the future.
61.	Lande, R. Extinction thresholds in demographic models of terrestrial populations.
American Naturalist. 1987; 130:624-635.
62.	Lassiter, R. R. and and T.G. Hallam. Survival of the fattest: implications for acute
effects of lipophilic chemicals on aquatic populations. Environmental
Toxicology and Chemistry. 1990; 9:585-596.
Note: Pharmacokinetic model of uptake and effects of lipophilic chemicals on
fish; suitable for incorporation in individual-based population models.
63.	Law, R. A model for the dynamics of a plant population containing individuals
classified by age and size. Ecology. 1983; 64:224-230.
Note: First good integration of age and size in a population model.
64.	Law, R. and M. T. Edley. Transient dynamics of populations with age- and size-
dependent vital rates, Ecologyl990; 71:1863-1970.
Note: Analysis of Law's age/stage-structured model.
65.	Leonardsson, K. Long-term ecological effects of bleached pulp-mill effluents on
benthic macrofauna in the Gulf of Bothnia. Ambio. 1993; 22:359-362.
Note: Sublethal effects of contaminants on benthic predator-prey system/
combines modeling, lab experiments, field monitoring.
66.	Li, H. J. F. Franklin F. J. Swanson and T. A. Spies. Developing alternative forest
cutting patterns: a simulation approach. Landscape Ecology. 1993; 8:63-75.
Note: Examination of effects of different forest cutting patterns on habitat
fragmentation in managed forest landscapes.

-------
51
67.	Lindenmayer, D. B. and R. C. Lacy. Metapopulation viability of arboreal marsupials
in fragmented old-growth forests: comparison among species. Ecological
Applications. 1995;5:183-199.
Note: Comparison of metapopulation viability for different Australian possum
species; relation to life-history parameters.
68.	Lindenmayer, B. B., and R. C. Lacy. Metapopulation viability of Leadbeater's
possum, Gymnobelideus leadbeateri, in fragmented old-growth forests.
Ecological Applications. 1995; 5:164-182.
Note: Application of VORTEX to viability of G. leadbeateri.
69.	Lindenmayer, D. B. Clark T. W. Lacy R. C. Thomas V. C. Population viability
analysis as a tool in wildlife conservation policy: with reference to Australia.
Environmental Management. 1993; 17:745-758.
Note: Nice overview of PVA.
70.	Liu, J. An introduction to ECOLECON: a spatially-explicit model for ECOLogical
ECONomics of species conservatin in complex forest landscapes. Ecological
Modelling. 1993; 70:63-87.
Note: Extension of Pulliam's BACHMAP model to include (1) GIS-based
landscape characteristics and (2) timber harvest economics.
71.	Liu, J. J. B. Dunning and H. R. Pulliam. Assessing alternative management strategies:
an example coupling GIS with spatially-explicit models. Conservation
Biology, (in press).
Note: Spatially-explicit population model linked to GIS.
72.	Loza, H. J. W. E. Grant J. W. Stuth and T. D. A. Forbes. Physiologically based
landscape use model for large herbivores. Ecological Modeling. 1992; 61:227-
252.
Note: Links physiologically-based model of cow foraging behavior to range
landscape model. Method potentially applicable to wildlife foraging in
agriculture-dominated landscapes.
73.	Madenjian, C. P. and S. R. Carpenter. Individual-based model for growth of young-
of-the-year walleye: a piece of the recruitment puzzle. Ecological
Applications. 1991; 1:268-278.
Note: Model of year-class production in walleye, modeling approach is similar
to the DeAngelis et al. smallmouth bass model.

-------
52
74.	Maurer, B. A. Dipodomys populations as energy-processing systems: regulation,
competition, and hierarchical organization. Ecological Modelling. 1990*
50:157-176.
Note: Two-species, energetics-based compartment model. Good application
of Odum-style ecology.
75.	McCauley, E. W. W. Murdoch R. M. Nisbet and W. S. C. Gurney. Thy physiological
ecology of Daphnia: development of a model of growth and reproduction.
Ecology. 1990;71:703-715.
Note: Development and testing of a model of the growth and reproduction of
individual Daphnia. A companion paper (Gurney et al. 1990) used the model
as a foundation for an individual-based Daphnia population model.
76.	McKelvey, K. B. B. R. Noon R. H. Lamberson. Conservation planning for species
occupying fragmented landscapes: the case of the northern spotted owl. IN:
Biotic Interactions and Global Change, Karieva, P. M. ,J. G. Kingsolver, and
R. B. Huey (eds.). Boston, MA: Sinauer; 1992; pp. 424-450.
77.	Messier, F. Ungulate population models with predation: a case study with North
American moose. Ecology. 1994; 75:478-488.
Note: moose-wolf population models and results.
78.	Meyer, J. S. and M. S. Boyce. Life historical consequences of pesticides and other
insults to vital rates. IN: Wildlife Toxicology and Population Modeling,
Kendall, R. J. and T. E. Lacher, Jr. (Eds.) Boca Raton, Florida: Lewis
Publishers; 1994; pp. 349-363.
Note: Life-table analysis of bird populations using the Euler equation:
estimation of sensitivity of lambda to changes in fecundity and survivorship,
given different age/size structures. Some new references.
79.	Miller, N. L. Stability analysis of toxic substances within aquatic ecosystems and
their effect on aquatic populations. Ecological Modelling. 1992; 60:151-165.
Note: 11-compartment model of vinyl chloride in an aquatic ecosystem;
includes Lotka-Volterra planktivorous fish model.
80.	Mittelbach, G. G. and Osenberg, C. W. Stage-structured interactions in bluegill:
consequences of adult resource variation. Ecology. 1993; 74:2381-2394.
Note: Classic study of bluegill population biology, emphasizing differences in
habitat selection and growth rates between juveniles and adults. Uses two-
stage population model.

-------
53
81.	Morgan, M. J. and Kiceniuk, J. W. Effect of fenitrothion on the foraging behavior of
juvenile Atlantic salmon. Environmental Toxicology and Chemistry. 1990;
9:489-495.
Note: Effects of sublethal exposure on foraging behavior (including dose-
response relationships). Suitable for inclusion in individual-based population
models.
82.	Murdoch, W. W. Population regulation in theory and practice. Ecology. 1994;
75:271-287.
Note: Review of theory and field results on population regulation;
reconciliation of Nicholson/Bailey with Andrewartha/Birch; argument that
stage-based/natural-history-related approaches are most fruitful avenue to
pursue at this time. Great references to recent literature.
83.	Murphy, D. D. and B. R. Noon. Integrating scientific methods with habitat
conservation planning: reserve design for Northern Spotted Owls. Ecological
Applications. 1992;2:3-17.
84.	O'Connor, R. J. Population patterns and process parameters - issues in integrating
monitoring and models. IN: Wildlife Toxicology and Population Modeling,
Kendall, R. J. and T. E. Lacher, Jr. (eds.) Boca Raton, Florida: Lewis
Publishers; 1994; pp. 283-300.
Note: Good qualitative discussion of density-dependence and habitat-
dependent population dynamics in birds. Good references to bird population
dynamics literature.
85.	Oliver, George R. and Laskowski, Dennis A. Development of Environmental
Scenarios for Modeling the Fate of Agricultural Chemicals in Soil.
Environmental Toxicology and Chemistry. 1986; 5:225-231.
Note: Used computerized soil survey information to identify soils most likely
to be encountered in major use areas. A climatic model generated rainfall
patterns for the study areas and provided the probability statement for the
environmental assessment. Approach applicable to development of spatially-
explicit population models for pesticide risk assessment.
86.	Oritsland, N. A. and Markussen, N. H. Outline of a physiologically based model for
population energetics. Ecological Modelling. 1990; 52:267-288.
Note: Physiological model of harp seals. Thermoregulatory requirements,
growth, feeding requirements. Linked to population via Leslie matrix.

-------
54
87.	Ostendorf, B. and J. F. Reynolds. Relationships between a terrain-based hydrologic
model and patch-scale vegetation patterns in an arctic tundra landscape.
Landscape Ecology. 1993; 8:229-237.
Note: Model predicts hydrology, which is then correlated with vegetation
data.
88.	Pulliam, H. R. Incorporating concepts from population and behavioral ecology into
models of exposure to toxins and risk assessment. IN: Wildlife Toxicology and
Population Modeling, Kendall, R. J. and T. E. Lacher Jr. (eds.) Boca Raton,
FL: Lewis Publishers; 1994; pp. 13-26.
Note: Review of applicability of foraging theory and habitat structure models
to the development of population exposure and effects models. Includes
metapopulation theory. Case study using Pulliam's study of Bachmann's
sparrow on the Savannah River Site.
89.	Pulliam, H. R. J. B. Dunning Jr. and J. Liu. Population dynamics in complex
landscapes: a case study. Ecological Applications. 1992; 2:165-167.
Note: Spatially explicit Population Model pf Bachman's Sparrow population
on DOE Savannah River Site.
90.	Reckhow, K. H. and S. C. Chapra. Confirmation of water quality models. Ecological
Modeling 20:113-133.
Note: Philosophy of hypothesis testing: suggestions of statistical techniques
for comparing model output to observational data. Good application of
Bayesian approach - possibly applicable to testing of pesticide effects models.
91.	Rice, J. A. L. B. Crowder and K. A. Rose. Interactions between size-structured
predator and prey populations: Experimental test and model comparison.
Transactions of the American Fisheries Society. 1993; 122:481-491.
92.	Rice, J. A. T. J. Miller K. A. Rose L. B. Crowder E. A. Marschall A. S. Trebitz and
D. L. DeAngelis. Growth rate variation and larval survival: Implications of an
individual-based size-dependent model. Canadian Journal of Fisheries and
Aquatic Sciences. 1993; 50:133-142.
93.	Rose, K. A. J. H. Cowan E. D. Houde and C. C. Coutant. Individual-based modeling
of environmental quality effects on early life stages of fish: a case study using
striped bass. American Fisheries Society Symposia, 1993; 14:125-145.
94.	Rose, K. A. S. W. Christensen and D. L. DeAngelis. Individual-based modeling of
populations with high mortality: A new method for following a fixed number
of model individuals. Ecological Modelling. 1993; 68:273-292.

-------
55
95.	Rose, K. P. and J. H. Cowan Jr. Individual-based model of young-of-the-year striped
bass population dynamics. I. Model description and baseline simulations.
Transactions of the American Fisheries Society. 1993; 122:415-438.
Note: Description of individual-based striped bass population model; basic
structure analogous to DeAngelis et al. smallmouth bass model, but the
principal emphasis is on early larval survival.
96.	Samuels, William B. and Ladino, Anthony. Calculations of Seabird Population
Recovery from Potential Oilspills in the Mid-Atlantic Region of the United
States. Ecological Modeling. 1983; 21:63-84.
Note: Model of herring gull and common tern population recovery from
potential oilspill damage in the US mid-Atlantic Outer Continental Shelf
(OCS) oil leasing area. Reviewed in detail by Barnthouse (1993).
97.	Schaefer, E. W. Jr. Comparative avian toxicology: What is its role in predicting and
monitoring the effects of agricultural pesticides? IN: Wildlife Toxicology and
Population Modeling, R. J. Kendall and T. E. Lacher, Jr. (eds.) Boca Raton,
Florida: Lewis Publishers; 1994; pp. 321-337.
Note: Review of comparative toxicology studies. Concludes that (1) species
sensitivites to the same pesticide vary over 2-3 orders of magnitude, (2) there
is no reliably "most sensitive" species, (3) therefore, extrapolation is
impossible [this was also at one time the consensus opinion among fish
toxicologists]. Instead, increase post-registration monitoring is needed.
98 Simberloff, D. The contribution of population and community biology to
conservation science. Annual Review of Ecology and Systematics. 1988;
19:473-511.
Note: Critical review of relevance of theoretical models to conservation of
small populations; calls for more autecological studies.
99. Slade, N. A. Models of structured populations: age and mass transition matrices. IN:
Wildlife Toxicology and Population Modelingt Kendall, R. J. and T. E.
Lacher, Jr. (eds.) Boca Raton, Florida: Lewis Publishers; 1994; pp. 189-199.
Note: Overview of age and size-structured matrix models. One example.
Some new references.
100. Sterbacek, Z.; Skopek, V., and Zavazal, V. A composite landscape ecology prognostic
expert system - COLEPES. Part I. System philosophy and design. Ecological
Modelling. 1990; 50:145-156.
Note: Description of expert system that uses quantitative environmental
models as subcomponents - few details provided.

-------
56
101.	Tipton, Alan R.; Kendall, Ronald J.; Coyle, John F., and Scanlon, Patrick F. A Model
of the Impact of Methyl Parathion Spraying on a Quail Population. Bulletin of
Environmental Contamination and Toxicology. 1980; 25:586-593.
Note: Novel application of matrix population model to acute pesticide
exposures. Reviewed in depth by Barnthouse (1993).
102.	Turner, M. G. 1989. Landscape ecology: the effect of pattern on process. Annual
Review of Ecology and Systematics. 1989; 20:171-197.
Note: General review of landscape ecology.
103.	Turner, M. G. G. J. Arthaud R. T. Engstrom S. J. Hejl J. Liu S. Loeb and K.
McKelvey. Usefulness of spatially explicit population models in land
management. Ecological Applications. 1995; 5:12-16.
Note: Overview of use of SEPMs in land management.
104.	Turner, M. G. Y. Wu W. H. Romme and L. L. Wallace. A landscape simulation
model of winter foraging by large ungulates. Ecological Modelling. 1993;
69:163-184.
Note: Description of spatially-explicit population model for elk and deer in
Yellowstone National Park.
105.	Turner, M. G. Y. Wu W. H. Romme L. L. Wallace and A. Brenkert. Simulating
winter interactions among ungulates, vegetation, and fire in northern
Yellowstone Park. Ecological Applications. 1994; 4:472-496.
Note: Uses spatially explicit population model to evaluate influence of fire
and climatic variability on survival of ungulate populations in Yellowstone
National Park.
106.	Turner, Monica G; R, Costanza, and F. H. Sklar. Methods to evaluate the
performance of spatial simulation models. Ecological Modelling. 1989; 48:1-
18.
Note: Comparison of different methods of quantifying spatial patterns.
107.	Tyler, J. A. and K. A. Rose. Individual variability and spatial heterogeneity in fish
population models. Review in Fish Biology and Fisheries. 1994; 491-123.
108.	Urban, D. L. and T. M. Smith. Microhabitat pattern and the structure of forest bird
communities. American Naturalist. 1989; 133:811-829.

-------