We evaluated evolutionary relationships of taxa within silky pocket mice of the Perognathus fasciatus species group, composed of 3 species, P. fasciatus, P. flavescens, and P. apache. These species are distributed throughout the Great Plains, Wyoming Basin, Colorado Plateau, and northern Chihuahuan Desert biogeographic regions in North America. We tested a previously postulated hypothesis of Pleistocene species divergence and introgression by analyzing mitochondrial DNA (mtDNA) and amplified fragment length polymorphisms (AFLPs). Both mtDNA and AFLP data support several genetic lineages in the fasciatus species group that are geographically structured. Molecular clock estimates reject a Pleistocene speciation hypothesis in favor of a deeper, more complex evolutionary history of initial divergence in the Miocene followed by secondary diversification beginning in the middle Pliocene and progressing through the Pleistocene. Results support recognition of an additional species within the group. Temporal and spatial congruence between the mtDNA clades in the fasciatus species group and other codistributed species of Perognathus appear to support a hypothesis of concerted diversification throughout the Chihuahuan Desert, Colorado Plateau, and Great Plains.
The focus of this study is on the phylogeographic diversification of the Perognathus fasciatus species group of silky pocket mice, which is distributed throughout the North American arid grasslands (Fig. 1) and is 1 of 4 phylogenetically well-delineated species groups (along with the parvus, longimembris, and flavus groups) that comprise the genus Perognathus (Alexander and Riddle 2005; Hafner et al. 2007). Osgood (1900), in his taxonomic revision of pocket mice, grouped the species of Perognathus into species groups primarily for convenience, although he recognized the close relationships of the species within each group. Williams (1978a) supported the close evolutionary relationships within Osgood's species groups based on karyotype diploid number and separated the flavus species group from the fasciatus species group (sensu Osgood 1900). Osgood suggested a close relationship between the flavus and longimembris species groups, a view upheld by recent mitochondrial DNA (mtDNA) evidence (Alexander and Riddle 2005; Hafner et al. 2007). Williams (1978a) further suggested a close relationship between the fasciatus and parvus species groups, but this relationship is yet to be supported robustly in recent studies.
Geographie ranges of species in the fasciatus species group following taxonomy of Hoffmeister (1986). Light shaded = Perognathus fasciatus, dark shaded = P. apache, stippled = P. flavescens. Lined areas represent major mountain ranges. Black circles show approximate collecting localities of specimens examined. Numbers refer to localities listed in Appendix I.
The fasciatus species group currently includes at least 2 recognized species of pocket mice, P. fasciatus and P. flavescens. Although Williams (1978b) subsumed P. apache under P. flavescens, proposing that detailed morphological and karyotypic data did not support the recognition of 2 species, Hoffmeister (1986) continued to recognize P. apache as a separate species, suggesting that the data of Williams (1978b) were not conclusive. For clarity we follow Hoffmeister in referring to populations within the intermountain basins of the North American cordillera as P. apache and populations east of the Front Range of the Rocky Mountains in the central and southern Great Plains as P. flavescens (Fig. 1), except when noted.
Williams (1978b) proposed that the ancestral species giving rise to the extant fasciatus species group was distributed across the northern and central Great Plains during the penultimate Pleistocene interglacial period. Under this model, as climates cooled during the latest glacial period, this species tracked habitats to the south, where the range was fragmented across the trans-Pecos region of western Texas. He envisioned this region as a transition zone: to the east P. fasciatus originated in the higher elevation habitats of the Edwards Plateau of central Texas, and to the west P. flavescens (including apache) originated in the Chihuahuan Desert of northern Chihuahua and southern New Mexico. As climates warmed again, P. fasciatus tracked expanding shrub and grassland habitats into the northern Great Plains, and P. flavescens remained in the northern Chihuahuan Desert, later also expanding northward onto the Colorado Plateau and into the Great Plains, less far north than P. fasciatus. This model predicts a late Pleistocene diversification between P. flavescens and P. fasciatus. Furthermore, the model predicts little genetic diversity between populations of P. flavescens in the Great Plains and P. apache in the intermountain basins, given the predicted recent coalescence of these lineages and predicted introgression near their contact zone (Williams 1978b).
An alternative hypothesis for biotic diversification across western North American arid lands is gaining support, in part from recent studies of a variety of heteromyid taxa. This hypothesis portrays a deeper history of geographic evolution in western North American arid lands, with diversification of species groups rooted in the late Miocene and species diversification continuing through the Pliocene and the Pleistocene (Hafner et al. 2007; Jezkova et al. 2009; McKnight 2005; Riddle et al. 2000a, 2000b, 2000c). A study of particular relevance to developing plausible hypotheses for geographic evolution of the fasciatus species group traced geographic evolution of the flavus species group using molecular clock analyses with both mitochondrial and nuclear data (Neiswenter and Riddle 2010). That study proposed a causal association between the diversification of basal lineages within the flavus species group and the initial development and expansion of arid grasslands. The authors identified several geographically distinct and reciprocally monophyletic phylogroups that were hypothesized to have arisen by vicariance between the Chihuahuan Desert, Great Plains, and Colorado Plateau bio-geographic regions. This distribution and set of postulated biogeographic events include a broad area of sympatry between the fasciatus and flavus species groups (Fig. 2), which provides rationale for developing a hypothesis for divergence within the fasciatus species group. First, we postulate that basal species group divergence began in the late Miocene coincident with the initial rapid spread of arid grasslands (Retallack 2001) and associated biota (Neiswenter and Riddle 2010). Second, we hypothesize a later bout of diversification as lineages comprising separate geographic isolates on the Colorado Plateau (including the southern Rocky Mountains), and in the northern Chihuahuan Desert and Great Plains, began diverging during the Pliocene or possibly early Pleistocene.
Distribution of Perognathus fasciatus species group (gray) and flavus species group (stippled) showing the area of sympatry in 3 regions.
We tested the above hypotheses, which outline specific spatial and temporal relationships within the fasciatus species group, within a modern molecular phylogeographic framework. We used mtDNA and amplified fragment length polymorphisms (AFLPs) to infer the distribution of evolutionary lineages across the topographically complex shrub-steppe and grassland communities over which the group is distributed (Fig. 1). Specifically, we developed mitochondrial phylogenetic hypotheses, estimated the timing of major divergences, and correlated the lineages with geologic and climatic events. We used a molecular clock to estimate divergence times for the major lineages within the fasciatus species group and evaluated the Pleistocene time frame of diversification between P. fasciatus and P. flavescens postulated by Williams (1978b). We also evaluated the extent to which mtDNA provides evidence of sex-biased gene flow between populations of P. apache and P. flavescens by testing for concordance between the mtDNA and AFLP data. Finally, we compared the evolution of 2 arid grassland rodent groups that co-occur across several biogeographic regions, the fasciatus and flavus species groups (Fig. 2), to evaluate whether diversification in each group might have been influenced by a similar suite of historical geologic or climatic events.
Materials and Methods
Specimens were sampled throughout the range of the fasciatus species group (Fig. 1). Samples field collected for this study were handled according to standards set forth by the American Society of Mammalogists (Gannon et al. 2007). Voucher skins and skeletons were collected when possible; however, in several instances samples, where provided from other researchers or state permits, allowed only the collection of nondestructive ear clips. Sampling was supplemented with tissues loaned from various natural history museums and personal collections (Appendix I). An attempt was made to include representatives from throughout the geographic range of each species to capture the phylogeographic diversity of the entire species group; however, we were unable to include representatives from every described subspecies.
Sequencing.—We sequenced the protein coding cytochrome oxidase III mitochondrial gene (COIII) from 1 to 7 (X̄ = 3) individuals per general locality. The primers 8618 and 9323 (Riddle 1995), and in some cases a primer specific to samples of P. apache (Table 1), were used for polymerase chain reaction amplification using the following protocol: an initial denaturation at 95 °C for 4 min, followed by 95 °C for 1 min, 55°C for 1 min, and 70°C for 1 min repeated 30 times, and a final extension step at 70°C for 5 min. Sequencing was performed in 1 of 2 ways. Several of the mtDNA sequences included in the analyses originated in Nickle (1994) and were sequenced using the protocol of Allard et al. (1991). All others were sequenced on an ABI 3130 (Applied Biosystems, Foster City, California) following the manufacturer's protocol. These sequences were checked for arbitrary base calls in SEQUENCHER 4.8 (Gene Codes Corporation, Ann Arbor, Michigan). Sequences were aligned using MEGA4 (Tamura et al. 2007). Nucleotide and haplotype diversity were estimated in DnaSP version 5 (Librado and Rozas 2009).
Primers used for amplified fragment length polymorphisms (AFLPs) and sequencing. Asterisk (*) denotes 56-FAM dye-labeled primer.
Phylogenetic trees were constructed using Bayesian analysis in MRBAYES (version 3.1.2—Huelsenbeck and Ronquist 2001). We performed several initial runs for each data set with different chain temperatures and branch length priors to confirm good mixing and convergence. The final analysis was run with a temperature of 0.05 and branch lengths set to 10. The GTR + G model was chosen based on likelihood ratio test results from MODELTEST (Posada and Crandall 1998). For each analysis we performed 2 independent runs with 4 chains each (1 hot and 3 cold), ran analyses for 4,000,000 generations, and summarized the last 10,000 trees of each run (20,000 trees total) using a 50% majority rule consensus tree using the posterior probabilities for clade support.
Maximum-likelihood analyses were performed using the program TREEFINDER (Jobb 2008) and the GTR + G model of evolution. Node support for the maximum-likelihood analysis was assessed using 100 bootstrap replicates. With the exception of the model of evolution, default values were used in all maximum-likelihood analyses.
In the molecular clock analyses we included COIII sequences downloaded from GenBank for several heteromyid genera and species outside of the fasciatus species group, including Dipodomys nelsoni, Chaetodipus baileyi, C. formosus, C. hispidus, C. eremicus, P. longimembris, P. parvus, P. flavus, and P. merriami (Appendix I). We used published fossil data (Hafner et al. 2007) to calibrate the molecular clock at a node outside of the fasciatus species group. We selected a representative from each of the mtDNA clades (see ' “Results”) and conducted a relaxed, uncorrelated lognormal molecular clock analysis in BEAST version 1.4.7 (Drummond and Rambaut 2007). Fossil calibration was based on the oldest know fossil that is a taxonomically reliable representative of the subfamily Perognathinae, estimated at 20–22 million years ago (mya— Hafner et al. 2007; James 1963). We conservatively placed this date at the base of the clade that includes Chaetodipus + Perognathus, as was done by Hafner et al. (2007), because the fossil record does not distinguish between these 2 genera. We used the HKY +1 + G model and constrained the Perognathinae and the fasciatus species group clades to be monophyletic. Several runs were conducted to ensure stationarity. The final analysis was 106 generations, sampling every 1,000 generations, and results were summarized after a 10% burn-in in TRACER version 1.4 (Rambaut and Drummond 2007) and FIGTREE version 1.1.2 (Rambaut 2008).
Fragment analyses.—We followed the protocol designed by Vos et al. (1995) for AFLP amplification. Briefly, total genomic DNA was restricted using EcoRI and MSEI enzymes, and known sequences were ligated to the restriction cut sites (Table 1). Restriction-ligated DNA fragments were amplified 1st in a preselective amplification and then used for subsequent selective amplifications using combinations of fluorescently labeled primers for the EcoRI primer and unlabeled MSEI primers (Table 1). Selective amplifications were sequenced at the genomics facility at the University of Nevada, Reno, on an ABI 3430 (Applied Biosystems). Selective amplification reactions were conducted twice to confirm the recovery of identical peak profiles. AFLP profiles were scored automatically in GeneMapper (Applied Biosystems) using a peak height threshold of 100 for peaks between 100 and 500 base pairs (bp). Final calls were confirmed by eye, and only unambiguous peaks were used in final analyses. Selective primers screened but not used are available from the authors.
To assess potential gene flow between P. apache and P. flavescens we used the clustering program STRUCTURE version 2.2.3 (Falush et al. 2007). We 1st ran exploratory analyses for a range of genetic groups (k), burn-in, and chain length. Initial runs were used to confirm good mixing and stationarity and to determine appropriate burn-in and chain length. We followed the guidelines in the user's manual for determining k qualitatively. Exploratory results suggested strong evidence for population structuring in the AFLP data set that was qualitatively similar to the mtDNA results.
We used the following methodology in STRUCTURE to assess nuclear gene flow between mtDNA populations of P. flavescens and P. apache. We a priori assigned each individual of either P. flavescens or P. apache to 1 of 3 populations, based on the results of the mtDNA analyses (see “Results”) to test for introgression of the nuclear genome among the mtDNA clades. Final analyses were run using prior population information (USEPOPINFO = 1) to identify possible hybrids. The RECESSIVEALLELES option was set to 1, burn-in was 105, and chain lengths were 106 generations. To determine the extent of introgression we defined 2 additional parameters, the probability that an individual is an immigrant (v) or has an immigrant ancestor in the last G generations. We set GENSBACK = 2 to assess immigration back to an individual's grandparents (2 generations) and ran the analyses for multiple values of v (MIGRPRIOR = 0.01, 0.05, and 0.1) to cover a range of plausible migration scenarios (Pritchard et al. 2000).
Sequencing analyses.—Eighty samples from the P. fasciatus species group from 29 general localities were sequenced for COIII. The resulting alignment consisted of 576 bp, of which 178 sites were parsimony informative. No gaps, insertions, or deletions were detected. Sequences are deposited in GenBank under accession numbers JF811601–JF811680.
Results of the Bayesian and maximum-likelihood phylo-genetic analyses were congruent and supported 4 major lineages (Fig. 3) with considerable uncorrected pairwise percent divergence between them (Table 2). The basal node of the tree corresponds to approximately 18% divergence between P. fasciatus and the rest of the group. Two of the major clades correspond with P. apache. The apache North clade is distributed in the northern Colorado Plateau north of the San Juan River in Utah and the apache South clade is distributed in the southern Colorado Plateau-northern Chihuahuan Desert. The 2 apache clades do not collectively form a single monophyletic clade; rather, the 4th major clade is sister to the apache South clade and corresponds with the distribution of P. flavescens. Haplotype and nucleotide diversity values within each of the 4 major clades are summarized in Table 3.
Distribution of major mitochondrial DNA (mtDNA) clades and amplified fragment length polymorphism (AFLP) groups in the Perognathus fasciatus species group are shown in relation to the major mountain chains (lined area on map). The phylogenetic tree at left is the consensus tree from the Bayesian analyses of mitochondrial DNA and is identical to the supported topology obtained for the maximum-likelihood (ML) tree (not shown). Numbers at nodes represent Bayesian posterior probability and ML bootstrap values, respectively. The bar graph shows the results of the exploratory AFLP analysis in STRUCTURE version 2.2.3 (Falush et al. 2007) for k = 4. Each bar represents the probability of a single individual belonging to 1 of the 4 groups. Shades of the bars corresponds to the shades of the symbols for each of the 4 mtDNA clades to show general congruence of mtDNA and AFLP groupings.
Haplotype and nucleotide diversity values for each mitochondrial clade of Perognathus.
Haplotype diversity (SD)
Nucleotide diversity (SD)
Three of the major clades have further structuring within them. The fasciatus clade contains 3 well-supported subclades, 1 currently restricted to the Wyoming Basin, a 2nd from the Front Range of the Rocky Mountains, and the 3rd widely distributed across the northern Great Plains and Wyoming Basin (Fig. 4A). The apache North clade is further divided north and south of the Colorado River in eastern Utah, and the apache South clade is divided east and west of the Chuska Mountains along the northern Arizona–New Mexico border (Fig. 4B).
Phylogenetic results of the mitochondrial DNA Baysian analysis showing more detailed geographic structuring within A) Perognathus fasciatus in the Wyoming Basin and northern Great Plains and B) the 2 apache clades in relation to the Colorado and San Juan rivers on the Colorado Plateau. Numbers at nodes are posterior probabilities and maximum-likelihood bootstrap support, respectively. Lined areas represent major mountain ranges.
Under the fossil-calibrated molecular clock the divergence between P. fasciatus and the ancestor to the other taxa in the group is estimated to have occurred in the late Miocene, approximately 7.4 my a (95% highest posterior density interval = 4.8-9.9 mya; Fig. 5). The mean time to most recent common ancestor for the apache-flavescens clade is 3.6 mya (95% highest posterior density interval = 2.3–5.0 mya), and the divergence between the flavescens clade and apache South is estimated at 2.5 mya (95% highest posterior density interval = 1.5-3.5 mya). Within the major clades further diversification is estimated to have occurred during the middle Pleistocene. The estimated mean mutation rate under the fossil calibrated clock is 0.034 ± 2.5 × 10−4 substitutions site−1 million years−1, and the likelihood estimate is −4,684.78 ±0.1.
Chronogram from relaxed molecular clock analyses showing molecular dating of major lineages in the Perognathus fasciatus species group. Numbers at nodes are median values in millions of years, and dark bars represent 95% highest posterior density (HPD) intervals. Timescale is estimated for visual purposes.
Fragment analyses.—The AFLP profiles for 2 primer combinations were developed for 68 individuals from the fasciatus species group. Some individuals were not included because DNA or tissue was no longer available for some of the samples obtained from Nickle (1994). A total of 189 variable sites were scored for the Mse + AGC primer, with an average of 59 alleles present per individual. The Mse + ATC primer resulted in a total of 157 variable sites being scored, with an average of 53 alleles present per individual. Exploratory analyses suggested 4 groups (mean In likelihood = −9,560), with membership corresponding closely to the 4 major mtDNA clades reported above (Fig. 3).
The results of the gene flow analysis suggest that a few individuals might have had immigrant ancestry in the past 2 generations (Table 4). With the highest probability of migration, v = 0.1, 3 individuals show a high probability of having an immigrant grandparent and 1 individual is not assigned strongly to any group. When migration is assumed to be low, v = 0.01, only 2 of these individuals were shown to have a high probability of immigrant ancestry. All other individuals were assigned to their respective mtDNA clades with a probability > 0.90 in all analyses.
Results from the STRUCTURE (Falush et al. 2007) analysis of gene flow between Perognathus flavescens and P. apache, showing the possible source population and ancestry in individuals with <0.90 probability of no immigrant ancestry under different migration priors. Individuals are labeled by their museum numbers or other number if no museum voucher is available; see Appendix I for more information, v is the probability that an individual is an immigrant or has immigrant ancestry. No immigrant ancestry is the probability that the nuclear genomic ancestry of the individual is from the same region as the mitochondrial DNA (mtDNA) from that individual. The other immigrant columns show the probability that the individual has ancestry from the possible immigrant source up to 2 generations ago. Rows do not add to 1 because of a small probability of ancestry from other populations. Bold numbers are probabilities < 0.50.
Possible immigrant source
No immigrant ancestry
The biogeographic hypothesis (and resulting taxonomy) suggested by Williams (1978b), largely based on morphology, underestimates the evolutionary diversity within thefasciatus species group that is revealed by the molecular sequence divergence. We identified 4 major mtDNA clades within the fasciatus species group that also are recovered using nuclear DNA. Additionally, further geographic and genetic structuring occurs within the 4 mtDNA clades. Using the fossil-calibrated molecular clock, diversification within the fasciatus species group began during the last half of the Miocene and continued into the Pleistocene.
Fossil and molecular evidence suggest massive alteration of mammalian diversity throughout the Cenozoic (Blois and Hadley 2009; Kohn and Fremd 2008; Riddle 1995; Webb 1977). Widespread changes in the distribution of arid biomes, including the expansion of arid grasslands and shrublands throughout the western North American lowlands, occurred in concert with several bouts of global cooling and drying (Axelrod 1985; Retallack 1997, 2001). Retallack (1997) used paleosols, stable isotopes, and fossil evidence to develop a model of the evolution of the grassland biomes, which depicts 3 stages of successive drying and cooling beginning with the shift from dry tropical forest to savannas with desert shrub and bunchgrasses around the Eocene-Oligocene boundary. The 2nd stage includes the expansion of sod-forming short-grass prairies during the middle Miocene (15 mya). The latest bout of cooling and increased aridity, beginning around the late Miocene (5–7 mya), involved the expansion of C4 grasslands and desert scrub.
Molecular clock analyses suggest the major lineages of the fasciatus species group initially began diverging during the 3rd climatic episode of the late Miocene, coincident with expanding arid grasslands throughout North America. The timing of diversification and habitat use (Manning and Jones 1988; Monk and Jones 1996; Williams 1978b) of the species in the fasciatus species group is consistent with an hypothesis outlined for the flavus species group (Neiswenter and Riddle 2010), which is codistributed with the fasciatus species group across parts of the Colorado Plateau, northern Chihuahuan Desert, and Great Plains (Fig. 2). The expansion of the fasciatus species group throughout the northern latitudes of North America might have began in the late Miocene as the ancestor to the group followed the expanding arid-adapted C4 grasslands. This expansion likely was caused by the decrease in carbon dioxide concentrations (Retallack 2001), increased seasonality of precipitation and wildfires (Osborne 2008), or some combination of these and possibly other factors (Kohn and Fremd 2008). The initial divergence within the fasciatus species group could have arisen as northern and southern isolates in the Wyoming Basin–northern Great Plains and Colorado Plateau–southern Great Plains. P. fasciatus is found only in higher-elevation grasslands in the southern portions of its geographic distribution along the Front Range of the Rocky Mountains, suggesting that it is adapted to cooler climates compared with P. flavescens or P. apache.
Following the north-south split within the fasciatus species group regional diversification continued throughout the Pliocene and Pleistocene. The apache North clade is estimated to have diverged during the middle Pliocene, coincident with extensive geological uplift and volcanism in the Colorado Plateau region throughout the Pliocene (Raymo and Ruddiman 1992; Sahagian et al. 2002). The molecular clock analysis points to a late Pliocene-Pleistocene time for the most recent common ancestor between the flavescens clade and apache South clade across the southern Rocky Mountains. A similar estimated timing of the Great Plains divergence within the codistributed flavus species group (merriami Chihuahuan Desert versus Great Plains, 6.0–3.3 mya) might represent the response of these taxa to a common event, the culmination of faulting and beginning of epeirogenic uplift along the Rio Grande Rift that resulted in the closing of a savanna corridor that connected populations to the east and west (Axelrod and Bailey 1976; McMillan et al. 2002; Morgan et al. 1986). In conjunction with the geologic activity the late Pliocene transition (approximately 3.2–2.7 mya) marks a distinct period of cooling in the Northern Hemisphere (Sosdian and Rosenthal 2009) that might have promoted regional adaptations of the local biota to the changing environmental conditions within each of these biogeographic regions, further contributing to diversification of the associated biota. These results are consistent with the hypothesis of Neiswenter and Riddle (2010) that a large portion of the phylogeographic diversity in these arid grassland species is a result of lineage diversification from geologic and climatic phenomena prior to the major glacial cycles of the Pleistocene, laying the foundation for within-region population structuring beginning in the middle Pleistocene.
Several of the major clades have further geographic structuring that is consistent with a scenario of persistence of discrete lineages in separate Pleistocene refugia. Each of the major clades has further structuring that is estimated to have begun in the middle Pleistocene. The middle Pleistocene transition, approximately 1.2–0.7 mya, is marked by Milankovich cycles shifting from the dominant 41,000-year obliquity cycles of the early Pleistocene to longer more extreme 100,000-year cycles (Sosdian and Rosenthal 2009). The longer, more extreme cycles might have further isolated populations of each of the clades in the fasciatus species group as they shifted their distribution, tracking their preferred habitat in response to the ever-changing climate. In North America much of the northern Great Plains was covered by glaciers during the colder climate cycles, so the habitable area available to P. fasciatus likely was reduced to areas along the Front Range of the Rocky Mountains and within the Wyoming Basin, which could have served as refugia during the Pleistocene, resulting in the current genetic structure recovered within this clade. The Front Range and Wyoming Basin might have served as refugia for other codistributed lowland taxa, such as grasshopper mice (Onychomys leucogaster—Riddle and Honeycutt 1990; Riddle et al. 1993), because these regions could have been buffered from severe climate changes due to their topographic complexity. Additionally, within both apache South and apache North are clades whose diversification can be explained by the persistence of multiple Pleistocene refugia throughout the basins of the Colorado Plateau. Because the Great Plains is topographically less complex than the intermountain basins, it is plausible that only a single refugium was available to the flavescens clade during the Pleistocene, which would explain the lack of similar substructure within this clade. A population genetic approach with more detailed sampling of individuals and genes for each of these species and other sympatric taxa is necessary to evaluate further details for presence, size, and locations of each of these postulated refugia.
Our nuclear data support the mtDNA groupings and lend credibility to the evolutionary history documented with the maternally inherited mtDNA, although a small number of individuals were identified as having immigrant ancestry. Using nuclear data, we assigned all but 4 individuals to their respective mtDNA clades with high probability in the STRUCTURE analysis, even when assuming the highest migration rate. Williams (1978b) suggested that P. flavescens and P. apache probably were introgressing across the trans-Pecos region of western Texas and southeastern New Mexico, an area where both species are fairly uncommon. Some indication of local introgression exists at each of the contact zones between the 3 mtDNA clades that comprise P. apache and P. flavescens under the assumed migration probabilities. To the extent that the range of migration priors used in these analyses reflect the true range of dispersal probability in these species, we can identify the probability of nuclear immigration among the mtDNA clades identified. For example, 2 individuals (NMMNH 3259 and NMMNH 3258) with flavescens mtDNA located near the trans-Pecos and southern Rocky Mountains at localities 13 and 15 (Fig. 1; Appendix I) have a high probability of having an immigrant grandparent, particularly under the higher migration scenarios. Additionally, 1 of the individuals with apache North mtDNA (MSB 76895; Appendix I) is predicted to have an apache South nuclear DNA component. This individual is from near the San Juan River (locality 25; Fig. 1, Appendix I) in Utah, close to the probable contact zone between apache North and apache South mtDNA clades. One of the samples with apache North mtDNA (LVT 9907) shows a possible immigration ancestry with flavescens but only under the highest migration prior. This sample is not near a contact zone with flavescens populations, making an introgression hypothesis less likely. Furthermore, the individual does not have a high probability of coming from any population under the highest migration prior. This individual might have retained some ancestral polymorphisms or homoplasies, or both, that make it more difficult to assign to its respective mtDNA population under a high migration scenario.
If nuclear introgression is occurring between mtDNA populations, it is geographically limited and uncommon. Perognathus are relatively small rodents that likely have restricted dispersal abilities (Manning and Jones 1988; Monk and Jones 1996; Williams 1978b), making the lower value assumed for the migration prior (0.01) a more likely representation of the true value. Only 2 individuals are predicted to have immigrant ancestry under this assumption, although only 3 individuals have a high probability of immigrant ancestry under the highest migration prior. Regardless, in each instance the introgression of nuclear DNA to mtDNA populations is predicted to have occurred at least 2 generations ago (i.e., no F1 hybrids were found). Moreover, the same localities have other individuals that are not predicted to have immigrant ancestry. Other localities also are found in the same general area of the localities with predicted immigrants and within Williams (1978b) proposed area of introgression (e.g., locality 20), with multiple individuals that have no immigrant ancestry. In light of this we propose that the congruent mtDNA and AFLP groups are maintaining their genetic distinction but suggest further investigation in each of the potential hybrid areas to better describe the nature and extent of the contact zones.
Morphologically based taxonomy might not be appropriate for delineating species diversity in Perognathus. We identified at least 4 genetic lineages that likely satisfy the requisites of a variety of species concepts, for example the Genetic Species Concept (as it applies to mammals—Bradley and Baker 2001) and Genealogical Concordance Concept (Avise and Ball 1990). Species designation was not an objective of this study, and we acknowledge that a more detailed analysis of clade boundaries with the inclusion of all recognized subspecies is necessary to evaluate fully the specific status of members in this group. However, several studies have shown that current taxonomy (largely based on skeletal morphology) grossly underestimates the molecular diversity within the genus Perognathus (Alexander and Riddle 2005; McKnight 2005; Neiswenter and Riddle 2010). Species of Perognathus exhibit extensive plasticity in external morphology that can be causally or directly linked to environmental conditions experienced by the individual. This is evidenced by extreme intraspecific color variation depending on the color of substrate on which an individual is found (e.g., P. apache melanotis) and skeletal variation associated with climatic conditions. Williams (1978b) showed that morphological variation tended to follow basic ecogeographic rules in Perognathus; mice from colder environments are larger, and the variation in relative measurements of auditory bullae and rostral size are associated with wetter environments.
We hypothesize that the apparent introgression between P. flavescens and P. apache across the trans-Pecos region is due to the convergence of morphology within similar environments; warmer, drier deserts and grasslands in the southern portions of the ranges of both species contain mice that have converged morphologically. This pattern could be more widespread within the genus Perognathus than previously recognized, being apparent in different species and in different supposed transition zones. For example, across the same region P. flavus and P. merriami previously were thought to hybridize (based on intermediate morphology) through P. m. gilvus (Wilson 1973) but probably coexist as separate species that are morphologically similar, based on levels of molecular divergence (Brant and Lee 2006; Neiswenter and Riddle 2010). Additionally, Osgood (1900) believed that P. callistus (currently synonymized with P. fasciatus) from the Wyoming Basin was intermediate between P. fasciatus and P. apache, resembling P. apache from the Uinta Basin in skull characteristics (presumably due to intermediate climate) but similar in color to P. fasciatus from the Great Plains. A reanalysis of morphological data from Williams (1978b) in light of the molecular results reported here is warranted to evaluate this hypothesis.
Conclusions and future directions.—A general model for the evolution and diversification of a North American arid grasslands biota is beginning to emerge. This study tested hypotheses regarding the diversification of arid grasslands and shrublands across several ecoregions in western North America. Although no formal statistical comparative phylo-geographic analysis was used, we found congruence with previous results from the flavus species group that indicate that these 2 independent lineages of small-bodied pocket mice likely followed expanding arid grassland habitat as it emerged in the late Miocene and diversified throughout the Pliocene and Pleistocene across the topographically complex North American cordillera. We recognize the need to include multi-gene data sets and more codistributed species within a comparative phylogeographic framework within this system, and also the need for a more detailed sampling and statistical approach, to evaluate the Pleistocene refugiai hypotheses outlined above. Finally, species limits within the fasciatus species group should be reevaluated by incorporating all available evidence from every subspecies along with a denser sampling at each of the clade boundaries.
We thank the following institutions and people for supplying tissue samples: Angelo State Natural History Collection (R. C. Dowler and L. Ammerman), Denver Museum of Nature and Science (J. Demboski), Museum of Texas Tech University (R. J. Baker and H. J. Garner), New Mexico Museum of Natural History (D. J. Hafner and P. Gegick), Museum of Southwestern Biology (J. A. Cook, M. Bogan, and C. Parmenter), Royal Alberta Museum (B. Weimann), P. Stapp, R. A. Sweitzer, and Sternberg Museum (J. R. Choate and C. J. Schmidt). We thank the systematics discussion group at University of Nevada, Las Vegas, for helpful comments and discussions, D. J. Hafner for his irreplaceable logistic support and intellectual input, and K. Mailland for laboratory assistance. This project was supported, in part, through funds provided by National Science Foundation grants DEB-0237166 to BRR and DEB-0236957 to D. J. Hafner, a Major Research Instrumentation grant DBI-0421519 to the University of Nevada, Las Vegas, and grants from the Graduate and Professional Students Association at University of Nevada, Las Vegas, the American Museum of Natural History Theodore Roosevelt Fund, American Society of Mammalogists grants-in-aid program, and T&E Inc. (Gila, New Mexico) to SAN.
Specimens examined in the Perognathus fasciatusspecies group and outgroup species. The 1st column corresponds to numbered localities in Fig. 1. Abbreviations are as follows: ASNHC and ASK—Angelo State Natural History Collection; LVT—University of Nevada, Las Vegas, tissue collection; MHP—Sternberg Museum; MSB and NK—Museum of Southwestern Biology; NMMNH—New Mexico Museum of Natural History; PF—personal collection of Paul Stapp; TTU and TK—The Museum, Texas Tech; RAM—Royal Alberta Museum; ZM—Denver Museum of Nature and Science.
Canada, Alberta, 7.5 miles S, 5 miles E of Cavendish
Canada, Alberta, 7 miles S, 2 miles E of Cavendish
Canada, Alberta, 1 mile S, 2 miles E of Cavendish
Canada, Alberta, 0.75 miles S, 2 miles E of Cavendish
Canada, Alberta, Canadian Forces Base Suffield, near Medicine Hat
Arizona, Coconino County, 7 miles N Cameron
Arizona, Navajo County, Petrified Forest National Park, 1.0 miles S, 0.4 miles
E Rainbow Forest Museum
Colorado, Custer County, 9 miles NE Silver Cliff
Colorado, Moffat County
Colorado, Ward County, Pawnee NGL
Colorado, Weld County, 4 miles S Roggen
Kansas, Dickinson County, 2 miles N, 3 miles W Abilene
Kansas, Finney County, 4 miles S Holcomb
Montana, Roosevelt County, 9 miles SE Bainville
North Dakota, Ransom County, Sheyenne NGL
Nebraska, Sheridan County, 27 miles N Lakeside
New Mexico, Chaves County, 5 miles W Kenna
New Mexico, De Baca County, 16 miles S, 3 miles E Taiban
New Mexico, Lea County, 20 miles W Hobbs
New Mexico, McKinley County, 1 miles N NM HWY 53 on Zuni
New Mexico, Mora County, 6 miles N Logan
New Mexico, San Juan County, 38 miles S Farmington
New Mexico, Socorro County, Sevilleta National Wildlife Refuge, Rio Salado
Texas, Ward County, Monahans Sandhills State Park, 1 mile N headquarters building
Utah, San Juan County, 16 miles N Monticello
Utah, Uintah County, 6 miles S Bonanza
Utah, Emery County, 7 miles S Green River
Utah, Grand County, 10 miles N Moab
Utah, San Juan County, Mexican Hat
Wyoming, Carbon County, 10 miles S Seminoe Dam
Wyoming, Natrona County, 25 miles NW Independence Rock
Wyoming, Sweetwater County, 10 miles SE Eden
Wyoming, Sweetwater County, 28 miles N Green River on CR5
Wyoming, Sweetwater County, 18.1 miles S Bitter Creek
Wyoming, Sweetwater County, 25 miles S Bitter Creek
1990. Principles of genealogical concordance in species concepts and biological taxonomy. Pp. 45–67 in Oxford surveys in evolutionary biology (FutuymaD., AntonovicsJ., eds.). Oxford University Press, Oxford, United Kingdom. Vol. 7.
. 2006. Studies of morphological and molecular phylogenetic divergence in spiders (Araneae: Homalo-nychus) from the American southwest, including divergence along the Baja California peninsula. Molecular Phylogenetics and Evolution 38:470–487.
. 2010. Isolation and introgression in the intermountain west: contrasting gene genealogies reveal the complex biogeographic history of the American pika (Ochotona princeps). Journal of Biogeography 37:344–362.
. 2010. Range-wide genetic differentiation among North American great gray owls (Strix nebulosa) reveals a distinct lineage restricted to the Sierra Nevada, California. Molecular Phylogenetics and Evolution 56:212–221.
1963. Paleontology and nonmarine stratigraphy of the Cuyama Valley Badlands, California. Part I. Geology, faunal interpretations and systematic descriptions of Chiroptera, Insectivora, and Rodentia. University of California Publications in Geological Sciences 45:i–iv, 1-171.
1995. Molecular biogeography of the pocket mice (Chaetodipus and Perognathus) and grasshopper mice (Onychomys): the late Cenozoic development of a North American aridlands rodent guild. Journal of Mammalogy 76: 283–301.
. 2000a. Phylogeography and systematics of the Peromyscus eremicus species group and the historical biogeography of North American warm regional deserts. Molecular Phylogenetics and Evolution 17:145–160.
. 2000b. Comparative phylogeography of Baileys' pocket mouse (Chaetodipus baileyi) and the Peromyscus eremicus species group: historical vicariance of the Baja California peninsula. Molecular Phylogenetics and Evolution 17:161–172.
. 1993. Mitochondrial DNA phylogeography in northern grasshopper mice (Onychomys leucogster)—the influence of Quaternary climatic oscillations on population dispersion and divergence. Molecular Ecology 2: 183–193.