The genus Myotishas undergone significant taxonomic revision since the advent of DNA sequencing techniques. Prior morphological examination of Myotishas indicated as many as 4 subgenera correlated with foraging strategies. Recent studies using mitochondrial DNA (mtDNA) sequence data have questioned the validity of these subgenera and have indicated that several taxa may require reevaluation as to their position within Vespertilionidae. Nevertheless, no study has used large-scale nuclear DNA sequencing to examine relationships within Myotis. We generated 4,656 base pairs (bp) of nuclear intron (PRKC1, STAT5A, and THY) and exon (APOB, DMP1, and RAG2) sequence data in addition to 2,866 bp of mtDNA sequence data to test previously hypothesized subgeneric groupings of Myotis. We included 21 species of Myotis from all morphological subgenera previously suggested, representatives of all subfamilies and tribes currently recognized in Vespertilionidae, and multiple representatives of all other families currently included in the superfamily Vespertilionoidea. We also included a representative of the rare African genus Cistugo, because significant doubt exists about its familial position. Our phylogenetic analyses did not support the morphologically defined Myotis subgenera and confirm that morphological similarities among Myotis are the result of convergent evolution. Divergence estimates derived from the total data set were concordant with previous studies, suggesting a middle Miocene trans-Beringian dispersal from Asia colonized North America, with subsequent South American colonization and diversification prior to the formation of the Isthmus of Panama 3–4 million years ago. Myotis latirostris fell outside of Myotis, and the high genetic distance separating it from other Myotis suggested that M. latirostris represented a distinct genus. The genus Cistugo, previously a subgenus within Myotis, fell basal to all vespertilionids, with a high genetic distance separating it from Vespertilionidae. We conclude that Cistugo should constitute a distinct family within Vespertilionoidea.
Conflicts between morphological and molecular data are abundant in the literature (Lee 2001; Patterson 1987), and evolutionary relationships within the order Chiroptera are no exception. Using mitochondrial DNA (mtDNA) sequence data, Ruedi and Mayer (2001) revealed significant convergent evolution within Myotis that led them to propose a convergent evolutionary hypothesis for morphological similarities. In resolving higher-level chiropteran relationships and elucidating evolutionary origins of echolocation, Eick et al. (2005) found that the majority of morphological characters they examined were homoplastic and therefore taxonomically misleading. As is evident from these studies, diverse data sets are often necessary to obtain robust and accurate assessments of evolutionary relationships.
The chiropteran family Vespertilionidae is the 2nd most speciose mammalian family, with approximately 407 species and an essentially worldwide distribution (Simmons 2005). Because of extensive convergent evolution, lack of taxonomically informative characters, and likely a rapid initial diversification, resolving evolutionary relationships within Vespertilionidae, based on morphology, has been difficult. Since the advent of DNA sequence–based phylogenetic techniques, significant taxonomic revisions have been made to Vespertilionidae, with many of the morphology-based phylogenies modified (Hoofer and Van Den Bussche 2003). Some of the most significant taxonomic changes included elevation of the subfamily Miniopterinae to family status (Miniopteridae—Hoofer and Van Den Bussche 2003; Miller-Butterworth et al. 2007), reorganization and elevation of the tribe Myotini to subfamilial status (Myotinae), and identification of many problematic assemblages at multiple taxonomic levels (e.g., nonmonophyly of the genus Eptesicus and validity of several tribes within Vespertilioninae).
Not only have the phylogenetic affinities between Cistugo and Myotis been problematic, but Myotis itself has been a source of significant systematic confusion. This genus is composed of approximately 100 species, distributed across all continents except Antarctica (Simmons 2005) and displays high levels of ecological and behavioral diversity. Original designations based on morphology proposed 3 subgeneric groupings (Myotis, Leuconoe, and Selysius) corresponding to modes of flight and feeding guilds (Findley 1972). Although these subgeneric groupings have been challenged based on dental characters (Godawa Stormark 1998; Menu 1987), many classifications of vespertilionid bats continue to recognize these subgenera. Koopman (1993), 1994) added Cistugo to the subgenera of Findley (1972), thus producing 4 subgeneric groupings. Multiple studies using DNA sequences have shown that these subgenera are not natural assemblages that reflect evolutionary history; instead, the phylogenetic relationships reflect biogeographic relationships, with major clades corresponding to continental landmasses and regions therein (Hoofer and Van Den Bussche 2003; Ruedi and Mayer 2001; Stadelmann et al. 2004, 2007). Adding to the difficulties in revealing the evolutionary history of Myotis, in the most recent systematic assessment, M. latirostris fell basal to the Myotis clade (although with relatively weak support), indicating that this species might represent either the oldest extant lineage of Myotis or even a distinct genus (Stadelmann et al. 2007). With 1 exception, previous phylogenetic studies of Myotis used only mtDNA sequences. Stadelmann et al. (2007) analyzed 1,148 base pairs (bp) of the nuclear RAG2 gene; however, RAG2 produced few phylogenetically informative characters, therefore adding little to the mtDNA phylogeny. Given the contrasting evolutionary relationships of Cistugo, Myotis, and M. latirostris based on morphology and mitochondrial sequence data, additional insight from nuclear sequence data could clarify these relationships and help stabilize the taxonomy of Vespertilionidae.
Because the majority of recent studies support elevation of Cistugo to generic status but were unable to determine whether Cistugo should be treated as a member of Vespertilionidae or as a member of a distinct family within Vespertilionoidea (Bickham et al. 2004; Eick et al. 2005; Stadelmann et al. 2004), our 1st objective was to test the validity of the Myotis subgenera Leuconoe, Myotis, and Selysius as defined by Findley (1972) and the validity of biogeographical grouping of species based on mtDNA sequence data (Bickham et al. 2004; Ruedi and Mayer 2001; Stadelmann et al. 2004, 2007). Our 2nd objective was to assess the phylogenetic hypothesis that Cistugo represents a distinct family within Vespertilionoidea. Additionally, Stadelmann et al. (2007) suggested that M. latirostris represented either the basal lineage of Myotis or possibly a distinct vespertilionid genus. Therefore, our 3rd objective was to assess the phylogenetic hypotheses of Stadelmann et al. (2007) regarding the uniqueness of M. latirostris. Although the questions associated with our objectives have been examined in several recent molecular studies (Bickham et al. 2004; Ruedi and Mayer 2001; Stadelmann et al. 2004, 2007), all of those studies examined portions of the mitochondrial genome. Because of its maternal inheritance and therefore differential accumulation of mutations relative to the nuclear genome, phylogenetic analysis of mtDNA only can result in a phylogeny not indicative of the true evolutionary history for a given taxon (Avise 1994). To address our 3 objectives, we generated DNA sequence data from 3 nuclear exons (APOB, DMP1, and RAG2) and 3 nuclear introns (PRKC1, STAT5A, and THY) from a taxonomically diverse sampling of Vespertilionoidea and combined these nuclear DNA sequence data with mitochondrial ribosomal DNA sequences previously generated for these same taxa (Hoofer and Van Den Bussche 2003; Van Den Bussche and Hoofer 2004).
Materials and Methods
We included 80 ingroup taxa representing Vespertilionidae, Molossidae, Miniopteridae, and Natalidae, the 4 families that molecular studies have indicated comprise the superfamily Vespertilionoidea (Eick et al. 2005; Teeling et al. 2005; Van Den Bussche and Hoofer 2004). Within Vespertilionidae, we included representatives of all vespertilionid subfamilies and tribes indicated in Hoofer and Van Den Bussche (2003); see Appendix I). Finally, we included representatives of Emballonuridae, Phyllostomidae, Mormoopidae, Noctilionidae, Thyropteridae, and Myzopodidae as outgroup taxa.
Total genomic DNA was extracted from heart, liver, kidney, or muscle tissue samples following standard protocols (Longmire et al. 1997). Three nuclear exons (apolipoprotein b [APOB], dentin matrix acidic phosphoprotein 1 [DMP1], and recombination activating gene 2 [RAG2]) and 3 nuclear introns (protein kinase C iota [PRKC1], signal transducer and activator of transcription 5A [STAT5A], and thyrotropin [THY]) were targeted using previously designed primers (Appendix II). Polymerase chain reaction amplifications were carried out in 30-µl reactions containing 200–500 ng of DNA, 0.14 mM of each deoxynucleoside triphosphate, 6 µl of 10× buffer, 3.5 mM of MgCl2, 0.8 mg/ml of bovine serum albumin, 0.15 µM of each primer, 1 unit Taq polymerase, and double distilled water to volume. The general thermal profile consisted of an initial denaturation of 94°C for 3 min, followed by 35 cycles of 94°C for 1 min, annealing for 1 min (see Appendix II for annealing temperatures), and 72°C for 1–2 min. A final elongation of 72°C for 30 min ensured all reactions ran to completion. In addition, we amplified the 12S rRNA, tRNAVal, and 16S rRNA regions of the mitochondrial genome for taxa not previously examined following the methods of Van Den Bussche and Hoofer (2000). Double-stranded products were purified using the Wizard SV Gel PCR Prep DNA Purification System (Promega, Madison, Wisconsin), and both strands of the purified polymerase chain reaction products were sequenced using Big Dye 1.1 chain terminators and an ABI 3130 Genetic Analyzer (Applied Biosystems, Inc., Foster City, California).
The mitochondrial and 6 nuclear data sets were aligned independently using Clustal X software (Thompson et al. 1997), and the resulting multiple alignments were imported into MacClade (Maddison and Maddison 2000) where each alignment was visually inspected and manually optimized. All of the nuclear alignments and the mitochondrial data contained insertion–deletion (indel) events, and gaps were introduced (either by Clustal X or manually) to optimize the alignment. All indel regions were examined carefully, and any position in the alignment where character state was not assigned confidently for all taxa was excluded from all analyses. Pairwise genetic distances were calculated in PAUP* (Swofford 2003) using maximum-likelihood (ML) model parameters estimated by MODELTEST version 3.06 (Posada and Crandall 1998).
A likelihood mapping analysis (Strimmer and von Haeseler 1997) as implemented in TREEPUZZLE 5.2 was conducted to quantify and compare the phylogenetic signal for the mtDNA and concatenated nuclear data sets. Incongruence between individual genetic markers (i.e., mtDNA versus RAG2, etc.) was determined by running maximum-parsimony (MP), ML, and Bayesian phylogenetic analyses for each data set independently and employing the 90% conflict criterion (De Queiroz 1993). This revealed no conflicting, well-supported nodes among data sets (well-supported nodes requiring ≥70% bootstrap support for MP and ML analyses and ≥0.95 posterior probability for Bayesian analyses). For further analyses we used 2 data sets: combined nuclear data and nuclear and mtDNA combined. Representatives of Phyllostomidae, Myzopodidae, Thyropteridae, Mormoopidae, Noctilionidae, and Emballonuridae were selected to serve as out-groups. We did not analyze the mtDNA data set separately because the data set differed from that analyzed by Hoofer and Van Den Bussche (2003) by only a few taxa. Analyses of the nuclear and combined nuclear and mitochondrial data sets were conducted as described below.
Maximum-likelihood and MP methods were carried out using PAUP* (Swofford 2003), whereas Bayesian phylogenetic analyses were conducted using MrBayes version 3.1.2 (Huelsenbeck and Ronquist 2001). For ML analyses the most appropriate model of nucleotide substitution was evaluated using MODELTEST version 3.06 (Posada and Crandall 1998), and trees were constructed using nearest neighbor interchange branch-swapping. For MP analyses, trees were generated using equal weighting and the heuristic search option, the maximum number of trees retained set at 500, tree-bisection-reconnection branch-swapping, and 25 random additions of input taxa. Reliability of clades from the ML and MP analyses was evaluated via bootstrap analysis. For the ML analyses, we performed 100 iterations with a heuristic search and nearest neighbor interchange branch-swapping. For the MP bootstrap analysis, we performed 1,000 iterations with a heuristic search that included 25 random additions of taxa and tree-bisection-reconnection branch-swapping. Bayesian analyses were performed using 4 simultaneous Markov chains run for 5 × 106 generations, with random, unconstrained starting trees. The analyses employed the GTR + I + Γ model of nucleotide substitution; values for model parameters were not defined a priori but were treated as unknown variables with uniform priors. Trees were sampled every 100 generations and “temperature” was set at 0.02. Resulting burn-in values (the point at which the model parameters and tree scores reached stationarity) were determined empirically by evaluating likelihood scores. All runs were checked for sufficient mixing, stable convergence on a unimodal posterior, and effective sample sizes (Drummond et al. 2002) > 100 for all parameters using TRACER version 1.4 (Drummond and Rambaut 2003).
For the combined data set we estimated node ages using the Markov chain Monte Carlo sampling method implemented in the program BEAST version 1.4.8 (Drummond and Rambaut 2007). To compare the null molecular clock model versus the alternative model, in which each branch is allowed its own unique rate, we used the likelihood ratio test (Felsenstein 1981). In the Bayesian framework Bayes factors (Kass and Raftery 1995; Newton and Raftery 1994; Suchard et al. 2001), as implemented in the program TRACER version 1.4 (Drummond and Rambaut 2003), were used to compare the relaxed uncorrelated exponential clock and relaxed uncorrelated lognormal clock. These analyses used the GTR + I + Γ model of sequence evolution and a Yule speciation tree prior (Yule 1924). Fossil calibrations were used to place a prior on 2 nodes. A minimum age of 30 million years ago (mya) was used for the Phyllostomidae–Mormoopidae divergence with a uniform distribution (Teeling et al. 2005) because the oldest fossils uniting this group are found in the Whitneyan 30– 32 mya (G. S. Morgan, University of Florida, pers. comm.). Therefore, the maximum age of the Mormoopidae–Phyllostomidae divergence (the maximum of the uniform distribution) was set at the Eocene–Oligocene boundary (34 mya), because actual divergences occurred some time before the fossil formed. The 2nd calibration was a minimum of 37 mya for the split between Vespertilionidae and Molossidae (Teeling et al. 2005), because verified vespertilionid and molossid fossils have been found from the middle Eocene (McKenna and Bell 1997). We used a lognormal prior distribution (offset = 37.0, X̄ = 0, SD = 1.3) on this calibration to encapsulate the entire middle Eocene in the prior.
Using the above calibrations as point estimates, we generated a chronogram using r8s version 1.60 (Sanderson 2003) to provide a starting tree for the dating analysis. This was done by allowing r8s to rescale the branch lengths on the phylogram resulting from phylogenetic analysis of the concatenated nuclear and mitochondrial sequence data to time rather than substitutions per site. For the BEAST analysis an initial run of 30,000,000 generations with 10% burn-in was run to optimize operators. The final analysis consisted of 2 separate runs of 30,000,000 generations, each with 10% burn-in. Results of these final 2 runs were log combined to obtain final estimates of divergence, and all runs were checked for sufficient mixing, stable convergence on a unimodal posterior, and effective sample sizes (Drummond et al. 2002) > 100 for all parameters using TRACER version 1.4 (Drummond and Rambaut 2003).
All sequences generated in this study were submitted to GenBank (see Appendix I for accession numbers), and each alignment is available from TreeBase (http://treebase.org). Results of the likelihood-mapping analysis indicated phylogenetic signal was nearly identical for the mtDNA and nuclear data sets with 94.9% and 96.0% of quartets resolved, respectively. Phylogenetic analyses of the separate nuclear and combined data sets revealed few conflicting supported nodes. All MP and ML bootstrap proportions are referred to as MPBS and MLBS, respectively, and Bayesian posterior probabilities are referred to as PP.
Nuclear analyses.—Concatenation of the APOB, DMP1, PRKC1, RAG2, STAT5A, and THY DNA sequences for each taxon resulted in a total of 4,656 aligned positions. Because of potential violation of the assumption of positional homology, 513 positions were excluded from phylogenetic analyses. Of the remaining 4,143 positions, 2,352 were variable, and 1,592 were parsimony informative. The likelihood ratio test implemented in MODELTEST indicated the GTR model of evolution with a proportion of invariant sites (I) and gamma distributed among-site rate variation (Γ) was most appropriate. For the combined nuclear ML analysis, model parameter values were set to those estimated by MODELTEST and were as follows: I = 0.1278; shape parameter α of the gamma distribution = 1.2077; base frequencies = 0.2767, 0.2384, 0.2431, 0.2418; R-matrix = 1.1001, 3.1682, 0.7205, 0.7205, 3.8490.
Maximum-parsimony analysis of the concatenated nuclear sequences resulted in 2 equally parsimonious trees of 7,126 steps (consistency index [CI] = 0.4598, retention index [RI] = 0.7092), and the ML analysis resulted in a single optimal tree (−lnL = 42,034.92810; Fig. 1). The nuclear Bayesian analysis reached stationarity at approximately 185,000 generations, so only trees following a conservative burn-in of 200,000 generations were analyzed. All parameters sampled in the Bayesian analysis converged on a stable, unimodally distributed posterior, indicating mixing was sufficient. ML and Bayesian phylogenies were identical and differed from a nuclear MP consensus phylogeny only in the placement of Miniopteridae and of the long branch leading to Eptesicus dimissus. MP analysis resulted in the miniopterids being sister to the molossids, with moderate support (70% MPBS), and ML and Bayesian phylogenies resulted in the miniopterids being sister to Vespertilionidae, with moderate support (76% MLBS, 0.97 PP). Because statistical support (albeit slight) was higher for the Miniopteridae–Vespertilionidae sister relationship, that relationship is shown in Fig. 1. ML and Bayesian analyses placed E. dimissus basal to the genus Pipistrellus, with significant statistical support (70% MLBS, 1.0 PP) and not with other species of Eptesicus. MP analysis placed E. dimissus sister to Hypsugo cadornae (90% MPBS), with that pair then being basal to bats in the tribes Nycticeiini, Pipistrellini, and Vespertilionini (sensu Hoofer and Van Den Bussche 2003) with strong support (89% MPBS).
Maximum-likelihood phylogram (—lnL = 42,034.92810) based on the GTR + I + Γ substitution model for the concatenated nuclear data set. The number of asterisks indicates support by 1 (*), 2 (**), or all 3 (***) phylogenetic inference methods. Clades are considered supported when bootstrap proportions are ≥70% or the Bayesian posterior probability is ≥0.95, or both. Outgroup taxa and representatives of non-Myotinae vespertilionids, Molossidae, and Natalidae were reduced to single branches for presentation. See Appendix I for specific taxa.
Combined nuclear and mitochondrial analyses.—The combined nuclear and mitochondrial data set resulted in 7,522 aligned positions. Because of potential violation of the assumption of positional homology, 1,339 positions were excluded from phylogenetic analyses. Of the remaining 6,183 aligned positions, 3,388 were variable, and 2,454 were parsimony informative. The likelihood ration test implemented in MODELTEST indicated the GTR + I + Γ model of nucleotide substitution most appropriately fit our data. For the ML analysis model parameter, values were: I = 0.2420; α = 0.6458; base frequencies = 0.3239, 0.2041, 0.2191, 0.2529; R-matrix = 1.4108, 4.3556, 0.8461, 1.0084, 8.2028.
Maximum parsimony analysis resulted in 10 equally parsimonious trees of 12,864 steps (CI = 0.3275, RI = 0.6360) and the ML analysis resulted in a single optimal tree (−lnL = 75,061.54406; Fig. 2). The Bayesian analysis reached stationarity at approximately 180,000 generations, so only trees following a conservative burn-in of 200,000 generations were analyzed. All parameters sampled in the Bayesian analysis converged on a stable, unimodally distributed posterior, indicating mixing was sufficient. The ML and Bayesian phylogenies of the combined nuclear and mitochondrial data were identical and differed from an MP consensus phylogeny only in the placement of Miniopteridae. Again, Bayesian and ML statistical support (76% MLBS, 0.95 PP, respectively) for the Miniopteridae–Vespertilionidae sister relationship was slightly higher than that for the Miniopteridae–Molossidae sister relationship returned in the MP analysis (70% MPBS). As a result, only the ML–Bayesian phylogeny is shown (Fig. 2).
Maximum-likelihood phylogram (—lnL = 75,061.54406) based on the GTR +I + Γ substitution model for the concatenated nuclear and mitochondrial DNA data set. The number of asterisks indicates support by 1 (*), 2 (**), or all 3 (***) phylogenetic inference methods. Clades are considered supported when bootstrap proportions are ≥70% or the Bayesian posterior probability is ≥0.95, or both. Outgroup taxa and representatives of non-Myotinae vespertilionids, Molossidae, and Natalidae were reduced to single branches for presentation. See Appendix I for specific taxa.
Molecular dating.—The likelihood ratio test significantly rejected the molecular clock for the combined data set (2ΔL = 118.647; P = 0.002). For Bayes factors, a value of logeB10 > 2 is taken to be positive support for the alternative model (in this case, the relaxed uncorrelated lognormal clock), and a logeB10 >10 indicates very strong support. Bayes factors indicated that the relaxed uncorrelated lognormal clock was favored over the relaxed uncorrelated exponential clock (logeB10 = 4.822). However, because this value was not >10, we conducted dating analyses using both the relaxed uncorrelated exponential clock and the relaxed uncorrelated lognormal clock. Divergence values for the 2 models were nearly identical, with differences growing slightly as estimates approached the root. Because of the overall concordance between these 2 analyses and the slight favoring of the lognormal model according to Bayes factors, only the results of the relaxed uncorrelated lognormal clock analysis are shown (Fig. 3).
Chronogram resulting from the relaxed uncorrelated lognormal clock molecular dating analysis of the combined data set conducted in BEAST version 1.4.8. Shaded bars represent the 95% highest posterior density interval for divergence estimates. Divergence estimates correspond to the mean node ages in units of millions of years before present. Noctilionidae, Molossidae, Phyllostomidae, Natalidae, and non-Myotinae vespertilionid lineages were collapsed for presentation. See Appendix I for specific taxa.
We estimate the age of divergence between the Old World and New World Myotis to have occurred approximately 13.4 mya (Fig. 3). The time to most recent common ancestor for the Old World clade was approximately 10.9 mya and for the New World clade was approximately 9.1 mya. Within the New World clade, neotropical and Nearctic Myotis radiated approximately 7.5 mya. M. latirostris diverged from the rest of Myotis approximately 18 mya, and Cistugo diverged from all other vespertilionids 34 mya. The time to most recent common ancestor for Vespertilionidae was 27.1 mya. Within Vespertilioninae (sensu Hoofer and Van Den Bussche 2003), divergence estimates should be approached with caution, because many of the intertribal relationships were unresolved in our analyses.
Diagnosis, description, and comparisons.—Description of this new family largely follows the description of Cistugo by Thomas (1912). Members of Cistugidae are small, pipistrelle-like bats (forearm length 32—36 mm; mass approximately 4 g) and possess 2–4 glands in the wing plagiopatagium just posterior to the forearm (Seamark and Kearney 2006:figs. 4 and 5) that are not present in any other vespertilionoid bat. These glands have been described as reduced in C. lesueuri or even absent from museum specimens (Smithers 1983), but the wing glands typically become apparent when dried specimens are wetted (Herselman and Norton 1985; Roberts 1951; Shortridge 1942). Members of Cistugidae possess 38 teeth, similar to Myotis, with 2 pairs of small and 1 pair of larger premolars on each jaw (Roberts 1951:plate 10); lower molars are myotodont (Menu and Sigé 1971). Species in Cistugidae lack distinguishing dental morphology compared with Myotis, and therefore Vespertilionidae, but given the nondescript nature of Vespertililonidae (Koopman 1994; Tate 1942), this is not surprising.
Cytogenetically, species in Cistugidae have a diploid number of 2n = 50 chromosomes and are completely distinct from all Myotis (2n = 44—Rautenbach et al. 1993) and other vespertilionoids (Natalidae, 2n = 36; Molossidae, 2n = 48; Miniopteridae, 2n = 46; Vespertilionidae, 2n = 26–58— Baker and Jordan 1970; Vollem and Heller 1994; Zima and Horacek 1985), except for Eptesicus (2n = 50—Bickham et al. 2004). Eick et al. (2005) cited the presence of multiple indels in 2 introns (SPTBN and PRKC1) that further distinguish species of Cistugidae from Vespertilionidae. We also recovered a single 18-bp indel in exon 6 of the DMP1 gene not present in any other species of Vespertilionoidea we examined.
Geographical distribution.—Currently restricted to southern Africa south of 15°S latitude. The distribution of C. lesueuri is known from scattered records from across South Africa in northern portions of Western Cape and southeastern Northern Cape, eastern Free State, and Lesotho (Herselman and Norton 1985; Lynch 1994; Watson 1998). C. seabrae is distributed from southwestern Angola through western Namibia and into the northern portion of the Namakwa District, and western portions of the Siyanda District of Northern Cape Province, South Africa (Seamark and Kearney 2006; Smithers 1983).
Based on strong support in all phylogenetic methods, the single representative species of Cistugo fell outside Vespertilionidae, which was congruent with results from past systematic studies (Bickham et al. 2004; Eick et al. 2005; Stadelmann et al. 2004). As was apparent in both the nuclear and combined nuclear and mitochondrial phylograms (Figs. 1 and 2), Cistugo was significantly divergent from Vespertilionidae. Although this significant divergence of Cistugo from Vespertilionidae has been noted previously, none of the previous studies included sufficient representation of Vespertilionoidea to evaluate if Cistugo represented a distinct family or was the most basal lineage of Vespertilionidae. Because previous studies have suggested that Cistugo may represent a distinct family, we included representatives of Miniopteridae, Molossidae, and Natalidae to evaluate distinctiveness of Cistugo. Corrected estimates of ML divergence based on our combined nuclear and mitochondrial data set (Table 1) indicated an average divergence of 22.17% between Cistugo and Vespertilionidae. Cistugo was similarly diverged from the other 2 most closely related vespertilionoid families, with average pairwise divergences of 22.82% with Miniopteridae and 20.16% with Molossidae. Good justification was obtained from our molecular results alone, or in combination with data from several other sources, for recognizing Cistugo as a family distinct from Vespertilionidae.
Maximum-likelihood corrected pairwise differences among Cistugo and families represented in Figs. 2 and 3.
Vespertilionidae (excluding Cistugo)
Vespertilionidae (excluding Cistugo)
Rautenbach et al. (1993) reported that the karyotype of Cistugo contained an all-acrocentric autosomal complement and possessed a 2n = 50. That diploid number was different from all Myotis species examined to date (Zima and Horacek 1985) but was shared with Eptesicus. As discussed by Bickham et al. (2004), that Cistugo and Eptesicus share the same diploid number can be explained by this karyotype being the ancestral condition for vespertilionids, which was proposed originally by Stock (1983). Morphologically, C. seabrae and C. lesueuri are distinct from Vespertilionidae because they possess wing glands of unknown function (Thomas 1912). Although M. vivesi also possesses glandular structures on the wings, they are relatively reduced and occur at the center of the wing; wing glands of Cistugo are pronounced and nearer the forearm. Although their taxonomic sampling of vespertilionoids was insufficient to address the distinctness of Cistugo, Eick et al. (2005) concluded that Cistugo was distinct from vespertilionids. Based on their analysis of nuclear DNA sequences, Eick et al. (2005) found that all vespertilionids they examined were characterized by a unique deletion in the SPTBN intron that was absent in Cistugo, and further, the 2 species of Cistugo had a unique insertion in the PRKC1 intron that was not present in any vespertilionids. Thus, based on the totality of evidence, including karyotypic and morphologic data, DNA sequence data from mitochondrial cytochrome-b, 12S rRNA, tRNAVal, and 16S rRNA genes, and a combination of nuclear exons and introns, Cistugo is different from Myotis, and possesses a level of genetic distinctiveness from Vespertilionidae. Moreover, the level of genetic distinctiveness is equal to, or greater than, differences between Vespertilionidae and Miniopteridae or Molossidae and differences between Phyllostomidae and Mormoopidae. Therefore, we recognize Cistugo as constituting a separate family within Vespertilionoidea and a sister taxon to Vespertilionidae with Miniopteridae being sister to this group (Figs. 1 and 2).
For Myotis, results of our combined nuclear and mitochondrial analyses were generally congruent with previous mtDNA phylogenies (Bickham et al. 2004; Hoofer and Van Den Bussche 2003; Kawai et al. 2004; Ruedi and Mayer 2001; Stadelmann et al. 2004, 2007), with strongly supported monophyletic clades corresponding to Old World (with the exception of M. latirostris, addressed below) and New World species (Fig. 2) and a collapse of the previously recognized morphological subgenera. Within the Old World clade, no statistically supported relationships conflicted between the nuclear and combined nuclear and mitochondrial data sets. The addition of nuclear data provided identical relationships to those previously reported, although some previously unresolved relationships were resolved with the additional data. The Mediterranean species M. capaccinii fell out basal to the South Pacific species M. moluccarum and M. cf. browni, with strong statistical support in all inference methods. For the European M. myotis the combined nuclear and mitochondrial data set did not resolve its position within the Old World Myotis, but nuclear analyses alone placed it basal to the 2 African species, M. bocagii and M. welwitschii, with strong statistical support. Unfortunately, we were unable to test the hypothesized subclades that Ruedi and Mayer (2001) and Stadelmann et al. (2004) suggested for the Old World because our sampling of this group was too sparse.
Within the New World clade the Nearctic and neotropical subclades previously outlined by Ruedi and Mayer (2001), Hoofer and Van Den Bussche (2003), and Stadelmann et al. (2007) were recovered (Fig. 2). Monophyly of these 2 subclades was supported only in the Bayesian analysis for the combined nuclear and mitochondrial data set but in all 3 methods of the phylogenetic analysis of the nuclear data set. Some conflict also existed between the 2 data sets concerning relationships within the subclades. Two Nearctic sister species, M. velifer and M. yumanensis, are nested within the neotropical subclade with high support in the combined nuclear and mitochondrial analyses. In the analysis of the nuclear data set those 2 species were not sister to each other, with the position of M. velifer unresolved and M. yumanensis basal to the neotropical subclade with strong support from all inference methods (100% MPBS and MLBS, 1.0 PP). This position of M. yumanensis may have some historical biogeographic implications. Stadelmann et al. (2007) suggested a Palearctic origin for ancestors of the New World Myotis during the Miocene, with an initial crossing of the Bering Strait followed by a southern radiation, eventually reaching South America. Perhaps M. yumanensis represents the closest extant relative of the Nearctic ancestor that gave rise to all neotropical Myotis, and the more slowly evolving nuclear markers still retain this phylogenetic signal.
Taxonomic conflict also existed for M. fortidens, which occurs from southern Mexico into Guatemala (Simmons 2005). Biogeographically, this would place it in the neotropical group, a relationship that was supported in the analysis of the combined nuclear and mitochondrial data set, although statistically only by the Bayesian analysis (1.0 PP). For the nuclear data set, M. fortidens was nested within the Nearctic clade, but that relationship was supported statistically only with MP (82% MPBS). This species was not included in the more taxonomically dense mtDNA phylogenetic study of New World Myotis, conducted by Stadelmann et al. (2007). Hoofer and Van Den Bussche (2003) did include it in their study and found strong support for its position nested within the neotropical subclade. Our study, and that by Hoofer and Van Den Bussche (2003), included <50% of the described Myotis in the New World and a more thorough sampling of the 2 New World subclades could provide clarification for the phylogenetic position of M. fortidens.
One notable exception to the monophyly of the Old World Myotis was the East Asian M. latirostris, which was basal to all Myotis in both data sets. Its position was supported by all phylogenetic methods (Fig. 2). This species has been included in only 1 other molecular phylogenetic study, and its results also suggest, although with weak statistical support, that M. latirostris might constitute a distinct genus (Stadelmann et al. 2007). We used corrected ML distances on the combined data set to determine the extent of divergence between M. latirostris and the remainder of Myotis. Corrected average genetic distance between M. latirostris and all other Myotis was 8.88%. The average pairwise divergence among all Myotis sampled here (excluding M. latirostris) was 6.29%. Average pairwise divergences among some other more densely sampled genera included in this study were 4.19%, 7.41%, 5.67%, 9.19%, and 4.36% for Scotophilus, Pipistrellus, Eptesicus (excluding E. dimissus), Kerivoula, and Miniopterus, respectively. Khan (2008) suggested the presence of multiple divergent lineages within Kerivoula based on cytochrome-b sequence data, potentially corresponding to multiple genera and explaining the high value obtained for that genus. Further support for recognition of M. latirostris belonging to a genus distinct from Myotis comes from comparison of divergence values between closely related genera. We found 9.18% divergence between the closely related Harpiocephalus and Murina, a value comparable to that found between M. latirostris and the remainder of Myotis (8.88%). However, a more rigorous sampling of Myotis, more specifically East Asian taxa, and comparative morphological diagnosis are necessary to make a more thorough systematic determination.
Molecular dating analyses returned divergence estimates for Myotis highly concordant with those of past studies (Stadelmann et al. 2004, 2007), and estimates outside of Myotis also were concordant with previous studies (Eick et al. 2005; Teeling et al. 2005), indicating that divergences produced here are robust. Our divergence estimates indicate that the split between Old World and New World Myotis occurred during the middle Miocene, approximately 13 mya. During much of the Oligocene and early Miocene the tropical climate of the Eocene gave way to “modern” temperate climates, with the polar regions becoming dominated by ice. This transition marked the most significant cooling event of the Cenozoic era (Zanazzi et al. 2007) and was likely responsible for significant faunal and floral shifts (Haines 1999). In addition, an abrupt drop in sea level produced the Bering land bridge, allowing a significant faunal exchange between Asia and North America (Haines 1999; Wolfe 1994). In accordance with Stadelmann et al. (2007), our evolutionary timescale also supports a middle Miocene colonization of North America from Asia via the Bering Strait and subsequent diversifications in the late Miocene. Within the New World Myotis the divergence between the Nearctic and neotropical clades occurred 10–11 mya and is highly congruent with past estimates (Stadelmann et al. 2007). This suggests that the intervening body of water separating North and South America was not a substantial barrier for dispersal and that the formation of the Isthmus of Panama 3–4 mya (Collins et al. 1996) likely had little effect on distributions or diversification of Myotis.
For M. latirostris the dating analysis suggests a divergence from all other Myotis approximately 18 mya, 5 million years prior to the Old World–New World divergence. This is much more distant from the base of the Myotis radiation than the divergence date (approximately 13 mya) suggested by Stadelmann et al. (2007). This suggests that the climatic or geologic events that led to the initial diversification of all other Myotis are not likely the same as those contributing to the divergence of M. latirostris from the Myotis lineage. The divergence estimate for Cistugo also indicates an ancient divergence relative to the base of Vespertilionidae. We estimate that Cistugo diverged from the Vespertilionidae lineage approximately 34 mya, almost 7 million years before the base of the vespertilionid radiation (approximately 27 mya). This places the divergence of Cistugo from Vespertilionidae near the Oligocene–Eocene boundary and the end of an extended tropical climate. The base of the vespertilionid radiation appears to have occurred in the late Oligocene, coinciding with extensive climatic cooling and dramatic shifts in sea level (Ogg et al. 2008; Zanazzi et al. 2007).
In conclusion, the overall concordance between the nuclear and combined nuclear and mitochondrial data sets analyzed in this study, coupled with the concordance of our results with those of previous studies, indicate that the phytogenies presented here provide an accurate and robust representation of the true evolutionary history of these taxa. More specifically, the deep divergence between Cistugo and the vespertilionids indicates that this taxon represents a distinct family (Cistugidae), sister to Vespertilionidae. The presence of clades within Myotis that correspond to biogeographical regions rather than to morphological or behavioral assemblages indicates, as originally suggested by Ruedi and Mayer (2001), that convergent evolution has been extremely common during the diversification of Myotis, and that Leuconoe, Myotis, and Selysius are not valid as subgenera. This stands as yet another example of habitat and environmental constraints dictating evolutionary trajectories, illuminating the power of natural selection in manipulating morphological and behavioral diversity during a relatively rapid radiation. Finally, the large amount of sequence divergence between M. latirostris and all other Myotis indicates that this taxon most likely represents a genus distinct from Myotis. Previous hypotheses suggesting that Myotis originated in Asia and spread to its current distribution (Findley 1972; Menu 1987) could be substantiated with the phylogenetic position of the East Asian M. latirostris. Divergence estimates support previously hypothesized scenarios for the evolution of Myotis. In addition, the divergence estimate for Cistugo from the vespertilionid lineage lends further support that it is a taxon distinct from vespertilionids, likely arising under different paleoclimatic conditions than those to which basal vespertilionids were exposed.
We extend our sincere gratitude to S. E. Weyandt and M. Leslie for assistance in the laboratory generating some of the RAG2 DNA sequences during the initial stage of this project. We also thank G. Eick for providing primer sequences and advice on polymerase chain reaction profiles for amplification of some of the nuclear sequences. For loaning tissue samples we extend our sincere gratitude to the following persons and institutions, without whose generosity and support this study would not have been possible: R. J. Baker of the Natural Sciences Research Laboratory of the Museum of Texas Tech University; N. B. Simmons of the American Museum of Natural History; B. D. Patterson, L. R. Heaney, and W. T. Stanley of the Field Museum of Natural History; S. B. McLaren of Carnegie Museum of Natural History; M. D. Engstrom and B. Lim of the Royal Ontario Museum; the late T. L. Yates of the Museum of Southwestern Biology at the University of New Mexico; T. E. Lee, Jr., of Abilene Christian University; J. O. Whitaker, Jr., and D. S. Sparks of the Indiana State University Vertebrate Collection; R. L. Honeycutt and D. Schlitter of the Texas Cooperative Wildlife Collection at Texas A&M University; and P. J. Taylor of the Durban Natural Science Museum. Thanks are also extended to D. S. Jacobs, C. Schoeman, and B. Stadelmann, who helped collect Cistugo in South Africa. We also thank the personnel of the Oklahoma State University Recombinant DNA/Protein Resource Facility. Finally, we thank A. Gardner, R. Pine, and 2 anonymous reviewers for their helpful comments that made this a better manuscript. Financial support for this project was provided by National Science Foundation grants DEB-9873657 and DEB-0610844 to RAVDB and the G. & A. Claraz Foundation. Any opinions, findings, and conclusions or recommendations expressed in this material are those of the authors and do not necessarily reflect the views of the National Science Foundation.
Specimens examined.—For catalogued specimens with museum catalog numbers, a voucher is housed in collections at American Museum of Natural History (AMNH), Carnegie Museum of Natural History (CM), Field Museum of Natural History (FMNH), Muséum d'Histoire Naturelle de Genève (MHNG), Museum of Southwestern Biology at the University of New Mexico (MSB), Museum of Texas Tech University (TTU), Oklahoma State University Collection of Vertebrates (OSU), Royal Ontario Museum (ROM), Indiana State University Collection of Vertebrates (ISUV), or the United States National Museum of Natural History (USNM). For uncatalogued specimens, only a tissue collection number is given and tissue number designations are as follows: Texas Tech University tissue (TK), Carnegie Museum Special Project tissue (SP), Oklahoma State University tissue (OK), Muséum d'Histoire Naturelle de Genève tissue (IZEA), Museum of Southwestern Biology at the University of New Mexico tissue (NK), Royal Ontario Museum tissue (F), Rodney L. Honeycutt personal collection (RLH and 05M3), Durban Natural Science Museum (DM), Abilene Christian University tissue (ACU), Manuel Ruedi personal collection (M), and Dale W. Sparks personal catalog (DWS). A dash (—) denotes information unavailable and therefore missing. GenBank accession numbers for previously published sequences are indicated in boldface type; all others were generated in this study.
Tissue collection no.
Museum catalog no.
Paracou, French Guiana
Berbice District, French Guiana
Paracou, French Guiana
Cayenne, French Guiana
South Buganda, Uganda
Transvaal Province, South Africa
St. John, Dominica
Central Province, Papua New Guinea
Rift Valley Province, Kenya
Western Province, Kenya
Central Province, Papua New Guinea
Valais Province, Switzerland
Goodhouse, South Africa
Phongsali Province, Lao PDR
Paracou, French Guiana
Rift Valley Province, Kenya
Uthai Thani Province, Thailand
Phongsali Province, Lao PDR
KwaZulu-Natal Province, South Africa
Dong Nai, Vietnam
Kerivoula lenis (analyzed previously as K. papulosa)
Dong Nai, Vietnam
East Kalimantan, Indonesia
Phongsali Province, Lao PDR
Dak Lak, Vietnam
Phongsali Province, Lao PDR
Tanga Region, Tanzania
Myotis cf. browni (analyzed previously as M. muricola)
Mindanao Island, Philippines
Northern Province, Jordan
St. Joseph Parish, Dominica
Moi-Li County, Taiwan
Sao Paulo, Brazil
Myotis moluccarum (analyzed previously as M. adversus)
Bern Province, Switzerland
Paracou, French Guiana
Kasese District, Uganda
Eastern Province, Kenya
Maan Government, Jordan
Malakand Division, Pakistan
Mindanao Island, Philippines
Pipistrellus pygmaeus (analyzed previously as P. pipistrellus)
Primer sequences, range of annealing temperatures, and reference for original primer sequences used to amplify and sequence the 6 nuclear loci examined in this study. APOB, DMP1, and RAG2 are nuclear exons, whereas PRKC1, STAT5A, and THY are nuclear introns. Ex and In refer to external and internal primers, respectively. F and R refer to forward and reverse primers, respectively.
1942. Field notes on the first and second expeditions of the Cape Museums' mammal survey of the Cape Province; and descriptions of some new subgenera and subspecies. Annals of the South African Museum 36:27–100.
1942. Results of the Archbold expeditions. No. 47: review of the Vespertilioninae bats, with special attention to genera and species of the Archbold collections. Bulletin of the American Museum of Natural History 80:221–296.