Molecular Phylogenetics and Evolution 37 (2005) 91–103 www.elsevier.com/locate/ympev

Molecular phylogeny of Funnel-eared bats (Chiroptera: Natalidae), with notes on biogeography and conservation Liliana M. Dávalos ¤ Department of Ecology, Evolution and Environmental Biology, Columbia University, USA Division of Vertebrate Zoology, American Museum of Natural History, Central Park West at 79th Street New York, NY 10024-5192, USA Received 18 October 2004; revised 21 April 2005 Available online 20 June 2005

Abstract Two assumptions have framed previous systematic and biogeographic studies of the family Natalidae: that it comprises a few widespread species, and that extant lineages originated in Mexico and/or Central America. This study analyzes new sequence data from the mitochondrial cytochrome b and the nuclear Rag2, to clarify species boundaries and infer relationships among extant taxa. Fixed diVerences in cytochrome b coincide with published morphological characters, and show that the family includes at least eight species. One newly recognized species is known to live from a single locality in Jamaica, suggesting immediate conservation measures and underscoring the urgency of taxonomic revision. Among the three genera, Chilonatalus and Natalus form a clade, to the exclusion of Nyctiellus. This phylogeny and the geographic distribution of natalids, both extant and extinct, are hardly compatible with a Middle American origin for the group. Instead, extant natalids appear to have originated in the West Indies. The threat of Caribbean hurricanes early in their evolutionary history might account for the specialized cave roosting that characterizes all natalids, even continental species.  2005 Elsevier Inc. All rights reserved. Keywords: Cytochrome b; Rag2; Nyctiellus; Chilonatalus; Natalus; Caribbean; Antilles; West Indies; Neotropics; Species limits; Biogeography; Event-based biogeography

1. Introduction The Natalidae is a small family of neotropical bats characterized by funnel-like ears and a tail about equal in length to the head and body (Emmons, 1997). Phylogenetic relationships between natalids and other bat families are relatively well established: natalids are sister to the worldwide families Molossidae and Vespertilionidae (Hoofer et al., 2003; Teeling et al., 2003; Van Den Bussche and Hoofer, 2001; Van Den Bussche et al., 2003). In contrast, the number of extant and fossil species assigned to this family has been disputed for more than 50 years (Dal* Present address: Department of Biochemistry and Molecular Biophysics, 208 Life Sciences South, University of Arizona, Tucson, AZ 85721, USA. Fax: +1 520 621 3709. E-mail address: [email protected].

1055-7903/$ - see front matter  2005 Elsevier Inc. All rights reserved. doi:10.1016/j.ympev.2005.04.024

quest, 1950; Koopman, 1994; Morgan, 2001; Morgan and Czaplewski, 2003; Silva-Taboada, 1979; Varona, 1974). This taxonomic confusion has been an obstacle to studies in ecology, conservation, and biogeography. Two interrelated topics dominate discussion on the ecology of natalids: the adaptation to roosting in cave systems with high humidity (Silva-Taboada, 1979), and the evidence for local extinction throughout the Caribbean (Morgan, 2001). The natalids appear to tolerate variations in roost temperature (Silva-Taboada, 1979; Timm and Genoways, 2003), but require stable humidity, a condition available only in caves (Goodwin, 1970; Morgan, 2001; Tejedor et al., 2004). The extirpation of Natalus and Chilonatalus from some Bahamian islands, Middle Caicos, and Grand Cayman, and the near-extirpation of Natalus from Cuba, have been linked to environmental change leading to the loss of appropriate

92

L.M. Dávalos / Molecular Phylogenetics and Evolution 37 (2005) 91–103

microenvironments in caves (Morgan, 2001; Tejedor et al., 2004). Because estimates of the relative abundance of natalids depend on the species boundaries adopted, taxonomic confusion hampers conservation eVorts. Dispersal from the continent has been the explanation for the distribution of natalids (Baker and Genoways, 1978; Koopman, 1989; Morgan and Czaplewski, 2003), as shown in Fig. 1. The single point of contention in natalid biogeography is whether Natalus from the Greater Antilles (Table 1) originated in South America (Koopman, 1989), or in Mexico and/or Central America, Middle America (Baker and Genoways, 1978; Morgan and Czaplewski, 2003). The biogeographic account of Morgan and Czaplewski (2003) has been the only one informed by phylogenetic analyses. Because Morgan and Czaplewski (2003) studied only morphological characters within a subset of extant natalid populations, additional taxon, and character sampling are necessary to increase the geographic scope of the phylogeny and test their conclusions. In this study, the complete mitochondrial cytochrome b (1.14 kb) from Central American and Caribbean individuals was sequenced to assess the distinctiveness of allopatric populations and help resolve decades of dispute on the number of species in the genus Natalus (Dalquest, 1950; Koopman, 1994; Morgan, 2001; Varona, 1974). A 1.36 kb fragment of the nuclear recombination activating gene 2 (Rag2) was sequenced from populations deemed to be distinct based on the mitochondrial gene. These new mitochondrial and nuclear DNA sequences

were analyzed separately, and combined with published morphological data to estimate relationships among extant natalids. The resulting phylogenies were used to infer the geographic origin of Caribbean natalids. Table 1 Geographic distribution of all natalid species and subspecies recognized by Simmons (in press) and taxonomic sampling of this study Taxon

Geographic range

Sampled

Nyctiellus lepidus Chilonatalus micropus micropus Chilonatalus micropus macer Chilonatalus micropus brevimanus Chilonatalus tumidifrons Natalus jamaicensis Natalus major Natalus primus Natalus stramineus stramineus Natalus stramineus mexicanus

Cuba, Bahamas Jamaica, Hispaniola

Yes Yes

Cuba Providencia

No No

Bahamas Jamaica Hispaniola Cuba Lesser Antilles Baja California, Sonora (Mexico) Mexico, Central America NW Venezuela central Brazil to Espírito Santo NE Brazil Curaçao, Bonaire

Yes Yes Yes No Yes No

Natalus stramineus saturatus Natalus stramineus tronchonii Natalus stramineus espiritosantensis Natalus stramineus natalensis Natalus tumidirostris tumidirostris Natalus tumidirostris continentis Natalus tumidirostris haymani

NE Colombia, NW Venezuela Trinidad

Yes No No No No No Yes

Fig. 1. Map of the Caribbean depicting previously hypothesized natalid dispersal routes. Middle America to the Greater Antilles: natalids in the Oligocene or Miocene (Morgan and Czaplewski, 2003); Nyctiellus, Chilonatalus, or their common ancestor (Koopman, 1989); Natalus major, N. jamaicensis and N. primus (Baker and Genoways, 1978). South America through the Lesser Antilles: all Caribbean Natalus (Koopman, 1989), or Natalus stramineus (Baker and Genoways, 1978). Central America to South America: Natalus in the Pliocene (Morgan and Czaplewski, 2003).

L.M. Dávalos / Molecular Phylogenetics and Evolution 37 (2005) 91–103

2. Materials and methods 2.1. Taxon sampling To examine relationships among natalids all currently recognized species were included (Table 1), with the exception of the rediscovered Natalus primus (Simmons, in press; Tejedor et al., 2004). At least two individuals per species, from as many localities as available, were sequenced to capture the genetic variation of each taxon. All ingroup cytochrome b sequences were generated for this study (Table 2), while two recombination activating gene 2 (Rag2) sequences (GenBank Accession Nos. AY141024, AY141023) were obtained from a previously published study (Hoofer et al., 2003). For outgroup comparison and to root the trees, sequences from the vespertilionids Myotis velifer (GenBank Accession Nos. AF376870, AY141033) and Myotis riparius (GenBank Accession Nos. AF376866, AY141032), and the molossid Tadarida brasiliensis (GenBank Accession Nos. L19734, AY141019) were included in phylogenetic analyses. 2.2. Molecular data For all specimens, DNA was isolated from wing clip, liver or muscle tissue that had been preserved in

93

ethanol or lysis buVer in the Weld. DNA was extracted using a Qiagen DNeasy Tissue Extraction Kit (Qiagen) following the manufacturer’s protocol. The complete mitochondrial cytochrome b (1.14 kb) was ampliWed and sequenced from extracted DNA as described elsewhere (Dávalos and Jansa, 2004; Jansa et al., 1999). A 1.36 kb fragment of Rag2 was ampliWed and sequenced with primers described by Baker et al. (2000) with slight modiWcations as shown in Table 3. AmpliWcation products were sequenced with internal primers and the same primers used for PCR ampliWcation. Sequencing reactions were puriWed through a MgCl2-ethanol precipitation protocol and run on an ABI 3100 automated sequencer. Sequences were edited and compiled using Sequencher 4.1 software (GeneTable 3 Primers used in ampliWcation and sequencing of natalid Rag2 Name

5⬘-primer sequence-3⬘

NAT-RAG2F2 NAT-RAG2F2-I NAT-RAG2F1-I NAT-RAG2R1 NAT-RAG2R1-I NAT-RAG2R2 NAT-RAG2R2-I

5⬘-TTTGTTATTGTGGGTGGCTATCAG-3⬘ 5⬘-GGATTCCACTCCSTTTGAAGA-3⬘ 5⬘-ATACAGTCGAGGAAAGAGCATGG-3⬘ 5⬘-AGCCTGTTTATTGTCTCCTGGTATGC-3⬘ 5⬘-GMGGCAGGCAGTCAGCTAC-3⬘ 5⬘-GGAAGGATTTCTTGGCAGGAGT-3⬘ 5⬘-ACAGCATGTAATCCAGTAGC-3⬘

Table 2 Taxa and locality data of individuals studied Taxon

Locality

Tissue voucher

Cadaver voucher

Nyctiellus lepidus Nyctiellus lepidus Chilonatalus micropus Chilonatalus micropus Chilonatalus micropus Chilonatalus tumidifrons Chilonatalus tumidifrons N. stramineus stramineus N. stramineus stramineus

Crawling Cave, Industrious Hill, Cat Island, Bahamas Crown Cave, Dumfries, Cat Island, Bahamas St. Clair Cave, Polly Ground, St. Catherine, Jamaica St. Clair Cave, Polly Ground, St. Catherine, Jamaica St. Clair Cave, Polly Ground, St. Catherine, Jamaica Reckly Maze Cave, San Salvador (Watling Island), Bahamas Reckly Maze Cave, San Salvador (Watling Island), Bahamas 0.5 mi N Toucari, St. John Parish, Dominica Morne Ducos, 1.5 km NE jct rt D203 Grand-Bourg along Rt N9 50 m, Marie Galante, Guadeloupe 1.1 km S Calihaut (by road) small shallow cave in slope below HWY, St. Peter Parish, Dominica Laguna Noh-Bec, 2 km W of Noh-Bec, Quintana Roo, Mexico Colola (5 km N) Municipio Aquila, Michoacán, Mexico Rio Uyus, 5 km E San Cristóbal Acasaguastlán, El Progreso, Guatemala Rivas, Nicaragua Cave Las Cuevas Research Station, Belmopan, Belize St. Clair Cave, Polly Ground, St. Catherine, Jamaica St. Clair Cave, Polly Ground, St. Catherine, Jamaica La Entrada (de Cabrera), María Trinidad Sánchez, Dominican Republic Don Miguel (Platanal de Cotui), Sánchez Ramírez, Dominican Republic Tamana Cave, St. Andrew, Trinidad Tamana Cave, St. Andrew, Trinidad

AMCC 119283 AMCC 119271 AMCC 102717 AMCC 102718 TK9456 AMCC 121973 AMCC 121978 TK15661 SP10036

AMNH 275537 AMNH 275535 AMNH 274630 AMNH 274631 CM 44580

SP9393

CM 112376

FN30994 TK43151 F34011

ROM 97519 UAMI ROM 99652

F48058 T48 TK9421 TK9424 AMCC 103028

ROM 112172 NHM 2003.201 TTU 29110 TTU 29113 AMNH 275480

AMCC 103056

AMNH 275513

AMCC 119246 AMCC 119247

AMNH 275517 AMNH 275518

N. stramineus stramineus N. stramineus saturatus N. stramineus saturatus N. stramineus saturatus N. stramineus saturatus N. stramineus saturatus N. jamaicensis N. jamaicensis N. major N. major N. tumidirostris haymani N. tumidirostris haymani

TTU 31458 CM 112377

Taxonomy follows Simmons (in press). When only tissue vouchers are listed, DNA was extracted from wing punctures. N., Natalus. AMCC, Ambrose Monell Cryogenic Collection of the American Museum of Natural History; AMNH, cadaver voucher at the Mammalogy Department of the American Museum of Natural History; CM, Carnegie Museum of Natural History; NHM, Natural History Museum (London); SP, tissue collection of the Carnegie Museum of Natural History; T, tissue collection of the Natural History Museum (London); TK, tissue collection of the Museum of Texas Tech University; UAMI, Universidad Autónoma Metropolitana Iztapalapa.

94

L.M. Dávalos / Molecular Phylogenetics and Evolution 37 (2005) 91–103

Codes). Base-calling ambiguities between strands were resolved either by choosing the call on the cleanest strand or using the appropriate IUB ambiguity code if both strands showed the same ambiguity. Molecular sequences generated as part of this study have been deposited in GenBank under Accession Nos. AY62 1006–AY621028 (cytochrome b) and AY6044 63– AY604468 (Rag2). 2.3. Morphological data The morphological character matrix of Morgan and Czaplewski (2003) was appended to the molecular data to generate combined analyses of all characters available for the group. Tadarida brasiliensis, Natalus major, and N. stramineus stramineus were coded as all missing in the morphological partition. The morphological study did not include representatives of these taxa as they are deWned in Table 1. 2.4. Data analysis Protein-coding cytochrome b and Rag2 sequences were aligned by eye using Sequencher 4.1 (GeneCodes). Appropriate coding of cytochrome b and Rag2 sequences was veriWed by translation into amino acids using the mammalian mitochondrial code or the universal code, as appropriate, implemented in MacClade v. 4.06 (Maddison and Maddison, 2003). To describe the variation in cytochrome b among taxa, uncorrected pairwise distances were calculated using PAUP* v. 4.0b10 (SwoVord, 2002). Cytochrome b sequences were also examined for Wxed character diVerences among named Natalus taxa (Table 1). Analyses were conducted on four data sets: (1) the mitochondrial cytochrome b sequences; (2) the nuclear Rag2 sequences; (3) the concatenated cytochrome b and Rag2 data set or molecular data set; and (4) the combined molecular and morphological data set or total evidence data set. Parsimony analyses were performed for separate data partitions and on combined matrices using branch and bound searches as implemented in PAUP*. For each search, phylogenetically informative characters were treated as unordered and equally weighted. Clade stability was assessed using non-parametric jackknife and the Bremer support index (Bremer, 1994). Parsimony jackknife analyses included 1000 replicates; searches were heuristic with 100 replicates of random taxon addition followed by TBR branch swapping. Bremer values were calculated with the aid of AutoDecay v. 4.0.2 (Eriksson, 1999). The Templeton (1983) test implemented in PAUP* was used to assess the signiWcance in the diVerences in length between topologies obtained from diVerent data sets. Best-Wt maximum likelihood models for molecular data were selected using the Akaike Information Criterion

(AIC) as implemented in Modeltest v. 3.06 (Posada and Crandall, 1998). Maximum likelihood phylogenetic analyses of the diVerent molecular data sets were performed using PAUP*. To provide the most conservative test for a clock-like (rate-constant) model of evolution, a UPGMA tree based on Jukes-Cantor distances was calculated, and the likelihood score for the best-Wt model (as determined using Modeltest) with no clock enforced (log L1) vs. the same model with a clock enforced (log L2) were compared. The signiWcance of the diVerence in likelihood scores was determined by comparing ¡2 log
against a 2 distribution (df D Ntaxa ¡ 2). If the value for ¡2 log
was signiWcant, then a molecular clock could be rejected. The non-parametric Shimodaira and Hasegawa (1999) test with 10,000 RELL pseudoreplicates was used to examine the signiWcance of diVerences in topology given a maximum likelihood model. A parametric bootstrap was also used to compare alternative trees. The optimal model of sequence evolution for the molecular data set was estimated using Modeltest. Model parameters were then used to simulate 100 data sets using Seq-Gen v. 1.2.7 (Rambaut and Grassly, 1997) based on the optimal topology found for the concatenated molecular data. PAUP* was then used to optimize trees for each of the 100 simulated data sets with and without topological constraints. These constraints corresponded to alternative resolutions of the weakest node in the optimal topology. The diVerence between maximum likelihood scores for constrained and unconstrained trees using the concatenated molecular data was then compared to the distribution of diVerences based on simulations. Bayesian methods were used to estimate a phylogeny applying diVerent models of molecular evolution for each partition of the molecular data. This analysis featured two partitions, mitochondrial and nuclear DNA, and the model of sequence evolution was determined using Modeltest (see above). The values for model parameters were allowed to vary between partitions and were not speciWed a priori, but treated as unknown variables to be estimated in each analysis. Bayesian analysis was conducted using MrBayes v. 3.0b4 (Huelsenbeck and Ronquist, 2001), with random starting trees without constraints, four simultaneous Markov chains were run for 2,000,000 generations, trees were sampled every 100 generations, and temperature was set to 0.20. Resulting burn in values, the point at which the model parameters and tree score reach stationarity, were determined empirically by evaluating tree likelihood scores. Analyses were repeated in 4 separate runs of MrBayes to ensure that trees converged on the same topology and parameters. 2.5. Biogeographic analyses Two biogeographic methods were used to estimate the geographic distribution of hypothetical ancestors: dispersal-vicariance analysis DIVA v. 1.1 (Ronquist, 1997), and ancestral area analysis (Bremer, 1992). Dispersal-vicari-

L.M. Dávalos / Molecular Phylogenetics and Evolution 37 (2005) 91–103

ance analysis does not require an independent hypothesis of area relationships, but instead reconstructs the ancestral distribution at each of the internal nodes of a given phylogeny. This is accomplished by using optimization rules and set costs of one for extinction and one for dispersal. Vicariant and sympatric speciation carry no cost. Species distributions are therefore explained by assigning costs for each event in a way that biogeographic explanations imply the least possible cost. The estimation can be constrained to contain any minimum number of areas. Ancestral area analysis (Bremer, 1992) allows the identiWcation of the ancestral area of a clade, given geographic distribution information on the branches of a phylogeny. Each area can be considered a binary character with two states (present or absent) and optimized on the phylogeny. By comparing the numbers of gains and losses, it is possible to estimate areas most likely to have been part of the ancestral areas. Both methods were applied to the diVerent topologies obtained from character analyses. Ancestral area analysis was only used for the ancestral node of the extant Natalidae. Morgan and Czaplewski (2003) suggested that natalids reached the Caribbean early in their evolutionary history (Fig. 1). This prediction regarding the relative age of diver-

Fig. 2. Scatter plot of uncorrected sequence divergence in cytochrome b against taxonomic rank. Taxonomy follows Simmons (in press). Numerals indicate cytochrome b distance for outliers: 1: between Nyctiellus and Myotis riparius; 2: Natalus vs. Chilonatalus; 3: Natalus and Chilonatalus vs. Nyctiellus; 4: Natalus tumidirostris vs. N. stramineus stramineus; 5: between two species of Chilonatalus; 6: between two species of Myotis; 7: among Natalus saturatus.

95

gence between natalids and their outgroup was tested by generating conWdence intervals around the branch lengths of rate-constant phylogenies using a parametric bootstrap. Seq-Gen was used to simulate 100 data sets based on the optimal maximum likelihood parameters and topologies for each molecular partition (see above). PAUP* was then used to optimize trees and rate-constant branch lengths for each of the 100 simulated data sets under optimal maximum likelihood parameters. The resulting branch lengths were then tabulated and used to calculate the 95% conWdence interval around the nodes of interest.

3. Results 3.1. Sequence variation and saturation analysis 3.1.1. Cytochrome b Complete cytochrome b sequences were obtained for most taxa, with the exception of the 402 base pairs available for Tadarida brasiliensis (GenBank Accession No. L19734). A summary of the uncorrected pairwise divergences among individuals in diVerent taxonomic ranks is shown in Fig. 2. Saturation in substitutions was detected only for third codon positions in comparisons between each ingroup taxon and Myotis and Tadarida. Sequence examination for Wxed characters among sampled species and subspecies of Natalus (Table 2) uncovered Wxed character diVerences among all named taxa (Table 4). Cytochrome b sequences of Natalus major from two distant localities were identical (Table 2). Within cytochrome b, 416 (36%) of sites were variable and 349 (31%) were parsimony informative. The distribution of the parsimony-informative sites was dependent on codon position: 16.8% in Wrst, 3.4% in second, and 63.7% in third. The average base composition of sequences was skewed, with a strong bias toward adenine (45.0%) and cytosine (32.2%) in third position. Taxa diVered signiWcantly in base composition at this site (P D 0.000). Heterogeneity is presumed to mislead phylogenetic analyses because unrelated taxa with similar base composition might appear as related when using

Table 4 Character diVerences in cytochrome b, and pairwise absolute number of diVerences in Rag2 among taxa in the genus Natalus Named taxon

N1

Compared to

N2

Fixed character diVerences

DiVerences in Rag2 (N D 1)

N. jamaicensis N. major N. stramineus stramineus N. stramineus saturatus N. stramineus saturatus N. stramineus saturatus N. tumidirostris haymani

2 2 3 5 5 5 2

Natalus major Natalus jamaicensis All other Natalus All other Natalus Natalus Greater Antilles N. stramineus stramineus N. stramineus stramineus

2 2 10 8 4 3 3

17 15 15 4 7 21 4

1 1 3–6 5–8 7–8 5 3

Taxonomy follows Simmons (in press). See Table 1 for geographic distribution, and Table 2 for localities. N., Natalus.

96

L.M. Dávalos / Molecular Phylogenetics and Evolution 37 (2005) 91–103

methods that do not account for this bias (Lockhart et al., 1994). The most divergent taxon in GC content for the third codon bases of cytochrome b sequences (GC content average for all taxa D 34.9%, SD D 4.4) was Nyctiellus (22.5%). Without Nyctiellus the diVerences in base composition at third codon positions in the data set became non-signiWcant (P D 0.298). Extreme bias in composition could not distort phylogenetic analyses because no other taxon exhibits similar variation. 3.1.2. Rag2 A Rag2 fragment (»700 bp) of Chilonatalus micropus AMCC 102718 (Table 2) was identical to GenBank Accession No. AY141023 (C. micropus TK9454 from Jamaica). An equivalent Rag2 sequence from Natalus stramineus stramineus TK15661 (Table 2) was identical to GenBank Accession No. AY141024 (N. stramineus TK15660 from Dominica). Details on both published sequences were provided by Van Den Bussche et al. (2003). The distribution of the parsimony-informative sites was highly dependent on codon position: 10.7% in Wrst, 14.3% in second, and 75.0% in third. The average base composition of sequences was skewed, but these biases did not diVer signiWcantly across taxa for the whole gene, or for diVerent codon positions (P > 0.979).

3.2. Phylogenetic analyses Molecular data from diVerent genes were concatenated from the same individual for combined analyses, with the exception of Rag2 sequences corresponding to GenBank Accession Nos. AY141023 and AY141024. Parsimony analyses were conducted with all unordered and unweighted characters (Figs. 3 and 4). Maximum likelihood analyses were performed Wrst using the program Modeltest (see above). The models and parameters applied to the diVerent data sets are shown in Table 5. Since the parameters for the two genes were considerably diVerent (Table 5), Bayesian methods were used to obtain an estimate of phylogeny that accounted for two models of sequence evolution while using all available molecular data. Stationarity in parameter estimation was reached after 200,000 generations (burn in D 2000 trees). The resulting trees are summarized in Fig. 4. There is strong support for the monophyly of the family Natalidae in all data partitions and types of analyses (Figs. 3 and 4), in agreement with phylogenies based on morphological characters or other mitochondrial and nuclear genes (Hoofer et al., 2003; Morgan and Czaplewski, 2003; Van Den Bussche and Hoofer, 2001; Van Den Bussche et al., 2002; Van Den Bussche et al., 2003). The molecular data corroborate the intergeneric relationships

Fig. 3. (A) Strict consensus of two equally parsimonious trees resulting from analysis of cytochrome b (L D 845 steps, CI D 0.634, RI D 0.775). Numbers above branches are Bremer support indices; below branches are percent of 1000 50% jackknife pseudoreplicates. C. Chilonatalu, N.s. Natalus stramineus. (B) Phylogram resulting from maximum likelihood analysis using a rate-constant GTR + I + model of DNA evolution (¡ln L D 5086.06). Numbers above or below branches are percent of 300 50% non rate-constant jackknife pseudoreplicates. Thicker branches have 100% jackknife support.

L.M. Dávalos / Molecular Phylogenetics and Evolution 37 (2005) 91–103

Fig. 4. Most parsimonious cladogram resulting from analysis of concatenated cytochrome b and Rag2 sequences (L D 967, CI D 0.702, RI D 0.615). This topology is identical to the majority-rule consensus of 18000 phylograms (¡ln L D 7635; 95% conWdence interval 7626– 7643) obtained by Bayesian analyses of concatenated cytochrome b and Rag2 sequences using separate models of evolution for each gene. The resolution of the ingroup in this tree is identical to that obtained from parsimony analysis of concatenated molecular sequences and morphological characters (total evidence, L D 939, CI D 0.741, RI D 0.555). This tree is compatible, but more resolved than, the strict consensus of six equally parsimonious trees resulting from parsimony analysis of Rag2 (L D 163 steps, CI D 0.951, RI D 0.949) and the strict consensus of nine trees resulting from maximum likelihood analysis of Rag2 (¡ln L D 2879.07). This tree is not compatible with the single topology resulting from the maximum likelihood analysis of concatenated molecular sequences (¡ln L D 7874.22), where N.s. saturatus and N.s. stramineus–N. tumidirostris are sister taxa. Support values are shown for total evidence in the Wrst column, concatenated molecular sequences in the second column, and Rag2 in the third column. The top row shows Bremer support indices; second row is the percent of 1000 50% jackknife pseudoreplicates using parsimony; third row is the percent of 300 50% jackknife pseudoreplicates using non rate-constant maximum likelihood; and the last row shows Bayesian posterior probability. Models of sequence evolution used to analyze each partition are shown in Table 5. Asterisks indicate jackknife support values of 100%, or Bayesian posterior probabilities of 1.0. Dashes indicate the analysis was not performed for the partition.

among extant natalids (Fig. 4) obtained by Morgan and Czaplewski (2003). Chilonatalus and Natalus are sister taxa (Figs. 3 and 4). The Templeton (1983) test using all molecular data rejected an Antillean clade formed by Nyctiellus and Chilonatalus (P D 0.006). The Shimodaira

97

and Hasegawa (1999) test did not reject this alternative (P D 0.0618 with a molecular clock, P > 0.5277 without a molecular clock), but a subsequent test using only the cytochrome b sequence data did (P < 0.0001 with a molecular clock, P > 0.0001 without a molecular clock). The topology test using a parametric bootstrap of all molecular data was not signiWcant (P > 0.50), but a similar test using only the mitochondrial marker was (P < 0.01). In summary, the support for resolving intergeneric relationships derives from the mitochondrial partition. Because the resolution of Fig. 4 agrees with the morphological data, the nuclear sequences do not reject it (the Rag2 phylogeny is unresolved at this node), and it is signiWcantly better at explaining the mitochondrial data using all methods, this is the only tree discussed in subsequent biogeographic analyses. Parsimony analysis of the fast-evolving mitochondrial cytochrome b gene did not recover a sister relationship between the Chilonatalus micropus and C. tumidifrons (Fig. 3A), probably because these taxa diverged early in the history of the genus (see Fig. 3B). This is remarkable in light of the minimal external diVerentiation that prompted Hall and Kelson (1959) and Varona (1974) to suggest that Chilonatalus comprised a single species. The monophyly of Chilonatalus is moderately supported in analyses of Rag2 and combined molecular data (Fig. 4), as it is in the study of Morgan and Czaplewski (2003). Nyctiellus and Natalus are monophyletic (Fig. 3). Each species of Natalus is monophyletic, with the exception of N. stramineus (Fig. 3). There is moderate support for two pairs of sister species in this genus: stramineus stramineus and tumidirostris, and major and jamaicensis (Fig. 3). The relationship between these two clades and with respect to stramineus saturatus is not resolved (Fig. 3, but see Fig. 4). Three alternatives are viable given the molecular data: stramineus saturatus (Fig. 4), stramineus stramineus and tumidirostris (Fig. 3A), or major and jamaicensis (Fig. 3B) is sister to all other Natalus. Both the Templeton (1983) (P > 0.16) and Shimodaira and Hasegawa (1999) non-parametric tests (P > 0.26 with a molecular clock, P > 0.13 without a

Table 5 Models of molecular evolution and parameters selected for each molecular data set, see Table 2 for sequences Data

Model

R-matrix



I

¡2 log


df

P

cyt b ML cyt b Bayes

GTR + I +
GTR + I +


1.6304 1.746 (0.6–3.2)

0.5076 0.539 (0.5–0.6)

24.43 —

21 —

>0.05 —

Rag2 ML Rag2 Bayes

GTR + I GTR + I

— —

0.5396 0.481 (0.2–0.7)

7.89 —

9 —

>0.05 —

combined ML

GTR + I +


1.5, 5.0, 1.6, 0.0, 17.8 2.6 (0.6-5.0), 8.7 (3.1–16.6), 2.6 (0.7–5.5), 0.1 (0.0–0.5), 41.1 (14.4–78.2) 1.0, 4.9, 1.0, 1.0, 6.1 4.2 (0.8–10.0), 13.2 (3.1–30.7), 2.3 (0.4–5.7), 3.0 (0.4–7.4), 16.3 (3.5–37.4) 4.3, 8.1, 2.9, 0.7, 42.0

0.5714

0.5565

9.15

9

>0.05

ML, parameters used in maximum likelihood analyses; Bayes, parameters used in Bayesian analysis of concatenated data; GTR, general time reversible model; R-matrix, rate matrix parameter (with respect to G–T transversion); , shape parameter, I, proportion of invariant sites; ¡2 log
, 2 [log L1–log L2], where L1 is the likelihood without clock and L2 is the likelihood with clock. Bayesian parameters are followed by the 95% conWdence interval (in parentheses).

98

L.M. Dávalos / Molecular Phylogenetics and Evolution 37 (2005) 91–103

molecular clock) failed to reject the null hypothesis that the molecular data Wts the three topologies equally well. The topology test using a parametric bootstrap also had non-signiWcant results (P > 0.48). The three hypotheses of relationships among clades in Natalus can explain available molecular data. Two previous phylogenies of Natalus included three species each: jamaicensis (called major by the authors, but only Jamaican specimens were included), stramineus saturatus (called stramineus, but only specimens from Mexico and Belize were included) and tumidirostris (Morgan and Czaplewski, 2003), or stramineus stramineus (Arroyo-Cabrales et al., 1997). Morgan and Czaplewski (2003) found stramineus saturatus and tumidirostris formed a clade, to the exclusion of jamaicensis. This is consistent with the tree of Fig. 3B, and with the results of Arroyo-Cabrales et al. (1997) where saturatus and stramineus were sister taxa, excluding jamaicensis. Although the best estimate of relationships among the three main Natalus lineages is that of Fig. 4, the morphology- and allozyme-based topologies of Natalus remain viable. Published phylogenies have not tested the monophyly of Greater Antillean Natalus. The sequence data suggest Greater Antillean Natalus form a clade (Figs. 3 and 4).

3.3. Biogeographic analyses The results of dispersal-vicariance optimizations for three nodes of interest are shown in Fig. 5. Following Ronquist (1996) optimizations were constrained to two areas of endemism per node, and Primonatalus prattae was included to analyze early divergences. The phylogenetic position of this fossil taxon was based on morphological data only (Morgan and Czaplewski, 2003). Nyctiellus lepidus and Chilonatalus micropus were assumed to be a widespread species. Estimated ancestral distributions shown did not change if only C. m. micropus was analyzed. Unsampled extant populations of Natalus (i.e., South American Natalus stramineus, N. primus) were not used when estimating ancestral areas. This accounts for the exclusion of Cuba and South America from the estimated range of the ancestor of Natalus in Fig. 5(3). Ancestral area analyses optimized the most recent common ancestor of extant natalids to Cuba and/or the Bahamas, regardless the internal resolution of the genus Natalus. ConWdence intervals around the estimates of sequence divergence for relevant nodes are presented in Fig. 6.

Fig. 5. Estimated geographic distribution of ancestral ancestral nodes using dispersal-vicariance analysis (Ronquist, 1997). Optimizations were constrained to a maximum of two areas, but all solutions are shown. The three alternatives to the polytomy of Natalus were analyzed, and all result in the same estimates. The number at the bottom right of each panel corresponds to the node in the cladogram of natalids. Asterisks indicate localities where fossil natalids have been found: (1) early Miocene Primonatalus prattae (Morgan and Czaplewski, 2003); (2) late Quaternary Nyctiellus and Chilonatalus (Morgan, 1999); and (3) sub Recent Natalus (Morgan, 1999; Tejedor et al., 2004). Dots indicate other bat fossil localities of comparable age where natalids have not been found: (1) middle Miocene fauna of La Venta, Colombia (Czaplewski et al., 2003a); and (2), from north to south, Quaternary Gruta de Lolt, Mexico (Arroyo-Cabrales and Ray, 1997), late Quaternary fauna of Cebada Cave, Belize (Czaplewski et al., 2003b), and Quaternary Cueva del Gu charo, Venezuela.

L.M. Dávalos / Molecular Phylogenetics and Evolution 37 (2005) 91–103

99

Fig. 6. ConWdence intervals around observed sequence divergence obtained by parametric bootstrapping of rate-constant natalid phylogenies. (A) Estimates of divergence for mitochondrial cytochrome b gene, see Fig. 3B for base topology. (B) Estimates of divergence for nuclear Rag2. The base topology is compatible with that shown in Fig. 4, with collapsed branches where Rag2 Bremer support indices equal zero.

4. Discussion 4.1. How many species of Natalus are there? Taxonomy and conservation As many as six (Dalquest, 1950) and as few as one (Linares, 1971) species of Natalus have been recognized over the last few decades. In one of the more conservative assessments of species diversity within the genus, Koopman (1993) proposed a deWnition of Natalus stramineus that made this taxon widespread from Baja California to Paraguay and throughout the West Indies. Although Morgan (2001, 1989) has argued for separating Greater Antillean Natalus from other N. stramineus

sensu Koopman (1993), the conservative taxonomy is still used widely, e.g., Arroyo-Cabrales et al. (1997), Timm and Genoways (2003). In their phylogenetic study Morgan and Czaplewski (2003) recognized three Natalus species: major for Greater Antillean populations, stramineus for the remainder of the range, and tumidirostris for the northern South American populations characterized by swollen rostra and emarginated palates (Goodwin, 1959). Based on that phylogeny and on a morphological study of Greater Antillean taxa (Tejedor et al., in press), Simmons (in press) recognized Wve species of Natalus (Table 1). The results of sequencing the mitochondrial cytochrome b gene show that the species diversity within

100

L.M. Dávalos / Molecular Phylogenetics and Evolution 37 (2005) 91–103

Natalus is still underestimated. Fixed character diVerences were found among tumidirostris (Goodwin, 1959), jamaicensis (Goodwin, 1959), major (Miller, 1902), stramineus stramineus (Gray, 1838), and stramineus saturatus (Dalquest and Hall, 1949) from Mexico and Central America (Tables 2 and 4). These character diVerences in mtDNA could be the result of small sample sizes within Natalus (Table 4). Three lines of evidence suggest the observed diVerences in cytochrome b are not artifacts. First, site diVerences in the slowly evolving nuclear gene Rag2 accompany the mtDNA characters (Table 2). Second, the morphological study of Tejedor et al. (Goodwin, 1959) encompassed many more individuals than were available for sequencing and found species-level diVerences among Greater Antillean populations. Third, a study of 34 allozymes (ArroyoCabrales et al., 1997) found Wxed alleles at one or more loci for stramineus stramineus, stramineus saturatus and jamaicensis. The discontinuous variation in molecular and morphological characters reviewed so far suggests there is little gene Xow between the Wve Natalus populations studied (Table 1). Natalids require humid roosts, making large bodies of water into eVective barriers. The ensuing geographic isolation might explain diVerentiation among Greater Antillean Natalus. At present only stramineus stramineus and stramineus saturatus are not considered morphologically distinct species (Simmons, in press). Unsampled Natalus stramineus span Costa Rica east along northern South America to Bahia in Brazil, and south to Paraguay. Isolation by ocean barriers cannot readily explain the diVerentiation observed. It could be that stramineus stramineus and stramineus saturatus appear distinct as a result of sampling at the extremes of a continuum. If so, phylogeny reconstruction would show Natalus stramineus to be monophyletic with respect to other Natalus, unless all species in the genus evolved so recently that no hierarchical pattern is discernible. Natalus stramineus is not monophyletic, and hierarchical relationships among lineages within Natalus are evident based on both the relative branching order and support values of several nodes (Figs. 3 and 4). Natalus stramineus stramineus (Gray, 1838) and N. stramineus saturatus (Dalquest and Hall, 1949) should be recognized as species, named following the aforementioned taxonomy. These taxa are hereafter treated as full species. Natalus tumidirostris and N. stramineus are distinct in both cytochrome b and Rag2 sequences (Table 4), but the mitochondrial lineages are not reciprocally monophyletic (Fig. 3). Introgression through hybridization, or incomplete lineage sorting with retention of ancestral polymorphism have been invoked to explain such patterns in other mammals, e.g., HoVmann et al. (2003), Patton and Smith (1994). Future studies with denser sampling of cytochrome b and fast-evolving nuclear and/

or paternally inherited markers could distinguish among the processes involved. Molecular data remain to be sampled for two Chilonatalus and seven Natalus named taxa (Table 1). Unassigned Natalus ‘stramineus’ specimens are also known from Santa Cruz, Bolivia, and Concepción, Paraguay (Taddei and Uieda, 2001). Named populations of Natalus believed until recently to be conspeciWc (e.g., Arroyo-Cabrales et al. (1997), Koopman (1994), Morgan and Czaplewski (2003)) are in fact distinct (Table 4). It is likely that additional taxon sampling doubles the number of species recognized in the family considering the results of this study, the disjunct distribution of most subspecies (Koopman, 1994), and their diVerences in size (Goodwin, 1959; Ottenwalder and Genoways, 1982; Taddei and Uieda, 2001). Taxonomic revision has immediate implications for conservation. Two natalids are listed as endangered species (IUCN, 2003): Nyctiellus lepidus, extant in Cuba, and Cat (formerly San Salvador), Long and Little Exuma in the Bahamas (near-threatened); and Chilonatalus tumidifrons, extant in San Salvador (formerly Watling Is.), Andros and Great Abaco in the Bahamas (vulnerable). In contrast, Natalus jamaicensis was not listed because it was not recognized as distinct. St. Clair Cave, St. Catherine Parish is the only living record for this species (Goodwin, 1959; Goodwin, 1970), and the cave is not deliberately protected (Dávalos and Eriksson, 2003). In this context, taxonomic revision of unsampled populations is as urgent as the protection of the single localities known for extant Natalus jamaicensis and N. primus (Tejedor et al., 2004). 4.2. Natalid biogeography Biogeographic research on Caribbean bats has focused on Wnding the ultimate continental origin of all island populations (Baker and Genoways, 1978; GriYths and Klingener, 1988; Koopman, 1989). All hypotheses about the origin of extant natalids posit that the geographic range of their ancestor included Mexico and/or Central America (Baker and Genoways, 1978; Koopman, 1989; Morgan and Czaplewski, 2003). It is in this framework that Morgan and Czaplewski (2003) assert: “[natalids] presumably survived in ƒMiddle Americaƒ between the early Miocene and the late Pleistocene, although natalids have no fossil record during that time period.” An unstated premise constrains the biogeographic narrative; that continental Natalus must have descended from such ancient Middle American natalid stock. Otherwise the question arises: if continental natalids are unlikely representatives of the ancient lineage that gave rise to Caribbean natalids (Koopman, 1989), how did Natalus arise? To answer this question the geographic history of the natalids has to be examined using all available data.

L.M. Dávalos / Molecular Phylogenetics and Evolution 37 (2005) 91–103

The ultimate origin of natalids can be traced back to a continent, but where and when? Higher-level chiropteran relationships entail a North American origin. Molecular studies have revealed a close relationship between natalids, molossids, and vespertilionids, the latter two forming a clade (Hoofer et al., 2003; Teeling et al., 2003, 2002; Van Den Bussche et al., 2002, 2003). The earliest molossid fossil is Wallia from the middle Eocene of Saskatchewan (Legendre, 1985), and the earliest vespertilionid fossil is Stehlinia from the middle Eocene of Europe (McKenna and Bell, 1997). These extinct taxa date the divergence among the three lineages prior to the middle Eocene (>50MYA), placing the last common ancestor of natalids, molossids, and vespertilionids on the supercontinent formed by North America and Europe (Savage and Russell, 1983). The ancestor of these lineages might have been tropical; during the early Eocene tropical forests extended to today’s Montana and London, and broadleaf forests thrived within the Arctic circle (Prothero, 1994). The earliest known oVshoots of the natalid radiation appeared in Florida during the Oligocene and early Miocene (Morgan and Czaplewski, 2003). At the time Florida was probably tropical: emballonurids, mormoopids, molossids, and perhaps even phyllostomids lived in the region (Czaplewski et al., 2003c; Morgan and Czaplewski, 2003). The late Oligocene-early Miocene period was relatively warm worldwide. After the early Miocene the natalids of Florida went extinct, perhaps as a result of climate change. By the middle Miocene, the Earth was an “ice-house” (Mutti, 2000; Pearson and Palmer, 2002), and the Neotropics receded to their contemporary contours. The divergence between natalids and their extant close relatives is distinct from subsequent branching events (Fig. 6), perhaps as a result of these early extinctions. This is clearer when Rag2 is used to estimate molecular divergences (Fig. 6B). Phylogenetic analysis of morphological characters found that Primonatalus, one of the Florida fossils, was sister to extant natalids (Morgan and Czaplewski, 2003). The ancestor of Primonatalus and extant natalids probably ranged throughout the West Indies and Florida (Fig. 5). Up to this point there is no indication that early natalids ranged into Mexico and/or Central America. Geographic proximity to the West Indies (Fig. 1) is the main reason to postulate an ancient Middle American natalid. Is there evidence of such an ancestor, i.e., a fossil, and/or an early branch in the phylogeny? No, as Koopman (1989) hinted and Morgan and Czaplewski (2003) demonstrated for the Wrst time, Natalus evolved after a Wrst Caribbean lineage had already diVerentiated (Figs. 3 and 4). To postulate the existence of this ancient Middle American taxon is to propose that natalids expanded their range from Florida to Middle America, reached the Caribbean from the latter rather than the former, went extinct on the continent without leaving a trace, diversi-

101

Wed on the islands, and then recolonized the continent during the diversiWcation of the genus Natalus. It is more economical to propose that the ancestor of extant natalids survived in the northern West Indies (Fig. 5), not in Mexico or Central America. In the Antilles the natalids diversiWed giving rise to Nyctiellus Wrst, and then to Chilonatalus and Natalus (Figs. 4 and 6). The phylogeny of natalids and their geographic distribution raise a novel proposition: continental Natalus evolved from Caribbean ancestors (Fig. 5). Phylogeny aside, another piece of evidence supports the West Indian origin of all extant natalids; their roosting ecology. Some molossid and vespertilionid species roost in caves, but for natalids the humid microenvironment found only in long caves is a critical requirement (Morgan, 2001; Tejedor et al., 2004). It is tempting to speculate that the strict roosting requirements of natalids evolved over millions of years while subjected to Caribbean hurricanes. There are some data showing that Caribbean cave roosting bats fare better after a hurricane than tree roosters (Jones et al., 2001). Future studies can evaluate whether natalids show diVerent rates of survival after a hurricane when compared to other bats of similar size (the study cited was conducted in Puerto Rico, where no natalids have been found). Niche modeling could identify variables that condition the distribution of extant natalids, and ancestral reconstruction of those niches could be used to evaluate the Caribbean origin hypothesis. The data at hand indicate that natalids reached the Caribbean from North America early in their evolutionary history and subsequently colonized Mexico, Central America, and South America (Fig. 5). The precise routes and timing of these events, as well as the detailed geographic history of the genus Natalus, remain to be explored.

Acknowledgments This article is based upon work supported by the National Science Foundation under Grant No. 0206336, it is a contribution from the Monell Molecular Laboratory and the Cullman Research Facility in the Department of Ornithology, American Museum of Natural History, and has received generous support from the Lewis B. and Dorothy Cullman Program for Molecular Systematics Studies, a joint initiative of The New York Botanical Garden and the American Museum of Natural History. This research was supported by the Department of Mammalogy of the American Museum of Natural History, the NASA Grant No. NAG5-8543 to the Center for Biodiversity and Conservation at the American Museum of Natural History, and the Department of Ecology, Evolution and Environmental Biology at Columbia University. Fieldwork was supported by

102

L.M. Dávalos / Molecular Phylogenetics and Evolution 37 (2005) 91–103

the Ambrose Monell Cryogenic Collection and the Department of Mammalogy, both at the American Museum of Natural History, the Center for Environmental Research and Conservation and the Department of Ecology, Evolution and Environmental Biology, both at Columbia University, Elizabeth Dumont’s NSF grant, and the Explorers’ Club (New York). The author is currently supported by a NIH grant in bacterial genomics to H. Ochman at the University of Arizona. For comments, collecting permits, Weld or lab assistance, tissue loans, and/or any other kind of support I thank F.K. Barker, A. Corthals, J.L. Cracraft, N. Czaplewski, M. Delarosa, R. DeSalle, A. Donaldson (NEPA, Jamaica), K. Doyle, E. Dumont, R. Eriksson, J. Feinstein, N. Gyan (Wildlife Section, Trinidad), R. Harbord, S.A. Jansa, S. Koenig, J. Mercedes, G.S. Morgan, T. Nicole, and V. Outten (Ministry of Agriculture and Fisheries, Bahamas), A.L. Porzecanski, A. Rodríguez, R.O. Sánchez (Vida Silvestre, República Dominicana), P. Schickler, M. Schwartz, N.B. Simmons, E. Teeling, A. Tejedor, and A. Wright. This work would have been impossible without the intellectual and material legacy of Karl F. Koopman at the American Museum of Natural History.

References Arroyo-Cabrales, J., Ray, C.E., 1997, Revisión de los vampiros fósiles (Chiroptera: Phyllostomidae, Desmodontinae) de México. In: ArroyoCabrales, J. Polaco, O.J. (Eds.), Homenaje al Profesor Ticul Álvarez. Instituto Nacional de Antropología e Historia, México, DF, pp. 69–86. Arroyo-Cabrales, J., Van Den Bussche, R.A., Sigler, K.H., Chesser, R.K., Baker, R.J., 1997. Genic variation of mainland and island populations of Natalus stramineus (Chiroptera: Natalidae). Occasional Papers, Museum of Texas Tech University 171, 1–9. Baker, R.J., Genoways, H.H., 1978. Zoogeography of Antillean bats. Special Publication, Academy of Natural Sciences of Philadelphia 13, 53–97. Baker, R.J., Porter, C.A., Patton, J.C., Van Den Bussche, R.A., 2000. Systematics of bats of the family Phyllostomidae based on RAG 2 DNA sequences. Occasional Papers Museum of Texas Tech University 202, 1–16. Bremer, K., 1992. Ancestral areas: a cladistic reinterpretation of the center of origin concept. Systematic Biology 41, 436–445. Bremer, K., 1994. Branch support and tree stability. Cladistics 10, 295–304. Czaplewski, N.J., Krejca, J., Miller, T.E., 2003b. Late quaternary bats from Cebada Cave, Chiquibul Cave System, Belize. Caribbean Journal of Science 39, 23–33. Czaplewski, N.J., Masanaru, T., Naeher, T.M., Shigehara, N., Setoguchi, T., 2003a. Additional bats from the middle Miocene La Venta fauna of Colombia. Revista de la Academia Colombiana de Ciencias Físicas. Exactas y Naturales 27, 263–282. Czaplewski, N.J., Morgan, G.S., Naeher, T., 2003c. Molossid bats from the late Tertiary of Florida with a review of the Tertiary Molossidae of North America. Acta Chiropterologica 5, 61–74. Dalquest, W.W., 1950. The genera of the chiropteran family Natalidae. Journal of Mammalogy 31, 436–443. Dalquest, W.W., Hall, E.R., 1949. A new subspecies of funnel-eared bat (Natalus mexicanus) from eastern Mexico. Proceedings of the Biological Society of Washington 62, 153–154.

Dávalos, L.M., Eriksson, R., 2003. New and noteworthy records from ten Jamaican bat caves. Caribbean Journal of Science 39, 140–144. Dávalos, L.M., Jansa, S.A., 2004. Phylogeny of the Lonchophyllini (Chiroptera: Phyllostomidae). Journal of Mammalogy 85, 404–413. Emmons, L.H., 1997. Neotropical Rainforest Mammals: A Field Guide, second ed. University of Chicago Press, Chicago. Eriksson, T., 1999. AutoDecay, version 4.0. Bergius Foundation, Royal Swedish Academy of Sciences. Goodwin, G.C., 1959. Bats of the subgenus Natalus. American Museum Novitates 1977, 1–22. Goodwin, R.E., 1970. The ecology of Jamaican bats. Journal of Mammalogy 51, 571–579. Gray, J.E., 1838. A revision of the genera of bats (Vespertilionidae), and the description of some new genera and species. Magazine of Zoology and Botany 2, 483–505. GriYths, T.A., Klingener, D., 1988. On the distribution of Greater Antillean bats. Biotropica 20, 240–251. Hall, E.R., Kelson, K.R., 1959. The Mammals of North America. Ronald Press, New York. HoVmann, F.G., Owen, J.G., Baker, R.J., 2003. mtDNA perspective of chromosomal diversiWcation and hybridization in Peters’ tent-making bat (Uroderma bilobatum: Phyllostomidae). Molecular Ecology 12, 2981–2993. Hoofer, S.R., Reeder, S.A., Hansen, E.W., Van Den Bussche, R.A., 2003. Molecular phylogenetics and taxonomic revision of the infraorder Yangochiroptera (Chiropteran: Mammalia). Journal of Mammalogy 84, 809–821. Huelsenbeck, J.P., Ronquist, F., 2001. MRBAYES: Bayesian inference of phylogenetic trees. Bioinformatics 17, 754–755. IUCN. 2003. 2003 IUCN Red List of Threatened Species. Jansa, S.A., Goodman, S.M., Tucker, P.K., 1999. Molecular phylogeny and biogeography of the native rodents of Madagascar (Muridae: Nesomyinae): a test of the single-origin hypothesis. Cladistics 15, 253–270. Jones, K.E., Barlow, K.E., Vaughan, N., Rodríguez-Durán, A., Gannon, M.R., 2001. Short-term impacts of extreme environmental disturbance on the bats of Puerto Rico. Animal Conservation 4, 59– 66. Koopman, K., 1993. Order Chiroptera. In: Wilson, D.E., Reeder, D.M. (Eds.), Mammal Species of the World, a Taxonomic and Geographic Reference. Smithsonian Institution, Washington, DC, pp. 137–242. Koopman, K., 1994. Chiroptera: systematics. Handbuch der Zoologie 8, 1–217. Koopman, K.F., 1989. A review and analysis of the bats of the West Indies. In: Woods, C.A. (Ed.), Biogeography of the West Indies, Past, Present, and Future. Sandhill Crane Press, Gainesville, Florida, pp. 635–644. Legendre, S., 1985. Molossidés (Mammalia, Chiroptera) cénozoiques de l’Ancien et du Nouveau Monde; statut systématique; intégration phylogénique de données. Neues Jahrbuch für Geologie und Paläontologie, Abhandlungen 170, 205–227. Linares, O.J., 1971. A new subspecies of Funnel-eared bat (Natalus stramineus) from western Venezuela. Bulletin of the Southern California Academy of Science 70, 81–84. Lockhart, P.J., Steel, M.A., Hendy, M.D., Penny, D., 1994. Recovering evolutionary trees under a more realistic model of sequence evolution. Molecular Biology and Evolution 11, 605–612. Maddison, D.R., Maddison, W.P., 2003. MacClade 4: Analysis of Phylogeny and Character Evolution, version 4.06. Sinauer Associates McKenna, M., Bell, S., 1997. ClassiWcation of Mammals above the Species Level. Columbia University, New York. Miller, G.S., 1902. Twenty new American bats. Proceedings of the Academy of Natural Sciences of Philadelphia 54, 389–412. Morgan, G., 2001. Patterns of extinction in West Indian bats. In: Woods, C.A., Sergile, F.E. (Eds.), Biogeography of the West Indies. CRC Press, Boca Raton, pp. 369–407.

L.M. Dávalos / Molecular Phylogenetics and Evolution 37 (2005) 91–103 Morgan, G.S., 1989. Fossil Chiroptera and Rodentia from the Bahamas, and the historical biogeography of the Bahamian mammal fauna. In: Woods, C.A. (Ed.), Biogeography of the West Indies: Past, Present, and Future. Sandhill Crane Press, Gainesville, Florida, pp. 685–740. Morgan, G.S., Czaplewski, N.J., 2003. A new bat (Chiroptera: Natalidae) from the early Miocene of Florida, with comments on natalid phylogeny. Journal of Mammalogy 84, 729–752. Mutti, M., 2000. Bulk o18O and o13C records from Site 999, Colombian Basin, and Site 1000, Nicaraguan Rise (latest Oligocene to middle Miocene); diagenesis, link to sediment parameters, and paleoceanography. Proceedings of the Ocean Drilling Program, ScientiWc Results 165, 275–283. Ottenwalder, J.A., Genoways, H.H., 1982. Systematic review of the Antillean bats of the Natalus micropus-complex (Chiroptera: Natalidae). Annals of Carnegie Museum 51, 17–38. Patton, J.L., Smith, M.F., 1994. Paraphyly, polyphyly, and the nature of species boundaries in Pocket Gophers (Genus Thomomys). Systematic Biology 43, 11–26. Pearson, P.N., Palmer, M.R., 2002. Atmospheric carbon dioxide concentrations over the past 60 million years. Nature 406, 695–699. Posada, D., Crandall, K.A., 1998. MODELTEST: testing the model of DNA substitution. Bioinformatics 14, 817–818. Prothero, D.R., 1994. The Eocene–Oligocene Transition: Paradise Lost. Columbia University, New York. Rambaut, A.E., Grassly, N.C., 1997. SEQ-GEN: an application for the Monte Carlo simulation of DNA sequence evolution along phylogenetic trees. Computer Applications in the Biosciences 13, 235–238. Ronquist, F., 1996. DIVA Manual, version 1.1. Uppsala University Ronquist, F., 1997. Dispersal-vicariance analysis: a new approach to the quantiWcation of historical biogeography. Systematic Biology 46, 195–203. Savage, D.E., Russell, D.E., 1983. Mammalian Paleofaunas of the World. Addison-Wesley, London. Shimodaira, H., Hasegawa, M., 1999. Multiple comparisons of log-likelihoods with applications to phylogenetic inference. Molecular Biology and Evolution 16, 1114–1116. Silva-Taboada, G., 1979. Los Murciélagos De Cuba. Editorial Academia, La Habana. Simmons, N.B., in press. Order Chiroptera. In: Wilson, D.E., Reeder, D.M. (Eds.), Mammal species of the World: A Taxonomic and

103

Geographic Reference. Johns Hopkins University Press, Baltimore. SwoVord, D.L., 2002. PAUP*. Phylogenetic Analysis Using Parsimony (¤ and Other Methods). version 4.0b10. Sinauer Associates. Taddei, V.A., Uieda, W., 2001. Distribution and morphometrics of Natalus stramineus from South America (Chiroptera, Natalidae). Iheringia Serie Zoologia 91, 123–132. Teeling, E.C., Madsen, O., Murphy, W.J., Springer, M.S., O’Brien, S.J., 2003. Nuclear gene sequences conWrm an ancient link between New Zealand’s short-tailed bat and South American noctilionoid bats. Molecular Phylogenetics and Evolution 28, 308–319. Teeling, E.C., Madsen, O., Van Den Bussche, R.A., de Jong, W.W., Stanhope, M.J., Springer, M.S., 2002. Microbat paraphyly and the convergent evolution of a key innovation in Old World rhinolophoid microbats. Proceedings of the National Academy of Sciences of the United States of America 99, 1431–1436. Tejedor, A., Silva-Taboada, G., Rodriguez Hernandez, D., 2004. Discovery of extant Natalus major (Chiroptera: Natalidae) in Cuba. Mammalian Biology 69, 153. Tejedor, A., Tavares, V.D.C., Silva-Taboada, G., in press. Taxonomic revision of Greater Antillean bats of the genus Natalus. American Museum Novitates. Templeton, A.R., 1983. Phylogenetic inference from restriction endonuclease cleavage site maps with particular reference to the evolution of humans and apes. Evolution 37, 221–244. Timm, R.M., Genoways, H.H., 2003. West Indian mammals from the Albert Schwartz collection: biological and historical information. ScientiWc Papers, Natural History Museum the university of Kansas 29, 1–47. Van Den Bussche, R.A., Hoofer, S.R., 2001. Evaluating the monophyly of Nataloidea (Chiroptera) with mitochondrial DNA sequences. Journal of Mammalogy 82, 320–327. Van Den Bussche, R.A., Hoofer, S.R., Hansen, E.W., 2002. Characterization and phylogenetic utility of the mammalian protamine P1 gene. Molecular Phylogenetics and Evolution 22, 333–341. Van Den Bussche, R.A., Reeder, S.A., Hansen, E.W., Hoofer, S.R., 2003. Utility of the dentin matrix protein 1 (DMP1) gene for resolving mammalian intraordinal phylogenetic relationships. Molecular Phylogenetics and Evolution 26, 89–101. Varona, L.S., 1974. Catálogo de los mamíferos vivientes y extinguidos de las Antillas. Academia de Ciencias de Cuba, Havana.