Sigmodon hispidus, the hispid cotton rat, has a wide geographic distribution and has been the focus of numerous ecological studies; however, patterns of genetic divergence within this species remain largely unknown. Amplified fragment length polymorphism (AFLP) and mitochondrial cytochrome-b DNA sequence analyses were employed to document patterns of nuclear and mitochondrial DNA (mtDNA) divergence within this species. AFLP analysis of specimens from Georgia, Louisiana, Texas, Oklahoma, Kansas, and Arizona demonstrated that the previously recognized genetic discontinuity dividing S. hispidus into eastern and western mtDNA lineages also was evident in the nuclear genome. The contact zone between these 2 lineages was located in eastern Texas and hybridization was confirmed. The width of the hybrid zone was found to be 3 orders of magnitude greater than the estimated dispersal distance. This, in combination with an absence of linkage disequilibrium within the hybrid zone and its unimodality, suggests that selection is not a strong factor in the maintenance of the hybrid zone.
The hispid cotton rat (Sigmodon hispidus), 1st described in 1825 (Say and Ord 1825), has been the focus of much research including ecological, biomedical, and systematic studies. Since the original description of S. hispidus, several species have been elevated from within what appeared to be a widespread and morphologically homogeneous taxon. Zimmerman (1970) used a combination of cytogenetic and morphological data to conclude that S. hispidus was actually a combination of 3 independent gene pools with no evidence of interbreeding, and elevated S. arizonae and S. mascotensis to species level. Peppers and Bradley (2000) subsequently elevated S. toltecus and S. hirsutus to species status based on extensive cytochrome-b sequence divergence.
The most recent systematic assessment (Carroll et al. 2004) describes S. hispidus as occurring only or mostly in the United States, ranging from the Atlantic Coast west to the disjunct population (S. h. eremicus) located in the lower Colorado River valley of California, Arizona, and Mexico. Northward expansion of S. hispidus has been documented (Farney 1975; Genoways and Schlitter 1967), and the current northern limit of its distribution reaches into southern Nebraska, Kentucky, and Virginia. Within this distribution, Carroll et al. (2004) documented 2 distinct mitochondrial DNA (mtDNA) lineages separating S. hispidus into an eastern form and a western form. Average percent sequence divergence between these lineages was 4.9%. Because of the moderate level of sequence divergence, Carroll et al. (2004) did not recommend elevation of these lineages to species status and suggested that additional studies be conducted before formal taxonomic revisions be made.
The objective of this study is to use amplified fragment length polymorphism (AFLP) and mtDNA analyses to determine if the genetic discontinuity dividing S. hispidus into eastern and western mtDNA lineages is evident in the nuclear genome, to determine if hybridization occurs between these lineages, and to characterize the nature of hybridization. AFLP has several advantages over other molecular techniques, including the low occurrence of artifacts, high resolution, and large number of easily generated markers (Mueller and Wolfenbarger 1999). The large number of genomic loci that are simultaneously amplified using AFLP offset the limitations presented by the dominant nature of AFLP fragments and also override the effects of historical lineage sorting that may factor in single-gene analyses (Creer et al. 2004). AFLP has been used in studies of genetic variation within populations (Mueller and Wolfenbarger 1999) and in investigating the congruence of mtDNA gene trees and species trees, an application that well suits the current study. Although AFLP has been underutilized in animal studies (Bensch and Akesson 2005), the strengths of the technique have led to its increased use in ecological and evolutionary studies.
Materials and Methods
Specimens were initially obtained from 7 localities (Fig. 1; Appendix I) within the United States using Sherman live traps (H. B. Sherman Traps, Inc., Tallahassee, Florida) and tissue loans from frozen tissue collections. These localities, along with sample sizes (n) are as follows: Telfair County, Georgia (n = 3); East Baton Rouge Parish, Louisiana (n = 8); Sabine County, Texas (n = 8); Erath County, Texas (n = 8); Osage County, Oklahoma (n = 6); Ellis County, Kansas (n = 8); and Yuma County, Arizona (n = 8). An initial analysis of these populations prompted us to refocus our investigation in an attempt to locate the contact zone between the eastern and western lineages to determine if hybridization occurs. Specimens were then obtained from a transect in Anderson, Cherokee, and Angelina counties in eastern Texas (Fig. 1) where the contact zone was thought to occur (R. Bradley and J. Patton, pers. comm.). For purposes of discussion, Sabine County, Texas, was included in the transect because of its close proximity and will be referred to as site 10. The transect was oriented northwest to southeast and consisted of 10 trapping locations (sites 1–10) with sample sizes of 8, 1, 6, 3, 3, 10, 8, 7, 8, and 8, respectively (Fig. 1; Appendix I). Guidelines of the American Society of Mammalogists (Animal Care and Use Committee 1998) were followed for the capture, handling, and care of specimens.
The AFLP protocol was modified from Vos et al. (1995). Two hundred nanograms total genomic DNA was digested for 3 h at 37°C with 20 units of AseI, 20 units of EcoRI, and appropriate volumes of restriction enzyme buffer. Ligations were performed by adding 75 pmoles of each EcoRI and AseI adapter (Table 1), appropriate volumes of ligation buffer, 3 units T4 DNA ligase, and 12 µl of H2O to restriction digestions and incubating for 16 h at 16µC. Ligation products were diluted by adding 160 µl of 10 mM Tris (pH 8.5).
A subset of ligated fragments was amplified using a pre-selective polymerase chain reaction. Amplifications were carried out in 50-µl reactions containing 10 µl of diluted ligation product, 15 pmoles of both preselective primers (Table 1), IX buffer, 1.5 mM MgCl2, 0.8 mM deoxynucleoside triphosphates, and 2.5 units of Taq DNA polymerase. Amplification conditions included an initial step of 72°C for 60 s followed by 20 cycles of 94°C for 50 s, 56°C for 60 s, and 72°C for 120 s. Forty microliters of the preselective polymerase chain reaction products were diluted in 720 µl of 10 mM Tris (pH 8.5) and used to conduct the selective polymerase chain reaction amplifications using 5 selective primer pairs (Table 1). Total reaction volumes (25 µl) contained 5 µl of diluted preselective product, 5 pmoles of both selective primers, IX buffer, 1.5 mM MgCl2, 0.8 mM deoxynucleoside triphosphates, and 1.25 units of Taq DNA polymerase. The thermal profile for selective reactions was as follows: 24 cycles of 94°C for 50 s, 65–56.6°C (0.7°C reduction for 2nd through 13th cycle) for 60 s, and 72°C for 120 s.
|EcoRI-CAT*||5′-ACTGCGTACC AATTCC AT-3′|
|AseI-TGT||5′-GATG AGTCCTG AGTAATTGT-3′|
|EcoRI-CAT*||5′-ACTGCGTACC AATTCC AT-3′|
|AseI-TGT||5′-GATG AGTCCTG AGTAATTGT-3′|
The selective EcoRI primers used in each of these reactions were fluorescently labeled for detection by a Beckman-Coulter CEQ8000 Automated DNA Analysis System (Beckman-Coulter, Inc., Fullerton, California). Beckman-Coulter software was used to size fragments based on an internal size standard. Only fragments that could be unambiguously scored as present or absent were included in the data set, which consisted of a table of ones indicating the presence of a fragment of a certain size, and zeros indicating its absence. Replicate data were generated for each individual to verify presence or absence of fragments. Discrepancies between replicates were rare, but when observed, either a 3rd replicate was generated to verify the fragment profile or the ambiguous fragment was disregarded in all specimens.
Patterns of divergence among populations were determined using principal coordinate analysis (PCoA) and analysis of molecular variance (AMOVA) in GenAlEx, version 6, software (Peakall and Smouse 2006). Pairwise ΦPT values also were calculated. The hypothesis of migration–drift equilibrium and subsequent isolation by distance among localities within each of the 2 lineages was assessed by means of a Mantel test as implemented in GenAlEx using Nei's gene divergence index. STRUCTURE 2.1 (Pritchard et al. 2000) was implemented to detect patterns of population structuring using the no admixture model as recommended for AFLP data. Length of the burn-in period and number of iterations was set at 10,000. Five independent runs were performed for each value of K and the resulting log-likelihoods were averaged. Analyses of linkage disequilibrium were conducted using the programs LIAN 3.1 (simulation option—Haubold and Hudson 2000) for the complete AFLP data set and DIS (written by J. Mallet) for fixed AFLP fragments. Newhybrids 1.1 beta (Anderson and Thompson 2002) was used to infer hybrid class membership for transect specimens by calculating the posterior probability of an individual falling into 1 of 6 hybrid classes (pure western, western backcross, F1, F2, eastern backcross, or pure eastern).
Partial cytochrome-b sequences (492 base pairs [bp]) were obtained using the polymerase chain reaction primers L14724 and H15906 (Irwin et al. 1991; Zheng et al. 2003) and were aligned using CLUSTALX (Thompson et al. 1997). In order to assign individuals to either the eastern or western mtDNA lineage, a neighbor-joining tree was constructed from uncorrected p-distances using MEGA 3.1 (Kumar et al. 2004). Uncorrected percent sequence divergence values within and among lineages were calculated using MEGA 3.1.
A total of 185 AFLP fragments were scored, with 32.4% (60 fragments) being polymorphic. Four of the 60 polymorphic fragments showed a notable pattern of fixed presence or absence that could be used as diagnostic markers to distinguish eastern and western lineages. Interestingly, all fixed markers were present in the eastern lineage and absent in the western lineage. No markers were unique to S. h. eremicus (Yuma County, Arizona).
The first 2 axes of the PCoA (excluding sites 2–9 of the transect) explained 76.6% of the variation in the AFLP data set. Individuals clustered into 3 nonoverlapping groups (Fig. 2), with 1 group containing all individuals from Telfair County, Georgia; East Baton Rouge Parish, Louisiana; and transect site 10. This group corresponds to the eastern mtDNA lineage identified by Carroll et al. (2004). A 2nd group contained individuals from transect site 1, Erath County, Texas; Osage County, Oklahoma; and Ellis County, Kansas. Individuals representing S. h. eremicus formed the 3rd group, which was somewhat closer to the western group. These latter 2 groups correspond to the western mtDNA lineage (Carroll et al. 2004; R. S. Pfau, in litt.).
Analyses were then conducted for the eastern and western lineages separately (excluding S. h. eremicus) in order to reveal finer-scale population structuring. The first 2 axes of the PCoA explained 52.5% and 46.7% of the variation within eastern and western lineages respectively. Analysis of the eastern localities (Telfair County, Georgia; East Baton Rouge Parish, Louisiana; and site 10) revealed 3 distinct clusters corresponding to the 3 localities (Fig. 3A). Individuals from western localities (site 1; Earth County, Texas; Osage County, Oklahoma; Ellis County, Kansas; and Yuma County, Arizona) exhibited much less divergence (Fig. 3B), but localities separated by the greatest geographic distance (site 1 and Ellis County, Kansas) formed nonoverlapping clusters, with individuals from Osage County, Oklahoma and Erath County, Texas being placed between and overlapping with these clusters. AMOVA revealed significant genetic divergence among localities within both lineages (P = 0.010 for both lineages). For the eastern lineage, pairwise ΦPT values range from 0.313 between Telfair County, Georgia, and East Baton Rouge Parish, Louisiana, to 0.486 between Telfair County, Georgia, and site 10. For the western lineage, pairwise ΦPT values range from 0.117 between Osage County, Oklahoma, and site 1 to 0.326 between Ellis County, Kansas, and Erath County, Texas. Estimation of posterior probabilities of K (number of populations) using STRUCTURE indicated that 2 populations (P(K|X) > 0.92) best explained the data for the eastern localities clustering individuals from Telfair County, Georgia, with East Baton Rouge Parish, Louisiana, exclusive of individuals from site 10. A single population best explained the data for the western localities (P(K|X) > 0.98). There was no significant pattern of isolation by distance for either lineage (eastern lineage: R2 = 0.0454, P = 0.56; western lineage: R2 = 0.1911, P = 0.15).
Results of Newhybrids analysis (Fig. 4) indicates that hybridization does occur at the contact zone between the 2 lineages. Individuals from transect sites 1 and 2 were classified as pure western. Sites 3–5 contain individuals that have a high probability of being pure western or western backcross. Sites 6 and 7 contain individuals with high likelihoods of being western backcross, eastern backcross, F2, and pure eastern. All individuals from sites 8 and 9 have a high likelihood of being pure eastern with the exception of 1 individual, which was classified as eastern backcross, whereas those from the easternmost transect site (site 10) were classified as pure eastern. No individuals within the center of the hybrid zone were classified as nonhybrids. Based on these results, the width of the hybrid zone is at least 130 km, this being the greatest distance between localities containing hybrid individuals (sites 3 and 9). Sites 1 and 10, which contain no hybrid individuals as determined by Newhybrids analysis and PCoA, are separated by 210 km; thus, the width of the hybrid zone (at this locality) likely does not exceed this distance. If the transect is not exactly perpendicular to the hybrid zone, these are slight overestimates. Interestingly, the 3 individuals from Georgia had an equal likelihood of being pure eastern or eastern backcross despite their distance from the contact zone.
None of the localities outside of the hybrid zone (including sites 1 and 10) exhibited a statistically significant probability of linkage disequilibrium based on the complete AFLP data set using LIAN (P ranged from 0.9347 to 0.1318). Of the 3 sites within the hybrid zone having sufficiently large sample sizes to test for linkage disequilibrium (sites 3, 6, and 7) only sites 6 and 7 were close to being statistically significant (P = 0.0659 and 0.0687, respectively). Pairwise tests of linkage disequilibrium using the program DIS for the 4 diagnostic AFLP fragments failed to demonstrate that these markers were in linkage disequilibrium within the hybrid zone.
Based on 492 bp of the cytochrome-b gene, 21 haplotypes were identified among 90 individuals (GenBank accession numbers DQ644040-DQ644128; Appendix I). A neighbor-joining tree (not shown) grouped these haplotypes into 2 clades representing the eastern and western mtDNA lineages. Average sequence divergence (uncorrected p-distance) between the 2 clades was 3.6%. Fourteen haplotypes were identified among the 41 individuals comprising the western clade (excluding S. h. eremicus), whereas only 7 haplotypes were identified among the 49 eastern individuals. Average sequence divergence was 0.6% among the 14 western haplotypes and 0.7% among the 7 eastern haplotypes. Within the hybrid zone as defined by AFLP data there are individuals representative of both mtDNA lineages. Furthermore, the mtDNA cline and the nuclear cline are geographically coincident (Fig. 4). Eastern mtDNA haplotypes were not observed in sites 1–3 and western haplotypes were not observed in sites 8–10 despite some individuals in site 9 being classified as hybrid backcross based on AFLP data.
Amplified fragment length polymorphism analysis supports the division of S. hispidus into eastern and western lineages as originally reported by Carroll et al. (2004) using mtDNA data. Because AFLP markers are representative of the entire genome, it can be concluded that these 2 lineages represent the evolutionary history of the species as a whole and not just the mitochondrial genome. AFLP analysis has proven valuable in supporting and clarifying mtDNA phylogenies in other studies (Creer et al. 2004; Giannasi et al. 2001), and although single-gene phylogenies can provide misleading relationships at times because of stochastic lineage sorting (Avise 2004), this does not appear to be the case with S. hispidus. AFLP analysis also supports a previous study demonstrating the genetic divergence of S. h. eremicus (Pfau et al. 2001). These results contrast with that of McClenaghan (1980), who found that this population was not differentiated from those of the main geographic distribution. Given that no AFLP fragments are unique to this population, whereas 4 fragments are unique to the eastern lineage, it is likely that the eastern lineage has been separated from the western lineage for a longer period of time than has the disjunct population (S. h. eremicus). Examination of cytochrome-b data confirms that S. h. eremicus is part of the western lineage (Carroll et al. 2004; R. S. Pfau, in litt.).
Principal coordinate analysis, AMOVA, and STRUCTURE show a greater degree of divergence among localities within the eastern lineage relative to the western. The continuity of suitable habitat in the western half of the distribution likely allows for freer dispersal throughout the region, creating a more homogeneous population genetic structure. The patchiness of suitable habitat in the southeastern United States may limit gene flow, thus increasing the effects of genetic drift. It is possible that the Mississippi River contributes to this effect, given that pairwise ΦPT values and STRUCTURE showed Telfair County, Georgia, and East Baton Rouge Parish, Louisiana, both east of the Mississippi River, to be the least divergent. The Mississippi River also has been shown to be a substantial barrier to gene flow for tree squirrels (Moncrief 1993). The hypothesis of migration-drift equilibrium can be rejected because of an absence of isolation by distance. If documented range expansions (Farney 1975; Genoways and Schlitter 1967) are a reflection of long-term, demographic processes within this species, it is not surprising that populations are not in migration-drift equilibrium.
The location of the contact zone between eastern and western lineages does not coincide with the subspecies boundaries designated by Hall (1981). Hall (1981) indicated that the range of S. h. hispidus to the east and S. h. texensis to the west abut along the Sabine River, whereas results presented here demonstrate that the center of the contact zone between eastern and western lineages occurs approximately 120 km west of the nearest currently recognized subspecies boundary. However, the subspecies boundaries defined by Hall (1981) are based on such incomplete sampling that they would likely not be recognized by today's more stringent criteria for recognizing patterns of biological diversity.
Analysis of the transect across the contact zone in eastern Texas indicates that hybridization does occur between the 2 lineages (Fig. 4). Individuals within the hybrid zone have high likelihoods of being western backcross, eastern backcross, or F2 hybrids. The cline defined by AFLP data is geographically coincident with that of mtDNA data (Fig. 4), and the four diagnostic AFLP markers (which do not appear to be linked) are coincident as well. Coincident clines can be indicative of strong selection acting on loci across the genome (Barton and Bengtsson 1986; Barton and Gale 1993; Durrett et al. 2000; Marshall and Sites 2001; Searle 1993). If selection is weak, locus-specific selection or drift or both may result in differing shapes, widths, and locations of clines. However, coincident clines also are expected to occur in the absence of selection as long as time since contact is relatively recent (<500 generations— Durrett et al. 2000), not a likely scenario given the width of the hybrid zone (see below). However, the lack of statistically significant levels of linkage disequilibrum is not consistent with strong selection acting across the genome as is expected under the tension zone model (Barton and Hewitt 1985). It is possible that a more rigorous analysis of cline location and shape, possible only with larger sample sizes and additional sampling localities across the hybrid zone, would reveal that mtDNA and AFLP clines are not exactly coincident and concordant—the pattern expected in the absence of selection.
The width of the hybrid zone is between 130 and 210 km, which could be considered wide relative to the dispersal abilities of this species. Several studies have documented dispersal abilities of S. hispidus. Cameron and Kincaid (1982) and Layne (1974) showed that individual monthly movements averaged 52.9 m and 57.3 m, respectively. Studies that released individuals into unfamiliar territories observed incidences of longer-distance dispersal. Maximum distances moved as reported by Bowne et al. (1999) and DeBusk and Kennerly (1975) were 760 m and 1,500 m, respectively, although most individuals moved substantially less, than this. The width of the hybrid zone is more than 3 orders of magnitude greater than the estimates of dispersal distance reported by Cameron and Kincaid (1982) and Layne (1974). Hybrid zones that are this wide relative to dispersal distance likely are influenced by weak selection, if at all (Sotka and Palumbi 2006). For a tension zone, rate of dispersal (variance of parent-offspring distance— Barton and Hewitt 1985) can be estimated to within an order of magnitude using the formula σ = √s(w), where σ is standard deviation of dispersal distance, s is selection intensity, and w is width of the hybrid zone. Selection values as low as 0.001-0.01 give dispersal rates of 13–4 km/generation; thus, even if σ = 1 km, it is unlikely that selection against hybrids is greater than 0.001.
Classification of hybrid zones based on the distribution of hybrid classes can further clarify relationships between the hybridizing taxa (Harrison and Bogdanowicz 1997; Jiggins and Mallet 2000). The hybrid zone of S. hispidus appears to be unimodal in that only hybrid individuals occur in the center of the hybrid zone. Bimodal hybrid zones, in which parental forms predominate and hybrid forms are rare, are maintained by stronger isolating mechanisms than those found in unimodal hybrid zones and indicate that speciation is nearly complete. It has been suggested that unimodal hybrid zones ultimately may lead to the fusion of lineages or extinction of one (Howard 1993; Liou and Price 1994; Paterson 1978); however, unimodal hybrid zones do exist in which selection against hybrids has been demonstrated (Hewitt et al. 1987; Szymura and Barton 1991). In this case, continued selection would then be expected to cause a shift toward bimodality (Dobzhansky 1940; Howard 1993).
The geographic location of the hybrid zone in eastern Texas coincides roughly with the post oak savannah-piney woods ecotone (Gould 1962), as noted by Carroll et al. (2004). Many hybrid zones coincide with ecotones, and various explanations for this have been given including environmental clines, tension zones, and bounded hybrid superiority (Hewitt 1988). Environmental clines result when each parental type is fitter in different environments. Tension zones result when there is a balance between dispersal of parental forms into the hybrid zone and endogenous selection against hybrids and thus are independent of environmental conditions (Barton and Hewitt 1989). Under this model, hybrid zones near areas of environmental transition tend to migrate and come to rest at the area of lowest density because of asymmetrical dispersal. The bounded hybrid superiority model states that hybrids are more fit within the ecotone than either parental type. Hewitt (1988) discusses the difficulties in distinguishing between these models and few studies have provided convincing evidence of 1 model over the others. It was hypothesized by Carroll et al. (2004) that the separation between eastern and western lineages is due to differing ecological preference, wherein the eastern lineage occupies forest habitat, whereas the western lineage prefers grassland, as would be predicted by the environmental cline model. Under this model one would expect genes not linked to habitat preference to flow freely across the hybrid zone (Hewitt 1988). The coincidence of mtDNA and unlinked, diagnostic AFLP markers within this hybrid zone suggests that this model may not be applicable—unless all of these loci are under the same selective pressures. Width of the hybrid zone (3 orders of magnitude greater than estimated dispersal distance), estimates of selection intensity required to maintain a hybrid zone of this width, absence of linkage disequilibrium, and absence of parental and F1 hybrid classes suggest that selection is either very weak or nonexistent—thus ruling out the possibility that this is a tension zone maintained by selection against hybrids. Clarification regarding the nature of hybridization between these 2 lineages must await examination of multiple transects along the contact zone, particularly in the northern- and southernmost portions of the species range where environmental gradients are different from that of the current study. For example, if S. hispidus has indeed recently colonized the northern portions of its geographic distribution, and if the width of the hybrid zone is due to neutral diffusion and not selection, one would expect that the width of the hybrid zone would be narrower to the north than the south.
Interestingly, individuals from Georgia were identified as having an equal probability of being eastern backcross or pure eastern (Fig. 4) and in the PCoA, individuals from Georgia group closer to the western lineage than do the other eastern individuals (Fig. 2). These individuals had a higher likelihood of being eastern backcross than did those from site 10, which is within 40 km of the hybrid zone. The reason for this is not obvious and additional samples are needed, especially from the northern portions of the distribution of S. hispidus, to determine why Georgia specimens showed a higher probability of hybrid class membership than did other eastern specimens.
The significant genetic discontinuity documented between these 2 lineages prompts questioning regarding their taxonomic status. No species concept is completely objective, and all require that a judgment be made in cases involving hybridization (Helbig et al. 2002). Additionally, species designations can often differ depending on the species concept being applied. Following the phylogenetic species concept, it could be argued that 2 species comprise the distribution of S. hispidus as the eastern and western lineages are independent groups of organisms “within which there is a parental pattern of ancestry and descent” (Cracraft 1983:170). However, following species concepts that take into account genetic and reproductive isolation, S. hispidus could be diagnosed as 1 species or 2 depending on the extent of hybridization. It has been proposed that when a unimodal hybrid zone exists the 2 populations should be considered 1 species (Mallet 1995). However, Baker and Bradley (2006) recommend species status if the integrity of the 2 gene pools is maintained regardless of hybridization. Although there is no question that the eastern and western mtDNA lineages are distinct evolutionary groups it is not so clear that the 2 gene pools are maintaining their integrity. It is likely that agreement will never be made regarding the taxonomic rank of these lineages, but we recommend that arguments for or against species status await a more complete characterization of patterns of hybridization and introgression along the length of the contact zone. At the very least, the subspecies designations as described by Hall (1981) will need to be revised to reflect this new understanding of patterns of biological diversity within S. hispidus.
We thank the Texas Tech University Natural Science Resource Laboratory, Louisiana State University Museum of Natural Science, K. McBee, R. Van Den Bussche, and G. Wilson for tissue loans. We also thank E. Sookoor and S. Senter for assistance in the field. P. Sudman and H. Rathburn provided valuable comments on an earlier version of this manuscript. This research was funded by the Tarleton State University Research Committee.
Specimens examined.—The 103 specimens examined in this study are listed by collection locality and museum number. GenBank accession numbers for specimens with cytochrome-b DNA sequences are given in parentheses. Abbreviations for museum numbers are as follows: Natural Science Research Laboratory, Museum of Texas Tech University (TK); Museum of Natural Science, Louisiana State University (M); Collection of Tissues, Oklahoma State University (OK); and Tarleton State University (TSU).
East Baton Rouge Parish, Louisiana.—M1285, M1286, M1288 (DQ644051), M1290 (DQ644052), M1291, M1293, M1294, M1296 (DQ644053).
Ellis County, Kansas.—OK5829 (DQ644054), OK5830 (DQ644055), OK5834 (DQ644056), OK5835 (DQ644057), OK5836 (DQ644058), OK5841 (DQ644059), OK5845 (DQ644060).
Erath County, Texas.—TSU1343 (DQ644068), TSU1344 (DQ644069), TSU1345 (DQ644070), TSU1346 (DQ644071), TSU1347 (DQ644072), TSU1348 (DQ644073), TSU1349 (DQ644074), TSU1350 (DQ644075).
Osage County, Oklahoma.—OK5879 (DQ644062), OK5880 (DQ644063), OK5881 (DQ644064), OK5882 (DQ644065), OK5884 (DQ644066), OK5886 (DQ644067).
Telfair County, Georgia.—TSU 1351 (DQ644047), TSU1352 (DQ644048), TSU1353 (DQ644049).
Site 1, Anderson County, Texas.—12 miles N, 22 miles W Palestine: TK116831 (DQ644076), TK116832 (DQ644077), TK116833 (DQ644078), TK116834 (DQ644079), TK116835 (DQ644080), TK116836 (DQ644081), TK116837 (DQ644082), TK116838 (DQ644083).
Site 2, Anderson County, Texas.—5 miles N, 13 miles W Palestine: TK1.16840 (DQ644084).
Site 3, Anderson County, Texas.—4 miles N Palestine: TK 116843 (DQ644085), TK116844 (DQ644086), TK116845 (DQ644087), TK116846 (DQ644088), TK116847 (DQ644089).
Site 4, Anderson County, Texas.—2.5 miles S, 4 miles E Elkhart: TK116848 (DQ644090), TK116849 (DQ644091), TK116850 (DQ644092).
Site 5, Cherokee County, Texas.—8.5 miles W Alto: TK116851 (DQ644093), TK116852 (DQ644094), TK116853 (DQ644095).
Site 6, Cherokee County, Texas.—1 miles N, 3 miles W Wells: TK 116854, TK116855 (DQ644096), TK116856 (DQ644097), TK116888 (DQ644098), TK116889 (DQ644099), TK116890 (DQ644100), TK116891 (DQ644101), TK116892 (DQ644102), TK116893 (DQ644103), TK116894 (DQ644104), TK116895 (DQ644105).
Site 7, Angelina County, Texas.—5 miles N, 6 miles W Lufkin: TK116913 (DQ644113), TK116914 (DQ644114), TK116915 (DQ644115), TK116916 (DQ644116), TK116917 (DQ644117), TK116918 (DQ644118), TK116919 (DQ644119), TK116920 (DQ644120).
Site 8, Angelina County, Texas.—6.5 miles E Lufkin: TK116856 (DQ644097), TK116903 (DQ644106), TK116904 (DQ644107), TK116905 (DQ644108), TK116906 (DQ644109), TK116907 (DQ644110).
Site 9, Angelina County, Texas.—2 miles S, 2 miles E Zavalla: TK116921 (DQ644121), TK116922 (DQ644122), TK116923 (DQ644123), TK116924 (DQ644124), TK116925 (DQ644125), TK116926 (DQ644126), TK116927 (DQ644127), TK116928 (DQ644128).
Site 10, Sabine County, Texas.—4 miles S Pineland: TSU1354 (DQ644040), TSU1355 (DQ644041), TSU1356 (DQ644050), TSU1357 (DQ644042), TSU1358 (DQ644043), TSU1359 (DQ644044), TSU1360 (DQ644045), TSU1361 (DQ644046).