Identifying geographic barriers that define genetic structure within a species is crucial in formulating an effective conservation plan. The identification of appropriate management units is critical for the protection and recovery of the gopher tortoise Gopherus polyphemus, which have declined across their entire range. Previous molecular work at various spatial scales has identified distinct population assemblages of the gopher tortoise. The goal of this study was to assess the genetic structure in gopher tortoises through a more complete sampling of the federally listed as threatened portion of the range and evaluate the extent of genetic isolation imposed by several potential geographic barriers. We sequenced a 712–base-pair portion of a mitochondrial gene (NADH dehydrogenase 4) for 322 individuals from 42 sites across the range. We found two major assemblages of haplotypes separated by a modest phylogenetic break (average uncorrected p distance = 0.015). The biogeographic barrier that best explained the geographic partitioning of genetic variation was the Apalachicola–Chattahoochee rivers and not the one used to delimit the federally listed as threatened portion of the range (Tombigbee–Mobile). However, the presence of distinct (group 1 and 2) haplotypes on either side of Apalachicola–Chattahoochee rivers indicates that the two lineages experienced historical isolation and divergence, after which they came back into contact. If one were to define genetic units of conservation for gopher tortoises, then the Apalachicola–Chattahoochee rivers delineation would be the most appropriate based on the analysis of molecular variance of the mitochondrial sequence data. However, a model that combines the Apalachicola–Chattahoochee and Tombigbee–Mobile rivers as geographic breaks was the second-best model in this analysis, which suggests that the federally listed as threatened portion of the range also contains important geographic structure. Thus, we recommend that making management decisions on the basis of mitochondrial data alone is premature, and that prior to any status review additional work that examines finer scale patterns of genetic structure by using microsatellite loci is required.
The gopher tortoise Gopherus polyphemus (Figure 1) is restricted to the Southeastern United States, where the species' distribution is intrinsically linked to the longleaf pine Pinus palustris ecosystem over most of its range. The gopher tortoise is an epigean species (i.e., a species living on or near the surface of the ground), which constructs numerous subterranean burrows throughout the landscape. These burrows play a critical role in the ecosystem by providing microhabitat for over 300 species (Jackson and Milstrey 1989). In the 1980s, populations of gopher tortoises were estimated to have declined by 80% since the 1800s (Auffenberg and Franz 1982). More recent research reports population declines are still occurring throughout the species' range (McCoy and Mushinsky 1992; Mushinsky et al. 2006; Waddle et al. 2006; Hammond 2009), including on managed (e.g., prescribed burns), protected lands (McCoy et al. 2006). Despite range-wide declines, only gopher tortoises west of the Tombigbee and Mobile rivers (Figure 2) are federally listed as threatened under the U.S. Endangered Species Act (ESA 1973, as amended; USFWS 1987). It was recently determined that gopher tortoises in the remaining portion (nonlisted) of the species' range also warrant listing as threatened (USFWS 2011). However, higher priority listing actions precluded this action and gopher tortoises in the nonlisted range of the species were added to the “candidate species” list (as defined by subparagraph (C) of section 4(b)(3) of ESA 1973, as amended; USFWS 2001).
The importance of identifying phylogeographic structure within a species is crucial in formulating an appropriate management strategy. The failure to manage genetically unique populations or regions could precipitate a loss of genetic diversity and local adaptations that are essential for the evolutionary potential of the species (Hilborn et al. 2003; Luck et al. 2003). In recognition of the value of genetically unique populations, the 1996 amendment to the ESA of 1973 called for including the protection of unique populations of a species by designating them as distinct population segments (DPS) on the basis of discreteness, significance, and conservation status (USFWS and NOAA 1996). Although the USFWS listed gopher tortoise as threatened pursuant to the ESA in only a portion of its range (Louisiana, Mississippi, and Alabama west of the Tombigbee and Mobile rivers) prior to the legal concept of DPS, its listing was based on several other factors pertaining to its conservation status, such as “the present or threatened destruction, modification, or curtailment of its habitat or range,” “overutilization for commercial, recreational, scientific, or educational purposes,” “disease or predation,” “the inadequacy of existing regulatory mechanism,” and “other natural or manmade factors affecting its continued existence” (USFWS 1987). Interestingly, the entire distribution of the gopher tortoise (i.e., the six southeastern states of Louisiana, Mississippi, Alabama, Georgia, Florida, and South Carolina) was recognized by USFWS as having these same issues even at the time the populations west of the Tombigbee and Mobile rivers were listed (USFWS 1987). Thus, the populations that are currently federally protected as threatened represent only a relatively small portion of this distribution where populations are experiencing declines.
Many species with large distributions possess intraspecific genetic structure (Avise 2000), and the gopher tortoise is no exception. Several studies, although focused on the nonlisted portion of the range of the gopher tortoise, have detected significant population structure with both mitochondrial DNA (mtDNA; Osentoski and Lamb 1995) and nuclear microsatellites (Schwartz and Karl 2005). In peninsular Florida, populations of gopher tortoises exhibited genetic structure that was associated with two systems of xeric upland ridges (Atlantic Coastal Ridges and a central ridge system consisting of Mt. Dora, Lake Wales, and Bombing Range ridges; Branch et al. 2003). Other species exhibit genetic structure associated with these ridge systems as well (Clark et al. 1999; Branch et al. 2003). To date, the study with the broadest range of sampling, Osentoski and Lamb (1995), had only one site within the federally listed as threatened portion of the range in Louisiana. Even with limited sampling within the federally listed as threatened portion of the range, Osentoski and Lamb (1995) identified the Apalachicola River as the delineation between two assemblages of populations, but with some overlap. Thus, a comprehensive range-wide phylogeographic study has not been conducted, which limits our ability to comprehensively define important conservation units within the species.
The goal of this study was to conduct a more thorough assessment of population structure in the gopher tortoise by including multiple populations from Mississippi and western Alabama that fall within the federally listed as threatened portion of the range. In particular, we wanted to investigate the extent to which several river systems act as biogeographic boundaries to define genetic structure across the species' range. These data will allow us to more explicitly evaluate the extent of genetic isolation and divergence among populations, and thereby aid federal and state agencies in making decisions on the legal protection of, and conservation or management efforts for, gopher tortoises.
Collections and sequencing
Tissue samples (i.e., blood or shell pieces) were obtained under the appropriate permits by either trapping efforts by the authors or donations made by various researchers (see Acknowledgments for a complete list). This collaborative effort yielded 322 individuals from 42 sites throughout the species' range. For the adult gopher tortoises captured by the authors, Tomahawk Model 18 Live Traps (81.28 × 25.4 × 30.48 cm) and custom-designed (71.12 × 35.56 × 27.94 cm) traps from Tomahawk were used. A blood sample (0.5–1 mL) was collected from the femoral or brachial veins using 23-gauge needles and 1-mL syringes. Each blood sample was stored in a 1.5-mL vial with approximately 0.5 mL of tissue-preservation buffer (Seutin et al. 1991). Total genomic DNA was extracted from the tissue samples using the DNeasy extraction kit (QIAGEN Inc., Valencia, CA).
We used the polymerase chain reaction to amplify a portion of the mitochondrial NADH dehydrogenase 4 (ND4) gene. Initial amplifications and sequencing were conducted using the ND4 primers reported by Spinks and Shaffer (2005). Based on sequences from these, we designed internal primers (5′-AAACTTGGAGGATACGGCATT-3′ and 5′-CCCTTAAAAGTGAGGGAGCTG-3′). In a total volume of 25 µL or 50 µL, polymerase chain reaction conditions consisted of 50 mM KCl, 10 mM Tris-HCl (pH 8.3), 0.01% gelatin, 1.0 mM MgCl2, 200 µM dNTPs, 0.19 units of Taq polymerase (Promega, Madison, WI), 0.3 µM of the forward and reverse primer, 20–100 ng of template DNA, and water to the final volume. The cycling conditions consisted of an initial 1 min denaturing step at 95°C followed by 30 cycles of 1 min at 95°C, 1 min at 55°C and 3 min at 72°C. A final elongation step of 7 min at 72°C completed the cycle. Amplifications were cleaned with Exo-Sap (USB Corp. Cleveland, Ohio), and then used as a template in a cycle-sequencing reaction using the ABI BigDye Terminator v 1.1 cycle-sequencing kit (Foster City, CA). All sequencing reactions were cleaned with sephadex (Princeton Separations, Adelphia, NJ) prior to gel runs at the Iowa State University DNA Sequencing and Synthesis Facility. Sequence data were edited and aligned using Sequencher v4.1 (GeneCodes Co., Madison, WI).
To assess relationships among haplotypes and their frequency and geographic distribution across the landscape, we created a haplotype network using TCS (Clement et al. 2000). We calculated pairwise uncorrected p distances (i.e., proportion of base-pair differences) between all haplotypes using PAUP* 4.0b10 (Swofford 2002). For each site, we used Arlequin 3.11 (Excoffier et al. 2005) to calculate genetic diversity indices, including haplotype diversity (h) and nucleotide diversity (π). Tests of demographic history were also performed in Arlequin 3.11. A mismatch distribution (Rogers and Harpending 1992) was used to test for the signal of a historical population expansion for all individuals across the range, as well as within each of the major geographic groups defined by the data. We also conducted Tajima's D statistic (Tajima 1989) and Fu's Fs (Fu and Li 1993; Fu 1997), which test for selective neutrality, but significantly negative values are an indication of a population expansion.
Phylogenetic relationships among haplotypes identified in this study, as well as sequences from GenBank representing three gopher tortoises (AY673483–AY673485) and three Bolson tortoises (Gopherus flavomarginatus; AY673473–AY673475) as the outgroup, were inferred using maximum parsimony (MP), maximum likelihood (ML) and Bayesian analyses. Prior to analysis, we truncated all sequences to a final length of 609 base-pairs. This resulted in a loss of 103 base-pairs from the 3′ end of our sequences, but this region was not variable in any of our samples. The MP and ML analyses were both performed using PAUP* 4.0b10 (Swofford 2002). The MP analysis used a branch-and-bound search, and the initial upper bound was calculated by stepwise addition. The ML analysis was based on HKY model. A Bayesian inference of the phylogeny was performed using MrBayes v. 3.2.1 (Ronquist and Huelsenbeck 2003), also using the HKY model. Tree space was explored starting with a random tree and employing two independent runs of four Markov chains of 1,000,000 generations, each sampled every 100 generations with the first 2,500 trees discarded as burn-in. The most appropriate models of sequence evolution for the ML and Bayesian analyses were selected by ModelTest v. 3.5 (Posada and Crandall 1998) and MrModeltest v. 2.3 (Nylander 2004), respectively. Phylogenetic support was assessed through bootstrapping (Felsenstein 1985) with 1,000 rounds of resampling for the MP and ML analyses. Support for clades identified in the Bayesian analysis (posterior probabilities) was measured by creating a majority-rule consensus of the 7,500 saved trees.
We performed an analysis of molecular variance (AMOVA; Excoffier et al. 1992) as implemented in Arlequin 3.11 (Excoffier et al. 2005) to examine geographic patterns of genetic variation. For the AMOVA, we constructed six models that varied in how they partitioned populations based on potential geographic barriers (i.e., river drainages). Three of the models partitioned the distribution of gopher tortoises into two groups based on the USFWS's delineation of the Tombigbee–Mobile rivers (model 1; USFWS 1987) or biogeographic breaks corresponding to the Apalachicola drainage (models 2 and 3; i.e., Apalachicola–Flint and Apalachicola–Chattahoochee) as seen in the literature (Avise et al. 1979; Swift et al. 1985; Bermingham and Avise 1986; Pauly et al. 2007). A split involving peninsular Florida as distinct from the rest of the range (model 4) was also tested based on the literature showing unique genetic structuring within peninsular Florida (Osentoski and Lamb 1995; Clark et al. 1999; Branch et al. 2003; Schwartz and Karl 2005). For this model, we used the Suwannee and Satilla rivers as delineations. The final two models were run with three groups using either the Tombigbee–Mobile and peninsular Florida (model 5) or the Tombigbee–Mobile and Apalachicola–Chattahoochee (model 6) as delineations. The significance of each AMOVA was tested via 1,000 permutations.
To avoid any potential bias in using a priori group delineations in the AMOVA, we also conducted a spatial analysis of molecular variance (SAMOVA; Dupanloup et al. 2002) using SAMOVA v1.0, which defines geographically proximate groups of individuals that also maximizes differentiation (ΦCT) among regions. We tested values of K (i.e., the number of groups) ranging from 2 to 3 with 100 simulated annealing processes to compare how SAMOVA partitioned the populations relative to the partitions tested in the AMOVA.
We obtained sequences for a 712–base-pair region of ND4 for 322 individuals from 42 sites across Mississippi, Alabama, Georgia, South Carolina, and Florida (Table 1; Figure 2). Of these, we found 13 unique haplotypes (Table 2; GenBank Accession numbers JF298788–JF298800; Data S1, Archived Material in Dryad, http://dx.doi.org/10.5061/dryad.4hs71t6t). The statistical parsimony network (Figure 3) detected two groups (group 1: haplotypes 1–4 and group 2: haplotypes 5–13) that were separated by a modest phylogenetic split (i.e., average uncorrected p distance between groups = 0.015; Table 3). All three phylogenetic analyses produced the same overall topology and we present the ML tree (−lnL = 1,060.45) as being representative of the phylogeny (Figure 4). Two clades were present in the phylogeny that corresponded to the same groups identified in the haplotype network. The majority of these clades were well-supported, although the Bayesian posterior probability value for one clade was rather low (Figure 4). Two of the gopher tortoise sequences from GenBank (after truncation) were identical to the most frequent haplotype (#9). No other structure was evident with any degree of phylogenetic support within either of the two clades.
The two groups of haplotypes identified in the network and phylogenetic tree were roughly partitioned, but the geographic distribution of haplotypes did not clearly suggest a particular biogeographic barrier that would have produced the phylogenetic break between groups. Sites at the edges of the range to the west (i.e., Mississippi) and east (including peninsular Florida) possessed haplotypes primarily from group 1 and group 2, respectively. However, several sites across the panhandle and northern Florida and western Georgia possessed haplotypes from both groups (Figure 5).
Except for the sites in Florida and Georgia having haplotypes from both groups, we typically found extremely low mtDNA genetic diversity throughout the range (Table 1). The average haplotype diversity across all sites was 0.133 and the average nucleotide diversity was 0.00054. Within the nonlisted portion of the range, the peninsular Florida region contained the most unique haplotypes (i.e., six) with two haplotypes (#10 and #13) differing by 3–4 nucleotide substitutions from the most common haplotype (#9) found in group 1. Although there were more haplotypes within the nonlisted region, genetic diversity indices did not significantly differ (Wilcoxon, χ2 = 0.008–0.015, P = 0.90–0.93) between listed and nonlisted portions of the range.
Each of the groupings of populations examined using the AMOVA models explained a significant portion of the molecular variance. However, the Apalachicola–Chattahoochee delineation (model 3) partitioned more of variance among groups than any of the other models (81.09%; P < 0.001; Table 4). The USFWS's delineation (i.e., Tombigbee–Mobile) of groups explained 59.26% of the molecular variance (P < 0.001), but this was still less than explained by either of the two Apalachicola models (Table 4). The peninsular Florida group was also a poor model because it only explained 57.08% the molecular variance (Table 4). Even models 5 (combination of models 1 and 4) and 6 (combination of models 1 and 3) explained less variation than model 3 (61.79% and 78.65%, respectively).
The SAMOVA explained similar amounts of molecular variance compared to the best AMOVA models for values for K of 2 (85.06%) and 3 (85.89%; Table 4). At a K of 3, the SAMOVA placed the Indian River Co., Florida (n = 2) and Jonathan Dickenson State Park, Florida (n = 7) sites into their own group. For both values of K, SAMOVA partitioned the samples on the basis of the Apalachicola–Chattahoochee rivers delineation (Figure 6).
For all sites combined, the mismatch distribution (P = 0.03), Tajima's D (D = 0.996; P = 0.07), and Fu's Fs test (Fs = 4.33; P = 0.43) failed to support a model of population expansion. Separate mismatch distribution analyses on the groups of sites, as defined by the Apalachicola—Chattahoochee rivers suggested a history of population expansion (P = 0.07 and P = 0.14 for sites west and east of these rivers). However, for each group of sites, the values of Tajima's D and Fu's Fs were not significantly negative, which does not match the expectation of population expansion. For both populations, Tajima's D was nearly significant (D = −1.365, P = 0.07 and D = −1.303, P = 0.07 for sites west and east of these rivers), but Fu's Fs were clearly not significant (Fs = −0.055, P = 0.53 and Fs = −0.867, P = 0.43 for sites west and east of these rivers).
Osentoski and Lamb's (1995) study revealed two assemblages of gopher tortoises, which they attributed to a biogeographic break imposed by the Apalachicola River. Our work, with more extensive sampling across Alabama and the federally listed as threatened portion of the range, clearly supported a similar delineation of these two major groups. The Apalachicola River has long been considered to play an important role in the biogeography of southeastern freshwater fishes (Swift et al. 1985) and to influence phylogeographic structure in numerous freshwater and terrestrial species (e.g., Avise et al. 1979; Bermingham and Avise 1986; Pauly et al. 2007). The phylogenetic break observed in our mtDNA haplotype network (average of 1.5% sequence divergence) falls within the range of sequence divergence values observed by Osentoski and Lamb (1995) in their mtDNA RFLP (1.1%) and control-region sequence (2.1%) data. Osentoski and Lamb (1995) also used a biogeographic event within the desert tortoise clade to calibrate a molecular clock and date their gopher tortoise split to the early Pleistocene.
Although the Apalachicola River apparently served as a historical biogeographic boundary as far back as the Pleistocene, it is not necessarily a barrier to more recent gene flow. It appears that both groups apparently experienced a period of population expansion following the Pleistocene isolation as evidenced by the results of the mismatch analysis, which could explain the overlap in the geographic distribution of haplotypes in group 1 and group 2 (Figures 5 and 6). Our data and Osentoski and Lamb (1995) both showed geographic overlap between the two lineages of haplotypes on both sides of the Apalachicola River. Osentoski and Lamb (1995) accepted this as a natural occurrence at the boundaries of the regions, and we found several sites proximate to the Apalachicola River where haplotypes from both groups were present (e.g., sites 15, 18, and 22–24). But, because gopher tortoises are commonly relocated within their range (Seigel and Dodd 2000), Osentoski and Lamb (1995) explained two cases of haplotypes located well outside of the appropriate region as being an artifact of these sorts of translocations. However, our data indicate a much broader zone of contact (Figures 4 and 5) with group 2 haplotypes found across the Florida panhandle (sites 12, 16, and 17) and a group 1 haplotype found in extreme eastern Florida (site 28). The one individual with a group 1 haplotype at site 28 may very well represent a translocation. However, the number of other sites with haplotypes from both groups suggests that translocations are not the only explanation for this overlapping pattern and recent dispersal likely has also played a role. An alternative explanation for the overlapping geographic distribution of haplotypes from the two groups is that this reflects incomplete lineage sorting. However, if this were the case, one might expect more sites from the central portion of the range to have retained haplotypes from both groups.
Peninsular Florida was the most diverse in terms of the number of haplotypes, with six haplotypes restricted to this region. Osentoski and Lamb (1995) and Schwartz and Karl (2005) identified population structure within this region that shares a pattern with other species (Clark et al. 1999; Branch et al. 2003). Peninsular Florida consists of two systems of “patchy” xeric uplands, which create a mosaic of isolated habitats. This unique geography, accompanied by sea-level fluctuations throughout the Pliocene and Pleistocene, could have led to multiple vicariance events (isolation produced by a geographical barrier; Webb 1990). Although thus far we have been interpreting our work in terms of two major assemblages (i.e., groups 1 and group 2 which are west and east of the Apalachicola River, respectively), the SAMOVA with K = 3 did detect genetic structure within peninsular Florida. However, the SAMOVA with K = 3 only explained slightly more of the among-group molecular variance than with a K = 2 (85.89% vs. 85.06%). Furthermore, our third peninsular Florida group corresponded to two sites (#38 and 39) in east-central Florida and was not congruent with Osentoski and Lamb's (1995) peninsular Florida genetic structuring. This lack of congruence could be an artifact of our sampling through the unknowing inclusion of translocated individuals and–or a lack of intense sampling along both xeric upland ridges.
The current federally listed as threatened portion of the gopher tortoises' range only includes the area west of the Tombigbee and Mobile rivers (USFWS 1987). Based on our data and that of Osentoski and Lamb (1995), the boundary currently recognized as defining the listed portion of the range does not coincide with the boundary best explaining the geographic partitioning of mitochondrial genetic variation. If one were to attempt to define a genetically based unit of conservation, the results of the AMOVA and SAMOVA suggest that the mitochondrial groups (1 and 2) as defined by the Apalachicola–Chattahoochee rivers would be the most appropriate delineations and would capture a major portion of the deeper evolutionary history of the gopher tortoise. However, we do not interpret this result to mean that the current federal listing is not warranted. As currently defined, USFWS's delineation protects only a portion of group 1, but the federally listed as threatened region is still genetically unique in that only group 1 haplotypes are found in all of the populations west of the Tombigbee and Mobile rivers. In fact, the AMOVA model (#6) that includes both the Tombigbee–Mobile and Apalachicola–Chattahoochee rivers is the second-best model for explaining the geographic partitioning of genetic variation. Lastly, it is worth adding that the amendment to the ESA (USFWS and NOAA 1996) that provides for the listing of a DPS includes criteria in determining the discreteness of a population other than genetic differences such as physical, physiological, morphological, ecological, or behavioral differences and the importance of the population segment to the species as a whole. Although, the federal listing of gopher tortoises took place before the legal concept of DPS, one could argue that populations west of the Tombigbee and Mobile rivers do represent a historically important portion of its range.
The conclusions of any genetic study need to be put into the context of the molecular marker employed, which in this case was mtDNA. Mitochondrial DNA in animals has many useful properties that have made it a commonly used tool in phylogenetics and population genetics (reviewed by Avise 2004). However, an important caveat about mtDNA is that it is maternally inherited and reflects but one aspect of the evolutionary history of a species. Nuclear markers such as microsatellites are biparentally inherited, have high mutation rates, and typically demonstrate high levels of variability, which makes them exceptionally useful for population-level studies. In gopher tortoises, microsatellites have already been employed in a variety of situations with a conservation context, such as measuring multiple paternity and reproductive success (Moon et al. 2006; Tuberville et al. 2010) and levels of genetic variation (Ennen et al. 2010; Richter et al. 2011). Importantly, microsatellites can also be used to detect population structure at finer spatial scales than mtDNA. For example, Schwartz and Karl (2005) provided important data for developing management strategies by identifying at least eight genetically distinct groups of gopher tortoise just in Florida and southeastern Georgia. In comparison, our study and Osentoski and Lamb (1995), using mtDNA only, reported two and three groups throughout the entire distribution, respectively. Although population structure at the landscape level has yet to be reported using microsatellites (Sinclair et al. 2010; Richter et al. 2011), this is not to say that it might not exist at certain spatial scales in gopher tortoises. Thus, while our mtDNA data provide insight into the deeper evolutionary history of gopher tortoises across their range, there is still a need to expand upon existing studies and conduct a broader scale survey of population structure using nuclear markers, such as microsatellites. We are currently pursuing such an assessment focusing on the listed region and areas outside of Florida in the nonlisted portion of the range.
Based on the considerations outlined above, our opinion is that other data need to be assembled to better inform the USFWS's management decisions concerning listing status during the next review period. Although based only on mtDNA data, our findings and those of Osentoski and Lamb (1995) provide support for a DPS that is delineated by the Apalachicola–Chattahoochee rivers and could allow protection of populations west of this delineation under the ESA. However, we feel that any decision that would affect the current listing of gopher tortoises at this time would be premature without additional molecular studies, in particular an analysis of population structure based on highly variable nuclear markers such as microsatellites. The mtDNA data suggest that the currently listed portion of the range does capture some important genetic structure, and our ongoing research using microsatellites also indicates that the currently listed populations of gopher tortoises appear to be genetically unique and therefore important for future management decisions as well (D. Gaillard, unpublished data). We suspect that, once a more comprehensive look at genetic structure across the range is assembled, at minimum the current ESA listing will be supported. It is also possible that the federally protected as threatened portion of the species' range may need to be expanded eastward, and–or that other portions (i.e., peninsular Florida) may warrant separate protection. However, we do recognize that the determination of DPS status under the ESA can and should reflect a variety of considerations in addition to genetic structure.
To cite this archived material, please cite both the journal article (formatting found in the Abstract section of this article) and the following recommended format for the archived material.
Ennen JR, Kreiser BR, Qualls CP, Gaillard D, Aresco M, Birkhead R, Tuberville TD, McCoy ED, Mushinsky H, Hentges TW, Schrey A. 2012. Data from: Mitochondrial DNA assessment of the phylogeography of the gopher tortoise, Journal of Fish and Wildlife Management, 3(1):110–122. Archived in Dryad Digital Repository: http://dx.doi.org/10.5061/dryad.4hs71t6t
Data S1. AMOVA and SAMOVA data files from 42 sites across Mississippi, Alabama, Georgia, South Carolina, and Florida.
We would like to thank the USFWS and the Gopher Tortoise Council for providing support for this project. We thank B. Jones and T. Smith for assistance in securing funding for tissue collection through the Mississippi Department of Wildlife, Fisheries, and Parks and USACE-ERDC-CERL (W9132T-06-2-0021), respectively. Partial support for sample collection and manuscript preparation was provided by the Department of Energy under Award Number DE-FC09-07SR22506 to the University of Georgia Research Foundation. We thank N. Sharp, D. Hartley, L. McCoy, C. Walters, K. Shelton, and R. Bolt for support with tissue collection. All samples were collected under the appropriate permits by the authors, and all laboratory work for this project was conducted under USM IACUC protocol number 193-003. We thank M. Stromer and L. Porter of S&ME, Inc. for GIS support. We also thank the anonymous reviewers and the Subject Editor of JFWM for their helpful suggestions and comments.
The use of trade, firm, or product names is for descriptive purposes only and does not imply endorsement by the U.S. Government.
Ennen JR, Kreiser BR, Qualls CP, Gaillard D, Aresco M, Birkhead R, Tuberville TD, McCoy ED, Mushinsky H, Hentges TW, Schrey A. 2012. Mitochondrial DNA assessment of the phylogeography of the gopher tortoise. Journal of Fish and Wildlife Management 3(1):110-122; e1944-687X. doi: 10.3996/102011-JFWM-063
The findings and conclusions in this article are those of the author(s) and do not necessarily represent the views of the U.S. Fish and Wildlife Service.