Professional Documents
Culture Documents
Population Genetic Assignment of Confiscated Gopher Tortoises
Population Genetic Assignment of Confiscated Gopher Tortoises
ABSTRACT As gopher tortoises (Gopherus polyphemus) increasingly become threatened throughout their range in Florida, USA, the need
for management and conservation will intensify. Here we evaluate the forensic applicability of genetic assignment tests based on microsatellite
genotypic data to 1) accurately assign individuals in our genetic database to the sample location or population of origin and 2) determine the
origin of 6 confiscated tortoises. Overall, we could correctly assign 90% of the individuals in the database to their population of origin, but we
were unable to determine the source of the confiscated tortoises. However, these individuals are unlikely to have come from any of our sampled
sites and all 6 may have come from the same population. This approach can be used by law enforcement personnel to identify the origin of
confiscated tortoises as well as by developers and wildlife managers to determine the genetic appropriateness of potential recipient populations
when it is necessary to relocate individuals. (JOURNAL OF WILDLIFE MANAGEMENT 72(1):254–259; 2008)
DOI: 10.2193/2006-243
KEY WORDS Florida, forensic, genetic assignment, gopher tortoise, Gopherus polyphemus, Program STRUCTURE, relocation.
The gopher tortoise (Gopherus polyphemus) is found from the and Stobeck 1998). From a management perspective, this is
extreme southern edge of South Carolina to the south- particularly useful when it is necessary to relocate animals
eastern corner of Louisiana, and throughout most of with a desire to maintain the genetic architecture of a
Florida, USA. Populations of gopher tortoises are becoming species. From a law enforcement perspective, molecular
increasingly threatened by extirpation throughout their genetic techniques are powerful forensic tools for determin-
range (Auffenberg and Franz 1982, Diemer 1986, Bury ing the species identity or source population for confiscated
and Germano 1994), primarily due to real estate develop- individuals (Shivji et al. 2002, Cassidy and Gonzales 2005).
ment. As this trend likely will continue, the need for Highly variable nuclear microsatellites are successfully
conservation of this species will intensify. Presently, the used in the identification of individuals and populations
gopher tortoise is federally listed in Appendix II of the (Immel et al. 1999, Boyd et al. 2001). Assignment tests
Convention on International Trade in Endangered Species designed for use with genetic data can identify unique
of Wild Fauna and Flora (Inskipp and Gillett 2003), and at genetic signatures for each population (or sampling
the state level as Threatened in every state where it occurs location). Subsequently, individuals of unknown origin can
except in Florida, where it is listed as a Species of Special be assigned with specific probabilities to populations using a
Concern by the Florida Fish and Wildlife Conservation genetic database (Pritchard et al. 2000). Analysis of
Commission. Although recently, the gopher tortoise was molecular variance approaches (AMOVA; Excoffier et al.
approved for Threatened status in Florida, contingent upon 1992, Schneider et al. 2002) using estimates of population
a management plan being approved (Florida Fish and genetic diversity can cluster groups in a hierarchical fashion,
Wildlife Conservation Commission 2006). Many technical thereby identifying genetically cohesive groups. Thus,
advances (Geographic Information Systems, spatial statis- combining the 2 approaches allows the assignment of
tics, DNA fingerprinting, etc.) now offer innovative ways of individuals to local breeding groups, groups to populations,
assisting wildlife managers in the difficult task of assessing and populations to species.
and maintaining threatened populations. Particularly, ge- Our goals for this project were 2-fold. First, we wished to
netic data is useful in assessing the degree of population determine how accurately assignment tests based on micro-
subdivision in geographically widespread species (Avise satellite data determine the location of origin of unknown
1994). These data also are useful in assessing the degree individuals. To do this, we tested how frequently individuals
of evolutionary distinctiveness of presumed isolated pop- in our database with known location information would be
ulations or subspecies (Avise 2000). By accurately defining successfully assigned to the correct sample. Second, we
evolutionary significant units, wildlife managers are better wanted to apply this approach to determine the likely source
able to apportion limited conservation resources to unique population of 6 tortoises confiscated from an individual
and at-risk groups. Genetic approaches also can be used to smuggler by agents of the South Carolina Department of
assign individuals to specific populations or population Natural Resources (SCDNR).
subsegments in species that have genetically differentiated
and geographically localized population segments (Waser STUDY AREA
1
Present address: 2012 Molecular Biology, Iowa State University, Gopher tortoises were collected from locations throughout
Ames, IA 50011, USA Florida (except the Panhandle) and in southern Georgia,
2
E-mail: skarl@hawaii.edu USA. Tortoises were captured in xeric scrub, longleaf pine
Sample Population
Assignment Assignment
Collection location and sample abbreviationa n HO SD A a FST Prob SD % Location FST Prob SD %
GA
Savannah River Ecology Laboratory (SREL)b 17 0.27 0.23 19 0 0.488 0.88 0.14 100 SREL 0.488 0.90 0.13 100
Moody Air Force Base, Lowndes County (MB) 14 0.46 0.22 33 2 0.309 0.78 0.22 93 MB 0.309 0.81 0.20 93
Jones Research Center, Baker County ( JR) 20 0.53 0.22 30 2 0.336 0.90 0.05 100 JR 0.336 0.91 0.04 100
FL
Big Shoals State Park, Columbia County 10 0.56 0.25 35 3 0.197 0.33 0.29 20 North FL 0.156 0.71 0.26 85
Cedar Key Scrub Preserve, Levy County, and 12 0.51 0.30 36 3 0.224 0.52 0.33 67
Ichetucknee Springs State Park,
Columbia County
Ashton Biological Preserve, Alachua County 18 0.45 0.24 35 1 0.198 0.42 0.27 56
Goldhead Branch and Cecil Field state parks, Duval 13 0.40 0.21 35 2 0.126 0.57 0.34 69
and Clay counties
Brooker Creek Preserve and Fort Cooper State Park, 21 0.45 0.22 27 2 0.224 0.58 0.25 86 Middle FL 0.197 0.63 0.26 75
Hillsborough and Citrus counties
Lake Louisa State Park, Lake County 15 0.36 0.21 24 1 0.190 0.31 0.20 53
Highlands Hammock State Park, 19 0.41 0.19 23 0 0.266 0.47 0.28 68 South FL 0.277 0.74 0.21 93
Highlands County
Cayo Costa State Park, Lee County 21 0.34 0.17 22 0 0.327 0.69 0.22 90
Jonathan Dickenson State Park, Martin County ( JD) 20 0.45 0.27 39 7 0.152 0.82 0.12 100 JD 0.152 0.85 0.11 100
Totals or averages 200 0.43 0.08 77 23 0.63 0.30 77 0.76 0.23 90
a
Sample location abbreviations correspond to Schwartz and Karl (2005).
b
This population was originally collected from coastal GA (McIntosh County) and was relocated to Savannah River Ecology Laboratory, Aiken, SC, USA.
(Pinus palustris), and sandhill habitat or where there is Park, Oldenburg Mitigation Park, and the University of
sandy, well-drained soil. Scrub habitat is characterized by South Florida Eco Area; Schwartz and Karl 2005). We also
low-growing shrubs (oaks [Quercus spp.], palmettos [Serenoa eliminated sample locations that had very small sample sizes
repens], etc.) with or without a canopy of pine trees. (i.e., Wakulla Springs State Park and Everglades National
Longleaf pine and sandhill habitats are similar, except they Park; Schwartz and Karl 2005) and individuals that were
are dominated by tall longleaf pines and open, grassy shown to be migrants that had moved or been moved by
groundcover. humans into different populations (Schwartz and Karl
2005). Finally, we combined sampling locations that had
METHODS small sample sizes and were not statistically genetically
Blood samples from 6 confiscated tortoises (referred to as different: Brooker Creek Preserve and Fort Cooper State
confiscated individuals 1 to 6) were sent to us from Park, Hillsborough and Citrus counties, Florida; Goldhead
SCDNR. We isolated DNA using a standard phenol and Branch and Cecil Field state parks, Duval and Clay
chloroform technique (Herrmann and Frishchauf 1987, Karl counties, Florida; and Ceder Key Scrub Preserve, Levy
et al. 1992) and assayed the test individuals at the same 9 County, and Ichetucknee Springs State Park, Columbia
microsatellite loci used to develop our gopher tortoise County, Florida (Table 1). The final database consisted of
genetic database (Schwartz and Karl 2005). General 200 individuals from 12 sampling locations and is a subset of
information on the number and frequency of alleles and individuals in the original database that provides a reliable
heterozygosity estimates for these loci is found in Schwartz genetic signature of the population structure prior to human
(2003) and Schwartz et al. (2003). disturbances (Schwartz and Karl 2005). We captured and
Gopher tortoises have been relocated extensively in certain blood-sampled animals for this study under a Florida Fish
areas of Florida primarily for management purposes, and and Wildlife Service permit (WV01274) and The University
some of our sample locations were known to be recipients of of South Florida Institute for Animal Care and Use permit
relocated tortoises. As these relocation events can obscure (1742).
the true (historical) genetic signature of the populations, for We estimated pairwise FST values between all sample
this exercise, we modified the original genetic database of locations and populations including a post hoc confiscated
Schwartz and Karl (2005) by removing sample locations that group consisting of the 6 tortoises from SCDNR. We
were demonstrated to be recipients of translocated individ- verified that the hierarchical levels of genetic subdivision of
uals (i.e., Wekiwa Springs State Park, Boyd Hill Nature the modified database, based on FST estimates using an
Analysis
an additional individual from GC assigned to the con- did. The 2 out of the 72 known individuals tested that
fiscated population in analysis IV. assigned to the confiscated population were both from the
more northern GC sample location, which is consistent with
DISCUSSION the general more northern association of the confiscated
In our evaluation of the forensic applicability of genetic individuals. Even so, when trying to assign individuals to
assignment tests based on microsatellite genotypic data, we sample locations (i.e., analyses I and III), the estimated allele
could correctly assign 90% of the individuals in the database frequencies in the database may not represent the true allele
to their population of origin. We were, however, unable to frequencies in the sampling locations because of too few
determine the source of the 6 confiscated tortoises. These individuals in the group. If our allele frequency estimates are
individuals, however, are unlikely to have come from any of not accurate, the assignment probabilities for the confiscated
our sampled sites and all 6 may have come from the same individuals may be over- or underestimates relative to the
population. Based on the internal consistency tests, the probabilities if the true allele frequencies were known.
individuals in the database assign with high probability back Although we were unable to confidently assign the
to their populations of origin and in many cases to specific confiscated individuals to a sample or population, we can
sample locations. From this, we are confident that the draw some specific conclusions. First, assignment tests based
assignment program can effectively distinguish the samples on genetic data from individuals included in this study are
and populations and would robustly assign the confiscated robust, highly useful, and are likely to be beneficial for
individuals to the correct group if it were in the database. gopher tortoise conservation as they have been for other
The low probability of assignment of the confiscated species (Vazquez-Dominguez et al. 2001, Guinand et al.
individuals to any group in the database indicates that their 2004, Paetkau et al. 2004). Second, if the confiscated
true population of origin likely is not represented. individuals originated in Florida, they must be from an
Consistently assigning confiscated tortoises to their own unsampled population that is genetically quite different
population indicates that at least 5 of the 6 confiscated from the 12 geographically widespread sample locations we
tortoises are likely to have originated from a single have assayed. We think that this is unlikely because our
population somewhere out of the sampling range of this geographic coverage of Florida is fairly complete (with the
database. This conclusion is further supported by the finding exception of the Florida Panhandle area). These individuals
of 3 alleles not represented in the modified database in 3 of are probably not from Florida but come from the part of the
the 6 confiscated tortoises. With more samples from range that we have not sampled. Third, given the repeated
throughout the range (i.e., more northern locations), we assignment of the confiscated individuals to the North
probably could tell with confidence from where these Florida population (particularly northeast locations), a
individuals were taken. genetic survey of gopher tortoise populations along the east
In analyses III and IV, the consistent assignment of the coast likely would reveal the true population of origin. We
known database individuals (who had their source informa- cannot, however, confidently extrapolate from these data to
tion set to unknown) to the correct group of origin indicates predict more precisely where that population would be.
that the small number of confiscated individuals did not Finally, the biggest restriction to this study was the limited
itself positively bias the assignment of the confiscated reference genetic database with which to compare the
individuals to their own population. If so, we would expect individuals of unknown origin. We think that expanding the
that the known individuals also would have assigned to the database to include the Florida Panhandle, more of the east
confiscated population as often as the confiscated individuals coast, and the remainder of the non-Florida range would