The United States Environmental Protection Agency's Office of Research and
Development partially performed and funded the research described here. It is intended
for internal EPA use. Mention of trade names or commercial products does not
constitute endorsement or recommendation for use.
Acknowledgements: Mark Bagley, Bernie Daniel, Ross Highsmith, Suzanne Jackson,
John Martinson, Erik Pilgrim, Greg Toth, and Eric Waits provided helpful comments and
insightful discussions during the preparation of this document. Their contributions are
greatly appreciated.
Cover designed by Teresa Ruby, SRA International, Contractor to the EPA.
Photos used by permission (clockwise):
Swallow-tailed Hummingbird (Eupetomena macroura) and (on back cover) African Tulip Tree (Spathodea campanulata)
http://www.flickr.com/photos/flaviocb/333264564/in/set-72057594067118778/ Photographer: Flávio Cruvinel Brandão
Monarch butterfly, (Danaus plexippus) http://www.flickr.com/photos/pieceoflace/1124313115/in/dateposted/page2/
Photographer: Lacey pieceoflace@yahoo.com
Grey-headed Flying-fox (Pteropus poliocephalus) http://www.flickr.com/photos/chi-liu/141064143/ Photographer: chi liu
Bullfrog (Rana catesbeiana) http://images.nbii.gov/amphibians/nbii_t00036.jpg Photographer: John J. Mosesso (Public Domain)
National Biological Information Infrastructure (NBII), USGS Biological Informatics Office
ii
Expert panel recommendations
for designing a genetic monitoring program to assess long-term effects of
pesticides and plant incorporated protectants
on non-target organisms
Susan E. Franson1, John W. Bickham2, Andrew J. Bohonak3,
David M. Marsh4, and Amy G. Vandergast5
1
U. S. Environmental Protection Agency, National Exposure Research Laboratory,
Cincinnati, OH
2 Center for the Environment, Purdue University, West Lafayette, IN
3 Department of Biology, San Diego State University, San Diego, CA
4 Department of Biology, Washington and Lee University, Lexington, VA
5 U. S. Geological Survey, Western Ecological Research Center, San Diego, CA
U.S. Environmental Protection Agency
National Exposure Research Laboratory
Ecological Exposure Research Division
Molecular Ecology Research Branch
Cincinnati, Ohio
iii
TABLE OF CONTENTS
1. INTRODUCTION ..........................................................................................................
1
2. RATIONALE .................................................................................................................
4
3. FORMAT AND COMPOSITION OF PANEL ................................................................
3.1 Questions Posed to Panel ..................................................................................
3.2 Biographical Sketches of Panel Members ..........................................................
6
6
7
4. SUMMARY OF PANEL RESPONSES .........................................................................
4.1 What are the advantages and disadvantages of a population genetic approach
for large-scale monitoring compared to traditional census approaches? ..........
4.2 What information is likely to be obtained using a genetic monitoring approach
that would not be available from traditional monitoring approaches? ................
4.3 What are the most appropriate genetic measures to evaluate? .........................
4.4 Are particular species, groups of species, functional guilds, etc. more
informative than others and more cost-effective for monitoring long-term
ecosystem responses to changing agricultural practices? ................................
4.5 How should genetic monitoring be combined with other measures, methods,
and models to achieve the most efficient assessment of current and predicted
ecosystem responses to changing agricultural practices? ................................
4.6 What are the key knowledge gaps and research questions that must be
addressed before designing a genetic monitoring program? ............................
10
10
13
14
17
18
19
5. IMPLICATIONS FOR MONITORING DESIGN............................................................. 21
5.1 Population genetic monitoring ............................................................................ 21
5.2 Linking PIP exposure to population and community responses ......................... 22
6. REFERENCES ............................................................................................................ 24
APPENDIX A. Responses Received from Panel ..............................................................
A-1 Response of Dr. John W. Bickham ....................................................................
A-2 Response of Dr. Andrew J. Bohonak and Dr. Amy Vandergast ........................
A-3 Response of Dr. David M. Marsh .......................................................................
25
26
40
63
Appendix B. Curricula Vitae of Panel Members ............................................................... 85
Appendix C. Project Background and Questions Sent to Panel ....................................... 102
iv
1. INTRODUCTION
The potential risks posed to non-target organisms by genetically modified (GM) crops
remain a concern to the United States Environmental Protection Agency (EPA). Crops
with plant incorporated protectants (PIPs), generally are believed to be both more
effective and environmentally benign than the broad-spectrum pesticides they are
designed to replace. The pesticide typically is expressed throughout the life of the
plant, reducing the labor and expense of repeated pesticide applications and providing
an economic benefit to growers. A concomitant result is a reduction in the exposure of
non-target organisms from direct spraying, spray drift, or transport of pesticides through
the environment.
EPA's Office of Pesticide Programs (OPP) is responsible for evaluating and registering
pesticides under the Federal Insecticide, Fungicide, and Rodenticide Act (FIFRA) and
promotes pest management policies that result in better protection of human health and
the environment. Rigorous registration and review processes ensure individual
pesticides (including PIPs) do not pose unreasonable risks of harm to human health and
the environment. PIPs and the genetic material necessary for the plant to produce the
substance are subject to OPP's pesticide registration process, although the plant itself is
not. Results from recent laboratory and short-term field studies have confirmed that Btcrops do not pose a significant risk to non-target insect populations (e.g., Mendelsohn et
al. 2003; Hilbeck and Schmidt 2006; Romeis, Bartsch et al. 2006; Romeis, Meissle
2006; Marvier et al. 2007, and references therein).
The pesticide registration process includes an assessment of risk to non-target species.
Short-term studies conducted under controlled conditions are not necessarily predictive
of long-term population responses, and lengthy post-marketing monitoring of pesticides
generally is not required by OPP. There is currently no mechanism in place to evaluate
long-term exposures and the related effects to individuals, populations, or ecosystems.
The possibility of unintended long-term consequences will increase as GM crop
adoption increases, more crop types are modified to incorporate PIPs, and new variants
and novel transgenes that express alternate proteins are discovered and
commercialized. Potential concurrent exposure to multiple PIPs, other pesticides and
stressors, and different agricultural practices further complicate evaluation of risk to
non-target organisms.
Existing data may not be sufficient to clearly establish the degree to which GM crop
adoption, separate from other environmental and agricultural trends, reduces or
increases risk to human health and non-target organisms in agricultural and other
ecosystems at national or regional scales. A long-term and large-scale monitoring
program would provide high quality data for a comprehensive evaluation of long-term
responses of non-target organisms, such as beneficial insects, birds, and other animals
to PIP exposures. It would expand information for risk assessment beyond the
laboratory evaluation of direct toxicity to non-target organisms to long-term effects that
manifest at the level of populations of non-target species. Secondary trophic effects
can give rise to indirect effects that cascade through the entire ecosystem. Changes in
1
abundance or genetic diversity within a population or in species diversity within an
ecosystem can alter ecosystem function and ultimately affect its ability to provide
ecosystem services. A long-term and large-scale monitoring program would provide the
necessary data to detect these changes.
Results from a monitoring program that clearly links observed long-term ecological
effects, either detrimental or beneficial, to conventional pesticide exposure and GM crop
usage could be used to improve future risk assessments. This long-term monitoring
program would provide an ecological accountability tool to evaluate the effectiveness of
the registration process and efforts to reduce the use of conventional agricultural
pesticides. The resulting data would provide the American public, Congress, and the
international community with a means to measure the effectiveness of EPA policies and
regulations in protecting human health and the environment.
EPA’s Office of Research and Development (ORD) is investigating the potential for
using molecular genetic data and population genetic monitoring to provide additional
risk assessment methods and measures to support the comprehensive evaluation of the
ecological risks of PIPs. As a part of this investigation, ORD sought the input,
recommendations, and opinions of a panel of external experts regarding design
considerations for an efficient monitoring program to assess long-term exposure and
effects of PIPs and other pesticides on non-target organisms. This document
summarizes the guidance received from the panel members on the relevance and
application of molecular and population genetic methods for long-term monitoring.
Following this Introduction, Section 2 provides an outline of the rationale for
investigating molecular approaches to ecological monitoring. It gives a very brief
overview of traditional population monitoring approaches and describes some of the
inherent limitations of traditional census-based methods. Population genetic
parameters are discussed as possible surrogates for traditional population demographic
parameters, and the additional information available from molecular genetic data that
cannot be obtained from traditional census-based sampling is explored.
Section 3 describes the composition of the expert panel that was formed to provide
insights regarding fundamental issues and research questions that should be addressed
prior to designing and implementing a large-scale population genetic monitoring
program. Short biographical sketches of the panel members highlight their expertise
and research interests, which are especially relevant to assessing populations of nontarget organisms. This section helps the reader interpret the panel’s comments in
relation to research backgrounds and professional experience.
Section 4 summarizes the major conclusions and recommendations provided by the
panel. It is not meant to substitute for the wealth of knowledge contained in the
individual reports provided by the panelists (Appendix A). Rather, it serves to highlight
common themes and consensus recommendations of the entire panel, as well as
unique comments and diverse viewpoints of individual panelists. The reader is directed
2
to the individual reports in Appendix A for more in-depth discussion of the subject
matter.
Finally, Section 5 attempts to integrate and assess the key points and recommendations
of the panel as they relate to future monitoring design research. Some additional
considerations not raised by the panel are also discussed.
3
2. RATIONALE
Traditional ecological studies typically monitor populations using incidence data, census
counts, and local abundance estimates. Similarly, community assessments often use
species incidence, composition, and relative abundance to calculate various species
diversity measures. These methods can be effective for populations and species in a
limited geographic area and have been used successfully in hundreds of monitoring
programs to detect trends in the size of populations of particular species of conservation
concern (Marsh and Trenham 2008). However, traditional ecological and population
measures may not be feasible for regional evaluations that potentially include very large
populations of multiple species or taxonomic groups. The sampling and analysis effort
required to collect accurate and reliable data that are comparable across a variety of
landscapes and time points would be prohibitively resource intensive.
Traditional ecological monitoring approaches only provide a portion of the information
available to assess significant changes in populations. For example, the number of
individuals in a population may not be a sufficient measure of the sustainability of that
population. Numbers alone do not provide complete resilience to fluctuations in the
environment, guarding against species extinctions, extirpation of local populations, and
concomitant reductions in species diversity. The size of the breeding population and
connectivity among populations are also keys to population persistence. For large
organisms, intensive observation and sampling may provide an estimate of the number
of breeders within a population and dispersal dynamics among populations. However,
direct measurement would be nearly impossible for large geographic areas and nearly
unimaginable for small organisms.
Understanding and defining population1 boundaries is fundamental to estimating
individual population parameters, for determining inter-population dynamics within a
species, and for analyzing and interpreting these data for evolutionary research,
conservation management, or ecological risk assessment. Determining population
boundaries is problematic for traditional ecological methods. Intensive observation and
mark and recapture methods may delineate a boundary between separate groups of
animals, however, ascertaining that such groups are truly distinct and different enough
to constitute separate populations may not be possible based on morphological
characters alone. Recently documented cases of cryptic species in several taxonomic
groups including butterflies (Hebert et al. 2004), amphipods (Witt et al. 2006) and
parasitic flies (Smith et al. 2006, Smith et al. 2007) highlight the difficulty of species
identification based solely on visible morphological features. Defining populations within
a species based solely on morphological features is likely to be even more difficult.
1
The term 'population' is used differently by different authors in different contexts (Waples and Gaggiott
2006 and references therein). The underlying concept is of a cohesive subset of individuals that share
common attributes that differ from those of other such subsets. In population and evolutionary biology, a
population is typically defined as a group of individuals of the same species that occupy a geographically
defined area, are usually isolated to some degree from other similar groups, and that potentially
interbreed and share a common gene pool. In some cases, genetically distinct populations are not
isolated spatially but by demographic and life history strategies (e.g., migrating birds or marine
mammals).
4
Molecular genetic techniques and population genetic analysis provide the scientific
basis for overcoming several of the limitations of traditional census-based ecological
monitoring (Schwartz et al. 2007). Although these methods and the resulting data have
their own unique challenges and limitations, they have the potential to provide data
useful for assessing long-term population level effects of exposure to PIPs and
conventional pesticides on non-target species. In some cases, population genetic
monitoring may provide an alternative monitoring approach that is both more costeffective and more informative than traditional monitoring approaches. In other cases,
collecting population genetic data in conjunction with traditional ecological monitoring
could provide complementary data, enhancing the information available for risk
assessment/risk management decisions.
As described in greater detail in the remainder of this document, molecular genetic data
can be used to derive measures that serve as surrogates for traditional population
parameters. Abundance and migration/dispersal can be inferred from estimates of
effective population size of local populations and gene flow between them. Effective
population size2 is less responsive to short-term environmental fluctuations and large
variations in population size; therefore, the effort required to obtain sufficient sample
sizes for population genetic estimates may be far less than that required for reliable
estimates of population census size using traditional methods. In addition, genetic
diversity within populations and within species can provide data useful for assessing the
risk of local population extinctions and fragmentation. Genetic information from multiple
species can be combined to provide an overall measure of biodiversity at a site or within
a region. DNA signatures can increase the precision of species identification, identify
interspecific hybrids, and detect the presence of parasites or pathogens known to infect
a species of interest. Finally, molecular genetic data also provides a means to define
population boundaries, which could eventually be used in spatially-explicit risk
assessments.
2
The effective population size, Ne, is defined as the size of an idealized, theoretical population that loses
genetic variation due to genetic drift or inbreeding at the same rate as the study population. It is
considered a standardized measure of the breeding population size that corrects for unequal sex ratio,
variance in individual reproductive output, and temporal fluctuations in population size. Although related
to the census population size, N, which reflects ecological processes such as birth and deaths, predation,
and competition, Ne more clearly corresponds to evolutionary parameters.
5
3. FORMAT AND COMPOSITION OF PANEL
ORD sought the input of an expert panel regarding the potential use of population
genetic approaches to ecosystem monitoring and assessment. The literature on
monitoring population-level effects of genetically modified crops on non-target
organisms is extensive; however, much of the current literature deals exclusively with
agricultural systems, emphasizes potential effects on natural predators of crop pest
species, and is motivated by economic interests. To assemble a panel with a broader
perspective of ecological impacts and monitoring, we identified individuals with
publications in peer-reviewed journals and presentations at scientific meetings in the
areas of population monitoring, population genetics, genetic monitoring, molecular
genetics, ecotoxicology, and evolutionary genetics. More detailed information on the
research interests and expertise of these individuals was gathered from the web sites of
the institutions with which these individuals were affiliated. This information was used to
ensure that the panel would be diverse with respect to the application of molecular and
population genetics in their research and familiarity with different taxonomic groups.
Individuals were contacted and provided a brief background description of ORD's
interest in ecological monitoring of non-target organisms (Appendix C). All those
contacted expressed an interest in considering the addition of genetic approaches in
ecological monitoring, although several were unavailable due to prior commitments and
obligations.
3.1 Questions posed to panel
The panel was asked to provide insights and recommendations regarding the potential
use of population genetic approaches in ecosystem monitoring, design considerations
for an efficient monitoring program, and fundamental issues and research questions that
should be addressed prior to designing and implementing a large-scale population
genetic monitoring program. Input was requested in the form of individual reports as the
intent was to obtain multiple independent insights into the fundamentals of designing a
monitoring program, rather than a single course of action or specific design resulting
from compromises among panelists. No format was specified for these reports and no
constraints were placed on content, although an introductory section briefly describing
the scope of the report was suggested. The only guidance provided was that the report
should include, but not necessarily be limited to, information that would address the
following questions:
1. What are the advantages and disadvantages of a population genetic approach for
large-scale monitoring compared to traditional census approaches?
2. What information is likely to be obtained using a genetic monitoring approach that
would not be available from traditional monitoring approaches?
3. What are the most appropriate genetic measures to evaluate?
4. Are particular species, groups of species, functional guilds, etc. more informative
than others and more cost-effective for monitoring long-term ecosystem responses
to changing agricultural practices?
6
5. How should genetic monitoring be combined with other measures, methods, and
models to achieve the most efficient assessment of current and predicted ecosystem
responses to changing agricultural practices?
6. What are the key knowledge gaps and research questions that must be addressed
before designing a genetic monitoring program?
3.2 Biographical sketches of panel members
The panel was selected for their ecological and molecular expertise, their past
experience, and their ability to address a broad range of applications of population
genetics with diverse taxonomic groups. General expertise and experience of the panel
includes:
 All have been involved in monitoring projects.
 The geographic regions and habitats these researchers have studied include tropical
rain forests in Hawaii and Central America; coastal scrub in southern California;
temperate forests in the southeastern U. S.; freshwater streams and ponds in
southern California, the Rocky Mountains and New York; arctic oceans; and
contaminated sites in Azerbaijan and Superfund sites in the U. S., including the
Nevada Test Site.
 They have researched a variety of organisms, ranging from tiny arthropods to
enormous whales, and including a variety of aquatic invertebrates, terrestrial
arthropods, fish, amphibians, reptiles, birds, and mammals.
Short biographical sketches for the panel members are provided below. Curricula vitae
are provided in Appendix B.
Dr. John W. Bickham is the Director of the Center for the Environment and a Professor
in the Department of Forestry and Natural Resources at Purdue University, West
Lafayette, IN. As Director of the Center for the Environment, he is responsible for
promoting interdisciplinary environmental research projects among 138 faculty
participants representing 30 academic departments. His own research is designed to
address and understand the genetic processes that act at the levels of populations and
species and to apply this knowledge to the conservation and management of natural
resources. Dr. Bickham has published over 180 papers in peer-reviewed scientific
journals on topics including: genetics, evolution, and conservation of the endangered
Steller sea lion; genetic structure and diversity of a population of bowhead whales
harvested by Alaskan natives; and the molecular evolution of vespertilionid bats. In
ongoing studies of environmental mutagens at contaminated sites in the Republic of
Azerbaijan and several localities in the U. S. including Superfund sites, Dr. Bickham
uses a combination of cytogenetic, flow cytometric, and molecular analyses to elucidate
the processes by which genotoxic effects are produced in individuals and transmitted
through populations and species. Contaminant effects are compared at various levels
of organization, from the individual to the population, in species including frogs, tree
swallows, house mice, turtles, and mosquitofish.
Dr. Andrew J. Bohonak is an Associate Professor in the Department of Biology at San
Diego State University (SDSU), San Diego, CA. He is Vice-Chair and Director of
7
Undergraduate Advising and Curriculum, and has affiliations with the Evolutionary
Biology Program Area, the Ecology Program Area and the Bioinformatics and Medical
Informatics Research Center at SDSU. His long-standing interest is in population
genetics, particularly the study of dispersal, gene flow, and recent population history.
His laboratory uses microsatellites, inter simple sequence repeats (ISSRs),
mitochondrial DNA (mtDNA) and nuclear DNA (nDNA) sequencing technologies in
studies of molecular ecology, conservation genetics, landscape genetics, and other
areas of evolutionary biology. The objective underlying most of this research is to inform
management and conservation decisions using patterns of genetic variation. He has
taxonomic expertise with freshwater invertebrates, although students in his laboratory
have studied a wide variety of other taxonomic groups. The organisms represented in
his over 30 journal articles and technical reports include freshwater invertebrates,
marine vertebrates, mule deer, Jerusalem crickets, and invasive agricultural pest
species of tephritid flies. Dr. Bohonak has also developed software for population
genetic analysis, including IBDWS (Isolation By Distance Web Service).
Dr. David M. Marsh is an Associate Professor of Biology at Washington and Lee
University, Lexington, VA. He is also a mentor at Mountain Lake Biological Station, the
field station of the Biology Department at the University of Virginia, as part of the
Research Experiences for Undergraduates (REU) program sponsored by the National
Science Foundation. Dr. Marsh has published 20 research articles that represent his
research in three distinct but related topics. The first uses genetic and ecological
methods to study the dispersal of terrestrial salamanders in both continuous and
fragmented habitats. The results of these studies help explain how and why
salamanders disperse and how different patterns of land use promote or hinder
salamander movement. Investigating the influence of salamander behavior on dispersal
led to the second research topic, studying the effects of observer expectations and
biases in collecting data on animal behavior, particularly the ways that gender may
interact with these biases. The third research topic, the optimal design of monitoring
programs, was motivated by the critical need for monitoring data for modeling
population dynamics. As a sabbatical fellow at the National Center for Ecological
Analysis and Synthesis in Santa Barbara, CA, from October 2006 to July 2007, Dr.
Marsh investigated using simulation models to derive general, quantitative guidelines for
the design of population monitoring programs based on measurable population
parameters.
Dr. Amy Vandergast is a Geneticist with the U.S. Geological Survey, Western
Ecological Research Center, San Diego Field Station, San Diego, CA. She also serves
as an Adjunct Assistant Research Professor in the Department of Evolutionary Biology
at San Diego State University, San Diego, CA. Her main research focus is population
genetics and conservation of terrestrial invertebrates, the subject of a dozen research
publications. She used a combination of allozyme and mtDNA markers to investigate
population genetic differentiation in three species of spiders in isolated forest kipukas
(small remnant islands of forest between lava flows) in Hawaii. Dr. Vandergast has
more recently applied similar molecular techniques to study the effect of anthropogenic
landscape changes, including habitat destruction and fragmentation, in California. In
8
this research, she and her colleagues are exploring the utility of geographic information
systems (GIS) for genetic inquiries at both regional and local scales. At a regional level,
the concordance among patterns of genetic diversity and geographic features is being
investigated for several co-occurring species including invertebrates, herpetofauna, and
small mammals; resulting patterns are being examined in relation to current protected
lands in southern California. At a finer scale, the goal is to evaluate the impacts of
recent habitat fragmentation, caused by urban development, on the population genetic
structure of endemic and flightless Jerusalem crickets. In addition, Dr. Vandergast is
involved in several genetic investigations of rare and endangered species, including the
Alameda whipsnake, the narrow-headed gartersnake, the Western shovel-nosed snake,
the Coachella Valley fringe-toed lizard, and southern California fairy shrimp.
9
4. SUMMARY OF PANEL RESPONSES
Appendix A provides the unedited responses of the panel members. Varying
perspectives on the potential uses of population genetic approaches to monitoring and
ecological risk assessments are provided, an intended consequence of the format and
composition of the panel. Depth of coverage of the topics addressed and the emphasis
placed on particular components of monitoring and population genetics vary in ways
that reflect the differences in individual areas of expertise and research interests. Many
similarities among the responses are also noted, emphasizing the fundamental
concepts of population genetics.
As indicated, no particular report format was specified and no constraints were placed
on the topics the panel members could address. The questions posed to the panel
served only as a general indication of issues to be considered in their response.
Themes covered by the questions partially overlapped, so particular comments might
apply to multiple questions. As a consequence, the responses of the panelists are not
in a standard question and answer format. The questions however, serve as a
convenient way to organize this summary of the input, insights, and recommendations
provided by the panel. In general, responses that address multiple questions are
summarized only once. Major comments and recommendations appear initially in bullet
form, followed by an overall summary of the responses. Unique responses are
referenced to the individual panelist, as are contrasting viewpoints.
As a preface to the following sections, it should be mentioned that one recurrent theme
among the responses of the panelists is that the answer to each of the questions
depends on the objectives of the monitoring program. The design of a monitoring
program, and the appropriate approach, methods, measures, and sentinel species
should be determined by the goals, objectives, temporal and spatial scope, and specific
questions to be addressed. The panelists describe the range of options for population
genetic monitoring, various monitoring scenarios for which particular applications are
most appropriate, and real-world examples drawn from their experiences in ecological
and population genetics research, along with recommendations for research that would
enhance non-target monitoring. This summary can only briefly touch on the key points
made by the panel, and the reader is referred to the individual responses for in-depth
treatment of these issues.
4.1 What are the advantages and disadvantages of a population genetic
approach for large-scale monitoring compared to traditional census approaches?
Advantages mentioned by the panel include:
 The ability to estimate population parameters that cannot be estimated using a
traditional census-based approach.
 The ability to accurately quantify genetic variability at relatively low cost.
 The ability to delineate genetically distinct populations.
 The possibility of targeted monitoring of functional genes in populations of particular
concern; DNA is available for later use as new functional genes are identified.
10
 The possibility of retrospective monitoring to compare the current state of the
population with past status.
 Data can be used to estimate multiple parameters and endpoints.
 The ability to integrate effects of multiple stressors (complex mixtures) through both
time and space.
 Estimates of population size based on the effective population size, Ne, may be
more stable over time than estimates of the census population size, N, if species
undergo extreme fluctuations in abundance from generation to generation.
 Ecological processes can be inferred more efficiently than by direct estimates from
traditional methods.
 Relatively low cost.
Disadvantages mentioned by the panel include:
 Some population demographic parameters can only be estimated using traditional
census-based methods (e.g., true population size, fecundity, and mortality).
 No information is provided about the cause of an observed effect.
 Estimates of the effective population size, Ne, have large confidence intervals
making it difficult to detect changes through time.
 Only a few indicator species can be analyzed, which may or may not adequately
reflect ecosystem responses.
 Genetic monitoring methods have little or no track record.
 No direct comparisons of the precision and costs of a population genetic approach
versus traditional survey approaches are available.
 The cost of a genetic approach to monitoring species diversity and abundance may
be significantly greater than that of traditional census-based approaches.
Related comment:
 Costs of any monitoring program must be balanced against the novel information
that it provides.
The ability to estimate population parameters that cannot be estimated using a
traditional census-based approach is an obvious advantage. Section 4.3 specifically
addresses the information that can be obtained only from a population genetic
approach. Similarly, there are some population demographic parameters (true
population size, fecundity, and mortality) that can only be estimated using traditional
census-based methods, and the inability to estimate these is a disadvantage for the
population genetic approach.
For some population demographic parameters, census-based and population genetic
approaches provide similar, but not identical, information. The limits of inference, bias,
and measurement error associated with all estimation procedures become
disadvantages for either approach. The panel members did a thorough job in
describing the relative merits and limitations of estimates derived from both approaches,
the nuances of their interpretation, and the conditions under which each is most
appropriate. The goals, objectives, questions to be addressed, spatial and temporal
11
scale, and sampling logistics and design of a monitoring program determine what
parameters are used. Thus, the relative advantages and disadvantages of a population
genetic approach, versus a traditional approach ultimately, will depend upon the specific
monitoring program.
Bickham and Bohonak and Vandergast mention that an advantage of a molecular
genetics approach is that it allows targeted monitoring of functional genes in populations
of particular concern. Possibilities include genes that function in detoxification for
monitoring contaminated sites, traits associated with fitness for monitoring threatened
and endangered species or populations with unique genetic profiles, and genes that are
under selection for pesticide resistance, behavioral modifications, and life cycle timing.
Once the DNA is extracted for use for population genetic analysis it would be available
for later use as new functional genes are identified.
Use of population data to determine if a population experienced bottlenecks in the past
improves the interpretation of the current status of that population in the light of its
evolutionary history. Bickham and Marsh both mention that the availability of historical
museum collections allows for retrospective monitoring to compare the current state of
the population to its status in the past. Marsh adds that historical effective population
size estimated either from historical samples or coalescence theory could be used to
test hypotheses about effects of changing land use on particular species. Bickham
emphasizes the importance of systematics collections as a resource to aid the study of
biodiversity loss, by providing both a standard with which to compare current levels of
genetic diversity of sentinel species as well as information about historic trends in
genetic variability.
Bickham points out that while pesticides have multiple mechanisms of toxicity at the
somatic level, they lead to similar effects at the population level so population genetic
endpoints are not specific to the mode of action of toxicity. The advantage of population
genetics is that effects of complex mixtures are integrated through both time and space.
The disadvantage is that they do not provide information about what particular toxin(s)
caused the observed effect. Bickham also mentions that new mutations might be
discovered when using a molecular genetics approach.
The cost of a population genetic approach as compared to that of traditional censusbased approaches is mentioned by all panel members. Cost is one area where the
panelists expressed three decidedly different opinions. Bickham sees the ability to
quantify genetic variability at a relatively low cost as an advantage of a population
genetics approach. In contrast, Marsh considers the cost of a genetic approach to
monitoring species diversity and abundance to be significantly greater than that of
census-based approaches. Bohonak and Vandergast present a cost-benefit view for
comparing relative costs of population genetic and census-based monitoring. They
state that the costs of any monitoring program must be balanced against the novel
information that it provides, and that a population genetic approach for long-term
monitoring of non-target arthropods would be most cost effective (relative to information
content) when the target species have highly variable population sizes and can be
12
sampled and identified by personnel that do not possess highly specialized taxonomic
training.
Marsh remarks that genetic monitoring methods have little or no track record and that
there are almost no real-world examples of genetic monitoring outside of the fisheries
and wildlife literature. This assessment contrasts with examples and references cited
by Bickham and Bohonak and Vandergast. Marsh cites the lack of a direct comparison
of the precision and costs of a population genetic approach versus traditional survey
approaches as a disadvantage. He also notes the continuing debate as to whether a
few indicator species can accurately reflect environmental change; because there is no
agreement on indicator species for agricultural ecosystems in the U.S., a focus on a
small number of indicator species in a population genetic approach is problematic.
4.2 What information is likely to be obtained using a genetic monitoring
approach that would not be available from traditional monitoring approaches?
The panel mentions that the following information, likely to be obtained using a genetic
monitoring approach, is not available from traditional monitoring approaches:
 Delineation of the boundaries of genetically distinct populations.
 Determination of population structure, subdivision, and demographic partitioning.
 Accurate quantitative measures of genetic diversity within populations, between
populations, and within species.
 Empirical estimation of the rate at which diversity is being lost.
 Estimates of gene flow, migration rates, and population connectivity
 Estimates of the effective population size, Ne, and/or the number of breeders, Nb, in
the population.
 Improved accuracy of species identification using genetic signatures, especially in
taxonomically complex samples (e.g. microbial communities), or where identification
requires an expert (e.g., invertebrate larvae).
The panel notes that molecular genetic data is a means to define the boundaries of
genetically distinct populations and cluster individuals within those populations. This
allows the determination of population structure, subdivision, and demographic
partitioning. Bohonak and Vandergast describe a variety of analytical approaches used
in these determinations and the strengths and weakness of each. Bickham points out
that while population structure and demographic partitioning are typically applied to
geographically defined populations, they are equally applicable to migrating species
(e.g., whales, birds, butterflies) with temporally defined populations.
Estimates of gene flow and migration rates can be determined only from genetic data.
These parameters reflect evolutionary processes related to dispersal, an ecological
process. While the two are related, they are not equivalent, and, as Bohonak and
Vandergast point out, both could be of interest. Dispersal could be determined using
traditional mark and recapture methods, however, the panelist emphasize that the cost
and effort required to do so would be enormous and, under the best of circumstances,
would likely miss rare but significant, long-distance dispersal events. Marsh notes that
13
gene flow is probably the best way to understand population connectivity at a landscape
scale and that estimates of migration and gene flow would be critical for monitoring
where connectivity may change rapidly. Bickham notes that patterns of connectivity and
gene flow among populations allow inferences about ecological process more efficiently
than direct ecological measurement and illustrates this with an example of source-sink
dynamics among populations of marsh frogs near a contaminated site.
The panel members agree that accurately quantifying genetic diversity within
populations, between populations, and within species is possible with a molecular
genetic approach. Monitoring allelic diversity allows empirical estimation of the rate at
which diversity is being lost; the rate of change in allelic frequencies could be an early
warning signal for the potential loss of genetic diversity. Estimates of within-population
diversity and population connectivity provide information about the risk of extirpation of
local populations and, if tracked through time, provide an insight into the evolutionary
potential of species.
The panel notes that molecular methods can estimate the effective population size, Ne,
and/or the number of breeders, Nb, which are related to loss of genetic diversity through
drift and to extinction risk. These parameters are generally correlated with overall
population size, and in small populations in which loss of genetic diversity is an
important concern, Ne provides a direct measure of the magnitude of that threat.
Species identification is essential for any monitoring program. In traditional approaches,
this could be done in the field by sight or sound. Bickham and Bohonak and
Vandergast note that in population genetic monitoring, the use of genetic data for
species identification could improve accuracy, especially in taxonomically complex
samples (e.g. microbial communities) or where identification requires an expert (e.g.,
invertebrate larvae). In a similar vein, Marsh mentions that DNA collected from noninvasive sampling can be used to identify individual animals for genetic mark-recapture
studies and cites Schwartz et al. (2007) to provide additional examples of the use of
genetic identification of individuals in population monitoring (although he does not
recommend this approach).
Bickham discusses an extensive range of possible applications of genetic data to
population monitoring. These applications include identifying new mutations to detect
the direct effects of mutagenic compounds and comparing levels of genetic diversity
among genes inherited via different genetic transmission systems (mtDNA, autosomes,
X- and Y- chromosomes). Genetic data could be particularly relevant for monitoring the
effects of endocrine disruptors or species sensitive to particular toxicants.
4.3 What are the most appropriate genetic measures to evaluate?
The panel notes that the following genetic measures should be included in any
population genetic monitoring program:
 Estimates of population boundaries.
 Estimates of effective population size, Ne.
14



Estimates of intrapopulation genetic variation, population differentiation, and gene
flow.
Demographic and ecological data from traditional approaches.
Biodiversity estimates, including both species diversity and genetic diversity within
species.
This list represents a consensus only in that these parameters are mentioned by all of
the panelists. The importance of any particular parameter depends on the goals and
objectives of the program to be designed. Comments as to which parameters are most
appropriate varied among the panelists. Each panelist describes population genetic
measures that could be included in a population genetic monitoring approach and
provides guidance for the application of each in particular situations. The guidance
includes recommendations for research (see Section 4.6) that should be done prior to
designing a monitoring program to provide a scientific basis for determining the most
appropriate measures to evaluate.
One area of consensus is that microsatellites represent the current molecular marker of
choice to access variation in the nuclear genome. Bohonak and Vandergast provide a
thorough discussion of the categories and types of molecular markers and the
advantages and disadvantages of each, including relative cost of development and
ease of developing protocols for high throughput analysis. They mention that single
nucleotide polymorphisms (SNP)s or nDNA sequencing are not likely to be cost
effective at present, but suggest that some limited effort might be put into developing
nDNA sequence markers for specific genes of interest. Bickham makes similar
comments regarding molecular markers and recommends microsatellites as the best
place to start. He puts a bit more emphasis on SNP development and identifying target
functional genes for the sentinel species. The panelists are also in agreement that
mtDNA sequencing, likely a portion of the cytochrome c oxidase subunit I (COI) gene,
should be included.
Bohonak and Vandergast provide a brief introduction to population genetics that gives
the rationale for choosing which population genetic parameters are appropriate for
addressing particular questions, based on the underlying processes that determine
genetic diversity within and among populations. They, and also Marsh, describe the
methods that can be used to estimate various population genetic parameters, the
assumptions made in their estimation, the advantages and disadvantages of each, and
software programs that are particularly useful for population genetic data analysis.
Marsh also discusses several different traditional methods used in ecological
monitoring. Bickham provides more real-world examples of applications of population
genetics in addressing particular questions of interest.
The importance of any particular parameter depends on the goals and objectives of the
program to be designed. For example, if the objective of a monitoring program is to
document changes in species composition and abundance over time, a traditional
approach using presence/absence or count data may be both more valid and cost
effective than a population genetic approach. Genetic data may be usefully applied
15
(e.g., DNA identification, genetic mark and recapture) in an ecological study, even
though population genetics parameters are not of interest.
Bickham notes that an important potential ecosystem response to changing agricultural
practices will be changes in biodiversity, measured both by species diversity and
genetic diversity within species. Bohonak and Vandergast note that maintaining
biodiversity has moral and aesthetic significance, as well as being fundamental to the
sustainability of agroecosystems and the level of ecosystem services they provide. If
documenting biodiversity and tracking changes in biodiversity over time is the objective,
then population genetic parameters are an essential component of a monitoring
program. Quantifying population boundaries is a prerequisite for measuring population
genetic parameters.
Discussion of the utility of the effective population size, Ne, provides an example of how
the relative importance of a population genetic parameter depends on the goals and
objectives of the study. Ne is tightly linked to loss of genetic diversity and thus, is an
important parameter if the objective is to monitor changes in genetic diversity over time,
and particularly for populations at risk of extirpation. The importance of Ne as a
surrogate for population census size, N, depends on the characteristics of the
population(s) to be monitored. If the population is small, or population size is highly
variable, changes in Ne might be easier to detect than changes in N. However,
estimates of Ne can have very large confidence intervals, so that in large populations,
changes in N might be detected more quickly if error rates in field sampling were small
and had been quantified. Bohonak and Vandergast and Marsh provide detailed
discussion of the issues concerning the estimation and application of Ne as a surrogate
for population size.
Genetic data from multiple populations can be compared and used to estimate
population differentiation and gene flow. Ultimately, the genetic diversity within and
among populations, along with the genetic connectivity among them is what determines
the evolutionary potential and persistence of a species. Intrapopulation genetic
variation can be measured by a variety of metrics. These measures can be used to
make inferences about the historical size and connectivity of populations, based on
coalescence theory or by direct temporal comparison, if historic samples are available.
The historic size and connectivity provide a context within which current population
parameters can be interpreted.
The need for demographic and ecological data from traditional approaches is
emphasized by each of the panelists. Traditional ecological population measures and
population genetic parameters are complementary, thus the greatest inferential power
would be provided by determining both in parallel. More importantly, while population
genetic data can provide an in-depth understanding of genetic variation of a few
species, traditional ecological methods could provide a better understanding of species
diversity. Both are essential components of overall biodiversity.
16
4.4. Are particular species, groups of species, functional guilds, etc. more
informative than others and more cost-effective for monitoring long-term
ecosystem responses to changing agricultural practices?
The panel recommendations include:
 Species selection should be based on the goals and objectives of the program.
 Monitor four or more species that include different taxonomic groups, functional
roles, or ecological guilds.
 Species selection criteria should include:
o Demographic characteristics (geographic distribution and range, abundance,
dispersal ability, ease of collecting, short generation times, etc.)
o Intrinsic value as ecosystem service providers (pollinators, natural predators
of pest species) or of aesthetic or conservation value
o Reflective of broader ecosystem responses to environmental change
o Vulnerability of species or populations (relative sensitivity to PIPs or
pesticide(s), temporal and spatial coincidence to exposure, rare or
endangered status, etc.)
o Taxonomic certainty - easily identified in the field or in the lab by readily
available taxonomic specialists or by DNA identification
 Candidate species for consideration include:
o Insects (e.g., pollinators, predators)
o Honeybees
o Butterflies
o Aquatic invertebrates (stream or pond)
o Neotropical migrant birds
o Songbirds
o Pond amphibians
Rather than identify particular species that should be selected, the panel members all
discuss considerations for selecting species. Bickham notes that the goals and
objectives of the monitoring program must be considered in selecting sentinel species
and identifying potentially vulnerable species. He illustrates this with an example of
monitoring the Arctic cisco to answer a specific question about population structure.
Sentinel species in agricultural areas could similarly be selected based on relative
sensitivity to pesticides, value in providing ecosystem services, or demographic
characteristics.
Marsh indicates that the criteria for selecting taxa would be different for traditional
monitoring and genetic monitoring. Criteria for traditional monitoring would include a
large number of species that were relatively common and widespread, easily detected,
and likely to be influenced by the environmental factors of interest. For population
genetic monitoring, the criteria would include a few species that are known to reflect
broader ecosystem responses, occur in small semi-isolated populations, are easily
captured, and have short generation times. Marsh notes that the homogeneity of
agricultural landscapes makes meeting these criteria difficult, but suggests that species
that live in ponds and woodlots might be good candidates.
17
Bohonak and Vandergast provide an overview of known GM effects on non-target
arthropods that is particularly relevant to PIPs. They note that sensitivity to pesticides
and non-target effects of PIPs will vary across taxonomic groups and transgenic traits
(or strains), particularly if these traits target different pest species. Differential sensitivity
may also occur for secondary trophic effects on non-target organisms that are natural
predators of pest species.
Bohonak and Vandergast note that life-history characteristics of species and
populations should be considered. Very small, recently introduced, or frequently
bottlenecked populations may have too little genetic variation available to make
meaningful population estimates. Conversely, in very large populations, impacts may
not be detected unless, or until, they are severe. For highly mobile species, local
changes in genetic diversity may be swamped by gene flow. They recommend that
power analysis be used to select taxa, collection method, number of replicates, etc.
needed to detect significant changes in response variables such as abundance. In
addition, they recommend that four or more species with different attributes be included
for genetic monitoring.
4.5 How should genetic monitoring be combined with other measures, methods,
and models to achieve the most efficient assessment of current and predicted
ecosystem responses to changing agricultural practices?
The panel recommendations include:
 Plan population genetic studies in parallel with traditional ecological surveys.
 Make use of museum specimens and systematics collections where available.
Simply stated, the panel’s recommendation is to collect samples for genetic monitoring
at the same time that traditional count surveys are conducted. Combining information
from a molecular population genetic approach with traditional ecological monitoring
provides the most inferential power for assessments. Species composition and
abundance from count surveys of multiple species gives an index of species diversity.
When combined with genetic diversity within species, an overall indicator of biodiversity
can be constructed. In cases where the traditional ecological approaches and the
genetic monitoring approaches reflect similar population processes (e.g., population
census size versus effective population size, dispersal versus gene migration),
information from one approach can often validate estimates from the other; these
parallel estimates improve the analysis of population processes. Greater understanding
can be achieved with the use of museum specimens and systematics collections where
available. Such collections would allow the monitoring data to be put in a historical
perspective. Past diversity levels can be compared with current levels and historical
trends in genetic variability can be discovered.
18
4.6 What are the key knowledge gaps and research questions that must be
addressed before designing a genetic monitoring program?
Panel recommendations for further analysis or study include:
 Identify goals and objectives for monitoring.
 Identify candidate species.
 Conduct an initial large study for each species to be monitored.
o Field test collection methods.
o Refine/develop lab protocols.
o Compare traditional ecological surveys to population genetic approaches.
 Develop additional markers for selected species, including functional genes of
interest.
The responses include recommendations for the next steps in designing a monitoring
program based on a population genetics approach. These next steps can be grouped
into the general categories identified in the above list, although the emphasis placed on
particular issues varies among the panelists. Some of the key points and issues
mentioned by the panelists are summarized below.
Bickham's response is directed to what he identifies as 'technology gaps' rather than
knowledge gaps. He notes that while development of microsatellites, SNPs, and
targeted genes for sentinel species is needed, the techniques for marker development
are known. He suggests that once management or monitoring objectives are identified
and sentinel species selected, monitoring could begin initially by taking advantage of
established population genetic methods such as mtDNA sequencing. He mentions that
microsatellites are well established as an approach for accessing nuclear gene
variation, as are the methods for developing microsatellite markers for additional
species. He recommends that a secondary effort be put into SNP development,
followed by marker development for functional genes of particular interest for each
individual species.
Marsh recommends that before undertaking a genetic monitoring program, estimates of
the comparative cost and precision of traditional monitoring be made. Preferably, a
direct comparison would be made for selected populations where traditional monitoring
is ongoing (to provide count data) and archived tissue samples are available (for genetic
analysis). As an alternative, Marsh suggests that simulation studies could be used. He
further suggests that feasibility studies for candidate species be conducted to determine
the sampling effort required to get estimates with sufficient precision to detect significant
changes. He concludes that if he were setting up a large-scale, prospective monitoring
program, he would lean strongly towards more traditional methods. This would involve
randomly selecting as many sites as possible within each region of interest, possibly
stratifying on level of PIP use, and conducting point count transects for birds, line
transects for butterflies, aquatic surveys for either amphibians or invertebrates, and
baited traps for honeybees.
19
Bohonak and Vandergast provide several specific recommendations, including those for
development of molecular markers, determining the number of markers and sample
sizes required, and methods and software for population genetic data analyses. They
recommend a large initial study for each species selected to establish likely population
boundaries, field-test collection methods, refine lab protocols, and provide guidance for
ongoing monitoring protocols. The initial study should include multiple sites per
species, at spatial scales small enough so that unique gene pools are represented
within the study area. For arthropods, they recommend sampling of up to 50 individuals
per site unless that level of sampling would be detrimental to the population. The initial
study would include mtDNA sequencing (likely COI) and 20 microsatellite loci per
species. They further recommend highly detailed documentation and protocols be
established including those for sample storage, DNA extraction and PCR, and
microsatellite scoring.
20
5. IMPLICATIONS FOR MONITORING DESIGN
The summary in the previous section integrates the information from the individual
reports and highlights instances of consensus on key points and areas where
responses differ, with the intent of providing an accurate and unbiased representation of
the individual responses of the panelists. This section discusses some of the key
implications of the panel’s guidance for development of a future EPA monitoring
program that includes population genetic methods.
5.1 Population genetic monitoring
An essential first step in designing an effective monitoring program is ensuring that the
goals and objectives of monitoring are clear (Witmer 2005, Pollock et al. 2002,
Holthausen et al. 2005, Schwartz et al. 2007). Goals and objectives may be broad
initially and defined more specifically as the program evolves. Defining the spatial,
temporal, and taxonomic scope of monitoring early in the conceptualization of the
program will help target a pilot program to obtain the most useful information. From a
design standpoint, if the intent of the monitoring program is to make inferences, the
statistical population must be clearly defined so that a sampling frame can be specified
for probabilistic sampling.
A clear understanding of the scope of the monitoring program will aid in determining the
population of organisms to monitor. For example, a specific focus on PIP exposure and
effects may not require monitoring of native plants. In this context, it is important to
recognize that Bohonak and Vandergast focus their report exclusively on arthropods in
agricultural areas and recommend a minimum of four species that represent a range of
population size, dispersal ability, ecological niche and known susceptibility. From an
EPA perspective, the agroecosystem of interest is the broader ecosystem in which the
primary land use is agriculture, and includes the surrounding uncultivated landscape.
Thus, multiple species may need to be sampled to monitor different component habitats
(e.g., woodlots, streams, farm ponds, protected natural areas, urban areas, golf
courses, etc.) within the larger agroecosytem.
A pilot program, which begins by taking advantage of established methods currently
available, is recommended before implementing a full monitoring program. A full range
of traditional ecological and population genetic methods and measures should be
included in the pilot so that strengths, weaknesses, costs, complementarity,
redundancy, and novelty of alternative approaches can be evaluated. Protocols can be
tested both in the field and in the laboratory, and refined if necessary, to ensure they are
logistically feasible for implementation in a large-scale and long-term study. Measures
that provide little additional information relative to their cost should be eliminated once
that determination is made.
21
5.2 Linking PIP exposure to population and community responses
The major hurdle to designing a monitoring program to directly link PIP usage to a
reduction in pesticide usage and reduced risk to non-target organisms is the availability
of exposure data at a scale that informs population monitoring. Various test kits are
available to field-test for the expression of transgenic traits in plant material, including
PIPs. Similarly, it is possible to test soils for the presence of PIPs and chemical
pesticides, although differences in persistence and availability would need to be
considered in estimating exposure. Direct assays of plant material or soil could be used
to produce spatial and temporal exposure profiles, although the number of sample
points required to inform at a scale relevant to population modeling would likely make
this approach cost prohibitive.
Marvier et al. (2008) proposed that spatially explicit data on GM crop usage and variety
and pesticide applications be collected and made available by the U.S. Department of
Agriculture's National Agricultural Statistics Service (NASS). There are many issues
relating to privacy of individual landowners and potentially confidential business
information of farming enterprises and GM crop developers that would need to be
resolved before this could be implemented. An alternative would be to model the spatial
distribution of pesticide and GM crop usage from available data in much the same way
that atmospheric deposition models spatially integrate data from discrete sampling
points to provide continuous maps of exposure. Modeling exposure to GM crops and
pesticides would suffer from the additional challenge of starting with data that is
aggregated at different spatial scales. Data aggregated by county (or sometimes
township or state) are available for crops and some pesticides. Estimates of the
percent of the total acres planted in GM crops are also available, although information
on particular transgenic traits or strains generally is not available. Spatial models could
be developed to provide estimates of exposure at any geographic location, although the
confidence interval associated with that estimate is likely to be large.
Herbicide resistance is the most common trait expressed in genetically modified crops.
Although not a PIP and not directly under OPP’s purview, increased use of nonselective herbicides made possible through herbicide resistant crops may have an
indirect detrimental effect on insect populations that rely on weedy plants to provide
nectar sources, larval food sources and oviposition sites. This presents an interesting
challenge for monitoring populations of non-target arthropods, particularly predator
species whose natural prey are a combination of economic pests and alternate prey
species. With the widespread use of GM crops that incorporate both insect and
herbicide resistance, it will be difficult to ascribe an observed change in populations of
natural predators to either one or both exposures. Collecting data on the herbaceous
plants in the area near monitoring sites could provide habitat information that could be
used as a covariate in analyses and aid in interpretation of the results.
22
6. REFERENCES
Hebert, P.D.N., E.H. Penton, J.M. Burns, D.H. Janzen, and W. Hallwachs. 2004. Ten
species In one: DNA barcoding reveals cryptic species in the neotropical skipper
butterfly Astraptes fulgerator. Proceedings of the National Academy of
Sciences of the United States of America 101(41):14812-14817. doi:
10.1073/pnas.0406166101
Hilbeck, A. and J.E.U. Schmidt. 2006. Another view on Bt proteins – how specific are
they and what else might they do? Biopesticides International 2(1):1-50.
Holthausen, R., R.L. Czaplewski, D. DeLorenxo, G. Hayward, W.B. Kessler, P. Manley,
K.S. NcKelvey, D.S.Powell, L.F. Ruggiero, M.K. Schwartz, B. Van Horne, and C.D.
Vojta. 2005. Strategies for monitoring terrestrial animals and habitats. Gen. Tech.
Rep. RMRS-GTR-161. Fort Collins, CO: U.S. Department of Agriculture, Forest
Service, Rocky Mountain Research Station. 34p.
Marsh, D.M., and P.C. Trenham. 2008. Current trends in plant and animal population
monitoring. Conservation Biology 22(3):647-655. doi: 10.1111/j.15231739.2008.00927.x
Marvier, M., C. McCreedy, J. Regetz, and P. Kareiva. 2007. A meta-analysis of Bt
cotton and maize on nontarget invertebrates. Science 316(5830):1475-1477. doi:
10.1126/science.1139208
Marvier, M., Y. Carriere, N. Ellstrand, P. Gepts, P. Kareiva, E. Rosi-Marshall, B.E.
Tabashnik, and L.L. Wolfenbarger. 2008. Harvesting data from genetically engineered
crops. Science 320(5875):452-453. doi: 10.1126/science.1154521
Mendelsohn, M., J. Kough, Z. Vaituzis, and K. Matthews. 2003. Are Bt crops safe?
Nature Biotechnology 21(9):1003-1009. doi:10.1038/nbt0903-1003
Pollock, K.H., J.D. Nichols, T.R. Simons, G.L. Farnsworth, L.L.Bailey, and J.R. Sauer.
2002. Large scale wildlife monitoring studies: statistical methods for design and
analysis. Environmetrics 13:105-119. doi: 10.1002/env.514
Romeis, J., D. Bartsch, F. Bigler, M.P. Candolfi, M.M.C. Gielkens, S.E. Hartley, R.L.
Hellmich, J.E. Huesing, P.C. Jepson, R. Layton, H. Quemada, A. Raybould, R.I. Rose,
J. Schiemann, M.K. Sears, A.M. Shelton, J. Sweet, Z. Viatuzis, and J.D. Wolt. 2006.
Moving through the tiered and methodological framework for non-target arthropod risk
assessment of transgenic insecticidal crops. Proceedings of the 9th Symposium on the
Biosafety of Genetically Modified Organisms. September 24-29, 2006, Jeju Island
Korea, pp. 64-69.
23
Romeis, J., M. Meissle, and F. Bigler. 2006. Transgenic crops expressing Bacillus
thuringiensis toxins and biological control. Nature Biotechnology 24(1):63-71.
doi:10.1038/nbt1180
Smith, M.A., N.E. Woodley, D. H. Janzen, W. Hallwachs, and P.D.N. Hebert. 2006.
DNA barcodes reveal cryptic host-specificity within the presumed polyphagous
members of a genus of parasitoid flies (Diptera: Tachinidae). Proceedings of the
National Academy of Sciences of the United States of America 103(10): 3657-3662.
doi: 10.1073/pnas.0511318103
Smith, M.A., D.M. Wood, D. H. Janzen, W. Hallwachs, and P.D.N. Hebert. 2007. DNA
barcodes affirm that 16 species of apparently generalist tropical parasitoid files (Diptera,
Tachinidae) are not all generalists. Proceedings of the National Academy of
Sciences of the United States of America 104(12):4967-4972. doi:
10.1073/pnas.0700050104
Schwartz, M.K., G. Luikart, and R.S. Waples. 2007. Genetic monitoring as a promising
tool for conservation and management. Trends in Ecology and Evolution 22(1):25–33.
doi:10.1016/j.tree.2006.09.013
Waples, R.S., and O. Gaggiott. 2006. What is a population? An empirical evaluation of
some genetic methods for identifying the number of gene pools and their degree of
connectivity. Molecular Ecology 15(6):1419-1439. doi: 10.1111/j.1365294X.2006.02890.x
Witmer, G.W. 2005. Wildlife population monitoring: some practical considerations.
Wildlife Research 32(3):259-263. doi: 10.1071/WR04003
Witt, J.D.S., D.L. Threloff, and P.D.N. Hebert. 2006. DNA barcoding reveals
extraordinary cryptic diversity in an amphipod genus: implications for desert spring
conservation. Molecular Ecology 15(10):3073-3082. doi: 10.111/j.1365294X.2006.02999.x
24
APPENDIX A.
Responses Received from Panel
25
APPENDIX A-1
Response of Dr. John W. Bickham
26
REPORT TO:
Dr. Susan E. Franson
Molecular Ecology Research Branch
26 W. Martin Luther King Dr
Cincinnati, OH 45268
FROM:
Dr. John W. Bickham
Title:
Recommendations on the Design and Implementation of a Genetic Monitoring Program
to Assess Long-Term Effects of Plan-Incorporated Protectants and Chemical Pesticides
on Non-target Organisms.
August 1, 2008
27
Introduction
Let me begin by presenting a little of my background, interests and experiences
in ecotoxicology, population genetics, and monitoring programs. My training is in
evolutionary genetics and I became interested in ecotoxicology in the mid 1980’s
(McBee et al., 1987) because I wanted to learn how environmental contaminants
influenced evolutionary processes. My lab began to study how genetic mutations were
induced by environmental mutagens and this led to our pioneering the use of flow
cytometry as a biomarker of genetic damage in somatic cells of wildlife exposed to
pollution (Bickham, 1990). Over the course of 20 years or so my lab conducted many
investigations on fish and wildlife species from a wide variety of contaminated sites and
learned a great deal about how contaminants cause somatic DNA damage in natural
populations. At the same time I had an interest in trying to determine how the effects of
such stressors are expressed at a higher level of biological organization. This led to a
series of articles published from my lab on the population genetic effects of chemical
contaminants. In 1994 we named this area of investigation Evolutionary Toxicology
and noted that this field draws from the experimental designs and methods of
ecotoxicology, the methods of modern molecular population genetics, and the
theoretical framework of evolutionary genetics (Bickham and Smolen, 1994). So, this
new multidisciplinary field will guide our efforts to determine the impact of contaminants
on the genetic systems of organisms.
Over the years I have also had experience with monitoring programs including
ones that involve population genetics. Since 1991 I have studied the population
genetics of Steller sea lions, a species that has been in decline due to unknown
stressors since the 1960’s. Our studies of population genetics have contributed to the
recognition of multiple genetic stocks in this species which has directly influenced how
the species is managed (Baker et al., 2005). As part of our studies we have
documented patterns of genetic variation as well as levels of diversity, both of which are
important aspects of a monitoring program. I am also involved in a long-term genetics
study of the bowhead whale, a species that is harvested by the Native American
communities along Alaska’s North Slope. The harvest is managed by the International
Whaling Commission to which genetic subdivision and diversity levels represent key
endpoints in their calculations of harvest quotas (strike limits). So, as part of this effort I
am a member of the US delegation to the IWC’s Scientific Committee and every year I
attend the annual meeting and present data from our studies. Periodically the data are
reviewed and harvest quotas are established. The significance of this is that molecular
genetics data are gathered on every species of great whale and population genetics are
considered a basic component of the monitoring systems, as are other ecological
indicators like population trends, growth rates, health status, etc.
Another activity in which I am involved that is relevant to this report is my status
as an independent international consultant to BP in Azerbaijan. I advise them mainly on
their environmental monitoring program for their offshore facilities in the Caspian Sea,
but as well I review reports and work plans for their nearshore and onshore facilities.
BP has an extensive environmental monitoring program that includes the usual sorts of
28
endpoints such as contaminant levels in water and sediment. However, they also use
biodiversity indicators including species richness and abundance as well as some
genetic biomarker studies on fish. They do not employ population genetics studies.
The above listed experiences, and other related studies I have conducted, have
led me to think a great deal about the subject of this report (at least the broad aspects of
such a monitoring program, not specific to pesticides). It is somewhat disappointing to
me that is has taken so long for population genetics to find it’s way into our planning of
environmental monitoring programs. I guess the ecologists have been more forceful in
pushing their agendas in this regard, but at least now important monitoring programs
such as NSF’s NEON program and the International Whaling Commission are
incorporating genetic studies. It is my opinion that monitoring diversity levels within and
among populations will eventually become a typical component of monitoring programs.
But the questions remain; What endpoints are appropriate to study and how should
such a program be implemented?
29
What are the advantages and disadvantages of a population genetic approach for
large-scale monitoring compared to traditional census approaches?
There is a long list of advantages or positive aspects to the population genetics
approach. The field of molecular genetics is one of the most dynamic and rapidly
advancing areas of science and technology. Our ability to gather genetics data is
rapidly increasing. When I first became interested in the mutagenic effects of
contaminants on chromosomes, I never dreamed that we would someday have the
capability of gathering DNA sequence data from natural populations. Not only can we
now gather sequence data from natural populations, we are on the verge of being able
to gather genomics data from such populations! Clearly the immense capability to
accurately quantify genetic variability, and the relatively low cost of this activity, is one of
the biggest advantages of a population genetics monitoring system.
Such monitoring systems will be data rich. In fact our ability to collect data presently
outstrips our ability to analyze it so we have to be judicious especially in how we apply
the new genomics procedures. In my lab we are presently working on methods to
apply the new genomics sequencing methods coming online at Purdue University to be
able to sequence long stretches of specific chromosomes. Mammals, for example,
have four different genetic transmission systems; mtDNA (strict maternal inheritance),
the Y-chromosome (strict paternal inheritance), the X-chromosome (sex linkage with
hemizygosity in the male), and autosomes (bi-parental inheritance). Each of these
systems has different advantages for population genetic studies, including different
expected evolutionary rates and population genetics patterns. Whereas the mtDNA has
been extensively applied for more than 2 decades in mammalian population genetics,
Y-chromosome markers have only been applied in humans (Rosser et al., 2000) and a
few other mammalian species (Cathey et al., 1998). Because nuclear genes evolve
much slower it is necessary to have about 10X the amount of sequence data compared
to the extranuclear mtDNA. And because Y-chromosomes are subject to selective
sweeps they require even greater amounts. So, we are designing primers to allow us to
sequence 50,000 to 100,000 bp per individual of Y-chromosome, X-chromosome, and
targeted autosomal regions using a combined approach of long-range PCR
amplification and genomic sequencing. The identification of haplotypes would be done
by sequence analysis while the application to population sampling would depend upon
SNP analyses. This will allow for equivalent levels of genetic diversity to be compared
among these nuclear genes with the rapidly evolving mtDNA control region in
mammals. Such datasets would have tremendous impact on our understanding of
population and evolutionary genetics processes, and of course would be directly
applicable to monitoring programs as well.
With the population genomics approach described above, and even with more
traditional automated sequence methods, it is possible to target specific chromosomal
regions due to their transmission patterns, or specific genes because of their function.
For example, in contaminated areas where a mutagenic chemical or chemicals are of
particular interest, it might do well to target both neutral genetic markers as well as
genes with known functions that might be relevant, such as DNA repair genes. In other
30
cases, a monitoring program might be focused on detecting genetic bottleneck effects.
So, highly variable markers like the mtDNA control region, or nuclear microsatellites,
would be likely subjects. In a case where the sentinel species might be endangered
and already at low population numbers, it would make sense to include genes critical to
survival. These might include MHC genes, for example. Because environmental
contaminants have many mechanisms of toxicity, and pesticides in particular are
diverse in their effects, the list of potential target genes for study is immense. As more
and more species have their entire genomes sequenced, our ability to identify and
employ relevant target genes will continue to increase. This is an important aspect of
why the use of population genetics in monitoring programs makes good sense.
As I mentioned above, pesticides have multiple mechanisms of toxicity. One of
the advantages of genetics is that many of the population genetics endpoints to be
monitored, such as diversity levels, are not specific to any particular mode of toxicity. In
other words, a wide variety of toxic effects at the somatic level will lead to similar
population genetic effects such as diversity loss. This is both an advantage and a
disadvantage, depending upon the objectives of a monitoring program. Population
genetics won’t tell us which toxin within a complex mixture causes an observed effect.
Rather it will integrate effects both through time (generations) and space (gene flow).
Detailed questions regarding the relative contributions of contaminants to an observed
effect in a complex environmental setting require the use of multiple levels of
investigation including environmental chemistry, biomarkers, and genetics. Ecological
or demographic data round out the picture and are generally an indispensable part of
any monitoring program.
31
What information is likely to be obtained using a genetic monitoring approach
that would not be available from traditional monitoring approaches?
Of course traditional monitoring programs tell us nothing about genetic
diversity within species. The reason such diversity estimates are important in
monitoring is that there is a well established correlation between diversity loss and the
probability of extinction of a species or the extirpation of a population. The conservation
of genetic diversity within species has become a cornerstone of conservation biology.
However, genetics data also can be used to infer ecological processes in a
much more efficient way than can be directly observed by the methods of traditional
monitoring. In a study of marsh frogs from the highly contaminated city of Sumgayit,
Azerbaijan, Matson et al. (2006) used mtDNA sequence data to estimate levels of gene
flow on a regional scale. Those authors showed that the predominant direction of gene
flow was into Sumgayit, both from the north and the south. In other words, genetic
markers common in populations to the south of Sumgayit were found within the city, as
were genetic markers common to the north of Sumgayit. This pattern is indicative of
gene flow (immigration) into the city from both directions. But the near absence of
northern markers in southern populations and southern markers in northern populations
meant that frogs from Sumgayit do not emigrate in either direction. Thus, the city
represents an ecological sink, with marsh frogs unable to reproduce at levels great
enough to promote emigration. The effort and cost necessary to demonstrate this using
ecological methods (mark and recapture, for example) would be immense. So, genetics
has the advantage with respect to being able to detect certain kinds of effects that result
from contaminant exposure expressed over many generations. Another example of the
process was observed by Theodorakis et al. (2001) who showed using kangaroo rats a
similar pattern of skewed gene flow into ground zero atomic-bomb test sites at the
Nevada Test Site.
In addition to gene flow patterns, a related important aspect of genetic
patterns that can be deduced from population genetic studies is the delineation of
genetically defined populations. On a large geographic scale, these can correspond to
subspecies, stocks, or management units. On a smaller geographic scale it could relate
to local populations or demes. The determination of population structure from a
geographically defined sampling design is a well established area of population genetics
and the use F statistics and related methods are well accepted. From these statistics
gene flow patterns and population subdivision can be calculated. However, we have
also pioneered the use of statistical methods to investigate demographic structure within
a migrating population of bowhead whales where geographic sampling was not possible
(Jorde et al., 2007). Thus, even when geographic sampling is not possible, for example
for a population of migrating butterflies or birds passing through a pesticide treated
region, it is possible to investigate population structure or demographic partitioning of
the population by comparisons of the genetic relatedness of individuals with the timing
of their appearance at the monitoring site.
32
Yet another important application of genetics to a monitoring program is the
use of genetics data for species identification, especially of highly bio-diverse samples.
This could include the use of genetic methods for identification of microbial community
structure (von Mering et al., 2007) but also for determining species identity in samples
of insects or other invertebrate groups where identification otherwise requires the aid of
an expert. Thus, phylogenetic methods or DNA barcoding would be used to clarify the
makeup of a taxonomically complex sample. While this is not specifically population
genetic monitoring, it is related in its use of genetics procedures.
33
What are the most appropriate genetic measures to evaluate?
An advantage of genetics data is that multiple endpoints are obtained during the
course of the investigation. In our studies of the effects of chemical contaminants on
the genetic systems of wildlife, we have found the diversity estimates (heterozygosity,
nucleotide diversity, nucleon diversity, etc.), gene flow patterns, and estimators of
population subdivision to be useful. In addition, we have found that methods for
identifying new mutations are also important for assessing genetic impacts.
So, the impacts of contaminants on genetic systems include effects on
demographic features like gene flow patterns and diversity levels. Such effects are
ultimately the result of impacts on the ability of a population to effectively reproduce.
Reduced diversity levels are the result of population bottlenecks, the severity of which
can be estimated using existing software programs. Impacts on genetic diversity
represent a hidden population impact that can only be revealed through the methods of
population genetics. For example, one might observe the populations of marsh frogs at
Sumgayit and conclude that there has been no impact because of the high numbers of
animals there. Population genetic studies show a different picture in that diversity
estimators clearly show a reduction in diversity levels. Such effects can even be
irreversible in cases where there are no variable populations to provide immigrants.
The identification of new mutations can be used to detect the direct effects of
mutagenic contaminants. Matson et al. (2006) observed new mutations in marsh frogs
at the most contaminated site in Sumgayit. New mutations are always tip haplotypes,
as determined by a phylogenetic analysis, and in the case of mtDNA they might appear
in the heteroplasmic condition. Another feature they will likely show is very restricted
geographic distribution. Using a combination of population genetics and evolutionary
genetics we have identified the wetlands adjacent to the wastewater treatment plant in
Sumgayit as a source of new mutations for both marsh frogs (Matson et al., 2006) and
mosquitofish (unpublished).
34
Are particular species, groups of species, functional guilds, etc. more informative
than others and more cost-effective for monitoring long-term ecosystem
responses to changing agricultural practices?
The selection of sentinel species is dependant upon the goals of the monitoring
program and the characteristics of the area to be monitored. One important aspect is to
identify potentially vulnerable species. For example, BP in Alaska has for approximately
25 years monitored the populations of fish around the Endicott Causeway at Prudhoe
Bay. One species, the Arctic cisco migrates from their spawning areas east of Prudhoe
Bay in the Mackenzie River system past the causeway. Tiny young-of-the-year fish are
entrained in a current of relatively fresh water that is pushed from the Mackenzie Bay
along the Beaufort Sea coast by easterly winds. The young fish must arrive at the
mouths of the Colville and Sagavanirktok Rivers, as these are the only available
overwintering habitats along Alaska’s North Slope. If the causeway were to prevent this
passive migration it could result in the disappearance of the Alaskan population. This is
a sensitive issue because this species supports a fishery for the Eskimo community at
the Colville River. We conducted population genetics studies of the species and found
that the Alaskan population is a subset of the Arctic Red River and Peel River stocks
and likely represents only a small fraction of these fish (Bickham et al., 1989; Morales et
al., 1993). There is not a unique Alaskan subpopulation of fish and thus the population
of fish is not vulnerable to extirpation by any detrimental effect the causeway might have
on the fish movements. In this case, the Arctic cisco was identified as a species of high
vulnerability due to its unique life history and genetic studies were focused on it to
answer a specific question about population structure.
The selection of sentinel species around agricultural sites might be done in a
way similar to the Arctic cisco study cited above. Fore example, migrating populations
of Neotropical migrant bird species might be of particular importance. Or, populations of
insects that are particularly important either as pollinators, predators, or for other
valuable ecosystem services might be monitored. Yet other sentinel species might be
selected for their relative sensitivity to the pesticides, or for other features of their
demography like abundance, ease of collecting, or geographic distribution.
Representatives of important ecological guilds, such as pollinators, or aquatic insects,
might also be of value.
35
How should genetic monitoring be combined with other measures, methods, and
models to achieve the most efficient assessment of current and predicted
ecosystem responses to changing agricultural practices?
In my opinion the most important impact of our changing agricultural practices will be
the effects it has on biodiversity. Biodiversity, in this case as measured by species
diversity as well as genetic diversity within species, is likely to be reduced due to
increased use of pesticides and fertilizers, new cropping practices including the use of
genetically modified crops, and the reduction of wildlife habitat as it is cleared for new
farmland. The reason I rank biodiversity loss at the top of the list of potential measures
is that it can be irreversible, and hence is extremely important to prevent, and because it
is the first domino in a potential cascade of effects that will effect detrimentally
ecosystem services. As species or populations disappear, ecosystems become more
simplified and functions are impaired. While it is important to monitor such functions
directly, genetics will aid by informing us of the initial effects of the cascade.
One important resource that we have to aid us in the study of biodiversity loss
is represented by systematics collections (Suarez and Tsutsu, 2004). Systematics
collections of insects, plants and wildlife hold the information needed to put our
monitoring efforts in a historical perspective. We can not only assess the levels of
genetic variation in sentinel species by collecting from the extant populations, but we
can assess past diversity levels by studying their representatives housed in or
museums. Because these archived specimens are useful sources of DNA, they can tell
us of historic trends in genetic variability to which can be added information from living
populations.
36
What are the key knowledge gaps and research questions that must be
addressed before designing a genetic monitoring program?
Population genetics, evolutionary genetics, and conservation genetics today
are mature sciences. There really are no key knowledge gaps preventing us from
implementing genetic monitoring programs and as I have noted above such programs
already exist. It is really only a matter of identifying the management or monitoring
objectives and then selecting the appropriate sentinel species and genetic methods
around which to design the monitoring program. In my opinion, we are ready to begin
this and we should have begun years ago!
Initially, such a monitoring program will take advantage of the current and well
established population genetics methods available. The easiest method to implement
will be mtDNA sequencing. The use of universal primers suitable for many insect,
vertebrate and even plant species and the extensive literature that exists on this marker
make this method the most low-hanging of fruit. Microsatellites are also well
established as a method to assess nuclear gene variation but loci must be developed
for each species unless sentinels for which this method has previously been developed
are chosen. SNP analysis has an advantage over microsatellites in being more easily
reproduced from lab to lab, but again this method needs extensive development and is
difficult for species in which a genome sequence is not available. And finally, targeted
genes will need to be developed for the identified sentinel species. The need for some
development of genetics protocols doesn’t really represent a knowledge gap in that we
well understand how to do the development. It really represents a technology gap. But,
the knowledge and technology gaps as they exist can be easily overcome and it is
highly desirable to begin the work postulated in this report. I applaud the EPA for
beginning to consider population genetics as a monitoring tool!
37
References Cited
Bickham, J. W. 1990. Flow cytometry as a technique to monitor the effects of
environmental genotoxins on wildlife populations. Pp. 97-108, in In Situ
Evaluations of Biological Hazards of Environmental Pollutants (S. S. Sandhu et
al., eds.). Plenum Publ. Corp., New York, 277 pp.
Bickham, J. W., S. M. Carr, B. G. Hanks, D. W. Burton, and B. J. Gallaway. 1989.
Genetic analysis of population variation in the Arctic Cisco using electrophoretic,
flow cytometric, and mitochondrial DNA restriction analyses. Biol. Pap. Univ.
Alaska 24:112-122.
Bickham, J. W., and M. J. Smolen. 1994. Somatic and heritable effects of
environmental genotoxins and the emergence of evolutionary toxicology.
Environmental Health Perspectives 102, Suppl. 12:25-28.
Cathey, J. C., J. W. Bickham, and J. C. Patton. 1998. Introgressive hybridization and
nonconcordant evolutionary history of maternal and paternal lineages in North
American deer. Evolution, 52:1224-1229.
Jorde, P. E., T. Schweder, J. W. Bickham, G. H. Givens, R. Suydam, D. Hunter, and N.
C. Stenseth. 2007. Detecting genetic structure in migrating bowhead whales off
the coast of Barrow, Alaska. Molecular Ecology 16:1993-2004.
Matson, C. W., M. M. Lambert, T. J. McDonald, R. L. Autenrieth, K. C. Donnelly, A.
Islamzadeh, D. I. Politov, and J. W. Bickham. 2006. Evolutionary Toxicology
and Population genetic effects of chronic contaminant exposure on marsh frogs
(Rana ridibunda) in Sumgayit, Azerbaijan. Environmental Health Perspectives
114:547-552.
McBee, K., J. W. Bickham, K. C. Donnelly, and K. W. Brown. 1987. Chromosomal
aberrations in native small mammals (Peromyscus leucopus and Sigmodon
hispidus) at a petrochemical waste disposal site. I. Standard karyology. Arch.
Environ. Contam. Toxicol., l6:681-688.
Morales, J. C., B. G. Hanks, J. W. Bickham, J. N. Derr, and B. J. Gallaway. 1993.
Genetic analysis of population structure in Arctic cisco (Coregonus autumnalis)
from the Beaufort Sea. Copeia 1993:863-867.
Rosser, Z. H., et al. 2000. Y-Chromosomal Diversity in Europe Is Clinal and Influenced
Primarily by Geography, Rather than by Language. American Journal of Human
Genetics, 67:1526–1543.
Suarez, A. V., and N. D. Tsutsui. 2004. The Value of Museum Collections for Research
and Society. Bioscience 54:66-74.
Theodorakis, C. W., J. W. Bickham, T. Lamb, P. A. Medica, and T. B. Lyne. 2001.
Integration of genotoxicity and population genetic analyses in kangaroo rats
(Dipodomys merriami) exposed to radionuclide contamination at the Nevada Test
Site. Environmental Toxicology and Chemistry 20:317-326.
38
von Mering, C., P. Hugenholtz, J. Raes, S. G. Tringe, T. Doerks, L. J. Jensen, N. Ward,
and P. Bork. 2007. Quantitative Phylogenetic Assessment of Microbial Communities in
Diverse Environments. Science 315:1126-1130.
39
APPENDIX A-2
Response from Dr. Andrew J. Bohonak and Dr. Amy Vandergast
40
Genetically modified crops and pesticides in agricultural systems:
Population genetic approaches for monitoring impacts to non-target arthropods
Andrew J. Bohonak, Associate Professor and Coordinator for Evolutionary Biology
San Diego State University
Amy G. Vandergast, Geneticist
Western Ecological Research Center, San Diego Field Station, U.S. Geological Survey
Report for the Environmental Protection Agency: August 15, 2008
Scope
This report focuses on the use of population genetic approaches for monitoring impacts to
non-target arthropods in agricultural systems where pesticides and GM crops are used. We
specifically address:
 General issues surrounding the development and design of a population genetics monitoring
program
 Delineation of population boundaries
 Estimation of effective population size and comparison with traditional estimates of census
population size
 Quantifying population differentiation and estimating gene flow
 Choice of molecular markers
 Choice of taxa
 Although outside our areas of expertise, we also provide a review of known GM effects on
non-target arthropods.
I. Background
Genetically modified (GM) crops that express insecticidal proteins offer an alternative to
traditional pesticide use that may impart some environmental and health benefits. Since their
commercialization over 10 years ago, GM crops have been used by farmers in dozens of
countries, primarily to control lepidopteran and coleopteran pests (James 2003, Thies and Devare
2007). Commonly, insect resistant GM crops express Bacillus thuringiensis (Bt) deltaendotoxins, although a wide variety of other insecticidal compounds are currently being
developed (O'Callaghan et al. 2005 and references therein). A large number of laboratory and
field studies have been conducted on both potential and actual responses of non-target organisms
to Bt crops worldwide. (A simple search of {{Bt OR “Bacillus thuringiensis”} AND {nontarget
OR “non-target”} yielded 918 references in the ISI Web of Knowledge during August 2008.)
41
It is estimated that 52.6 million hectares of transgenic crops were planted globally as of
2001, 99% of which were in the United States, Canada, China or Argentina (James 2003). Insect
resistance (through expression of Bt endotoxins) is the second most commonly used trait in
commercial GM crops, after herbicide resistance (James 2003). A broader spectrum of
insecticidal GM crops are being developed or investigated, including plants that express protease
inhibitors, lectins, biotin-binding proteins, toxins from bacterial symbionts of entomopathogenic
nematodes, chitinases, spider venom peptides, plant defensins and hormones (reviewed in
O'Callaghan et al. 2005).
The benefits of using insect resistant GM crops are thought to include a reduction in the
amount of pesticides used to control pests, more effective control of some pests (because the
insecticides are expressed in all tissues at all times), a reduction in yield loss due to pests, and a
reduction in the secondary micotoxin infection of crops that can occur when plants are damaged
by pests (Hails 2000, Pimentel and Raven 2000, Thies and Devare 2007). All of these can have
positive impacts on human and environmental health. However, several potential risks of GM
crop use have also been identified. These include introgression of transgenes to wild relatives,
the evolution of insecticide resistance in pests, and the impacts to non-target species (e.g., Hails
2000, Wolfenbarger and Phifer 2000, Dale et al. 2002). The latter two concerns also exist with
applied pesticides. Accordingly, there have been calls for governmental regulation, risk
assessment and research to assess the long term ecosystem effects of the continually growing use
of transgenic crops (e.g., Snow et al. 2005, Hill and Sendashonga 2006).
Ultimately, the impacts of pesticide use and GM crops on non-target organisms will be
determined by a combination of laboratory toxicity tests, mesocosm experiments, short-term
field experiments, and long-term monitoring programs. In addition to the moral and aesthetic
significance of maintaining biodiversity, these impacts have practical benefits for pest
management, since non-target organisms include natural predators, parasites and pathogens.
Further, the sustainability of agroecological systems and the level of ecosystem services that they
provide (e.g., Tilman et al. 2002) depend on the maintenance of biodiversity.
The costs of any monitoring program must be balanced against the novel information that
it provides. The goals can range from simple presence-absence surveys, to quantitative studies
estimating treatment effects, to precise, repeated estimates of abundance that may be needed for
establishing temporal trends or performing a population viability analysis (PVA). As we discuss
below, genetic techniques for long-term monitoring of non-target arthropods in an agricultural
setting would be most cost effective (relative to information content) when the target species
have highly variable population sizes, and can be sampled and identified by personnel that do not
possess highly specialized taxonomic training.
42
II. Introduction to population genetics
The most basic tenets of population genetics hold that patterns of genetic divergence
among populations and genetic diversity within populations are the product of four basic
microevolutionary forces:
1) Random drift (stochastic sampling error during gamete and offspring production)
2) Gene flow (dispersal and the subsequent incorporation of gene copies into the new gene
pool)
3) Mutation (creation of new alleles)
4) Natural selection (for genes that impact fitness, or those that are closely linked)
Genetic diversity at the individual level depends on these four factors, as well as:
5) Patterns of mating within the gene pool (random, dependent on phenotypic similarity, or
dependent on relatedness)
When one of these five factors changes over time, patterns of genetic diversity and divergence
also need to be interpreted in terms of:
6) Nonequilibrium conditions
Nonequilibrium conditions are important in many (and possibly most) species for some
ecological and evolutionary questions of interest, since both natural and human-dominated
landscapes are dynamic. The biases imposed by nonequilibrium conditions and the time required
to reach equilibrium vary widely, depending on the summary statistic or analysis (e.g., see
Bohonak and Roderick 2001).
Population genetics traditionally focused on Mendelian markers with discrete states,
frequency-based analyses, and qualitative differences among alleles. With the widespread
availability of DNA sequencing, faster computers and new analytical approaches, the population
geneticist’s toolbox is now considerably more complex. As discussed below, traditional analyses
such as contingency table tests and Wright’s (1931) F-statistics still hold a valuable place in
analyses of population structure and gene flow. However, many current studies focus on
recently diverged populations or highly connected landscapes using model-based estimates of
parameters such as gene flow, and individual-based algorithms for inferring gene pool
membership (e.g., Pritchard et al. 2000, Nielsen and Wakeley 2001).
In addition to a random sample of (presumably neutral) genes from the entire genome, a
genetic monitoring program for non-target organisms might also include traits under direct
selection (e.g., for pesticide resistance, behavioral modifications, life cycle timing). Selected
traits could be monitored using a variety of techniques, such as traditional quantitative genetics
breeding experiments, artificial selection studies, QTL mapping of the traits of interest, and
molecular studies of single-locus traits (Falconer 1989, Storfer 1996, Andreev et al. 1999, Burt
2002, Conner 2003). Although these would provide valuable additional information, it is our
understanding that such studies fall outside the scope of this invited review.
43
III. Population genetic parameters of interest for conservation, management and monitoring
The five primary genetic forces listed in the previous section all impact patterns of
genetic diversity in natural populations. However, 3) mutation is a relatively slow process, 4)
natural selection on traits related to pest management falls outside the bounds of this review, and
there is no a priori reason to believe that 5) mating patterns would be affected by pesticides or
GM crops. Thus, the parameters of interest are 1) effective population size Ne (which
determines drift) and 2) gene flow m. (Note that mutation is also considered in many approaches
to estimating these two parameters.) GM crops or application of pesticides could impact nontarget species by reducing the census population size (and therefore Ne), impacting movement
ability (quantified as m), or creating incipient “races” that are specially adapted to agricultural
settings or surrounding natural populations (effectively reducing m between gene pools in
differing habitat types).
In addition to quantifying the population parameters Ne and m, a monitoring program
should quantify population boundaries and estimate intrapopulation genetic diversity.
Conservation of adequate genetic variability ultimately facilitates evolutionary processes that
include adaptation to changing environments (Frankel 1974, Frankham 1995).
A. Preliminary analyses: defining population boundaries
The central issue of defining population (gene pool) boundaries necessarily precedes the
issue of estimating specific population parameters. Although some questions in population
genetics do not require that individuals be grouped prior to analysis (e.g., standard
phylogeography, NCA, paternity analysis), those questions do not apply to the monitoring
program of interest here. The term “population” is admittedly vague, and used in a variety of
ways in different contexts (briefly reviewed by Waples and Gaggiotti 2006). Throughout this
report, we consider a population to be equivalent to a gene pool, defined classically the set of
individuals that interbreed within a given area, where spatial position does not affect mate
choice. (Note that mate choice need not be random with respect to all factors, however.) When
species are distributed continuously rather than in discrete sites or patches, the boundaries of a
population or gene pool are only approximations.
In many situations, collection sites are predetermined by habitat patchiness and collection
protocols. When these sites are smaller than or equal to the population boundaries, both Hudson
et al. (1992) and Waples and Gaggiotti (2006) found that contingency table tests of {site x allele}
outperform more complicated methods of population assignment. Simulated data sets in the
Waples and Gaggiotti (2006) study provide some guidance for sampling effort to be employed.
When gene flow was high (25 individuals immigrating to each site per generation), statistical
power was extremely low if the sampling strategy included L = 10 low mutation loci
(independent genes) and n = 25 individuals per site. In contrast, panmixia was rejected 100% of
the time when L = 20 high mutation loci (e.g., microsatellites) and n = 50. A robust monitoring
protocol would meet or exceed these guidelines.
44
Alternative methods for assigning population boundaries (and indirectly making
inferences about gene flow) include analyses of F-statistics (between all pairs of sampling sites
or populations), and so-called “assignment tests”. Both should be included in a monitoring study
because they provide additional information about population structure, although Waples and
Gaggiotti (2006) found each to be somewhat less reliable than contingency tests for the narrow
question of “what is a population”. STRUCTURE (Pritchard et al. 2000) is the most widely used
assignment test, although Waples and Gaggioti found increased power with the earlier test of
Rannala and Mountain (1997). A variety of assignment tests are continually being developed; all
share the common goal of clustering individuals into populations based on individual multilocus
profiles, and the assumption of random mating. For example, the newer program of Chen et al.
(2007) claims to perform better than STRUCTURE, and other programs are available that relax
the random mating assumption (Wilson and Rannala 2003), or use information about the spatial
location of individuals (Guillot et al. 2005). Because each program has a unique algorithm and
set of underlying assumptions, they may provide different results (pers. obs.). We would
recommend that data be analyzed using STRUCTURE and at least one other program in this
genre.
B. Estimate effective population size Ne
Genetic drift consists of changes in allele frequencies due to sampling error. In any
population of finite size, this sampling error will cause gene frequencies to fluctuate from
generation to generation, with effects that are inversely proportional to the effective population
size, Ne. Effective population size is a mathematical abstraction used to correct for factors such
as unequal sex ratio, variance among individuals in reproductive output, and temporal
fluctuations in population size (Ewens 1982). More precisely, Ne is the number of individuals
that would be found in an ideal population that undergoes the same amount of drift as the real
population. (More technical definitions for Ne exist, but will not be considered in detail here.)
Ne is an important parameter to monitor for two reasons:
1) Ne is tightly linked to the loss of genetic variation over time. Populations that undergo
bottlenecks of reduced Ne and subsequently lose genetic variation may be less able to adapt
to changing environments, more likely to become fixed for maladaptive traits, and more
vulnerable to inbreeding effects, disease and extinction (e.g., O'Brien et al. 1985, O'Brien and
Evermann 1988, Frankham 1995, Frankham et al. 1999, Boakes et al. 2007). (Note,
however, that the relationship between population size and heritability in quantitative traits is
weak in most cases.)
2) Effective population size is a surrogate for the census population size N, which may be
difficult or expensive to monitor directly. Although Ne may provide a relative estimate of N
(to be compared over time, for example), it should not be considered a precise or unbiased
estimate of N. Based primarily on theory, Nunney (1991, 1993) suggested that the ratio of
Ne / N should be approximately 0.5, and rarely fall below 0.25. However, empirical data
show that this ratio varies widely among species, with a median value of only 0.11 - 0.14
45
(Frankham 1995, Palstra and Ruzzante 2008). In a particularly dramatic example, a heavily
fished population of New Zealand snapper with spawning stock numbers in the millions was
estimated to have an effective population size of less than 200 (Hauser et al. 2002).
For monitoring insects and other arthropods, the range of field techniques includes sweep
netting, pitfall traps, light traps, sticky cards and lures (pheromone or otherwise). Specific field
methods for monitoring census population size N in agricultural system insects and associated
statistical considerations are discussed in Duelli (1999), among other places, and are outside the
scope of this review. Methods for estimating Ne can be broadly categorized as follows (see
reviews by Waples 1989, Caballero 1994, Blackwell et al. 1995, Schwartz et al. 1998, Leberg
2005, Waples and Yokota 2007, Palstra and Ruzzante 2008):
1) Estimates that derive from calculation of the inbreeding coefficient ƒ from pedigree analyses,
or knowledge of a regular mating system.
2) Estimates that derive from ecological data: census population size, adjusted for unequal sex
ratio, variance in reproductive output, etc.
3) Estimates that derive from a snapshot of contemporary genetic diversity levels, including the
diversity statistics mentioned above.
4) Estimates that derive from time series analyses.
Only the final two would be appropriate for a program monitoring non-target arthropods in
agricultural systems. Because the program would presumably be conducted over many years,
the advantages of analyzing genetic data across the time series would be particularly high.
Estimates of Ne vs. N have unique advantages and limitations. Estimates of Ne more
clearly correspond to evolutionary parameters, while estimates of census population size would
more directly reflect ecological processes such as births and deaths, predation and competition.
Temporal trends in N might be easier to detect quickly than in Ne if error rates in field sampling
were small and had previously been quantified. On the other hand, temporal trends in Ne may be
easier to detect in systems where variability in population size is very high from generation to
generation, since effort to obtain a sufficient sample size for genetics (say, 30-100 individuals)
may be far less than that required in an ecological study. The greatest inferential power is gained
from monitoring both population parameters in parallel, since they reflect different processes.
Even when reliable estimates of N are too costly or time-consuming to obtain, estimates of catch
per unit effort are still recommended to help validate trends in Ne.
Any sampling scheme focused on estimates of Ne should be mindful of model
assumptions, which nearly always include a closed population (i.e., a single isolated gene pool).
Thus, as discussed above, properly defining population boundaries is an essential precursor to
estimating Ne from the data set. Models that assume discrete generations are often assumed,
which may be true for many non-target arthropods of interest. Sample sizes should be at least 30
per population, and preferably greater than 50. This would probably not be a problem under the
guidelines listed above (n = 50 individuals per site, and probably multiple sites per gene pool).
46
Further guidance on specific analyses should be obtained from the most recent reviews on this
topic (Waples and Yokota 2007, Palstra and Ruzzante 2008).
C. Estimate intrapopulation variation
The genetic data that would be gathered in a monitoring program should be analyzed
using standard summary statistics, to track trends over time and compare among populations.
These include
 He (expected heterozygosity; nearly identical statistics are also called “gene diversity”)
 Ho (observed heterozygosity; can only be calculated for codominant markers)
 A (the number of unique alleles in a sample; can be adjusted for sample size using a
rarefaction method)
For DNA sequence data, additional statistics include
  (the average difference in base pairs between any two randomly selected sequences)
 S (the number of segregating sites, or polymorphic base pair positions)
These are easily calculated using publicly available software packages.
More in depth analyses might focus on changes in genetic diversity patterns over time to
detect population bottlenecks (decreases in Ne). There are a number of statistical tests and other
approaches for studying bottlenecks (e.g., Rogers and Harpending 1992, Cornuet and Luikart
1996, Drummond et al. 2005). However, we question whether these would be very useful in a
population monitoring program, because non-target populations are unlikely to be in a drift-gene
flow equilibrium at the start of the program, and because the time lag between decreasing
population size and statistically detectable changes in population diversity may be too great for
management action. Bottlenecks severe enough to quickly impact genetic diversity would be
noticed more rapidly with a traditional ecological census monitoring program. A general rule of
thumb for “severe enough” might be Ne < 30 individuals for a short time, or Ne < 100 for many
generations.
D. Quantify population differentiation / estimate gene flow
In addition to changes in population boundaries and in effective population size, Bt crops
and pesticides may impact non-target species by decreasing gene flow among populations and
increasing differentiation. Such an effect would probably occur in combination with a reduction
in effective population size. However, it is theoretically possible that gene flow could be
reduced without lowering Ne. For example, pesticide use might impact the dispersive life history
stage most heavily, while density-dependent mortality keeps the local population size
approximately constant.
Population subdivision has classically been estimated as the genetic distance FST (Wright
1931), with more recent studies analyzing “isolation by distance” in scatterplots of genetic
distance vs. geographic distance for all possible pairs of sites or populations (Slatkin 1993,
47
Peterson and Denno 1998). Modified distances that accommodate DNA sequence or
microsatellite data (Excoffier et al. 1992, Slatkin 1995) are easily calculated using any number of
software applications (Jensen et al. 2005, Excoffier et al. 2006, Rousset 2008). Statistical power
in these analyses (particularly the isolation by distance analyses) is gained by increased numbers
of sampling sites more than in the analyses mentioned above, although a high number of loci
(preferably > 10) allows for error in FST to be estimated from bootstrapping over loci (Weir
1996).
It is possible that movement of individuals per se would be a variable of interest, rather
than simply the amount of population differentiation. If so, then estimation of dispersal and/or
gene flow could be conducted using a variety of methods. Each has its own advantages.
Dispersal estimates obtained ecologically (e.g., from mark and recapture studies) are limited by
the spatial and temporal scale of the study, and the considerable effort involved with arthropods.
However, the movement of individuals is ultimately the variable of interest for many questions
in population ecology. In contrast, gene flow impacts a species’ genetic structure over tens,
hundreds or even thousands of generations. Thus, traditional gene flow estimates (e.g., from
FST) may reflect rare but significant dispersal events that go undetected in an ecological study
(stressed by Slatkin 1985, 1994). It is impossible to know whether these events are no longer
occurring (and gene frequencies are no longer in equilibrium), or simply unobserved. For this
reason and others, great care must be taken when estimating gene flow using most available
methods including FST and its derivatives, and coalescent approaches (Bossart and Pashley
Prowell 1998, Whitlock and McCauley 1999, Bohonak and Roderick 2001, Slatkin 2005, but see
Bohonak et al 1998, Neigel 2002). We would recommend that exploratory analyses with
isolation by distance plots (Jensen et al. 2005) and hierarchical AMOVA (Excoffier et al. 2006)
be conducted to obtain qualitative inferences about patterns of gene flow. However, coalescent
approaches to estimating gene flow (Beerli and Felsenstein 2001, Nielsen and Wakeley 2001)
inherently average over time scales that are longer than frequency-based approaches, and would
probably be misleading in human-dominated agricultural landscapes.
At least two alternatives to indirect frequency- or coalescent-based estimates of gene flow
are available. First, one could estimate gene flow directly in a short-term study by releasing
genetically unique members of the species into the field and watching their genes spread in
subsequent generations (e.g., Grosberg 1991). Such an experiment would be outside the scope of
a typical monitoring program. Alternatively, assignment tests (reviewed above) can be used to
infer short-term gene flow by studying the mismatch between gene pool assignments for each
individual, and the locations in which they were sampled. (This is conceptually similar to spatial
analyses of parentage, such as Burczyk et al. 2006). The multilocus linkage disequilibrium
created by immigrants should persist only one to a few generations if the immigrant and its
descendants mate randomly; thus, assignment tests provide a short-term estimate of individual
movement (Waser and Strobeck 1998). The statistical power of assignment tests is proportional
to the amount of differentiation among gene pools. For highly connected populations in
48
relatively small agricultural landscapes, low signal would need to be countered with a large
number of highly polymorphic loci (as discussed in the previous section).
IV. Choice of molecular markers
A. Organellar markers
1. mtDNA sequencing
Mitochondrial DNA has been the workhorse for arthropod systematics and
population genetics for the past 20 years (see early review by Simon et al. 1994). The
mitochondrion exists in multiple copies per cell, is almost entirely protein-coding,
and has been very well characterized for all major taxa. In insects and other
arthropods, genes such as cytochrome oxidase I (CO I) are nearly always
polymorphic within populations, leading to their continued use in (we guess) well
over half of recently published studies. Although the mtDNA genome provides only
one large, linked locus, mtDNA is often paired with microsatellites or dominant
markers to provide information from a qualitatively different kind of marker.
Research and development costs for mtDNA sequencing in novel arthropods are close
to zero, and the per-sample operating cost for PCR and sequencing is very low (< $10
and quite possibly < $5, depending on volume and labor costs). Due to its ease of use
and the wide array of published studies to compare with, we would recommend that
mtDNA sequencing be paired with another class of markers in a monitoring program.
Population geneticists are for the most part acutely aware of the limitations of
using mtDNA, including its maternal inheritance, the possibility for paralogy from
nuclear copies of mitochondrial genes, the possibility of natural selection, and the fact
that only one gene cannot be used to accurately characterize the stochastic processes
of drift, mutation and gene flow. Nonetheless, mtDNA continues to hold an
important place in the field, particularly for inferences about recent population history
(e.g., see defense by Zink and Barrowclough 2008). So-called "DNA barcoding"
(Hebert et al. 2003) uses mtDNA genes (often CO I), although the focus in those
studies tends to be the rapid identification of taxonomic groups by nonspecialists.
B. Codominant markers
Codominant markers are those in which heterozygotes can be scored directly, having a
different banding pattern or chromatogram than homozygotes. This provides direct information
about individual heterozygosity, which can be interpreted in terms of natural selection or mating
patterns within populations. They are expected to provide more accurate estimates of allele
frequencies than dominant markers. As a corollary, codominant markers can be analyzed in
assignment tests with more information and fewer assumptions than dominant markers.
49
1. Allozymes
Due to their relatively low levels of variation, the subjective nature of allele
scoring, and the availability of more accurate molecular markers for comparable
costs, we would not recommend that allozymes be used in a population genetics
monitoring program.
2. Microsatellites
The molecular markers of choice in most conservation- and managementbased population genetics studies are microsatellites (reviewed by numerous authors,
e.g., Haig 1998, Loxdale and Lushai 1998, Balloux and Lugon-Moulin 2002).
Microsatellites are usually highly variable, presumably provide a random sample of
the genome, and can be developed for most new animal species using standardized
techniques. Typically, 8 or more microsatellite markers are used in a population
genetics study, and it is not uncommon for more than 15 to be used. Costs may run
from $15-$50 per sample, depending on the lab, labor costs and number of loci.
The limitations of microsatellites include the price of developing them in new
species. Private companies can provide such services at ≈ $12,500 for 15 loci with a
delivery time of 3-4 months. Academic labs that specialize in developing
microsatellites may be able to accomplish the same goals for half of these costs or
less, depending on the amount of experience and whether the work is performed by a
lab technician or an unpaid graduate student. However, the time required for
relatively untrained personnel (such as grad students) may be much longer (6-12
months). It is highly recommended that breeding experiments be conducted, and that
parents and offspring be analyzed for Mendelian inheritance at all loci. If this cannot
be done, it may still be possible to collect gravid inseminated females from the field,
and analyze mothers and offspring.
During the data collection phase, a great attention to detail is required to
achieve reproducible results (more than mtDNA sequencing, for example), with
numerous re-runs and controls that help to establish the genotyping error rate. (Note
that published papers that use microsatellites usually fail to adequately document
such efforts.) The use of museum specimens or noninvasive samples (e.g., hair, scat)
requires more extraordinary measures to assure accurate genotyping (reviewed by
Frantzen et al. 1998, Bonin et al. 2004, Broquet et al. 2007, and others).
3. Sequencing of single copy nuclear genes
The information content of DNA sequences exceeds that of microsatellites,
and facilitates a more accurate reconstruction of the mutational history of alleles in
lineages. For the same number of loci, intuition says that DNA sequence data should
outperform microsatellites in coalescent-based analyses, provided that they contain
50
comparable levels of variation. Even for allele-based analyses (where alleles are
characterized at the haplotypic level), one would expect less homoplasy in DNA
sequences than in microsatellites, provided the sequence length and level of variation
is sufficient. Rapid and relatively inexpensive analyses of nDNA sequences are
likely where the field is headed. However, per-sample costs are highly variable at
this time, and might range between $20 and $100 depending on the number of loci,
labor costs, etc.
Unique technical hurdles are necessary to develop and analyze nDNA markers
for use at the individual and population levels. Some single copy protein-coding
genes have been well characterized for specific arthropod groups and subsequently
used in population genetics studies (e.g., Villablanca et al. 1998, Bohonak et al.
2001). Also widely used are "universal" primers for some genes that sit in slowly
evolving areas of exons and amplify across introns (Palumbi 1996). However,
multiple copies of these genes sometimes exist, creating the potential for paralogous
copies to be sequenced. If multiple copies are recognized, each can sometimes be
targeted with a unique set of primers and analyzed separately (e.g., Mun et al. 2003).
Unlike nDNA sequencing for systematic studies, the biggest challenge for
population-level analyses is how to score phase in heterozygous individuals. To
obtain variation comparable to microsatellites, most or all individuals would be
heterozygous in a population study. Assuming more than one polymorphic site,
direct sequencing of PCR products does not allow one to determine whether, for
example, the "A" or "G" at one position pairs with the "C" or "T" in another.
Researchers have a number of indirect methods for inferring the underlying
haplotypes (including simple algorithms and computational approaches), but the best
information comes from separating the two gene copies in a heterozygous individual
through cloning or allele-specific primers. This adds to the time, costs and technical
expertise required for genotyping.
Due to the challenges involved with developing DNA sequencing protocol for
nuclear genes, most published population genetics studies use only 1-4 loci with
varying amounts of heterozygosity. In our view, it would be advantageous to instead
have 20 or more microsatellites in a long-term monitoring study, although some
limited efforts could be put into developing nDNA markers for specific genes of
interest.
C. Fragment-based anonymous dominant markers
Dominant markers are typically used in population genetics when the time and resources
are not available for microsatellite development. Their advantages include low protocol
development cost, low per-sample cost (≈ $10-$30 per sample), and access to dozens or even
hundreds of (presumably independent) markers. However, dominant markers are limited by low
51
repeatability across laboratories, or even in the same laboratory over different studies and
personnel. Also, they do not provide information about individual heterozygosity, since
homozygous dominant individuals display the same banding pattern phenotype as heterozygous
individuals. Thus, most population genetic analyses that use dominant markers assume random
mating (i.e., no inbreeding or outbreeding: mate choice does not depend on relatedness). At least
one recent program relaxes this assumption (Holsinger et al. 2002), which we suggest should be
included in any analyses of dominant markers.
If dominant markers (likely AFLPs or ISSRs) are to be used in a long-term
monitoring program, extreme care must be taken to assure consistent DNA quality for all
samples, continuity of laboratory personnel training, and the use of multiple positive
controls over time. The potential for low repeatability with dominant markers would
raise concern with any independent scientific reviewers. Scoring is most reliable when
conducted on a DNA sequencer, with size bins of 2 bp under the best conditions (pers.
obs.).
1. RAPDs (Random Amplified of Polymorphic DNA)
RAPD genotyping uses random primers that may anneal anywhere in the
genome. If small enough, DNA fragments between two primers are amplified by
PCR, visualized on a gel, and then scored by size. PCR conditions must be precise
enough so that a single base pair mismatch with the primer results in no product.
(Otherwise, bands of varying intensities may be present during scoring, with no easy
way to determine a threshold.) Due to notoriously low repeatability, very few
population geneticists still use RAPDs, and the field's premier journal Molecular
Ecology stopped publishing papers with these markers long ago.
2. AFLPs (Amplified Fragment Length Polymorphism)
AFLPs are more reliable than RAPDs, and have become the most common
dominant marker technique in population genetics (reviewed by Mueller and
Wolfenbarger 1999). Little preliminary protocol development may be necessary for a
researcher who has AFLP experience with other organisms. Briefly, whole
organismal DNA is digested with restriction enzymes, and adaptors (known
oligonucleotide fragments) are ligated to the ends. PCR primers match the adaptors
and extend into the sequence a few random base pairs, to reduce the number of
amplified fragments to a manageable number.
3. ISSRs (Inter Simple Sequence Repeats)
ISSRs (Gupta et al. 1994, Zietkiewics et al. 1994) are similar in some respects
to both RAPDs and microsatellites. One random primer is used to amplify random
DNA fragments, except that its content is a simple sequence repeat (e.g.,
ATATATATAT...) with one or a few different base pairs at the end from which
52
extension begins. Thus, the primer is anchored in a repeat region, amplifying
between adjacent repeats. Reliability is generally expected to meet that of AFLPs.
Although used primarily in plants, we have applied ISSRs to an insect and a
vertebrate in previous work (Lewallen and Bohonak 2005, Vandergast et al. in press).
D. Nucleotide-based dominant markers
1. SNPs (Single Nucleotide Polymorphisms)
SNP is a generic term for single base pair positions that are polymorphic
within the group of interest. SNPs are widely screened in human medical genetics
using a variety of genotyping methods that include DNA microarrays (some can assay
over 500,000 SNPs). Outside of humans and a few other model organisms, their
application to population genetics has been limited. This is likely due to the high
initial costs associated with high-throughput SNP screening, and the problem with
assigning phase to diploid individuals (see discussion under B3 above). The focus on
phase estimation has taken center stage within the field of human genetics, with the
"HapMap project" focused on moving past simple base pair scoring to reconstructing
entire haplotypes.
In our opinion, the use of high throughput SNPs (e.g., microarrays) in
monitoring projects such as that considered here may not be worth the initial
monetary investment, unless one wished to include both neutral markers and a suite
of genes likely to undergo natural selection (e.g., insecticide resistance).
V. Choice of taxa
In previous studies, the choice of which species to monitor or test for impacts of
transgenic crops and pesticide use has been based on criteria such as abundance, ease of
handling in the laboratory, taxonomic certainty, value to the agroecosystem (e.g.,
pollinators, natural enemies), and endangered status (O'Callaghan et al. 2005). Schmitz
et al. (2003) proposed a risk index for selecting taxa consisting of four criteria:
1) temporal and spatial coincidence to exposure, 2) feeding mode, 3) susceptibility to the
toxin under consideration, 4) rare or endangered status. Prasifka et al. (2008) further
advocate the use of power analyses to help select target taxa (and appropriate collection
methods, number of replications etc), that are likely be able to detect changes in response
variables such as abundance. The spatial and temporal scale of the study should also be
considered when selecting indicator taxa (McGeoch 1998).
A large number of studies to date have examined the effects of Bt and other
transgenic insecticidal crops on non-target organisms (mainly insects). These have been
conducted in laboratory and field settings, farm-scale evaluations and commercial crop
monitoring, with a wide variety of non-target organisms examined and response factors
ranging from individual mortality and growth to changes in taxon abundance. Multiple
53
comprehensive reviews are now available for this body of research (e.g., Wolfenbarger
and Phifer 2000, Obrycki et al. 2001, O'Callaghan et al. 2005, Marvier et al. 2007, Thies
and Devare 2007). Most of these reviews have found variable effects, depending on the
particular species studied, crop system, and GM strain. For example, a recent metaanalysis of 42 field experiments determined that non-target invertebrates were more
abundant in lepidopteran resistant Bt cotton and maize than in non-GM cotton and maize
sprayed with insecticides, but less abundant than in non-GM insecticide free fields
(Marvier et al. 2007). Reductions in abundance were most significant for non-target
lepidopterans and hymenopterans. In contrast, coleopteran resistant Bt maize appears to
have no statistically detectable affects on the abundance of non-target invertebrates when
compared to non-GM fields with or without pesticide application (Marvier et al. 2007).
Particular species or functional groups of non-target invertebrates may be considered
important to monitor because of their intrinsic value as ecosystem service providers (such as
pollinators and natural enemies), or due to aesthetic or conservation value. As with other taxa,
non-target effects on these groups vary across taxonomic groupings and particular strains of GM
crop. For example, honey bees have been widely studied due to their importance as pollinators
in both agricultural and natural systems. Generally, Bt pollen has been shown to have little or no
effect on honey bees, although some negative effects of protease inhibitors have been
documented at high dosages (O'Callaghan et al. 2005). For natural enemies of herbivorous
insects, O'Callaghan’s (2005) review suggested that GM effects depend on the potential for the
natural enemy to be exposed to the insecticidal protein expressed, its susceptibility to that
particular protein, and its reliance on susceptible herbivorous prey. For example, Bt plants
insecticidal to Lepidoptera are most likely to negatively impact non-target Lepidoptera
(O'Callaghan et al. 2005, Marvier et al. 2007), but may not affect other taxa of natural enemies
(e.g., Dutton et al. 2002). However, natural enemies could be impacted in sublethal ways by
consuming suboptimal Bt susceptible prey (Raps et al. 2001, Dutton et al. 2002). Additionally,
some predators show an aversion to prey raised on Bt in choice experiments (Meier and Hilbeck
2001). These results underscore the complexity of ecological and trophic effects that can occur
in field settings.
As with ecological monitoring, genetic monitoring should cover a range of taxa that span
different taxonomic groups, functional roles, susceptibility to the insecticide(s) under study, and
perceived value. However, genetic studies must be conducted on individual species, whereas
ecological studies are often conducted at higher taxonomic or functional group levels (e.g.,
Lepidoptera, parasitic Hymenoptera). This restricts the choice of taxa for genetic monitoring to
species with well-known taxonomic status that are easily identified 1) in the field, 2) in the
laboratory by readily available taxonomic specialists, or 3) using genetic tests such as DNA
barcoding that have already been validated.
Other considerations in choosing appropriate species include generation time, average
population size, and dispersal ability. For impacts such as isolation or severe population size
54
reduction, species with short generation times will accumulate genetic differences more rapidly.
This may be the most important criteria to consider, because the time scale for microevolution
generally exceeds that for human impacts. For genetic methods to be informative, the spatial
scale of the population (range, size and dispersal ability) must appropriately match the scale of
the study. For example, if the range of a single panmictic population exceeds the treatment plots
under investigation, effects are likely to be diluted over a larger geographic area. It is also
possible that very small, recently introduced or frequently bottlenecked populations will contain
too little genetic variation for meaningful analysis. (This may only be known after a preliminary
study is conducted.) Conversely, impacts to an extremely large population may not be detectable
using genetic techniques unless they are severe. Dispersal ability will also be correlated with the
amount of gene flow among sampled locations (Peterson and Denno 1998). For highly mobile
species, local changes to genetic diversity and structure may be swamped by gene flow. An
effective monitoring strategy would consider all of these factors, targeting four or more species
with different attributes.
VI. Synthesis
Monitoring programs focusing on pesticide or Bt effects in non-target arthropods
would presumably include response variables such as population size that could be
monitored with population genetic techniques. Other important variables, such as
fecundity, mortality and size could only be studied with direct surveys. However,
population genetic studies would offer several unique advantages.
 Population genetic studies could be used to monitor specific traits under natural
selection at a molecular level.
 They would allow measures of genetic variation to be tracked through time, providing
insight for the evolutionary potential of non-target species, and perhaps detecting
bottlenecks.
 They would allow inferences about population boundaries and connectivity patterns
that may be difficult to obtain directly.
 Genetic methods of estimating effective population size Ne may be more stable over
time than estimates of the census population size N, if species undergo extreme
fluctuations in abundance from generation to generation.
 The highest levels of inference would by pairing genetic studies with standard
ecological monitoring. For example, comparing Ne with N would greatly contribute
to an understanding of how many individuals are actually reproducing in the nontarget species.
Sampling recommendations include:
 Multiple target species should be selected (perhaps four or more) that range in
population size, dispersal ability, ecological niche and known susceptibility.
Taxonomic identifications should not hinder the monitoring program. Species with
55



rapid generation times, moderate population sizes and low to moderate dispersal
abilities will probably be most useful.
A large initial study will be needed for each species to establish likely population
boundaries, field-test collection methods, refine lab protocols, and provide guidance
for the continuing monitoring protocol. The initial study would include multiple sites
per species, at spatial scales small enough so that unique gene pools are represented
with many sites. 50 individuals per site would not be excessive, unless the target
species would be negatively impacted with that level of sampling effort.
A typical study of this type would include mtDNA sequencing (likely CO I) and
microsatellites. We suggest at least 20 microsatellite loci per species for a robust
monitoring program, which is likely to entail considerable development costs at the
beginning of the program. The additional use of anonymous dominant markers might
be considered if the per-sample costs are low enough. However, we would not
suggest the use of AFLPs instead of microsatellites, because the information content
of the latter is much better. SNPs or nDNA sequencing are unlikely to be cost
effective, unless loci likely to be under selection will be used.
Continuity of laboratory protocol must be assured. Highly detailed documentation
should be established at the beginning of the project, including sample storage,
extraction and PCR protocol. For microsatellites, very specific guidelines should be
established for scoring, taking into account stutter and allelic dropout. For
anonymous dominant markers, multiple individuals should be repeated throughout the
study to assure constant sizing information and repeatability of all bands.
56
Literature cited
Andreev, D., M. Kreitman, T. W. Phillips, R. W. Beeman, and R. H. French-Constant. 1999.
Multiple origins of cyclodiene insecticide resistance in Tribolium castaneum (Coleoptera:
Tenebrionidae). Journal of Molecular Evolution 48: 615-624.
Balloux, F., and N. Lugon-Moulin. 2002. The estimation of population differentiation with
microsatellite markers. Molecular Ecology 11: 155 -165.
Beerli, P., and J. Felsenstein. 2001. Maximum likelihood estimation of a migration matrix and
effective population sizes in n subpopulations by using a coalescent approach.
Proceedings of the National Academy of Sciences of the United States of America 98:
4563-4568.
Blackwell, B. F., P. D. Doerr, J. M. Reed, and J. R. Walters. 1995. Inbreeding rate and effective
population size: A comparison of estimates from pedigree analysis and a demographic
model. Biological Conservation 71: 299-304.
Boakes, E. H., J. Wang, and W. Amos. 2007. An investigation of inbreeding depression and
purging in captive pedigreed populations. Heredity 98: 172-182.
Bohonak, A. J., N. Davies, G. K. Roderick, and F. X. Villablanca. 1998. Is population genetics
mired in the past? Trends in Ecology and Evolution 13: 360.
Bohonak, A. J., N. Davies, F. X. Villablanca, and G. K. Roderick. 2001. Invasion genetics of
New World medflies: testing alternative colonization scenarios. Biological Invasions 3:
103-111.
Bohonak, A. J., and G. K. Roderick. 2001. Dispersal of invertebrates among temporary ponds:
are genetic estimates accurate? Israel Journal of Zoology 47: 367-386.
Bonin, A., E. Bellemain, P. Bronken Eidesen, F. Pompanon, C. Brochmann, and P. Taberlet.
2004. How to track and assess genotyping errors in population genetics studies.
Molecular Ecology 13: 3261-3273.
Bossart, J. L., and D. Pashley Prowell. 1998. Genetic estimates of population structure and gene
flow: limitations, lessons and new directions. Trends in Ecology and Evolution 13: 202206.
Broquet, T., N. Menard, and E. Petit. 2007. Noninvasive population genetics: a review of sample
source, diet, fragment length and microsatellite motif effects on amplification success and
genotyping error rates. Conservation Genetics 8: 249-260.
Burczyk, J., W. T. Adams, S. Birkes, and I. J. Chybicki. 2006. Using genetic markers to directly
estimate gene flow and reproductive success parameters in plants on the basis of naturally
regenerated seedlings Genetics 173: 363-372.
Burt, D. W. 2002. A comprehensive review on the analysis of QTL in animals. Trends in
Genetics 18: 488.
Caballero, A. 1994. Developments in the prediction of effective population size. Heredity 73:
657-679.
57
Chen, C., E. Durand, and F. Forbes. 2007. Bayesian clustering algorithms ascertaining spatial
population structure: a new computer program and a comparison study. Molecular
Ecology Notes 7: 747-756.
Conner, J. K. 2003. Artificial selection: A powerful tool for ecologists. Ecology 84: 1650-1660.
Cornuet, J. M., and G. Luikart. 1996. Description and power analysis of two tests for detecting
recent population bottlenecks from allele frequency data. Genetics 144: 2001-2014.
Dale, P. J., B. Clarke, and E. M. G. Fontes. 2002. Potential for the environmental impact of
transgenic crops. Nature Biotechnology 20: 567-574.
Drummond, A. J., A. Rambaut, B. Shapiro, and O. G. Pybus. 2005. Bayesian coalescent
inference of past population dynamics from molecular sequences. Molecular Biology and
Evolution 22: 1185-1192.
Duelli
, P., M. K. Obrist, and D. R. Schmatz. 1999. Biodiversity evaluation in agricultural
landscapes: above-ground insects. Agriculture, Ecosystems and Environment 74: 33-64.
Dutton, A., H. Klein, J. Romeis, and F. Bigler. 2002. Uptake of Bt-toxin by herbivores feeding
on transgenic maize and consequences for the predator Chrysoperla carnea. Ecological
Entomology 27: 441-447.
Ewens, W. J. 1982. On the concept of effective population size. Theoretical Population Biology
21: 373-378.
Excoffier, L., G. Laval, and S. Schneider. 2006. Arlequin: an integrated software package for
population genetics data analysis, v.3.01 for Windows.
http://cmpg.unibe.ch/software/arlequin3
Excoffier, L., P. E. Smouse, and J. M. Quattro. 1992. Analysis of molecular variance inferred
from metric distances among DNA haplotypes: application to human mitochondrial DNA
restriction data. Genetics 131: 479-491.
Falconer, D. S. 1989. Introduction to quantitative genetics. 3rd edition. Wiley, New York, USA.
Frankel, O. H. 1974. Genetic conservation: our evolutionary responsibility. Genetics 78: 53-65.
Frankham, R. 1995. Conservation genetics. Annual Review of Genetics 29: 305-327.
Frankham, R., K. Lees, M. E. Montgomery, P. R. England, E. H. Lowe, and D. A. Briscoe. 1999.
Do population size bottlenecks reduce evolutionary potential? Animal Conservation 2:
255-260.
Frantzen, M. A. J., J. B. Silk, J. W. H. Ferguson, R. K. Wayne, and M. H. Kohn. 1998. Empirical
evaluation of preservation methods for faecal DNA. Molecular Ecology 7: 1423-1428.
Grosberg, R. K. 1991. Sperm-mediated gene flow and the genetic structure of a population of the
colonial ascidian Botryllus schlosseri. Evolution 45: 130-142.
Guillot, G., A. Estoup, F. Mortier, and J. Cosson. 2005. A spatial statistical model for landscape
genetics. Genetics 170: 1261-1280.
Gupta, M., Y. S. Chyi, J. Romero-Severson, and J. L. Owen. 1994. Amplification of DNA
markers from evolutionarily diverse genomes using single primers of simple-sequence
repeats. Theoretical and Applied Genetics 89: 998-1006.
58
Haig, S. M. 1998. Molecular contributions to conservation. Ecology 79: 413-425.
Hails , R. S. 2000. Genetically modified plants - the debate continues. Trends in Ecology and
Evolution 15: 14-18.
Hauser, L., G. J. Adcock, P. J. Smith, J. H. Bernal Ramirez, and G. R. Carvalho. 2002. Loss of
microsatellite diversity and low effective population size in an overexploited population
of New Zealand snapper (Pagrus auratus). Proceedings of the National Academy of
Sciences of the United States of America 99: 11742-11747.
Hebert, P. D. N., S. Ratnasingham, and J. R. de Waard. 2003. Barcoding animal life:
Cytochrome c oxidase subunit 1 divergences among closely related species. Proceedings
of the Royal Society of London Series B Biological Sciences 270: S96-S99.
Hill, R., and C. Sendashonga. 2006. Conservation biology genetically modified organisms, and
the biosafety protocol. Conservation Biology 20: 1620-1625.
Holsinger, K. E., P. O. Lewis, and D. K. Dey. 2002. A Bayesian approach to inferring population
structure from dominant markers. Molecular Ecology 11: 1157-1164.
Hudson, R. R., D. D. Boos, and N. L. Kaplan. 1992. A statistical test for detecting geographic
subdivision. Molecular Biology and Evolution 9: 138-151.
James, C. 2003. Global review of commercialized transgenic crops. Current Science 84: 303309.
Jensen, J. L., A. J. Bohonak, and S. T. Kelley. 2005. Isolation by Distance, Web Service. BMC
Genetics 6: 13. v.2.1. http://ibdws.sdsu.edu/~ibdws/
Leberg, P. 2005. Genetic approaches for estimating the effective size of populations. Journal of
Wildlife Management 69: 1385-1399.
Lewallen, E. A., and A. J. Bohonak. 2005. The genetic population structure of the leopard shark
(Triakis semifasciata). American Society of Ichthyologists and Herpetologists. Tampa,
Florida. July 6-11, 2005.
Loxdale, H. D., and G. Lushai. 1998. Molecular markers in entomology. Bulletin of
Entomological Research 88: 577-600.
Marvier, M., C. McCreedy, J. Regetz, and P. Kareiva. 2007. A meta-analysis of effects of Bt
cotton and maize on nontarget invertebrates. Science 316: 1475-1477.
McGeoch, M. A. 1998. The selection, testing and application of terrestrial insects as
bioindicators. Biological Reviews 73: 181-201.
Meier, M. S., and A. Hilbeck. 2001. Influence of transgenic Bacillus thuringiensis cornfed prey
on prey preference of immature Chrysoperla carnea (Neuroptera: Chrysopidae). Basic
and Applied Ecology 2: 35-44.
Mueller, U. G., and L. L. Wolfenbarger. 1999. AFLP genotyping and fingerprinting. Trends in
Ecology and Evolution 14: 389-394.
Mun, J., A. J. Bohonak, and G. K. Roderick. 2003. Population structure of the pumpkin fruit fly
Bactrocera depressa (Tephritidae) in Korea and Japan: Pliocene allopatry or recent
invasion? Molecular Ecology 12: 2941-2951.
59
Neigel, J. E. 2002. Is FST obsolete? Conservation Genetics 3: 167-173.
Nielsen, R., and J. Wakeley. 2001. Distinguishing migration from isolation: A Markov chain
Monte Carlo approach. Genetics 158: 885-896.
Nunney, L. 1991. The influence of age structure and fecundity on effective population size.
Proceedings of the Royal Society of London Series B Biological Sciences 246: 71-76.
Nunney, L. 1993. The influence of mating system and overlapping generations on effective
population size. Evolution 47: 1329-1341.
O'Brien, S. J., and J. F. Evermann. 1988. Interactive influence of infectious disease and genetic
diversity in natural populations. Trends in Ecology and Evolution 3: 254-259.
O'Brien, S. J., D. E. Wildt, D. Goldman, C. R. Merril, and M. Bush. 1985. The cheetah is
depauperate in genetic variation. Science 221: 459-461.
O'Callaghan, M., T. R. Glare, E. P. J. Burgess, and L. A. Malone. 2005. Effects of plants
genetically modified for insect resistance on nontarget organisms. Annual Review of
Entomology 50: 271-292.
Obrycki, J. J., J. E. Losey, O. R. Taylor, and L. C. H. Jesse. 2001. Transgenic insecticidal corn:
beyond insecticidal toxicity to ecological complexity. Bioscience 51: 353-367.
Palstra, F. P., and D. E. Ruzzante. 2008. Genetic estimates of contemporary effective population
size: what can they tell us about the importance of genetic stochasticity for wild
population persistence? Molecular Ecology 17: 3428-3447.
Palumbi, S. R. 1996. Nucleic acids II: the polymerase chain reaction. Pages 205-247 in D. M.
Hillis, C. Moritz and B. K. Mable, editors. Molecular systematics. Sinauer Associates,
Sunderland, MA, USA.
Peterson, M. A., and R. F. Denno. 1998. The influence of dispersal and diet breadth on patterns
of genetic isolation by distance in phytophagous insects. American Naturalist 152: 428446.
Pimentel, D. S., and P. H. Raven. 2000. Bt corn pollen impacts on nontarget Lepidoptera:
Assessment of effects in nature. Proceedings of the National Academy of Sciences of the
United States of America 97: 8198-8199.
Prasifka, J. R., R. L. Hellmich, G. P. Dively, L. S. Higgins, P. M. Dixon, and J. J. Duan. 2008.
Selection of nontarget arthropod taxa for field research on transgenic insecticidal crops:
Using empirical data and statistical power. Environmental Entomology 37: 1-10.
Pritchard, J. K., M. Stephens, and P. Donnelly. 2000. Inference of population structure using
multilocus genotype data. Genetics 155: 945-959.
Rannala, B., and J. L. Mountain. 1997. Detecting immigration by using multilocus genotypes.
Proceedings of the National Academy of Sciences 94: 9197-9201.
Raps, A., J. Kehr, P. Gugerli, W. J. Moar, F. Bigler, and A. Hilbeck. 2001. Immunological
analysis of phloem sap of Bacillus thurigiensis corn and of the nontarget herbivore
Rhopalosiphum padi (Homoptera: Aphididae) for the presence of Cry1Ab. Molecular
Ecology 10: 525-533.
60
Rogers, A. R., and H. C. Harpending. 1992. Population growth makes waves in the distribution
of pairwise genetic differences. Molecular Biology and Evolution 9: 552-569.
Rousset, F. 2008. Genepop ’007: a complete reimplementation of the Genepop software for
Windows and Linux. Molecular Ecology Notes 8: 103-106.
Schmitz, G., D. Bartsch, and P. Pretscher. 2003. Selection of relevant non-target herbivores for
monitoring the environmental effects of Bt maize pollen. Environmental Biosafety
Research 2: 117-132.
Schwartz, M. K., D. A. Tallmon, and G. Luikart. 1998. Review of DNA-based census and
effective population size estimators. Animal Conservation 1: 293-299.
Simon, C., F. Frati, A. Beckenbach, B. Crespi, H. Liu, and P. Flook. 1994. Evolution, weighting,
and phylogenetic utility of mitochondrial gene sequences and a compilation of conserved
polymerase chain reaction primers. Annals of the Entomological Society of America 87:
651-701.
Slatkin, M. 1985. Gene flow in natural populations. Annual Review of Ecology and Systematics
16: 393-430.
Slatkin, M. 1993. Isolation by distance in equilibrium and non-equilibrium populations.
Evolution 47: 264-279.
Slatkin, M. 1994. Gene flow and population structure. Pages 3-17 in L. A. Real, editor.
Ecological genetics. Princeton University Press, Princeton, USA.
Slatkin, M. 1995. A measure of population subdivision based on microsatellite allele frequencies.
Genetics 139: 457-462.
Slatkin, M. 2005. Seeing ghosts: the effect of unsampled populations on migration rates
estimated for sampled populations. Molecular Ecology 14: 67-73.
Snow, A. A., D. A. Andow, P. Gepts, E. M. Hallerman, A. Power, J. M. Tiedje, and L. L.
Wolfenbarger. 2005. Genetically engineered organisms and the environment: Current
status and recommendations. Ecological Applications 15: 377-404.
Storfer, A. 1996. Quantitative genetics: A promising approach for the assessment of genetic
variation in endangered species. Trends in Ecology and Evolution 11: 343-348.
Thies, J. E., and M. H. Devare. 2007. An ecological assessment of transgenic crops. Journal of
Development Studies 43: 97-129.
Tilman, D., K. G. Cassman, P. A. Matson, R. Naylor, and S. Polasky. 2002. Agricultural
sustainability and intensive production practices. Nature 418: 671-677.
Vandergast, A. G., E. A. Lewallen, J. Deas, A. J. Bohonak, D. B. Weissman, and R. N. Fisher. in
press. Loss of genetic connectivity and diversity in urban microreserves in a southern
California endemic Jerusalem cricket (Orthoptera: Stenopelmatidae: Stenopelmatus
“santa monica”). Journal of Insect Conservation.
Villablanca, F. X., G. K. Roderick, and S. R. Palumbi. 1998. Invasion genetics of the
Mediterranean fruit fly: variation in multiple nuclear introns. Molecular Ecology 7: 351365.
61
Waples, R. S. 1989. A generalized approach for estimating effective population size from
temporal changes in allele frequency. Genetics 121: 379-392.
Waples, R. S., and O. Gaggiotti. 2006. What is a population? An empirical evaluation of some
genetic methods for identifying the number of gene pools and their degree of
connectivity. Molecular Ecology 15: 1419-1439.
Waples, R. S., and M. Yokota. 2007. Temporal estimates of effective population size in species
with overlapping generations Genetics 175: 219-233.
Waser, P. M., and C. Strobeck. 1998. Genetic signatures of interpopulation dispersal. Trends in
Ecology and Evolution 13: 43-44.
Weir, B. S. 1996. Genetic data analysis II: methods for discrete population genetic data. Sinauer
Associates, Sunderland, MA, USA.
Whitlock, M. C., and D. E. McCauley. 1999. Indirect measures of gene flow and migration: FST ≠
1/(4Nm + 1). Heredity 82: 117-25.
Wilson, G. A., and B. Rannala. 2003. Bayesian inference of recent migration rates using
multilocus genotypes. Genetics 163: 1177-1191.
Wolfenbarger, L. L., and P. R. Phifer. 2000. The ecological risks and benefits of genetically
engineered plants. Science 290: 2088-2093.
Wright, S. 1931. Evolution in Mendelian populations. Genetics 16: 97-159.
Zietkiewics, E., A. Rafalski, and D. Labuda. 1994. Genome fingerprinting by simple sequence
repeat (SSR)-anchored polymerase chain reaction amplification. Genomics 20: 176-183.
Zink, R. M., and G. F. Barrowclough. 2008. Mitochondrial DNA under siege in avian
phylogeography. Molecular Ecology 17: 2107-2121.
62
APPENDIX A-3
Response from Dr. David M. Marsh
63
Monitoring the Effects of Plant Incorporated Protectants on Non-Target Species: Prospects
for a Genetic Approach
David Marsh
Associate Professor of Biology
Washington and Lee University
Lexington VA 24450
Introduction
Traditionally, animal and plant populations have been monitored with presence/absence
surveys, count surveys, or through the direct estimation of population density. With
presence/absence surveys, data are usually collected over a large number of sites. Then, by
repeating surveys at regular time intervals, one can ask whether the proportion of sites containing
the organism is increasing or decreasing over time (McKenzie et al. 2005). Count data are
obtained by counting individual organisms or some feature associated with organisms, such as
tracks, scat, or burrows. Count data are often treated as a population index that is assumed to be
correlated with actual population size. Count data, like presence/absence data, may be collected
over a number of sites, and one can ask about trends at individual sites or across the study area
(Gibbs et al. 1998). Finally, monitoring is sometimes focused on estimating true population size.
Calculations of true population size usually combine some form of count survey with an estimate
of the detectability of individual organisms during surveys (White 2005). The most common
approach is mark-recapture analysis, in which a sample of the population is marked and then
individual detectability is estimated from multiple recapture sessions.
Each of these three approaches to monitoring is widely used, though each approach has some
considerable drawbacks. Presence/absence surveys generate limited information, and thus
provide very low power for detecting increases or decreases in populations (Pollak 2006).
Furthermore, because one needs to register complete disappearances (rather than just declines) in
order for sites to change in status, presence/absence surveys have the odd property that more
careful surveys actually result in reduced power to detect population trends (Joseph et al. 2006).
In contrast to presence/absence surveys, count surveys can produce useful information about
population abundance. Because count surveys can show increases or decreases in populations,
count surveys tend to yield higher power than do presence/absence surveys. The main drawback
of counts is that their relationship to true population size is often unknown. This can be a critical
gap: for example, if organisms become more visible as population size decreases, a count or
population index may show few signs of change in spite of large decreases in population size. In
some cases, e.g. scat surveys and track surveys, comparison of population indices to more
intensive density estimation has clearly shown population indices to be poor surrogates for actual
population size.
Estimates of population density (and detectability) are clearly superior to count surveys for
monitoring any given population. However, density estimation is far more labor intensive than
are count surveys since estimating density and detectability requires multiple visits to each site
and, in many cases, capturing and marking individual organisms. As a result, mark-recapture
64
based monitoring programs will almost always cover fewer sites than count or presence/absence
surveys. The trade-offs between precision at a local scale and the ability to monitor many
populations across the landscape have not been rigorously studied. Nevertheless, it is likely that
when population indices are decent reflections of population size, the benefits of landscape-scale
coverage will almost certainly exceed the costs of reduced precision. Put another way, rigorous
density estimation is useful when there are only a few populations to be monitored (e.g. highly
endangered species) but are not optimal for landscape-scale monitoring programs.
A recent review by Schwartz et al. (2007) touted a distinct new set of monitoring approaches
based on population genetics. These genetic monitoring approaches take several distinct forms.
First, DNA samples, often collected with non-invasive methods (e.g. from scat, hair, or feathers),
can be used to identify individual animals. In most cases, these identifications are used in
“genetic mark-recapture”, where DNA sequences are treated as marks and matched with
recaptured samples from additional surveys. DNA mark-recapture methods are generally used
with species that cannot be easily captured or marked, particularly large mammals like bears
(Boulanger et al. 2004) and coyotes (Prugh et al. 2005). A related approach is to use the rate of
accumulation of new genotypes to estimate the total number of individuals in the population
(“rarefaction”, e.g. Frantz and Roper 2006). However, heterogeneity in capture rates limits the
utility of this approach, and it is not routinely recommended (Lukacs and Burnham 2005).
Second, one can use population genetic techniques to test hypotheses about the history of a
population. Typically these approaches use coalescence theory to estimate the historical
effective population size or to look for evidence of past population bottlenecks (e.g. Beaumont
1999; Estoup et al. 2001). Because inferences about population history can be compared with
the current state of a population, these techniques could be thought of as a form of “retrospective
monitoring” (Schwartz et al. 2007). Finally, genetic monitoring can refer to the tracking of
population genetic parameters over time. Parameters such as effective population size, allelic
diversity, and rates of gene flow can be estimated over several time periods to test for changes in
population status. This form of monitoring may be prospective when one measures parameters
at regular time intervals. It can also be retrospective, when tissue samples are available from
archived collections.
In this report, I comment on the suitability of genetic monitoring for examining the effects of
plant-incorporated protectants on non-target species in agricultural landscapes. I focus on
genetic monitoring in the sense of prospective tracking of population parameters over time, and
to a lesser extent, retrospective approaches using archived samples or inferences from
coalescence theory. I do not discuss genetic mark-recapture, as I do not believe that species of
interest within agricultural landscapes would require this approach. I also do not discuss
rarefaction for estimating population size, as this approach would not be necessary when one can
choose the species to be monitored.
I first discuss the advantages and disadvantages of genetic monitoring in comparison to
traditional survey approaches. One of the principle advantages of genetic monitoring is that
allows one to estimate parameters that are not estimable with traditional approaches, so I also
discuss these parameters in the first section. Second, I consider the most informative parameters
to estimate using genetic monitoring methods. Third, I suggest some taxa that might be good
65
choices for monitoring (genetic or otherwise) and some of the advantages and disadvantages of
each. Fourth, I consider strategies for combining information from genetic monitoring programs
with other available information on the status of non-target species. Fifth, I detail some of the
knowledge gaps and scientific issues that need to be addressed before monitoring can be
implemented. Finally, I end with a summary and my personal opinions about the feasibility of
genetic monitoring.
A few comments about my background and expertise may be of help in interpreting my
comments and suggestions. I have recently done a lot of work with simulation models to
develop landscape-scale approaches to monitoring (with count surveys). As a separate line of
research, I have been involved in a number of population genetic studies of salamanders. So, in
a broad sense I consider myself well-qualified to deal with most of the issues involved in genetic
monitoring. However, my population genetics work is collaborative – I typically design the
studies, oversee the collection of samples, and analyze the data we obtain. I do not have much
experience as a laboratory geneticist, so I do not discuss technical issues associated with marker
development, amplification, scoring and binning alleles, etc. Waits and Leberg (2001), Paetkau
(2003), and Lukacs and Burnham (2005) provide good overviews of these kinds of issues.
Additionally, I have primarily worked in forest ecosystems, so my knowledge of agricultural
ecology is somewhat limited. For this reason, I pay more attention to statistical consideration
than functional consideration when suggesting taxa for monitoring. Finally, my knowledge of
native plants in agricultural areas is close to nil, so I have not made any attempts to comment on
the possibility of monitoring non-target plant species.
Advantages and Disadvantages of Genetic Monitoring
The advantages and disadvantages of genetic monitoring are best considered in direct
comparison to traditional approaches. Schwartz et al. (2007) consider some of the advantages of
genetic monitoring compared to approaches based on density estimation. This is somewhat
misleading, since the vast majority of monitoring programs are based on population indices (e.g
counts) rather than rigorous estimation of population density (Marsh and Trenham 2008). I thus
compare genetic monitoring to traditional approaches using population indices, and, to a lesser
extent, multi-species presence/absence surveys.
One of the main advantages of genetic monitoring is that it allows the estimation of parameters
that cannot be estimated using traditional approaches. Perhaps the most important of these
parameters in Ne , the genetic effective population size. Ne is directly related to the rate of loss of
genetic diversity through drift. At a minimum, Ne could be considered a population index, as it
should generally be correlated with overall population size. However, for small populations in
which loss of genetic diversity is a real concern, Ne provides a direct meaure of the magnitude of
this threat. Many conservation biologists have argued that demographic considerations will
generally trump genetic consideration with respect to extinction risk (Lande 1988), but there is
some empirical evidence that Ne can be related to extinction risk (Saccheri et al. 1998).
Additional parameters that can only be estimated through genetic monitoring include allelic
diversity and gene flow. Monitoring allelic diversity direct could allow one to empirically
determine the rate at which diversity is being lost. The rate of change in allele frequencies
66
(Ftemporal) could provide an even earlier warning signal for the potential loss of genetic diversity
(Luikart et al. 1999). With respect to gene flow, methods for estimating gene flow and migration
rates from microsatellite data have been getting better and better (e.g. Beerli and Felsenstein
2001). In contrast, one cannot get this information at all from population indices. And even with
extensive mark-recapture studies, it is usually quite difficult to get good estimates of dispersal,
since long-distance dispersal is rare and most long distance dispersers will be missed. Thus,
estimates of gene flow are probably the best way to understand population connectivity at a
landscape scale. In the context of prospective monitoring, one could estimate gene flow over
multiple time periods and examine the changes in connectivity that result from some change in
the environment.
Another major advantage of genetic monitoring is that it potentially allows for retrospective
monitoring, that is, comparing the current state of a population to the state sometime in the past.
This kind of analysis is sometimes performed when tissue, blood samples, hair, or feathers are
available and when DNA can be extracted from these samples. Whether these kinds of samples
are available will be case-by-case, and historical data can also be potentially used to inform
count or presence/absence surveys (see Shaffer et al. 1998 for review). However, when
historical samples are available, comparison of current to past status can be of interest both
scientifically and from a management perspective (Miller and Waits 2003). Even without
historical data, statistical methods based on coalescence theory can be used to make some
general inferences about past population status. These inferences most commonly concern the
“historical Ne” (that is, Ne averaged over a very long period of time: e.g. Alter et al. 2007), but
they can also be used to compare different scenarios for colonization and population growth
(Estoup et al. 2001).
On the other hand, population genetic methods also present some real difficulties in comparison
with population indices or presence/absence surveys. The most obvious of these is that genetic
monitoring methods have little or no track record. Thus, while it is reasonable to assert that
these methods hold a lot of promise, there are very few real-world examples of genetic
monitoring, and almost none outside of the fisheries and wildlife literature. Furthermore, to my
knowledge, no one has carried out a direct comparison of the precision and costs of genetic
monitoring as compared with traditional survey approaches. Most examples of genetic
monitoring in the literature use genetic approaches either because historical samples happen to
available (e.g. Miller and Waits 2003) or because population counts are extremely difficult to
obtain for the species of interest (e.g. Rudnick et al. 2005). So, even though population indices
may have their problems, there has been a great deal of research on these problems that allow
their effects to be minimized. With genetic monitoring, some limitations have been suggested in
the literature (Beaumont 2003), but others are certainly waiting to be discovered.
An additional disadvantage of genetic approaches is that they necessarily focus on a small
number of species (since sample collection and genetic methods will always be species-specific).
Whether or not a few indicator species can accurately reflect environmental change is a subject
of continued debate in the scientific literature. Without wading into these disputes, it is safe to
say that there are no agreed upon indicator species for agricultural ecosystems in the U.S. In
contrast to genetic approaches, traditional count surveys often require limited additional effort to
collect data on multiple species. For example, point counts can be used to simultaneously record
67
the presence of multiple bird species, transect counts can be used to count multiple
Lepidopterans, and call surveys can be used to document the presence of multiple amphibian
species. Because of this, it is easy to use traditional surveys to monitor species diversity and
trends in individual species’ abundance. Moreover, one can look for trends across broad
numbers of species without having to guess a priori which species will be the most sensitive to
environmental change.
Cost is likely to be an important consideration with genetic monitoring. I hesitate to speculate on
absolute costs because these will vary widely from one lab to the next. But it is always cheaper
to spend a day in the field counting organisms then to spend a day in the field collecting tissue
samples, then take these samples back to the lab, isolate several microsatellites and sequence
them. One could probably expect to carry somewhere between 5 and 50 population counts for
what it would cost to process a set of DNA samples. Ne estimates present an additional
difficulty. Ne is most commonly estimated using temporal methods in which one takes two
samples separated in time (Waples 1989; Wang 2001; Wang and Whitlock 2003; Waples 2007)
and uses the change in allele frequency as a measure of the magnitude of drift (and thus,
population size). That means that 2 samples are required for a single Ne estimate and a minimum
of 3 samples are required in order to test for change over time. The costs here are borne more in
terms of time than simply in terms of the additional sample. Since the precision of an Ne
estimate is directly related to the number of generations between samples (Waples 1989),
samples used in temporal estimates of Ne are typically taken at least 10 years apart (often much
farther apart). For prospective monitoring, this would mean that a decade would need to elapse
before one would have a baseline Ne estimate and the another 10 years before one could ask
about trends. Obviously, this would not be optimal for rapid detection of population trends.
Ne can instead be estimated from the magnitude of linkage disequilibrium (Hill 1981) and this
method requires only a single sample. However, the sample size must be much larger than the
samples (~50) typically used with temporal methods (Leberg 2005). Ardren and Kapuscinski
(2003) for example, were able to estimate Neb for a steelhead trout population based on LD, but
the confidence intervals on this estimate ranged from 9 or 10 to infinity for two of their three
samples. The LD method is generally considered to be less precise than the various temporal
methods and it is therefore not routinely used (Leberg 2005).
Although Ne estimates from temporal methods are more precise, there is good to ask whether
they are precise enough to be much value for monitoring programs aimed at detecting population
trends. Table 1 shows a collection on Ne estimates with confidence intervals for a range of taxa.
These estimates do not represent a complete or systematic search of the literature, but they were
in no way pre-selected for being good or bad estimates. Several points can be taken from this
table. First, confidence intervals on Ne estimates are typically quite large. In most cases, Ne
values would need to increase or decrease by 50-80% in order for a change to be considered
statistically significant. The most precise estimates of Ne in the Table are from the Wang and
Whitlock (2003) method, which simultaneously estimates Ne and migration rate. However, at
least one set of authors using this method (Hoffman et al. 2004) concluded that these estimates
were unreasonable and that the method required too much information about the origin of
migrants to be used effectively. Second, larger Ne estimates often result in confidence intervals
that include infinity. The reason for this is that in larger populations, drift is very weak and can
68
be very difficult to distinguish from sampling or scoring errors. In his original description of the
temporal method, Waples (1989) argued that it may be difficult to distinguish large populations
from infinitely large populations. This means that monitoring of Ne is only feasible when
effective population sizes are small.
Are estimates from traditional count surveys any better than estimates of Ne? As part of a
previous project, I estimated the magnitude of observation error in count data for a variety of
organisms with data taken from the Global Population Dynamics Database (NERC 1999). To
convert all measurements to a common (and admittedly ad hoc) scale, I transformed both Ne
confidence intervals and observation error estimates from count data to give the percentage of
the count or Ne estimate that is spanned by the 95% confidence intervals. For example, if the
count or Ne estimate was 100 and the confidence intervals went from 80 to 130, the “percent
span” would be 50 (20% down and 30% up). If the count or Ne estimate was 200 and the
confidence intervals went from 100 to 400, the percent span would be 150 (50% down and 100%
up). Figure 1 shows the results for both Ne estimates (Fig 1A) and for count data (Fig 1B).
While count data often give wide confidence intervals, in general they appear to give lower
confidence intervals than do Ne estimates. More to the point, there appear to be a large number
of organisms for which count data yield small observation errors, whereas it is hard to find Ne
estimates with a percent span of less than 50%.
This comparison is not really complete – monitoring programs also have to deal with process
error (i.e. background fluctuations in population size that are unrelated to the trends of interest)
and bias (as directional error). With respect to process error, it is possible that genetic
approaches could handle process error better than count surveys, since fluctuations are averaged
over the period between the first and second genetic samples. With respect to bias, one could
argue that either genetic or traditional methods could produce biased estimates. With genetic
methods, error in genotyping (or processing samples) will tend to bias estimates downwards,
since errors will look like drift. With count approaches, estimates can be either high or low,
depending on the particulars of the surveys. In any case, the issue is not really bias per se, but
whether bias changes with population density in a way that makes trend detection difficult. For
example, Ne might become relatively larger as population size decreases (i.e. genetic
compensation, Ardren and Kapuscinski 2003). Really, direct studies are needed that compare
genetic monitoring to traditional approaches before one could draw any conclusions about which
methods are more subject to bias. But regardless of bias, the possibility for reduced precision
from very labor-intensive methods should give one pause about using genetic monitoring when
other options are readily available.
Informative parameters
What would be the most informative parameters to estimate with genetic monitoring? Several
parameters that can be estimated, while very interesting from a scientific perspective, may have
limited utility for genetic monitoring. First, one can test whether or not past population
experienced a bottleneck and estimate the size of the bottleneck. However, in agricultural areas,
prior population sizes are most likely determined by prior land-use practices. If good land-use
records exist, one could test specific hypotheses about the effects of land use on certain species.
But this approach would be less likely to provide good information for any kind of generalized
69
monitoring program. Second, one can estimate the historical effective population size over long
periods of time. Again, in agricultural areas, this information might be of limited use unless one
is testing specific hypotheses about the effects of land use on particular species. Third, Schwartz
et al. (2007) suggest more monitoring of results derived from genetic analysis, such as the
clustering of individuals among populations. This would indeed be interesting, but would be
most appropriate for situations where some management intervention is expected to rapidly alter
how individuals disperse, for example, if a dam were to be removed or a corridor were created to
connect nearby habitat areas.
Estimating Ne. The most obvious parameter to monitor is Ne. As mentioned above, Ne is both an
index of overall population size and a biologically informative measure of population status with
respect to loss of genetic diversity. Moreover, a good statistical toolbox exists for estimating Ne
in species with a variety of life-histories, including species with overlapping generations (Waples
2007) or species that might disperse into or out of the study area (Wang and Whitlock 2003).
The main limitation with respect to monitoring Ne is the large confidence intervals that normally
result from its estimation (Waples 1989).
Estimating Nb. In some cases, Ne cannot be efficiently estimated but one can estimate Nb, the
number of breeders in the population. One important approach uses of allele frequency
differences between parents and offspring to estimate Nb (Leberg 2005). As a biological
parameter, Nb is less meaningful than is Ne, but Nb can be estimated much more quickly (Leberg
2005). Several of the studies in Table 1 (e.g. Jehle et al. 2001; Ardren and Kapuscinki 2003)
estimated Nb as well as Ne, and the confidence intervals were not any smaller (in some cases they
were larger). So, Nb provides some advantages in terms of time, but probably does not yield
more precise estimates.
Testing for changes in genetic diversity. There are several parameters than can be used for
estimating genetic diversity, among the most commonly used are A, allelic diversity and Ftemporal,
the change in allele frequencies over time. Allelic diversity is probably the most biologically
relevant of these parameters as it is a direct measure of the genetic diversity remaining in a
population. Ftemporal, however, is more sensitive, since alleles will tend to become more rare
before they are actually lost. Either of these quantities is a very reasonable thing to monitor, but
they are only really meaningful for small populations in which a loss genetic diversity is a very
real concern.
Gene flow or migration rate. A variety of techniques exist for estimating gene flow. These
techniques have greatly improved in recent years, at least in terms of what parameters can be
estimated with a modicum of assumptions. Program MIGRATE (Beerli and Felsenstein 2001),
for example, can estimate the number of dispersers between any two adjacent areas from
microsatellite data, even allowing for asymmetric dispersal or unsampled populations. The
caveat is that these sorts of estimates make the most sense when sampling areas can be viewed as
habitat islands which are connected by dispersal. In more continuous habitats, estimates of
migration and gene flow are likely to be helpful in a relative sense, but are less meaningful as
real biological parameters. Also, these parameters will be critical for monitoring where
connectivity is expected to change rapidly (e.g. road construction, logging, etc.).
70
Selecting suitable taxa – traditional monitoring
The criteria that one would use to select suitable taxa would vary quite a bit between a traditional
monitoring program and a genetic monitoring program. For a traditional monitoring program
one would want to select taxa that are relatively common and widespread, can be easily detected
during surveys, and that are likely to be influenced by the environmental factors of interest.
Furthermore, if using population counts or other indices, one might want to use taxa for which
indices are well-developed. For monitoring the effects of PIPs in agricultural areas, presumably
one would want species that may be sensitive to levels of pesticide use (or PIPs themselves).
Several groups might fit the bill here. First, one could imagine using point-counts for songbirds
to monitor the effects of PIPs. Songbirds have several advantages as study organisms. First,
point counts for songbirds (usually along transects) have been widely used to document
population trends throughout North America (Link and Sauer 1998). Second, songbirds are
easily observed (or heard) and are widespread, so that the same set of species could be monitored
across large geographic regions. Furthermore, one can easily record the presence or absence of
many species at the same time. Finally, most songbirds are generalist consumers of seeds or
insects – thus, they are likely to be sensitive to changes in levels of pesticide use. One would
need to be sure to not survey during migration periods (which may vary temporally from year to
year) but beyond that, bird surveys can be carried out largely at one’s convenience, as long as
that time period is consistent from one year to the next.
I also believe butterflies would make good candidates for a count-based monitoring program. As
consumers of nectar and leaves, butterflies would likely be subject to both direct and indirect
effects of Bt crops. Butterflies can be easily observed by walking line transects through sites of
interest, and multiple species can be observed at the same time. The Millenium Butterfly Survey
in Britain has already worked out many of the kinks with these kinds of surveys in agricultural
and oldfield habitats (Asher et al. 2001; Thomas 2005).
Finally, I think it would helpful to choose one taxon in an aquatic habitat, as pesticide runoff into
aquatic habitats can strongly impact these ecosystems. One possibility would be stream
invertebrates. Diversity surveys for stream invertebrates have a long tradition in environmental
biology, and methods are well-developed. Furthermore, individual invertebrate species that are
sensitive to disturbance and pollution have been well-categorized. Another option would be
dipnet surveys in farm ponds for invertebrates. Farm ponds are more much more common than
streams in many parts of the country (i.e. the flat parts), so invert surveys in farm ponds could be
carried out more consistently from one region to the next.
I also considered amphibians as candidates for aquatic surveys, but perhaps I know a little too
much about the problems associated with amphibian surveys. On the plus side, amphibians are
probably of greater conservation concern than are pond invertebrates, and they are known to be
adversely affected by pesticides (Hayes et al. 2006). Amphibians also tend to be widespread
and, using dipnet or call surveys (Crouch and Paton 2002), multiple species can be surveyed at
the same time. On the down side, call surveys must be carried out at night (which is often
inconvenient) and dipnet surveys can be difficult to time with the rapid breeding of species like
Wood Frogs. Perhaps the biggest problem with amphibian surveys is that pond amphibian
71
populations can fluctuate by huge amounts from one year to the next. This makes trend
detection notoriously difficult (Pechmann and Wilbur 1994).
Finally, with traditional count surveys, one might wish to monitor functional components of
agricultural ecosystems. Honeybees are an obvious choice here, and they can be sampled
effectively with baits. Nematodes are also interesting candidates, though problems with
identification and taxonomy might make implementation difficult. Monarchs would also be a
good candidate, given previous concerns (perhaps unfounded) about the effects of PIPs on
monarch survival.
Selecting suitable taxa – genetic monitoring
For genetic monitoring, the criteria for selecting taxa would be very different. Since one could
presumably only monitor a few species, it would be very important to know that these species
are reflective of broader ecosystem responses to changing agricultural practices. Additionally,
one would want species that occur in small populations that are reasonably isolated from one
another; otherwise the estimation of genetic parameters would be difficult. Furthermore, one
would want to select species that can be easily captured or where non-invasive approaches can
be used. Finally, one would want species with short generation times, as the precision of
temporal methods for Ne estimation is directly related to the number of generations between
samples.
It is worth considering the kinds of species on which genetic monitoring studies have previously
been carried out. A number of such studies have been conducted with anadromous fish (Hansen
et al. 2006), and these organisms appear to meet the above criteria rather well. Genetic
monitoring studies have also been carried out with grizzly bears (Miller and Waits 2003) and
Italian wolves (Lucchini et al. 2004). These also seem to be good choices, both in terms of their
suitability for genetic monitoring and because of the relative difficulty of monitoring these
species with traditional approaches.
For monitoring non-target species in agricultural landscapes, one is faced with some real
challenges with respect to species selection. Most agricultural landscapes are fairly
homogeneous. As a result, any species common enough to sample would also be expected to
have a very large effective population size. As discussed previously, this makes estimation of Ne
difficult, but it also makes it unlikely that one would see changes in allelic diversity over time.
Furthermore, I would be surprised if changes in pesticide use had substantial effects on
connectivity for any particular group of organisms, so monitoring gene flow or population
structure would probably not be a priority.
Given these challenges, I think pond amphibians would not be a bad choice for genetic
monitoring. They occur in small, but distinct habitat patches, and most agricultural landscapes
have ponds which could be sampled. Eastern frog species generally occur over broad swaths of
either the mid-Atlantic and Northeast, or across the Southeast. The Midwestern states also
generally share an assemblage of frogs.
72
Still, it would be important to also choose some additional species that are found in terrestrial
habitats, as these would be the most directly affected by changes in pesticide use. One would
need to identify taxa that are patchily distributed (in order to get manageable Ne values), but that
live in habitats that can be identified from existing maps or remote sensing. Perhaps there are
butterflies that are common to areas bordering cultivated lands. Or perhaps one could use birds
or small mammals in isolated woodlots that are adjacent to agricultural areas (small woodlots are
reasonably common in the Midwest). My cursory knowledge of agricultural ecosystems might
be the problem here, or it may be that there are simply too many competing needs with respect to
species-selection for genetic monitoring in agricultural areas.
To my mind, one of the most exciting recent advances in population genetics is the use of
coalescence theory to test hypotheses about population history (Beaumont 1999; Estoup et al.
2001; Alter et al. 2007). With a single sample, one can potentially test hypotheses about the
history of populations in a form of retrospective monitoring. However, given the dramatic
changes in agricultural practices over the past few hundred years, these kinds of approaches
would be unlikely to provide much information with respect to the effects of pesticide use.
Estimates from coalescence theory are great for distinguishing broad classes of hypotheses (e.g.
a very large historical Ne versus a recent bottleneck) but would be unlikely to distinguish, say,
the effects of agricultural practices in the 1960s from the effects of the 70s or 80s.
Combining the results of genetic monitoring with other sources of information
Genetic monitoring is not an all-or-nothing proposition. The results of any genetic monitoring
program could be used in conjunction with other sources of information, particularly data from
traditional monitoring programs. Complex risk assessment is not my field, though I am
reasonably familiar with meta-analytic techniques for synthesis of evidence.
One important issue in combining results is that the specific parameters that one would estimate
with genetic monitoring (e.g. Ne, A, or migration rate) are not equivalent to the parameters
estimated in traditional monitoring (e.g. N or dispersal rate). Ne to N ratios may vary
considerably over time, so really these are distinct quantities. Nb could vary even more depend
on how population density influences mating parameters. The migration rate that is estimated
from allele frequencies reflects migration followed by establishment and reproduction. In
contrast, the dispersal rate that one estimates with mark-recapture just determines movement, not
establishment and reproduction.
That said, when population size N is increasing, Ne will usually increase as well. In this case,
data on N and data on Ne both provide independent evidence that a population is recovering.
This is particularly true because the data used to estimate each are obtained in completely
different ways. Within this framework, one could think about assessing the evidence for or
against recovery from each data sourse. These sources of information could then be weighted by
the reciprocal of their standard errors, and potentially combined. It is worth noting that this kind
of synthesis is controversial in the meta-analysis literature. Some argue that it is reasonable to
combine information in this way as long as the hypothesis being tested is defined in a sufficiently
broad manner (which is my personal view). Others would argue that N and Ne are always
73
“apples and oranges” and that combining any information about the two would only muddle the
question of what is actually happening to these populations.
On a related note, if one wanted to do both traditional count surveys and genetic monitoring, it
would make a lot of sense to carry these out at the same time. That is, one could do count
surveys, but collect the samples for genetic analysis at the same time. This would tend to reduce
the labor costs that would be incurred with carrying out the two types of surveys separately. I
am doing this in some of my current salamander work - we are monitoring salamander
populations in cover-board arrays, but also taking samples of mtDNA for phylogenetic analysis.
Knowledge gaps and outstanding scientific issues
Some have touted the advantages of genetic monitoring (Hansen et al. 2006; Schwartz et al.
2007), whereas I have arrived at a more skeptical view. The basic problem here is that no one
has ever directly compared the two in terms of costs and the precision of estimates obtained.
From a scientific point-of-view, it would certainly be helpful to compare the two. The easiest
approach would be simulation studies in which one simulated the dynamics of populations and
the changes in allele frequencies from one generation to the next. One could then overlay
potential monitoring strategies on these dynamics, particularly the estimation of a population
index from count data (with observation error) and the estimation of Ne from drift and/or linkage
disequilibrium (or Nb from parent-offspring differences). One could then compare quantities
such as the power to detect a real increase or decrease in the population, or the precision obtained
per unit effort (or dollar spent). As I’ve suggested above, my intuition is that for most species, a
population index would be much more efficient, but the intuition of mid-level academics is no
substitute for actual evidence.
One step beyond this would be the direct comparison of traditional to genetic monitoring for
some selected populations. Personally, I would be very interested to see this sort of comparison
for pond amphibians, as their variable population sizes make the use of snapshot count surveys
rather difficult for monitoring effectively. As discussed above, I don’t think there would be any
way of doing this quickly in a prospective manner (it would take 5-10 years just to get a good
baseline estimate of Ne, though one could calculate Nb more quickly). Ideally, one would want
to find a set of sites where there was ongoing monitoring (to provide a source of count data) as
well as archived tissue samples that could be used for genetic analysis.
Additionally, I think some feasibility studies for candidate species would be in order. One would
want to find species that were common enough to be sampled but are found in patchy
populations in agricultural areas. One would then want to attempt to calculate effective
population sizes and migration rates for these populations to see how much effort would need to
be invested to get confidence intervals down to reasonable levels for prospective monitoring.
For most taxa, there are not vast numbers of Ne estimates in the literature, so these would be
considered interesting from a scientific perspective alone.
74
Conclusions and Suggestions
This report takes a somewhat skeptical view with respect to the prospects for genetic population
monitoring of non-target species in agricultural areas. For what it’s worth, I had no particular
bias against this approach until I thought about it seriously in the preparation of this report.
Ultimately, I think there are a few basic problems that would have to be overcome before genetic
monitoring could be effectively used for monitoring non-target species.
First, while great advances have been made in estimating population parameters from genetic
data, these estimates are rarely very precise. The published literature tends to focus on the
maximum likelihood (or moment) estimates of parameters, but for detecting population trends,
the confidence intervals are much more important. For Ne estimation, confidence intervals are
typically very large – reaching down to a few individuals in smaller populations or up to infinity
in larger ones.
These wide confidence intervals will lead to low power to detect changes in genetic parameters
over time. For most of the populations listed in table 1, a decline of 50% or an increase of 100%
might be necessary before a significant change could be noted. In their review, Schwartz et al.
(2007) listed the advantage of monitoring Ne, as “high power,” yet they provided no data to
substantiate this claim. The one study I found that examined statistical power was Hansen et al.
2006 who simulated population declines in North Sea houting and the ability of Ne monitoring to
detect these declines. They concluded that genetic monitoring was indeed effective for detecting
declines. However, they simulated declines of either 91 or 96% of the population. Any kind of
reasonable count survey would detect a decline of this magnitude at a much lower cost.
The second basic problem is that genetic population monitoring does not seem particularly wellsuited for the general goal of monitoring non-target species in agricultural areas. Genetic
monitoring is potentially useful when there is only a single species of interest, when individuals
are very difficult to count and when populations are spatially well-defined. These criteria will
mostly apply to endangered species, particularly things like marine mammals, large terrestrial
mammals, and anadromous fish (and these represent most of the uses of genetic monitoring thus
far).
In contrast, when monitoring pesticide impacts one would want to monitor multiple species in
different habitats. However, one would have the luxury of choosing species that could be easily
counted. Additionally, because agricultural habitats tend to stretch over broad areas, finding
discrete populations might be difficult, at least in terrestrial habitats. Some of the most
interesting uses of population genetics involve comparisons between current population
parameters and historical parameters, but because of the large-scale changes in agricultural
landscapes, these sorts of studies would not be particularly informative with respect to pesticide
impacts.
There are some legitimate scientific issues to be investigated with respect to genetic monitoring.
For one, it would be interesting to do some head-to-head comparisons of genetic versus
traditional monitoring of non-target species. This could be done both with simulation models
and empirically on a few select populations. Based on this information, one could carry out a
75
formal cost benefit analysis of different monitoring methods. This kind of thing might make a
good project for an EPA post-doc.
Additionally, if multiple sets of archived tissue samples (from which DNA could be obtained)
were available from any species over a broad area, these could be used for retrospective
monitoring. Retrospective approaches do promise some real time savings over traditional
monitoring, as one would not need to wait 10-20 years in order to gauge environmental impacts.
Nevertheless, if I were considering setting up a large-scale, prospective monitoring program, I
would lean strongly towards more traditional methods. Specifically, I would randomly choose a
large number of sites within each region of interest. Depending on patterns of PIP use, I might
also stratify monitoring sites into low PIP sites and high PIP sites to obtain a good representation
of each. I would then carry out point count transects for birds, line transects for butterflies, and
aquatic surveys for either amphibians or invertebrates. I might also use baited traps for
honeybees. I would have as many sampling sites as possible, even if that meant sampling each
site only every few years (trends would not be seen year-to-year in any case). The data obtained
from this kind of monitoring program could be used to test for trends for individual species or for
entire guilds, and to ask whether population trends are related to the intensity of PIP and/or
pesticide use. This approach may not be particularly inspiring, but it is highly likely to yield
meaningful conclusions about the effects of PIPs and pesticide use on non-target species.
76
References
Alter SE, Rynes E, Palumbi SR (2007) DNA evidence for historic population size and past
ecosystem impacts of gray whales. Proceedings of the National Academy of Science USA 2007:
15162-15167.
Anderson EC, Williamson EG, Thompson EA (2000) Monte Carlo evaluation of the likelihood
for Ne from temporally spaced samples. Genetics 156: 2109-2118
Araki H, Ardren WR, Blouin MS (2007) A potential bias in the temporal method for estimating
Nein admixed populations under natural selection. Molecular Ecology 16: 2261–2271.
Ardren WR, Kapuscinski AR (2003) Demographic and genetic estimates of effective population
size (Ne) reveals genetic compensation in steelhead trout. Molecular Ecology 12: 35–49.
Asher J et al. (2001) The Millennium Atlas of Butterflies in Britain and Ireland. Oxford
University Press, Oxford UK.
Beaumont MA (1999) Detecting population expansion and decline using microsatellites.
Genetics 153:2013–2029.
Beerli P, Felsenstein J (2001) Maximum likelihood estimation of a migration matrix and
effective populations sizes in n subspopulations by using a coalescent approach. Proceedings of
the National Academy of Sciences, USA 98: 4563–4568.
Berthier P, Beaumont MA, Cornuet JM, Luikart G (2002) Likelihood-based estimation of the
effective population size using temporal changes in allele frequencies: a genealogical approach.
Genetics 160: 741–751.
Boulanger J, Himmer S, Swan C (2004) Monitoring of grizzly bear population trends and
demography using DNA mark-recapture methods in the Owikeno Lake area of British Columbia.
Canadian Journal of Zoology 82: 1267-1277.
Crouch WB, Paton PWC (2002) Assessing the use of call surveys to monitor breeding anurans in
Rhode Island. Journal of Herpetology 36: 185-192.
Estoup A., Wilson IA, Sullivan C, Cornuet JM, Moritz C. 2001. Inferring population history
from microsatellite and enzyme data in serially introduced cane toads, Bufo marinus. Genetics
159:1671–1687.
Funk WC, Tallmon DA, Allendorf FW. 1999. Small effective population size in the long-toed
salamander. Molecular Ecology 8: 1633–1640.
Frantz AC, Roper TJ (2006) Simulations to assess the performance of different rarefaction
methods in estimating population size using small datasets. Conservation Genetics 7: 315-318.
77
Gibbs JP, Droege S, and Eagle P. (1998) Monitoring populations of plants and animals.
BioScience 48: 935–940
Hansen MM, Nielsen EE, Mensberg KLD (2006) Underwater but not out of sight: genetic
monitoring of effective population size in the endangered North Sea houting (Coregonus
oxyrhynchus). Canadian Journal of Fisheries and Aquatic Sciences 63: 780–787.
Hayes TB et al. (2006) Pesticide mixtures, endocrine disruption, and amphibian declines: are we
underestimating the impact? Environmental Health Perspectives 114: 40-50.
Hill WG (1981) Estimation of effective population size from data on linkage disequilibrium.
Genetical Research 38: 209–216.
Hoffman EA, Scheuler FW, Blouin MS (2004) Effective population sizes and temporal stability
of genetic structure in Rana pipiens, the northern leopard frog. Evolution 58: 2536–2545.
Jehle R, Wilson GA, Arntzen JW (2005) Contemporary gene flow and the spatio-temporal
genetic structure of subdivided newt populations (Triturus cristatus, T-marmoratus). Journal of
Evolutionary Biology 18: 619–628.
Jensen LF, Hansen MM, Carlsson J, Loeschcke V, Mensberg KLD (2005) Spatial and temporal
genetic differentiation and effective population size of brown trout (Salmo trutta) in small
Danish rivers. Conservation Genetics 6: 615–621.
Johnson JA, Bellinger MR, Toepfer JE, Dunn P (2004) Temporal changes in allele frequencies
and low effective population size in greater prairie-chickens. Molecular Ecology 13: 2617–2630.
Jorde P, Ryman N (1995) Temporal allele frequency change and estimation of effective
population size in populations with overlapping generations. Genetics 139: 1077–1090.
Jorde P, Ryman N (1996) Demographic genetics of brown trout (Salmo trutta) and estimation of
effective population size from temporal change in allele frequencies. Genetics 143: 1369–1381.
Joseph LN, Field SA, Wilcox C, Possingham HP. 2006. Presence-absence versus abundance data
for monitoring threatened species. Conservation Biology 20: 1679–1687.
Lande R (1988) Genetics and demography in biological conservation. Science 241: 1555-1560.
Leberg P (2005) Genetic approaches for estimating the effective size of populations. Journal of
Wildlife Management 69: 1385–1399.
Link WA, Sauer, JR. (1998) Estimating population change from count data: Application to the
North American Breeding Bird Survey. Ecological Applications 8: 258–268.
Lucchini V., Galov A, Randi E (2004) Evidence of genetic distinction and long-term population
decline in wolves (Canis lupus) in the Italian Apennines. Molecular Ecology 13: 523–535.
78
Luikart G et al. (1999) Temporal changes in allele frequencies provide estimates of population
bottleneck size. Conservation Biology 13: 523-530.
Lukacs PM, Burnham KP (2005) Review of capture-recapture methods applicable to noninvasive
genetic sampling. Molecular Ecology 14: 3909–3919
MacKenzie DI, Nichols JD, Royle JA, Pollock KH, Hines JE, Bailey LL (2005) Occupancy
estimation and modeling: inferring patterns and dynamics of species occurrence. Elsevier, San
Diego , California.
Marsh and Trenham (2008) Current trends in monitoring programs for animal and plant
populations. Conservation Biology 22: 647-655.
Miller CR, Waits LP (2003) The history of effective population size and genetic diversity in the
Yellowstone grizzly (Ursus arctos): Implications for conservation. Proceedings of the National
Academy of Sciences USA 100: 4334–4339.
NERC Centre for Population Biology, Imperial College (1999) The Global Population Dynamics
Database. http://www.sw.ic.ac.uk/cpb/cpb/gpdd.html
Paetkau D (2003) An empirical exploration of data quality in DNA-based population inventories.
Molecular Ecology 12: 1375–1387
Pechmann JHK, Wilbur HM. (1994) Putting declining amphibian populations in perspective:
natural fluctuations and human impacts. Herpetologica 50: 65–84.
Pollock JF (2006) Detecting population declines over large areas with presence-absence, time-toencounter, and count survey methods. Conservation Biology 20: 882–892.
Prugh LR, Ritland CE, Arthur SM, Krebs CJ (2005) Monitoring coyote population dynamics by
genotyping faeces. Molecular Ecology 14: 1585-1596.
Rudnick JA, et al. (2005) Using naturally shed feathers for individual identification, genetic
parentage analyses, and population monitoring in an endangered Eastern imperial eagle (Aquila
heliacal) population from Kazakhstan. Molecular Ecology 14: 2959-2969.
Saccheri I, Kuussaari M, Kankare M, Vikman P, Fortelius W, et al. (1998) Inbreeding and
extinction in a butterfly metapopulation. Nature 392: 491–494.
Schwartz MK, Luikart G, Waples RS (2007) Genetic monitoring as a promising tool for
conservation and management. Trends in Ecology & Evolution 22: 25–33.
Shaffer HB, Fisher RN, and Davidson C. (1998) The role of natural history collections in
documenting species declines. Trends in Ecology and Evolution 13:27–30.
79
Storz JF, Beaumont M (2002) Testing for genetic evidence of population expansion and
contraction: An empirical analysis of microsatellite DNA variation using a hierarchical Bayesian
method. Evolution 56, 154–166.
Thomas JA (2005) Monitoring change in the abundance and distribution of insects using
butterflies and other indicator groups. Phil. Trans. R. Soc. B 360: 339-357
Waits JL, Leberg PL (2000) Biases associated with population estimation using molecular
tagging. Animal Conservation 3: 191–199.
Wang J (2001) A pseudo-likelihood method for estimating effective population size from
temporally placed samples. Genetical Research 78: 243–257.
Wang J, Whitlock MC (2003) Estimating effective population size and migration rates from
genetic samples over space and time. Genetics 163: 429–446.
Waples RS (1989) A generalized approach for estimating effective population size from
temporal changes in allele frequency. Genetics 121: 379–391.
Waples RS, Yokota M (2007) Temporal estimates of effective population size in species with
overlapping generations. Genetics 175: 217–233.
White GC (2005) Correcting wildlife counts using detection probabilities. Wildlife Research 32:
211-216.
80
Table 1. Examples of confidence intervals for Ne values estimated from allele frequency data. Time refers to the time between samples for estimates based on
temporally-spaced samples. Sample size refers to the number of individuals sampled.
Species
Citation
Estimation Method
Years
Sample Size
Loci
Ne
95% CI
Long-toed salamander
Long-toed salamander
Long-toed salamander
Long-toed salamander
Long-toed salamander
Funk et al. (1999)
Funk et al. (1999)
Funk et al. (1999)
Funk et al. (1999)
Funk et al. (1999)
Waples
Waples
Waples
Waples
Waples
18 or 19
18 or 19
18 or 19
18 or 19
18 or 19
21-57
21-57
21-57
21-57
21-57
4
4
4
4
4
207
132
23
64
191
8-∞
5-∞
1-∞
2-∞
6-∞
Crested newt
Marbled newt
Marbled newt
Jehle et al. (2001)
Jehle et al. (2001)
Jehle et al. (2001)
Waples
Waples
Waples
9
9
12
8/75
23/145
90/43
8
5
5
12
13
10
5 – 48
5 – 43
6 – 20
Brown trout
Brown trout
Palm et al. (2003)
Palm et al. (2003)
Jorde & Ryman
Jorde & Ryman
20
20
636
1392
17
17
19
48
11-35
34-71
Leopard frogs
Leopard frogs
Leopard frogs
Leopard frogs
Leopard frogs
Leopard frogs
Leopard frogs
Leopard frogs
Leopard frogs
Leopard frogs
Leopard frogs
Leopard frogs
Leopard frogs
Leopard frogs
Leopard frogs
Hoffman et al. (2004)
Hoffman et al. (2004)
Hoffman et al. (2004)
Hoffman et al. (2004)
Hoffman et al. (2004)
Hoffman et al. (2004)
Hoffman et al. (2004)
Hoffman et al. (2004)
Hoffman et al. (2004)
Hoffman et al. (2004)
Hoffman et al. (2004)
Hoffman et al. (2004)
Hoffman et al. (2004)
Hoffman et al. (2004)
Hoffman et al. (2004)
Waples
Wang
Wang & Whitlock
Waples
Wang
Wang & Whitlock
Waples
Wang
Wang & Whitlock
Waples
Wang
Wang & Whitlock
Waples
Wang
Wang & Whitlock
30
30
30
30
30
30
22
22
22
30
30
30
30
30
30
20-50
20-50
20-50
20-50
20-50
20-50
20-50
20-50
20-50
20-50
20-50
20-50
20-50
20-50
20-50
7
7
7
7
7
7
7
7
7
6
6
6
6
6
6
588
324
21
420
205
15
1019
243
16
410
102
15
1820
469
21
378-1355
230-488
17-26
245-837
150-295
13-18
490 - ∞
165-395
14-19
222-940
71-152
13-17
660-∞
313-786
18-24
Italian wolves
Lucchini et al. (2004)
Waples
6
26/24
18
283
79-495
Steelhead trout
Steelhead trout
Steelhead trout
Ardren & Kapuscinski (2003)
Ardren & Kapuscinski (2003)
Ardren & Kapuscinski (2003)
Waples
Waples
Waples
8
9
17
50/50
50/50
50/50
8
8
8
228
344
233
108-320
150-494
119-300
81
Notes
allozymes
allozymes
allozymes
allozymes
allozymes
sampled each year w/ allozymes
sampled each year w/allozymes
population #1
population #1
population #1
population #2
population #2
population #2
population #3
population #3
population #3
population #4
population #4
population #4
population #5
population #5
population #5
(Table 1 cont.)
North Sea houting
Hansen et al. (2006)
Beaumont
22
39/50
12
577
297-3720
90% credibility interval
Yellowstone Grizzly
Yellowstone Grizzly
Yellowstone Grizzly
Yellowstone Grizzly
Miller & Waits (2003)
Miller & Waits (2003)
Miller & Waits (2003)
Miller & Waits (2003)
Waples
Anderson
Wang
Waples
69
69
69
69
256
256
256
256
8
8
8
8
100
105
110
95
45-220
90-240
60-240
65-200
data read from graph
data read from graph
data read from graph
data read from graph
Prairie chicken
Prairie chicken
Prairie chicken
Prairie chicken
Prairie chicken
Prairie chicken
Prairie chicken
Prairie chicken
Prairie chicken
Prairie chicken
Prairie chicken
Prairie chicken
Johnson et al. (2004)
Johnson et al. (2004)
Johnson et al. (2004)
Johnson et al. (2004)
Johnson et al. (2004)
Johnson et al. (2004)
Johnson et al. (2004)
Johnson et al. (2004)
Johnson et al. (2004)
Johnson et al. (2004)
Johnson et al. (2004)
Johnson et al. (2004)
Waples
Wang
Wang & Whitlock
Waples
Wang
Wang & Whitlock
Waples
Wang
Wang & Whitlock
Waples
Wang
Wang & Whitlock
48
48
48
48
48
48
48
48
48
48
48
48
125/181
125/181
125/181
125/181
125/181
125/181
125/181
125/181
125/181
125/181
125/181
125/181
6
6
6
6
6
6
6
6
6
6
6
6
189
214
19
108
100
15
205
222
26
130
140
16
122-271
135-358
13-40
70-155
67-149
10-28
134-293
150-330
18-50
84-186
93-213
11-29
population #1
population #1
population #1
population #2
population #2
population #2
population #3
population #3
population #3
population #4
population #4
population #4
82
Figure 1A. Magnitude of confidence intervals on Ne in a selection of the published literature.
“Percent span” refers to the percent of the Ne estimate that is spanned by the lower and upper
bounds of the 95% confidence intervals.
20
0.4
0.2
5
0.1
0
0.0
0
100
200
300
400
500
600
Percent span on Ne estimates
83
∞
Bar
10
y
Frequency
0.3
Proportion per Bar
15
Figure 1B. Magnitude of confidence intervals based on observation error in count surveys. Data
are compiled from a range of species (N= 392 datasets) from the Global Population Dynamics
database. “Percent span” refers to the percent of the population index that is spanned by the
lower and upper bounds of the 95% confidence intervals.
250
0.5
150
0.3
100
0.2
y
Frequency
0.4
50
0.1
0
0
0.0
100 200 300 400 500 600 700 800 >800
Percent span on count CIs
84
Proportion per Bar
200
Appendix B.
Curricula Vitae of Panel Members
85
JOHN W. BICKHAM
Director of the Center for the Environment and
Professor of Forestry and Natural Resources
Purdue University
West Lafayette, IN 47907
phone: 765-494-5146
fax: 765-496-2422
bickham@purdue.edu
Current Research:
Evolutionary Toxicology
Genetic Studies of Bowhead Whales
Genetic Studies of Steller’s Sea Lions
Professional Positions:
2006 - present
Director, Center for the Environment, Purdue University
2006 - present
Professor, Department of Forestry and Natural Resources,
Purdue University
1986 - 2006
Professor, Faculty of Genetics, Texas A&M University
1986 - 2006
Professor, Faculty of Toxicology, Texas A&M University
1986 - 2006
Professor of Wildlife and Fisheries Sciences,
Texas A&M University
1983 - 1986
Associate Professor of Wildlife and Fisheries Sciences,
Texas A&M University
1976 - 1983
Assistant Professor of Wildlife and Fisheries Sciences,
Texas A&M University
Education:
Texas Tech University, Lubbock, TX
University of Dayton, Dayton, OH
University of Dayton, Dayton, OH
Ph.D., Zoology, 1976
M.S., Biology, 1973
B.S., Biology, 1971
Other Professional Activities and Awards:
Present
Editorial board of Ecotoxicology
2003 - present
Member of the US delegation to the Scientific Committee of
the International Whaling Commission
2001 - 2003
Editorial board of Environmental Toxicology and Chemistry
1991
Donald W. Tinkle Award for Research Excellence,
Southwestern Association of Naturalists
1985 - 1987
President of the Southwestern Association of Naturalists
Recent Grants and Contracts:
National Science Foundation (NSF), Resolving Mammalian Phylogeny Using
Genomic and Morphological Approaches, $600,000 (2006-2011), Co-PI
with William Murphy (PI) and Rodney Honeycutt (Co-PI).
NOAA and Alaska Department of Fish and Game, Steller sea lion genetics,
$335,000 (2002-2007 project funding began in 1992)
North Slope Borough, Bowhead whale genetics, $50,000 (2003);
86
$300,000 (2004-2007)
“Technical assistance to the offshore environmental monitoring program”, contract from
BP Azerbaijan, 2008 $81,000 J.W. Bickham PI.
“Identification of Taxonomic Status of Some Mammal Species of Azerbaijan” CRDF
and Azerbaijan National Science Foundation, $40,000 (my part $8,000) I.K.
Rahkmatulina (Azeri PI) and J.W. Bickham (US PI) 2008-2009.
Selected peer-reviewed publications (from a total of 180):
Baird, A. B., D. M. Hillis, J. C. Patton and J. W. Bickham. 2008. Evolutionary history of
the Genus Rhogeessa (Chiroptera: Vespertilionidae) as revealed by mitochondrial DNA
sequences. Journal of Mammalogy 89:744–754.
Wood, C. C., J. W. Bickham, R. J. Nelson, C. J. Foote, and J. C. Patton. 2008.
Recurrent evolution of life history ecotypes in sockeye salmon: implications for
conservation and future evolution. Evolutionary Applications 1:207-221.
Kenow K. P., D. J. Hoffman, R. K. Hines, M. W. Meyer, J. W. Bickham, C. W. Matson,
K. R. Stebbins, P. Montagna, and A. Elfessi. 2008. Effects of methylmercury exposure
on glutathione metabolism, oxidative stress, and chromosomal damage in captivereared common loon (Gavia immer) chicks. Environmental Pollution
doi:10.1016/j.envpol.2008.06.009.
Jorde, P. E., T. Schweder, J. W. Bickham, G. H. Givens, R. Suydam, D. Hunter, and N.
C. Stenseth. 2007. Detecting genetic structure in migrating bowhead whales off the
coast of Barrow, Alaska. Molecular Ecology 16:1993-2004.
O’Corry-Crowe G., B. L. Taylor, T. Gelatt, T. R. Loughlin, J. Bickham, M. Basterretche,
K. W. Pitcher, and D. P. DeMaster. 2006. Demographic independence along
ecosystem boundaries in Steller sea lions revealed by mtDNA analysis: implications for
management of an endangered species. Canadian Journal of Zoology, 84: 1796-1809.
Hoffman, J. I., C. W. Matson, W. Amos, T. R. Loughlin, and J. W. Bickham. 2006. Deep
genetic subdivision within a continuously distributed and highly vagile marine mammal,
the Steller’s sea lions Eumetopias jubatus. Molecular Ecology 15:2821-2832.
Harlin-Cognato, A., J. W. Bickham, T. R. Loughlin, and R. L. Honeycutt. 2005. Glacial
refugia and the phylogeography of Steller’s sea lion (Eumatopias jubatus) in the North
Pacific. Journal of Evolutionary Biology doi:10.1111/j.1420-9102.2005.01052.x.
Baker, Alyson R., Thomas R. Loughlin, Vladimir Burkanov, Cole W. Matson, Robert G.
Trujillo, Donald G. Calkins, Jeffrey K. Wickliffe, and John W. Bickham. 2005. Variation
Of Mitochondrial Control Region Sequences Of Steller Sea Lions, Eumetopias jubatus:
The Three-Stock Hypothesis. Journal of Mammalogy 86:1075-1084.
87
Politov, D. V., J. W. Bickham, and J.C. Patton. 2004. Molecular Phylogeography of
Palearctic and Nearctic Ciscoes. Ann. Zool. Fennici 41:13-23.
Trujillo, R. G., T. R. Loughlin, N. J. Gemmell, J. C. Patton, and J. W. Bickham. 2004.
Variation in microsatellites and mtDNA across the range of the Steller sea lion,
Eumetopias jubatus. Journal of Mammalogy 85:12-20.
Wynen L. P., S. D. Goldsworthy, S. J. Insley, M. Adams, J. W. Bickham, J. Francis, J. P.
Gallo, A. R. Hoelzel, P. Majluf, R. W.G. White, and R. Slade. 2001.
Phylogenetic relationships within the eared seals (Otariidae: Carnivora): Implications for
the historical biogeography of the family. Molecular Phylogentics and Evolution 21:270284.
Politov, D. V., N. Yu. Gordon, K. I. Afanasiev, Yu. P. Altukhov, and J. W. Bickham.
2000. Identification of Palearctic coregonid fish species using mtDNA and allozyme
genetic markers. Journal of Fish Biology 57 (Supplement A):51-71.
Cronin, M. W., and J. W. Bickham. 1998. A population genetic analysis of the potential
for a crude oil spill to induce heritable mutations and impact natural populations.
Ecotoxicology 7:259-278.
Bickham, J. W., T. R. Loughlin, D. G. Calkins, J. K. Wickliffe, and J. C. Patton. 1998.
Genetic variability and population decline in Steller sea lions from the Gulf of Alaska.
Journal of Mammalogy 79:1390-1395.
Bickham, J. W., J. A. Mazet, J. Blake, M. J. Smolen, Y. Lou, and B. E. Ballachey. 1998.
Flow-cytometric determination of genotoxic effects of exposure to petroleum in mink and
sea otters. Ecotoxicology 7:191-199.
Bickham, J. W., T. R. Loughlin, J. K. Wickliffe, and V. N. Burkanov. 1998. Geographic
variation in the mitochondrial DNA of Steller sea lions: haplotype diversity and
endemism in the Kuril Islands. Biosphere Conservation 1:107-117.
Bickham, J. W., J. C. Patton, S. Minzenmayer, L. L. Moulton, and B. J. Gallaway. 1997.
Identification of Arctic and Bering ciscoes in the Colville River delta, Beaufort Sea coast,
Alaska. Pp. 224-228, in Fish Ecology in Arctic North America (J. Reynolds, ed.).
American Fisheries Society Symposium 19, Bethesda, MD, 345 pp.
Bickham, J. W., J. C. Patton, and T. R. Loughlin. 1996. High variability for controlregion sequences in a marine mammal: implications for conservation and maternal
phylogeny of Steller sea lions (Eumetopias jubatus). Journal of Mammalogy 77:95-108.
Bickham, J. W., C. C. Wood, and J. C. Patton. 1995. Variation in mitochondrial
cytochrome b sequences and allozymes in sockeye (Oncorhynchus nerka). Journal of
Heredity 86:140-144.
88
Cronin, M. A., W. J. Spearman, R. L. Wilmot, J. Patton, and J. Bickham. 1993.
Mitochondrial DNA variation in chinook and chum salmon detected by restriction
enzyme analysis of polymerase chain reaction (PCR) products. Can. J. Fish. Aquat.
Sci. 50:708-715
Morales, J. C., B. G. Hanks, J. W. Bickham, J. N. Derr, and B. J. Gallaway. 1993.
Genetic analysis of population structure in Arctic cisco (Coregonus autumnalis) from the
Beaufort Sea. Copeia 1993:863-867.
Lockwood, S. F., and J. W. Bickham. 1992. Genome size in Beaufort Sea coastal
assemblages of Arctic cisco. Trans. Amer. Fish. Soc. 121:13-20.
Lockwood, S. F., and J. W. Bickham. 1991. Genetic stock assessment of spawning
Arctic cisco (Coregonus autumnalis) populations by flow cytometric determination of
DNA content. Cytometry 12:260-267.
Lockwood, S. F., B. T. Seavey, R. E. Dillinger, Jr., and J. W. Bickham. 1991. Variation
in DNA content among age classes of broad whitefish (Coregonus nasus) from the
Sagavanirktok River delta. Canadian Journal of Zoology 69:1335-1338.
Bickham, J. W. 1990. Flow cytometry as a technique to monitor the effects of
environmental genotoxins on wildlife populations. Pp. 97-108, in In Situ Evaluations of
Biological Hazards of Environmental Pollutants (S. S. Sandhu et al., eds.). Plenum Publ.
Corp., New York, 277 pp.
Bickham, J. W., S. M. Carr, B. G. Hanks, D. W. Burton, and B. J. Gallaway. 1989.
Genetic analysis of population variation in the Arctic Cisco using electrophoretic, flow
cytometric, and mitochondrial DNA restriction analyses. Biol. Pap. Univ. Alaska 24:112122.
89
ANDREW J. BOHONAK
Associate Professor
San Diego State University
Department of Biology, MC 4614
5500 Campanile Drive
San Diego, CA 92182
phone: 619-594-0414
fax: 619-594-5676
bohonak@sciences.sdsu.edu
Current Research:
Population genetics
Conservation genetics
Freshwater biology
Professional Positions:
2008 - present
Vice-Chair and Director of Undergraduate Advising and
Curriculum, Department of Biology, San Diego State University,
San Diego, CA
2007 - 2008
Coordinator, Evolutionary Biology Program Area,
Department of Biology, San Diego State University, San Diego, CA
2006 - present
Associate professor, Department of Biology,
San Diego State University, San Diego, CA
2000 - 2006
Assistant professor, Department of Biology,
San Diego State University, San Diego, CA
1999 - 2000
Postdoctoral fellow, Division of Insect Biology,
University of California, Berkeley, CA
1998 - 1999
Postdoctoral fellow, Center for Conservation Research and
Training, University of Hawaii at Manoa, Honolulu, HI
Education:
Cornell University, Ithaca, NY
Allegheny College, Meadville, PA
Ph.D., Ecology and Evolutionary Biology, 1998
B.S. Biology, 1991
Current Grants and Contracts:
California Department of Fish and Game and U.S. Fish and Wildlife Service. Traditional
Section 6 Grant, NCCP focus. Development of a monitoring protocol to quantify
population sizes for the San Diego fairy shrimp.
EDAW, Inc. Dry season fairy shrimp survey at Marine Corps Base Camp Pendleton.
Selected peer-reviewed publications:
Vandergast, A. G., A. J. Bohonak, S. A. Hathaway, J. Boys and R. N. Fisher. 2008.
Are hotspots of evolutionary potential adequately protected in southern California?
Biological Conservation 141:1648-1664.
Viaud-Martinez, K. A., R. L. Brownell, Jr., A. Komnenou and A. J. Bohonak. 2008.
Genetic isolation and morphological divergence of Black Sea bottlenose dolphins.
Biological Conservation 141:600-1611.
90
Vandergast, A. G., E. A. Lewallen, J. Deas, A. J. Bohonak, D. B. Weissman, and R. N.
Fisher. (in press). Loss of genetic connectivity and diversity in urban microreserves in a
southern California endemic Jerusalem cricket (Orthoptera: Stenopelmatidae:
Stenopelmatus “santa monica”). Journal of Insect Conservation (in press).
Bohonak, A. J. 2008. Genetic drift in human populations, in Encyclopedia of Life
Sciences, cross-listed in Handbook of Human Molecular Evolution. John Wiley and
Sons, Ltd. (http://www.els.net/ [doi: 10.1002/9780470015902.a0005440.pub2]).
Lewallen, E. A., T. W. Anderson and A. J. Bohonak. 2007. Genetic structure of leopard
shark (Triakis semifasciata) populations in California waters. Marine Biology 152:599609.
Vandergast, A. G., A. J. Bohonak, D. B. Weissman and R. N. Fisher. 2007.
Understanding the genetic effects of recent habitat fragmentation in the context of
evolutionary history: phylogeography and landscape genetics of a southern California
endemic Jerusalem cricket (Orthoptera: Stenopelmatidae: Stenopelmatus). Molecular
Ecology 16:977-992.
Viaud-Martinez, K. A., M. Martinez Vergara, P. E. Gol’din, V. Ridoux, A. A. Öztürk, B.
Öztürk, P. E. Rosel, A. Frantzis, A. Komnenou and A. J. Bohonak. 2007.
Morphological and genetic differentiation of the Black Sea harbour porpoise Phocoena
phocoena. Marine Ecology Progress Series 338:281-294.
Zickovich, J. M. and A. J. Bohonak. 2007. Dispersal ability and genetic structure in
aquatic invertebrates: a comparative study in southern California streams and
reservoirs. Freshwater Biology 52:1982-1996.
Bohonak, A. J., M. D. Holland, B. Santer, M. Zeller, C. M. Kearns and N. G. Hairston, Jr.
2006. The population genetic consequences of diapause in Eudiaptomus copepods.
Archiv für Hydrobiologie 167:183-202.
Bohonak, A. J. 2005. Genetic drift, in Encyclopedia of Life Sciences, John Wiley and
Sons, Ltd. (http://www.els.net/ [doi: 10.1038/npg.els.0005440]).
Jensen, J. L., A. J. Bohonak and S. T. Kelley. 2005. Isolation by Distance Web
Service. BMC Genetics 6:13. (IBDWS web site – http://ibdws.sdsu.edu/~ibdws/).
Bohonak, A. J., B. P. Smith and M. Thornton. 2004. Distributional, morphological and
genetic consequences of dispersal for temporary pond water mites. Freshwater Biology
49:170-180.
Mun, J., A. J. Bohonak and G. K. Roderick. 2003. Population structure of the pumpkin
fruit fly Bactrocera depressa (Tephritidae) in Korea and Japan: Pliocene allopatry or
recent invasion? Molecular Ecology 12:2941-2951.
91
Bohonak, A. J. and D. G. Jenkins. 2003. Ecological and evolutionary significance of
dispersal by freshwater invertebrates. Ecology Letters 6:783-796. (Invited Review).
Bohonak, A. J. 2002. IBD (Isolation By Distance): A program for analyses of isolation
by distance. Journal of Heredity 93:53-154. (Software available for download from
http://www.bio.sdsu.edu/pub/andy/IBD.html).
Bohonak, A. J. 2002. RMA: Software for Reduced Major Axis Regression. Software
available for download from http://www.bio.sdsu.edu/pub/andy/RMA.html).
Bohonak, A. J., N. Davies, F. X. Villablanca, and G. K. Roderick. 2001. Invasion
genetics of New World medflies: testing alternative colonization scenarios. Biological
Invasions 3:103-111.
Bohonak, A. J., and G. K. Roderick. 2001. Dispersal of invertebrates among temporary
ponds: are genetic estimates accurate? Israel Journal of Zoology 47:367-386. (Special
Issue: Ecology of temporary pools).
Bohonak, A. J. 1999. Dispersal, gene flow and population structure. Quarterly Review
of Biology 74:21-45.
Bohonak, A. J. and H. H. Whiteman. 1999. Dispersal of the fairy shrimp Branchinecta
coloradensis (Anostraca): effects of hydroperiod and salamanders. Limnology and
Oceanography 44:487-493.
Bohonak, A. J. 1999. Effect of insect-mediated dispersal on the genetic structure of
postglacial water mite populations. Heredity 82:451-461.
Hairston, N. G., Jr., L. J. Perry, A. J. Bohonak, M. Q. Fellows, C. M. Kearns and D. R.
Engstrom. 1999. Population biology of a failed invasion: paleolimnology of Daphnia
exilis in upstate New York. Limnology and Oceanography 44:477-486.
Wissinger, S. A., A. J. Bohonak, H. H. Whiteman and W. S. Brown. 1999. Subalpine
wetlands in Colorado: habitat permanence, salamander predation, and invertebrate
communities. Pages 757-790 in D. P. Batzer, R. B. Rader and S. A. Wissinger, editors.
Invertebrates in Freshwater Wetlands of North America: Ecology and Management.
John Wiley and Sons, New York.
Bohonak, A. J. 1998. Genetic population structure of the fairy shrimp Branchinecta
coloradensis (Anostraca) in the Rocky Mountains of Colorado. Canadian Journal of
Zoology 76:2049-2057.
Bohonak, A. J., N. Davies, G. K. Roderick and F. X. Villablanca. 1998. Is population
genetics mired in the past? Trends in Ecology and Evolution 13:360.
92
Hairston, N. G., Jr., and A. J. Bohonak. 1998. Copepod reproductive strategies:
lifehistory theory, phylogenetic pattern and invasion of inland waters. Proceedings of
the Sixth International Conference on Copepoda. Journal of Marine Systems 15:23-34.
Roderick, G. K., N. Davies, A. J. Bohonak and F. X. Villablanca. 1998. The interface of
population genetics and systematics: invasion genetics of the Mediterranean fruit fly
(Ceratatis capitata). Pages 489-499 in M. P. Zalucki, R. A. I. Drew and G. G. White,
editors. Proceedings of the 6th Australasian Applied Entomological Research
Conference, University of Queensland, Brisbane, Australia. University of Queensland
Printery.
Roderick, G. K., N. Davies , A. J. Bohonak and F. X. Villablanca. 1998. Origins of
Medflies trapped in California. Pages 21-25 in J. G. Morse and R. V. Dowell, editors,
Proceedings of the Exotic Fruit Fly Research Symposium. Regents of the University of
California, Riverside, California.
McClure, M. and A. J. Bohonak. 1995. Non-selectivity in extinction of bivalves in the
late Cretaceous of the Atlantic and Gulf Coastal Plain of North America. Journal of
Evolutionary Biology 8:779-794.
Whiteman, H. H., S. A. Wissinger and A. J. Bohonak. 1994. Seasonal movement
patterns in a subalpine population of the tiger salamander, Ambystoma tigrinum
nebulosum. Canadian Journal of Zoology 72:1780-1787.
Technical Reports:
Sustainable Ecosystems Institute. 2008. Scientific review of the draft Northern Spotted
Owl recovery plan. Provided expert review of unpublished genetic studies.
Bohonak, A. J. and A. G. Vandergast. 2006. Biodiversity in the southern California
coastal ecoregion: Phylogeographic congruence and evolutionary processes. California
State Parks. December 20, 2006.
Bohonak, A. J. 2005. Genetic testing of the endangered fairy shrimp species
Branchinecta sandiegonensis. Final report to City of San Diego and US Fish and
Wildlife Service. (Appendix to the City of San Diego's Vernal Pool Inventory). August
12, 2005.
Viaud, K. A., A. J. Bohonak, R. L. Brownell, Jr., A. Komnenou and L. M. Mukhametov.
2004. Conservation status of Black Sea bottlenose dolphin (Tursiops truncatus):
Assessment using morphological and genetic variation. Report to ACCOBAMS
scientific committee (Agreement on the Conservation of Cetaceans of the Black Sea,
Mediterranean Sea and contiguous Atlantic area, concluded under the auspices of the
Convention on the Conservation of Migratory Species of Wild Animals (CMS)).
September 3, 2004.
93
Computer Programs:
Jensen, J. L., A. J. Bohonak and S. T. Kelley. updated 2004-2008. Isolation by Distance
Web Service. described in Jensen et al. (2005). http://ibdws.sdsu.edu/
Bohonak, A. J. updated 2001-2004. IBD: Isolation By Distance. described in Bohonak
(2002). http://www.bio.sdsu.edu/pub/andy/IBD.html
Bohonak, A. J. and G. K. Roderick. updated 2000-2002. ESP: Evolution in Simulated
Populations. described in Bohonak et al. (2001).
http://www.bio.sdsu.edu/pub/andy/ESP.html
Bohonak, A. J. updated 2001-2004. RMA: Software for Reduced Major Axis
Regression. http://www.bio.sdsu.edu/pub/andy/RMA.html
Bohonak, A. J. and K. van der Linde. updated 2004-2008. RMA: Software for Reduced
Major Axis regression (for Java). http://www.kimvdlinde.com/professional/rma.html
Bohonak, A. J. 2001. MANTEL: Software for Mantel Tests.
http://www.bio.sdsu.edu/pub/andy/MANTEL.html
94
DAVID M. MARSH
Department of Biology
Washington and Lee University
Lexington, VA 24450
phone: 540-458-8176
fax: 540-458-8012
marshd@wlu.edu
Current Research:
Effects of habitat fragmentation of amphibian populations
Optimal design of population and biodiversity monitoring programs
Modeling salamander population dynamics
Methods for monitoring amphibian populations
Professional Positions:
Assistant Professor of Biology, Washington and Lee University, 2000-2006
Associate Professor of Biology, Washington and Lee University, 2006-present
Sabbatical Fellow, National Center for Ecological Analysis and Synthesis, 2006-2007
Education:
University of California, Davis
University of Virginia
Ph.D., 2000
B.A., 1993
Fellowships and Grants:
National Center for Ecological Analysis and Synthesis Sabbatical Fellowship: Optimal
Design of Population Monitoring Programs, $30000 (2006-2007)
Mellon Foundation: Optimal Design of Population Monitoring Programs, $21000 (20062007)
National Science Foundation RUI: Fragmentation of terrestrial salamander populations
by forest roads: ecological and genetic effects, w/ Paul R. Cabe, $375,000 (20032006)
National Science Foundation REU supplements: $7200 (2004-2005)
ACS: Biodiversity and Conservation in the Western Ghats of India, $3000 (2004)
Glenn grants for research, Washington and Lee University, $8400 (2001-2005)
NSF Postdoctoral Fellowship in Bioinformatics ($120,000, declined)
Smithsonian Predoctoral Fellowship, 1999
NSF Predoctoral Research Fellowship, 1996-1998
Selected peer-reviewed publications:
Marsh, D.M., and Trenham, P.C. 2008. Current trends in monitoring programs for
animal and plant populations.. Conservation Biology 22:647-655.
Marsh, D.M., Page, R.B., Hanlon, T.J., Corritone, R., Little, E.E., Seifert, D.E., and
Cabe, P.R. 208. Effects of roads on patterns of genetic differentiation in red-backed
salamanders, Plethodon cinereus. Conservation Genetics 9:603-613.
Marsh, D.M. and Hanlon, T.J. 2007. Seeing what we want to see: confirmation bias in
animal behavior research. Ethology 113:1089-1098.
95
Marsh, D.M. 2007. Edge effects of gated and ungated forest roads on terrestrial
salamanders. Journal of Wildlife Management 71:389-394.
Marsh, D.M, Page, R.B., Hanlon, T. J., Bareke, H. Corritone, R. Jetter, N., Beckman,
N.G., Gardner, K., Seifert, D.E., and Cabe, P.R. 2007. Ecological and genetic
evidence that low-order streams inhibit dispersal by red-backed salamanders
(Plethodon cinereus). Canadian Journal of Zoology 85:319-327
Cabe, P.R., Page, R.B., Hanlon, T.J., Aldrich, M.E., Connors, L., and D. M.
Marsh. 2007. Fine-scale genetic population structure and gene flow in a terrestrial
salamander living in continuous habitat. Heredity 98:53-60.
Marsh, D. M., Milam, G. S., Gorham, N. A., and Beckman, N. G. 2005. Forest roads as
partial barriers to salamander movement. Conservation Biology 19: 004-2008
Adams, V. M, Marsh, D. M., and Knox, J. S. 2005. Importance of seed banks for
population viability and population monitoring of an endangered wetland herb.
Biological Conservation 124:425-436.
Marsh, D. M., Thakur, K. A., Bulka, K., and Clarke, L. B. 2004. Dispersal and
colonization through open fields by a woodland salamander. Ecology 85:3396-3405.
Marsh, D. M. and Beckman, N.G. 2004. Effects of forest roads on abundance and
activity patterns of terrestrial salamanders in the Southern Appalachians. Ecological
Applications 14:1882-1891.
Marsh, D. M. and Hanlon, T. E. 2004. Gender and observation bias in animal behavior
research: experimental tests with red-backed salamanders. Animal Behaviour 68:14251433.
Marsh, D. M. and Goichochea, M. A. 2003. Monitoring terrestrial salamanders: biases
due to frequent sampling and choice of cover objects. Journal of Herpetology 37:460466.
Marsh, D. M. 2001. Amphibian population fluctuations: a meta-analysis. Biological
Conservation 101:327-335.
Marsh, D. M. 2001. Behavioral and demographic responses of túngara frogs to
variation in pond density. Ecology 82:1283-1293.
Marsh, D. M. and Borrell, B. 2001. Flexible oviposition strategies in túngara frogs and
their implications for tadpole spatial distributions. Oikos 93:101-109.
Trenham, P.C. and Marsh, D.M. 2001. Amphibian translocation programs: reply to
Seigel and Dodd. Conservation Biology 16:555–556.
96
Marsh, D. M. and Trenham, P.C. 2001. Metapopulation dynamics and amphibian
conservation. Conservation Biology 15:40-49.
Marsh, D. M., Rand, A. S., and Ryan, M. J. 2000. Effects of inter-pond distance on the
breeding ecology of tungara frogs. Oecologia 122:505-513.
Fegraus E.H., and Marsh, D.M. 2000. Are newer ponds better? pond chemistry,
oviposition site selection, and tadpole performance in tungara frog, Physalaemus
pustulosus. Journal of Herpetology 34:455-465.
Marsh, D. M., Fegraus, E. H., and Harrison, S. 1999. Effects of pond isolation on the
spatial and temporal dynamics of pond use in the tungara frog, Physalaemus
pustulosus. Journal of Animal Ecology 68:804-814.
Marsh, D. M. and Pearman, P. B. 1997. Effects of habitat fragmentation on the
abundance of two species of Leptodactylid frogs in an Andean montane forest.
Conservation Biology 11:1323-1328.
Manuscripts:
Marsh, D.M. Evaluating methods for sampling stream salamanders across multiple
observers and habitat types. Journal of Herpetology, (in review).
Marsh, D.M., and Haywood, L.E. Plot and Transect Surveys for Terrestrial Amphibians.
Chapter 14 in Conservation and Ecology of Amphibians: A Handbook of Techniques.
Oxford University Press, (under contract).
Invited Seminars:
University of Kentucky, Lexington, KY. Department of Biology. September 2006.
Vassar College, Poughkeepsie, NY. Department of Biology. November 2005.
James Madison University, Harrisonburg, VA. Department of Biology. September 2005.
San Diego State University, San Diego, CA. Department of Biology. February 2005.
Virginia Tech, Blacksburg, VA. Department of Biology. September 2004.
University of Richmond, Richmond, VA. Department of Biology. April 2004.
University of Florida, Gainsville, FL. Department of Zoology. February 2003.
College of William and Mary, Williamsburg, VA. Department of Biology. May 2002.
Old Dominion University, Norfolk, VA. Department of Biology. May 2002.
University of Virginia, Boyce, VA. Blandy Experimental Farm, July 2001.
Grant Reviewer for:
National Science Foundation Population Biology Program
National Science Foundation Ecology Program
National Science Foundation International Program
Jeffress Memorial Trust
97
Manuscript Reviewer for:
American Naturalist, Amphibia-Reptilia, Applied Herpetology, Biodiversity and
Conservation, Biological Conservation, Canadian Journal of Zoology, Conservation
Biology, Copeia, Ecological Applications, Ecology, Ethology, Environmental
Conservation, Heredity, Herpetologica, Herpetological Conservation, Journal of Animal
Ecology, Journal of Applied Ecology, Journal of Herpetology, Journal of Neotropical
Herpetology, Journal of Tropical Ecology, Oecologia, Restoration Ecology, Wetlands
Other Service:
NSF Population Biology Panel Reviewer
External PhD Examiner, McGill University
Mountain Lake Biological Station REU mentor
DuPont Minority Research Program mentor
W&L Environmental Studies Program
Goldwater Scholarship Campus Representative
98
AMY G. VANDERGAST
U.S. Geological Survey
Western Ecological Research Center
San Diego Field Station
4165 Spruance Road, Suite 200
San Diego, CA 92101
Phone: (619) 225-6445
Email: avandergast@usgs.gov
Current Research:
Using GIS (Geographic Information Systems) in spatially explicit analyses of population
genetic structure
Phylogenetics and genetic effects of recent habitat fragmentation on Jerusalem crickets
in Southern California
Molecular systematics of North and Central American Jerusalem crickets
Conservation genetics of the federally endangered Alameda whipsnake
Rapid identification of southern California fairy shrimp species from cysts
Narrow-headed gartersnake population genetics
Assessing population genetic structure and genetic diversity in the western shovelnosed snake
Ecological and genetic effects of habitat fragmentation on spiders in Hawaiian kipuka
Professional Positions:
2004 - present
Geneticist, USGS Western Ecological Research Center,
San Diego Field Station, San Diego, CA
2004 - present
Adjunct Assistant Research Professor, Department of Biology,
San Diego State University, San Diego, CA
2002 - 2004
NSF Postdoctoral Fellow, Department of Biology,
San Diego State University, San Diego, CA
Spring 2002
Adjunct Faculty, Department of Biology,
Cuyamaca College, El Cajon, CA
Education:
University of California, Berkeley
University of Hawaii at Manoa
University of California, San Diego
Ph.D., 2002
M.S., 1998
B.S.,1995
Other Professional Activities and Awards:
2005
USGS STAR Award. WERC, San Diego Field Station
2002 - 2004
NSF Postdoctoral Fellowship in Bioinformatics
2001
Forest Entomology Award. Insect Biology, UC Berkeley
2000
Sigma Xi Grants in Aid of Research Award
2000
Edward Steinhaus Memorial Award. Insect Biology, UC. Berkeley
2000
Taussig Memorial Endowment Fellowship. Graduate Division,
University of California, Berkeley, CA
1997 - 2000
Graduate Fellowship, EECB, University of Hawaii
1997 - 1998
Student Research Award, EECB, University of Hawaii
99
1996 - 1997
Graduate Fellowship, Research Corporation, University of Hawaii
Current Grants and Contracts:
Determining population genetic structure and diversity within the Alameda Whipsnake
(Masticophis lateralis euryxanthus). USFWS/USGS Science Support Partnership
Grant; Sacramento USFWO.
Rapid Identification of southern California fairy shrimp from cysts. USFWS/USGS Quick
Response Program Grant.
Narrow-headed gartersnake population genetics. Arizona State Heritage Funds;
Arizona Game and Fish Department.
Use of genetic techniques to evaluate the impacts of urbanization and fragmentation on
wildlife from wide-ranges to sedentary species. USGS Natural Resources
Preservation Program.
Assessing population structure and genetic diversity in the Western shovel-nosed snake
(Chionactis occipitalis). Arizona Game and Fish Department.
Selected peer-reviewed publications:
Vandergast, A. G., E. A. Lewallen, J. Deas, A. J. Bohonak, D. B. Weissman, and R. N.
Fisher. 2008. Loss of genetic connectivity and diversity in urban microreserves in a
southern California endemic Jerusalem cricket (Orthoptera: Stenopelmatidae:
Stenopelmatus “santa monica”). Journal of Insect Conservation (in press). DOI
10.1007/s10841-008-9176-z.
Wood, D. A., J. M. Meik, A. T. Holycross, R. N. Fisher, and A. G. Vandergast. 2008.
Molecular and phenotypic diversity in the Western Shovel-nosed snake, with emphasis
on the status of the Tucson Shovel-nosed snake (Chionactis occipitalis klauberi).
Conservation Genetics (in press). DOI: 10.1007/s10592-007-9482-0.
Vandergast, A. G. 2008. Kipuka in R. G. Gillespie, and D. A. Clague, editors.
Encyclopedia of Islands. University of California Press, Berkeley, CA. (in press).
Weissman, D, A. G. Vandergast, and N. Ueshima. 2008. Jerusalem crickets in J.
Capinera, editor. Encyclopedia of Entomology. Kluwer Academic Publishers,
Dordrecht, The Netherlands. (in press).
Vandergast, A. G., A. J. Bohonak, S. A. Hathaway, J. Boys and R. N. Fisher. 2008.
Are hotspots of evolutionary potential adequately protected in southern California?
Biological Conservation 141:1648-1664.
Vandergast, A. G., A. J. Bohonak, D. B. Weissman and R. N. Fisher. 2007.
Understanding the genetic effects of recent habitat fragmentation in the context of
evolutionary history: phylogeography and landscape genetics of a southern California
endemic Jerusalem cricket (Orthoptera: Stenopelmatidae: Stenopelmatus). Molecular
Ecology 16:977-992.
100
Wood, D. A., Vandergast, A. G., and R. N. Fisher. 2006. Assessment of Genetic
Diversity in the Western Shovel-nosed Snake (Chionactis occipitalis), with Special
Emphasis on the Subspecific Status of the Tucson Shovel-nosed Snake (C. o. klauberi).
Prepared for Arizona Game and Fish Department, U.S. Geological Survey, San Diego
Field Station.
Vandergast, A.G., R.G. Gillespie and G.K. Roderick. 2004. Influences of volcanic
activity on the population genetic structure of Hawaiian Tetragnatha spiders:
fragmentation, rapid
population growth and the potential for accelerated evolution. Molecular Ecology
13:1729-1743.
Vandergast, A.G. and R.G. Gillespie. 2004. Effects of natural forest fragmentation on a
Hawaiian spider community. Environmental Entomology 33:296-1305.
Vandergast, A.G., and G.K. Roderick. 2004. Mermithid parasitism of Hawaiian
Tetragnatha spiders in a fragmented landscape (addendum). Journal of Invertebrate
Pathology 85:136-137.
Vandergast, A.G. and G.K. Roderick. 2003. Mermithid parasitism of Hawaiian
Tetragnatha spiders in a fragmented landscape. Journal of Invertebrate Pathology
84(2):128-136.
Vandergast, A.G. 2002. Ecology and genetics of habitat fragmentation in spiders of
Hawaiian kipukas. Dissertation, ESPM (Environmental Science Policy and
Management), Divison of Insect Biology. University of California, Berkeley, CA.
Vandergast, A.G. 1998. Changes in spider community structure across a habitat
boundary (Abstract in 23rd Annual Albert L. Tester Memorial Symposium (Manoa,
Hawaii, USA; April 5-7, 1998). Pacific Science 53(1):113-122.
101
Appendix C.
Project Background and Questions Sent to Panel
102
EPA’s Office of Research and Development (ORD) is interested in your input regarding
design considerations for an efficient monitoring program to help assess and predict
long-term effects of pesticides and insect resistant crops on non-target organisms. We
are assembling an expert panel to provide recommendations regarding potential use of
population genetic approaches in ecosystem monitoring.
We request that you draw upon your experience in ecological and population genetics
to provide ORD with insights regarding fundamental issues and research questions that
should be addressed prior to implementing a large-scale population genetic monitoring
program. It is important to note that we are not asking for an actual monitoring program
design, but for your expert recommendations regarding the most important topics for
consideration prior to developing such a program. The following list of questions should
be considered but are not meant to constrain your thinking:
7. What are the advantages and disadvantages of a population genetic approach for
large-scale monitoring compared to traditional census approaches?
8. What information is likely to be obtained using a genetic monitoring approach that
would not be available from traditional monitoring approaches?
9. What are the most appropriate genetic measures to evaluate?
10. Are particular species, groups of species, functional guilds, etc. more informative
than others and more cost-effective for monitoring long-term ecosystem responses
to changing agricultural practices?
11. How should genetic monitoring be combined with other measures, methods, and
models to achieve the most efficient assessment of current and predicted ecosystem
responses to changing agricultural practices
12. What are the key knowledge gaps and research questions that must be addressed
before designing a genetic monitoring program?
Background:
EPA's Office of Pesticide Programs (OPP) is responsible for evaluating and registering
pesticides under the Federal Insecticide, Fungicide, and Rodenticide Act (FIFRA).
Before a pesticide can be marketed in the United States, FIFRA requires an evaluation
to assure that it will not pose unreasonable risks of harm to human health and the
environment. Plant incorporated protectants (PIPs) and the genetic material necessary
for the plant to produce the substance are subject to OPP's pesticide registration
although the plant itself is not. The pesticide registration process includes an
assessment risk to non-target species. OPP also has a mandate to reduce pesticide
risk and actively promotes pest management policies that result in better protection of
human health and the environment.
Genetically modified crop plants that express pesticidal proteins derived from Bacillus
thuringiensis are believed to offer environmentally benign alternatives to broad
spectrum pesticides. Laboratory and short-term field studies have shown that Bt-crops
do not pose a significant risk to non-target insect populations Short-term studies
conducted under controlled conditions are not necessarily predictive of long-term
population responses that may occur following the extensive adoption of GM crops,
103
introduction of new variants of GM crops, and changes in usage patterns of
conventional broad-spectrum pesticides in complex interacting agricultural ecosystems.
Post-marketing monitoring is generally not required by OPP and there is currently no
mechanism in place to evaluate long-term population responses of non-target
organisms potentially exposed to multiple insect resistant crops, conventional
pesticides, other stressors, and different agricultural practices.
Adoption of insect resistant crops may provide environmental benefits by reducing the
use of conventional broad-spectrum pesticides and the risks they pose to human health
and the environment. Targeting pest species is expected to reduce the impact on nontarget organisms, maintain biodiversity, and result in less disruption of ecosystem
functions. In particular, maintaining populations of natural predator species could lead
to a positive feedback loop in which natural control of pests reduces the need for
conventional pesticides. While there is some evidence that GM crop adoption has
reduced application of conventional pesticides, direct environmental benefit has not yet
been demonstrated. Furthermore, existing data may not be sufficient to clearly
establish the degree to which GM crop adoption, separate from other environmental and
agricultural trends, has resulted in decreased detrimental effects on human health and
non-target organisms in agricultural and other ecosystems at national or regional
scales.
Long-term and large-scale monitoring program may be required to provide data for the
comprehensive evaluation of potential long-term effects of pesticides (including PIPs)
on non-target organisms. A monitoring program that clearly links GM crop usage to
conventional pesticide exposure and ecological effects would serve as an ecological
accountability tool for OPP to evaluate the effectiveness of pesticide registration
process and efforts to reduce risks to human health and the environment. It would
provide data on the effectiveness of EPA policies and regulations for society, Congress,
and OMB.
Monitoring programs can provide enormously useful information and are extraordinarily
variable in scope, purpose, duration, design, and success. The one unifying feature of
large-scale monitoring programs is the enormous cost involved. This is especially true
when the questions to be addressed are not targeted to particular species. When great
expense of traditional population monitoring approaches meet budgetary reality, the
scope of monitoring program may be constrained to the point that the information it can
provide may not justify the cost of implementation.
What is needed is a cost-effective approach for monitoring long-term population-level
responses to pesticides. Molecular and population genetics have the potential to
provide this cost-effective approach to assess ecological responses and demonstrate
the effectiveness of pesticide regulations. In addition, such data will contribute to the
development of population and landscape genetic models to provide spatially explicit,
testable predictions of long-term population responses. Combining these predictions
with information on the geographic distribution of pesticide exposures and other
environmental stressors will lead to improved risk assessments.
104
105
106