Nucleotide sequences from 94 individuals representing the Oryzomys palustris complex (O. palustris and O. couesi) were examined to assess phylogenetic relationships and taxonomic boundaries. Sequence data from the entire mitochondrial cytochrome-b (Cytb; 1,143 base pairs [bp]), a portion of exon 1 of the nuclear interphotoreceptor retinoid-binding protein (1,266 bp), and intron 2 of the alcohol dehydrogenase 1 (580 bp) genes were analyzed using phylogenetic methods (maximum parsimony and Bayesian inference). In the Cytb analysis, individuals recognized as O. palustris and O. couesi formed reciprocally monophyletic clades supporting their recognition as species; however, additional phylogenetically informative groups were present within each of the 2 nominate clades. In addition, levels of genetic divergence within the currently recognized taxa exceeded values normally associated with intraspecies variation. Together, the phylogenetic and genetic divergence data imply that consideration should be given to recognizing 4 additional species in this complex.
The taxonomic status of Oryzomys couesi (Coues’ rice rat) and O. palustris (marsh rice rat) in the southern United States and Central America has been problematic for over a century. Originally described as species in separate genera (Mus palustris [Harlan 1837] and Hesperomys couesi [Alston 1876]), Baird (1857) and later Thomas (1893), respectively, later placed the 2 taxa within Oryzomys. In his synopsis of the genus Oryzomys, Merriam (1901) treated couesi and palustris as separate species and placed them in a palustris-mexicanus group (along with 16 other species). In addition, Merriam (1901) divided this group into 2 subgroups based on morphological data: white-bellied forms (containing O. palustris and other species) and fulvous-bellied forms (containing O. couesi and others). Goldman (1918) relegated many of Merriam’s species to subspecific status under O. palustris or O. couesi. He envisioned the 2 taxa belonging to a palustris group, which he divided into 2 sections: palustris (containing only palustris) and couesi (containing couesi and 6 other species).
Goldman (1918) and Paradiso (1960) identified large amounts of morphological variation within populations of Oryzomys, implying that morphology alone is not useful for distinguishing among groups of this genus. This observation was highlighted by Hall (1960) who, in contrast to previous authors, treated couesi as a subspecies of palustris based on patterns of intergradation associated with cranial morphology and pelage color of O. c. aquaticus and O. p. texensis in extreme southern Texas (Willacy and Cameron counties). Although Hall recognized distinctions in pelage color between couesi from Cameron County and palustris from San Patricio County, he determined that San Patricio samples were intermediate between the more southwesterly distributed couesi and more northeasterly distributed palustris (Hall 1960). This assignment of couesi into palustris impacted more than southern Texas, in that Hall’s synopsis placed all subspecies of couesi from Mexico and Central America within palustris. This trend soon extended to other taxa recognized by Merriam (1901) and Goldman (1918) as all but 3 taxa (O. palustris, O. fulgens, and O. nelsoni) eventually were subsumed into palustris (Hall 1981; Handley 1966; Hershkovitz 1971; Jones and Lawlor 1965).
Benson and Gehlbach (1979) recommended that couesi be reelevated to species status based on a larger body size, differences in morphology of the X chromosome, and no apparent intergradation with palustris. Hall (1981:1179) agreed, indicating that his initial material was not adequate and that further studies were needed to address levels of intergradation in couesi and palustris. Schmidt and Engstrom (1994) contended that sampling by Benson and Gehlbach (1979) was flawed and that the karyotypic data were not sufficiently consistent to be used as diagnostic characters. However, Schmidt and Engstrom (1994) did concur that 2 species should be recognized based on electrophoretic data gathered from samples obtained from an area of sympatry in southern Texas and northern Tamaulipas, Mexico.
The 1st goal of this study was to use nucleotide sequence data to evaluate the assertions of Benson and Gehlbach (1979) and Schmidt and Engstrom (1994) that couesi and palustris should be recognized as separate species. Although it generally is accepted that couesi and palustris are unique species (Musser and Carleton 1993, 2005; Weksler 2003, 2006; Weksler et al. 2006), no study has used DNA sequence data specifically to evaluate this hypothesis. The 2nd goal was to examine genetic and karyologic variability within the geographic distributions of the 2 taxa. Although many studies have examined palustris and couesi in the eastern and southern United States (Benson and Gehlbach 1979; Hamilton 1955; Humphrey and Setzer 1989; Schmidt and Engstrom 1994; Spitzer and Lazell 1978), little is known about populations, subspecies, or related taxa in Mexico or Central America. At least 26 taxa have been proposed from Central America (Bole 1937; Burt 1934; Goldman 1912; Merriam 1901; Murie 1932) at the level of species or subspecies. Two of these (nelsoni and penninsulae) are believed to be extinct, and 1 (fulgens) is enigmatic in that the type locality is unclear, and the type specimen has not been adequately aligned with any other member of this group (Carleton and Arroyo-Cabrales 2009). Of the remaining 23 taxa, 7 were described as subspecies (apatelius, aztecus, Iambi, molestus, peragrus, pinícola, and rufinus), leaving 16 taxa originally described as species and contained within O. couesi. Of those 16, samples were included from at, or near, the type locality of 10. With the addition of all 6 of the named taxa for palustris, 16 of the 22 taxa putatively affiliated with couesi (16) and palustris (6) were examined.
Nucleotide sequence data were obtained from 3 independent gene regions (mitochondrial protein coding gene, nuclear exon, and nuclear intron). First, the mitochondrial cytochrome-b gene (Cytb) was examined throughout the geographic ranges of populations assigned to couesi and palustris. This gene has been used successfully to assess species-level relationships in Oryzomyini genera (Bonvicino and Moreira 2001; Langguth and Bonvicino 2002; Myers et al. 1995). Second, to control for biases generally associated with mitochondrial DNA sequences (Avise 1994; Hillis et al. 1996; Prychitko and Moore 2000), a portion of exon I of the nuclear interphotoreceptor retinoid-binding protein gene (Rbp3) and intron 2 of the alcohol dehydrogenase 1 gene (Adhl-12) were examined from a subset of individuals (n = 29) represented in the Cytb data set. These gene regions have been used to examine relationships within Sigmodontinae and Oryzomyini (Amman et al. 2006; D'Elia 2003; Longhofer and Bradley 2006; Weksler 2003, 2006), and should be useful in offsetting any biases in the mitochondrial data and in providing resolution at mid- and deeper-level nodes. Finally, karyotypes of 23 individuals of O. couesi from western and eastern Mexico and from Honduras were compared to examine chromosomal differences between populations.
Materials and Methods
Samples.—Tissue samples for 48 individuals of O. couesi and 46 individuals of O. palustris were collected from natural populations, following methods approved by the American Society of Mammalogists Animal Care and Use Committee (Gannon et al. 2007), the Texas Tech Animal Care and Use Committee, and the University of Miami Animal Care and Use Committee. Samples also were borrowed from museum collections (Appendix I; Fig. 1). For each taxon as many subspecies as possible were sampled (10 of 16 for couesi and 6 of 6 for palustris) with multiple individuals in most taxa: O. c. aquaticus (n = 4), O. c. azuerensis (n = 2), O. c. bulled (n = 1), O. c. couesi (n = 9), O. c. cozumalae (n = 1), O. c. goldmani (n = 5), O. c. mexicanus (n = 9), O. c. richardsoni (n = 5), O. c. teapensis (n = 6), O. c. zygomaticus (n = 6), O. p. coloratus (n = 3), O. p. natator (n = 2), O. p. palustris (n = 14), O. p. sanibeli (n = 3), O. p. planirostris (n = 2), and O. p. texensis (n = 22).
Map (modified from Hall  and Humphrey and Setzer ) depicting localities for specimens examined and species distributions (shaded areas and arrows). Circles represent sampling sites for Oryzomys palustris and O. texensis, squares represent O. couesi and O. mexicanus, and triangles represent Oryzomys species 1 and Oryzomys species 2. Minor clade designations are listed under taxon names.
DNA extraction, polymerase chain reaction amplification, and sequencing.—Genomic DNA was isolated from approximately 0.1 g of either liver, muscle, or tail clips using either a Qiagen extraction kit (Qiagen Inc., Valencia, California) or standard laboratory DNA extraction procedure. Polymerase chain reaction (Saiki et al. 1988) was used to amplify the complete 1,143 base pairs (bp) of Cytb for 94 samples, a 1,266-bp portion of exon 1 of Rbp3, and 580 bp of Adh1 -I2 for a subset of 29 samples (couesi: n = 19; palustris: n = 10) representing major clades estimated in the Cytb analyses. Reaction concentrations for Cytb (50-µl volume) included ≤300 ng genomic DNA, 0.07 mM dNTPs, 2.86 mM MgCl, 5 µl 10X buffer, 1.25 U enzyme (Go Taq, Promega, Madison, Wisconsin), and 0.286 µM of primers MVZ05 (Smith and Patton 1993) and CB40 (Hanson and Bradley 2008). Reaction concentrations for Rbp3 and Adh1-I2 (35 µl volume) included ≤300 ng genomic DNA, 0.07 mM deoxynucleoside triphosphates, 2.86 mM MgCl, 3.5 µl 10X buffer, 1.5 U AmpliTaq Gold enzyme (Applied Biosystems, Foster City, California), and 0.286 µM of primers Al (Stanhope et al. 1992) and B2 (Weksler 2003) for Rbp3 and Exon2F and 2340-2 (Amman et al. 2006) for Adh1-I2. For Cytb, polymerase chain reaction thermal profiles included an initial denaturation step at 95 °C (2 min), 30–40 cycles with denaturation at 95°C (45 s), annealing at 54°C (1 min), extension at 72°C (1 min, 30 s), and a final extension cycle of 72°C (8 min). For Rbp3, polymerase chain reaction conditions included an initial denaturation at 95°C (10 min), 26 cycles with denaturation at 95°C (25 s), annealing at 58°C (20 s), extension at 72°C (1 min, 15 s), and a final extension cycle of 72°C (2 min). For Adh1-I2, polymerase chain reaction conditions included an initial denaturation at 95°C (2 min), 35 cycles with denaturation at 95°C (30 s), annealing at 55°C (1 min), extension at 73°C (1 min, 30 s), and a final extension cycle of 73°C (15 min).
Amplicons were purified using a QIAquick PCR purification kit (Qiagen, Inc., Valencia, California) or ExoSap-IT enzymes (USB Corp., Cleveland, Ohio). Polymerase chain reaction amplicons then were sequenced using ABI Prism Big Dye Terminator v3.1 ready reaction mix (Applied Biosystems) and an ABI 3100-Avant or ABI 3730 automated sequencer (Applied Biosystems). Polymerase chain reaction primers were used in conjunction with internal primers for cycle sequencing: Cytb—OL14841, CMTCATGATGAAACTTCGATC; O400R (Hanson and Bradley 2008); O400R*, CYCCTCAGAATGATATTTGTCATGG; Fl (Whiting et al. 2003); O550R, ACTARRGCTGTRATAATAAATGG; O700H*, GGAAATATCATTCTGGTTTAATRTGTGC; O870R (Hanson and Bradley 2008); 700L (Peppers and Bradley 2000); and MVZ04 and MVZ45 (Smith and Patton 1993); Rbp3—A (Jansa and Voss 2000); 395R, TGACCACCAGTACATTGCCGAAGA; and D2 and E2 (Weksler 2003); Adh1-I2—410F, 350F, and 410R (Amman et al. 2006). Sequencing consisted of 95° s) denaturing, 50°nnealing, and 60°) extension. After 30–s, reactions were purified using Edge Biosystems columns (Gaithersburg, Maryland) and precipitated in isopropanol. Sequencher 4.1.4 software (Gene Codes, Ann Arbor, Michigan) was used to align and proof nucleotide sequences, and chromatograms were examined to resolve any discrepancies and to inspect sequences for heterozygous sites. MEGA4 software (Tamura et al. 2007) was used to check for stop codons and the presence of pseudogenes. DNA sequences were deposited in GenBank and accession numbers are listed in Appendix I; alignments were deposited in Treebase under accession number S2514.
Karyotyping.—Karyotypes were obtained from 23 individuals of O. couesi (Appendix I) following the methods described by Baker et al. (2003). At least 10 metaphase spreads from each individual were photographed with an Applied Imaging camera (Applied Imaging Systems, San Jose, California). These spreads were captured and arranged into karyograms using the Genus System 3.7 from Applied Imaging Systems and enhanced with Image Pro Plus 4.5.1 22 (Media Cybernetics, Inc., Bethesda, Maryland). Diploid numbers (2n) were recorded for all 23 individuals, and fundamental numbers (FN) were recorded for 13 individuals.
Data analyses.—Sequence data were evaluated in 3 steps using Holochilus sciureus, Melanomys chrysomelas, Sigmodontomys alfari, Nectomys squamipes, and Aegialomys xanthaeolus as outgroup taxa in all analyses. Additionally, Melanomys caliginosas and Nectomys apicalis were used as benchmark species for comparisons of genetic distances. These taxa were selected based on results of molecular studies by Weksler (2003), (2006) in which they formed a monohyletic group sister to the O. palustris complex within the Oryzomyini. Cytb sequences were analyzed using maximumparsimony and Bayesian inference methods. For parsimony analyses all nucleotide positions were treated as equally weighted, unordered, discrete characters with 4 possible states: A, C, G, and T. Uninformative characters were excluded, and optimal trees were estimated using the heuristic search method with tree-bisection-reconnection branch-swapping and stepwise addition sequence options in the software package PAUP* version 4.0bl0 (Swofford 2002). Nodal support of topologies was estimated using heuristic bootstrapping (Felsenstein 1985) with 100 iterations. The software MrModeltest (Nylander 2004) was used to estimate the most appropriate model of evolution (GTR+I+G) to be included in the Bayesian analysis performed in MRBAYES version 3.1.2 (Ronquist and Huelsenbeck 2003) with sequences partitioned by codon using site-specific gamma distribution allowing for invariable sites, and the following options: 4 Markov chains, 10 million generations, and sample frequency every 1,000th generation. The first 1,000 trees were discarded as burn-in based on stabilization of likelihood scores, and a consensus tree (50% majority rule) was constructed from the remaining trees using PAUP* (Swofford 2002). Nodal support (clade probability values) for topologies was inferred using clade probabilities estimated with MRBAYES 3.1.2 (Ronquist and Huelsenbeck 2003).
Individuals (n = 29) representing members of major clades (identified in the parsimony and Bayesian analyses of the Cytb data) were selected and the Rbp3 gene sequenced. Rbp3 sequences were analyzed using the parsimony methods and outgroups described above. Heterozygous sites were designated using the IUB polymorphic code. In the Bayesian analysis sequences were partitioned by codon, and the GTR+I+G model was run under the same conditions as the Cytb Bayesian analysis.
Individuals sequenced for Rbp3 also were sequenced for Adh1-I2. Adh1-I2 sequences were analyzed using the parsimony methods and outgroups as described above, except bases were not partitioned; however, gaps were coded as phylogenetically informative using FastGap (Borchsenius 2007) following the conservative method outlined in Simmons and Ochoterena (2000). Heterozygous sites were designated using the IUB polymorphic code. In the Bayesian analysis the GTR+G model was used under the same conditions as the Cytb and Rbp3 analyses.
A combined 3-gene data set for 29 individuals was analyzed using BEST 2.3 (Liu 2008). The data were partitioned by gene, and the respective model was applied to each genetic data set and run with the following options: 4 Markov chains, 60 million generations, and a sample frequency every 1,000th generation. The mitochondrial data were designated as haploid, and an annealing process was used during the initial 20% of generations. The initial 30,000 trees were discarded as burn-in, and a majority rule consensus tree was constructed using the remaining trees. In addition, pairwise genetic distances were estimated using Cytb data under the Kimura 2-parameter model of evolution (Kimura 1980). Average genetic distances were estimated for selected clades, and values were used to infer levels of genetic divergence between taxonomic groups. Comparisons of genetic distances between M. caliginosus and M. chrysomelas and between N. apicalis and N. squamipes were used as benchmarks for evaluating levels of genetic divergence between and within O. palustris and O. couesi following the rationale outlined in Bradley and Baker (2001) and Baker and Bradley (2006).
Nucleotide sequences from the Cytb gene were obtained from 94 individuals of O. couesi and O. palustris. Comparison of nucleotide substitutions revealed that transitions were 4.1 times more common than trans versions. In the parsimony analysis 251 informative characters were used to generate a tree with the following characteristics: length = 1,002 steps, consistency index = 0.4022, and retention index = 0.8998. Both the parsimony and Bayesian analyses (Fig. 2, with bootstrap values for parsimony) revealed 2 major clades labeled clade I and II. Clade I contained individuals referable to O. palustris, and clade II contained those referable to O. couesi. Within clade I, 2 minor clades (A and B) were estimated. Clade A contained samples from Alabama, Florida, South Carolina, and Virginia, whereas clade B included samples from Arkansas, Mississippi, Louisiana, Tennessee, Texas, Oklahoma, and Mexico, including specimens that were assignable to O. p. palustris (Tennessee and Mississippi). Within clade II, 4 minor clades (C–F) were estimated. Clade C contained samples from southern and western Mexico (Chiapas, Colima, Jalisco, Michoacán, Nayarit, and Oaxaca) and El Salvador, and clade D was composed of samples from Texas, eastern Mexico, Guatemala, Honduras, and Nicaragua. Clades E and F were composed of samples from Panama and Costa Rica, respectively. All major and minor clades had strong (90–100%) bootstrap support and were supported with posterior probabilities > 95%.
Phylogenetic tree obtained from the Bayesian analysis of 1,143 bases of the cytochrome-b gene. Major clades are depicted by Roman numerals (I and II), and minor clades are depicted by letters (A–F). Asterisks above branches represent clade probability values > 95%, whereas bootstrap support values > 50% (obtained from a parsimony analysis of the same data set) are shown below branches. Support values are not shown for relationships representing clades of 3 or fewer individuals unless that relationship represents a named group.
Nucleotide sequences for the Rbp3 gene from 29 individuals of Oryzomys were analyzed using parsimony and Bayesian methods. Comparison of nucleotide substitutions revealed that transitions were 2.0 times more common than trans versions. Parsimony analysis of 1,266 bp of Rbp3 used 17 informative characters to generate a tree with the following characteristics: length = 75 steps, consistency index = 0.4533, and retention index = 0.6555. A maximum-parsimony tree (not shown but with bootstrap values depicted in Fig. 3) contained both major clades (I and II) and 2 of the minor clades (E and F) estimated in the Cytb analyses. Bootstrap support was low (60–75%) to moderate (76–95%) for the major clades, and high (91–100%) for the 2 minor clades. Although 4 of the minor clades (A, B, C, and D) were not inferred, the Bayesian analysis for Rbp3 estimated support for clade I and for minor clades E and F (Fig. 3).
Phylogenetic tree obtained from the Bayesian analysis of 1,266 bases of the interphotoreceptor retinoid binding protein gene. Major clades are depicted by Roman numerals (I and II), and terminal branches are labeled according minor clades (A–F) from cytochrome-b analysis. Numbers above branches represent clade probability values > 95%, whereas bootstrap support values > 60% (obtained from a parsimony analysis of the same data set) are shown below branches.
Nucleotide sequences for the Adh1-I2 gene from the same 29 individuals sequenced for Rbp3 were analyzed using parsimony and Bayesian methods. Parsimony analysis of 580 bp of Adh1-I2 used 32 informative characters to generate a tree with the following characteristics: length = 56 steps, consistency index = 0.7500, and retention index = 0.9451. A maximum-parsimony tree (not shown but with bootstrap values depicted in Fig. 4) contained a clade comprised exclusively of individuals from clade II, including a supported minor clade E; and a clade (I′) that contained all individuals from clade I depicted in Fig. 2, minor clade F, and a clade (C′) composed of 4 individuals from the clade C depicted in Fig. 2. Bootstrap support was low (60–75%) for the clade containing clade I individuals and minor clade F and high (91–100%) for clades I′, II, C′ and E. Although 4 of the minor clades (A, B, C, and D) were not inferred, the Bayesian analysis for Adh1 -I2 estimated support for clades I′ and II and minor clades C′, E, and F (Fig. 4). Two individuals in clade C′ were collected at the same locality (Playa de Oro, Colima, Mexico) as an individual contained in clade II. Comparisons of Adh1-I2 sequences revealed that the individuals in clade C′ showed a palustris-like sequence at positions 93, 107, 126, 168, 171, 172, 173, 175, 180, 355, and 358, including a deletion at 144, and a couesi-like insertion at positions 58–61. In addition, the individuals in clade C shared unique bases at positions 119, 136, 137, and 200.
Phylogenetic tree obtained from the Bayesian analysis of 580 bases of the alcohol dehydrogenase gene. Major clades are depicted by Roman numerals (I′ and II) or letters (C′), and terminal branches are labeled according minor clades (A–F) from cytochrome-b analysis. Numbers above branches represent clade probability values > 95%, whereas bootstrap support values > 60% (obtained from a parsimony analysis of the same data set) are shown below branches.
To account for the gene heterogeneity found when comparing the 3 respective gene trees, a Bayesian estimate of species tree (BEST) analysis was performed to estimate a species tree independent of the individual gene trees. This analysis recovered a monophyletic Oryzomys (Fig. 5). It also recovered a clade (clade II′) composed of the clade II depicted in Fig. 2 minus minor clade F, plus a supported clade I. The relationships between clades I, II′ and F were unresolved, as were the relationships between clades C, D, and E. However, in contrast to the Adh1-I2 gene tree, the species tree did recover clade C within clade II′
Phylogenetic tree obtained from the BEST analysis based on combined data from the cytochrome-b gene, exon I of the nuclear interphotoreceptor retinoid-binding protein gene, and intron 2 of the alcohol dehydrogenase 1 gene. Each branch represents 1 of the species listed in the discussion. The major clades (I and II′) are listed as in the text; minor clades (A–F) are labeled relative to Fig. 2 with number of individuals in parentheses. Posterior probabilities >95% are listed above the branches.
Mean genetic distances for Cytb (Table 1) were estimated using the Kimura 2-parameter model of evolution (Kimura 1980). The mean genetic distance within clades ranged from 0.65% (clade B) to 2.06% (clade C), whereas values between clades ranged from 4.41% (C and D) to 12.6% (E and F). Genetic distances between species used as benchmarks (M. caliginosus-chrysomelas and N. apicalis-squamipes) were 7.48% and 7.52%, respectively.
Kimura 2-parameter (Kimura 1980) genetic distances calculated with MEGA4 (Tamura et al. 2007) from the cytochrome-b gene showing average percent (%) differences with standard deviations between and among groups.
Average % difference ± SD
Between Melanomys caliginosus and
7.48 ± 0.15
Within M. caliginosus
1.8 ± 1.29
Within M. chrysomelas
Between Nectomys apicalis and
7.52 ± 0.30
Between clades I and II
11.30 ± 0.80
Between clades I and E
11.37 ± 0.96
Between clades I and F
12.04 ± 0.99
Within clade A
0.65 ± 0.13
Within clade B
0.80 ± 0.14
Within clade C
2.06 ± 0.27
Within clade D
1.28 ± 0.17
Between clades A and B
6.05 ± 0.65
Between clades C and D
4.41 ± 0.49
Between clades C and E
7.20 ± 0.74
Between clades D and E
6.53 ± 0.16
Between clades C and F
11.93 ± 0.96
Between clades D and F
11.69 ± 0.98
Between clades E and F
12.06 ± 0.11
Comparision of karyotypes from 23 individuals from Mexico (eastern and western) and Honduras revealed 2n = 56 and from 13 individuals FN = 56–58. No detectable differences in karyotypes were identified respective to geographic origin or taxonomic assignment. In karyotyped specimens from both Mexico and Honduras the X chromosome was either acrocentric or subtelocentric. However, individuals collected in central Honduras had a much more pronounced arm in their subtelocentric X.
The DNA sequences obtained from the Cytb, Rbp3, and Adh1-I2 data sets produced somewhat similar topologies. O. palustris and O. couesi formed reciprocally monophyletic groups in the Cytb and Rbp3 analyses and supported previous hypotheses for species-level designations of these 2 taxa (Benson and Gehlbach 1979; Musser and Carleton 2005; Schmidt and Engstrom 1994; Weksler 2003, 2006; Weksler et al. 2006). Unfortunately, the nuclear markers did not exhibit sufficient genetic variation to provide resolution of most of the clades observed in the Cytb phylogeny. This was expected based on the number of phylogenetic ally informative characters in the different data sets (Cytb—299, Rbp3—31, and Adh1-I2—40). Additionally, all 3 markers showed gene tree heterogeneity, which would have led to inappropriate violations of the assumptions under typical concatenation. However, the BEST analysis was performed to account for this gene tree heterogeneity, branch length heterogeneity, and deep coalescence (Brito and Edwards 2009; Liu 2008; Liu et al. 2008) and supported the biogeographic subdivisions described below. In addition to the reciprocal monophyly, the genetic distance between O. couesi and palustris (11.3%), coupled with the apparent lack of evidence of hybridization or introgression in the small zone of sympatry (Schmidt and Engstrom 1994), provides additional support for recognition of 2 species. However, values of genetic divergence between minor clades and strong support for those clades within palustris and couesi suggest that current taxonomic arrangements should be reevaluated (see Genetic Species Concept— Baker and Bradley 2006). These taxonomic considerations are discussed below.
In the Cytb analyses (Fig. 2) and the BEST analysis (Fig. 5), O. palustris formed 2 unique genetic groups, with the 1st containing the subspecies coloratus, palustris, sanibeli, and planirostris (clade A) and the 2nd containing the subspecies texensis and western individuals of palustris (clade B). The genetic difference between clade A and B (6.05%) is comparable to those between currently recognized species within Melanomys and Nectomys (7.48% and 7.52%), arguing for the possible existence of 2 genetic species following the criteria outlined by Baker and Bradley (2006). Additionally, a sample from northwestern Alabama (MSB 81543, clade A) and a sample from eastern Mississippi (MSB81544, clade B), collected less than 97 km from each other, were placed in different clades and differed by a sequence divergence of 6.2% indicating a major genetic subdivision over a relatively small geographic distance.
Two taxonomic options are available for taxa in clades A and B. First, they can be recognized as a single species (O. palustris), which would follow the current taxonomy. The 2nd option would be to recognize the 2 clades as separate species. Under this scenario, clades A and B correspond to taxonomic entities recognized by Goldman (1918). The type specimen of O. palustris is from New Jersey (Harlan 1837) and presumably is represented by members of clade A, providing an appropriate name for specimens included in clade A. If members of clade B represent a species distinct from O. palustris, then O. texensis (Allen 1894) is the only available name.
Other lines of evidence in addition to the genetic data provided herein support the recognition of 2 species (palustris and texensis). First, Goldman (1918) initially described texensis based on it having a narrower skull than other subspecies of palustris. Second, Humphrey and Setzer (1989) were able to assign texensis to populations unique from the other specimens sampled based on a number of cranial measurements. Third, a phenetic analysis of allele frequency data (Schmidt and Engstrom 1994) depicted 2 well-differentiated clades (Texas and Georgia, respectively) with nearly fixed differences at 3 loci (SOD2 [superoxide dismutase EC#22.214.171.124], IDH1 [isocitrate dehydrogenase EC#126.96.36.199], and ADA [adenosine deaminase EC#188.8.131.52]). Although their sampling scheme was not designed to examine geographic variation, an eastern and western subdivision is apparent in the allozyme data.
The Apalachicola River has been suggested as a barrier to gene flow in other vertebrate organisms such as salamanders and pocket gophers (Avise 1992; Sudman et al. 2006) and initially was thought by the authors to serve a similar purpose in respect to Oryzomys. However, samples from both sides of the river (ANNERR and SJBP in Appendix I) were included in the taxonomic sampling, and both populations grouped in clade A. Although no obvious geographical barrier is apparent, potential geographic barriers exist in eastern Mississippi and northwestern Alabama and include the southeastern reaches of the Appalachian Mountains and the Tombigbee-Alabama and Tennessee rivers. Alternatively, the 2 species may represent populations that diverged in Pleistocene glacial refugia and have recently reexpanded their ranges. More detailed studies are required to determine the geographic limits of the 2 species and whether they hybridize. Another observation depicted by the genetic data is that the samples from eastern Mississippi and Tennessee extend the range limit of texensis east and north of that proposed by Hall (1981) by approximately 150 km, which illustrates the need for further sampling to determine the degree of genetic structure in the subspecies of the eastern United States.
Similar to the situation described for O. palustris, specimens corresponding to O. couesi (clade II) formed wellsupported clades (C–F). Additionally, genetic distances between clades C, D, E, and F supported the recognition of 4 unique genetic phylogroups within clade II. Four potential taxonomic options are available for couesi, resulting in the recognition of 1, 2, 3, or 4 species. Under the 1st option members of clades C, D, E, and F would be included in a single species (O. couesi) that ranges from southern Texas to northern Colombia. The 2nd option would be to recognize members of clade F (specimens from Costa Rica) as a species and retain members of clades C, D, and E under couesi. The 3rd option would be to recognize members of clades E (Panama) and F (Costa Rica) as separate species and restrict members of clades C and D to couesi. A 4th option would be to recognize members of clades C, E, and F as 3 separate species and restrict members of clade D to couesi.
Clade C was composed of specimens from western and southern Mexico and El Salvador, whereas clade D contained specimens from southern Texas, the eastern coast of Mexico, Honduras, and Nicaragua (including the type locality of couesi). The genetic distance between clades C and D (4.5%) was less than distances reported in other species comparisons. Additionally, no fixed karyotypic difference was detectable between individuals from clade C and D; rather, some individuals from the same locality had different FNs. However, considering that clade C historically was separated from clade D (Merriam 1901) based on a more massive skull, with stronger zygomata, more defined superciliary ridges, and heavier molars; unique hantaviruses correspond to the 2 groups (Chu et al. 2008; Milazzo et al. 2006); the 2 groups are separated by numerous mountain ranges (Sierra Madre Occidental, Sierra Madre Oriental, Sierra Madre de Chiapas, and Meseta Central de Chiapas) that serve as geographic barriers to gene flow; and sequence data and allozyme data show similar separation between the 2 clades, it appears that members of clades C and D may represent unique entities. Additionally, the lack of karyological differences does not preclude the presence of 2 species because also no consistent karyological difference is found between couesi and palustris (Benson and Gehlbach 1979; Bradley and Ensink 1987; Burton et al. 1987; Gardner and Patton 1976; Haiduk et al. 1979; Hsu and Benirschke 1969;Schmidt and Engstrom 1994), which suggests a conserved karyology (2n = 56, FN = 56–60) for the genus. The type locality for O. couesi (Alston 1876) is Coban, Guatemala, which was represented by specimens included in clade D. If clade C is recognized as a species, the appropriate name would be O. mexicanus (Allen 1897), the type locality for which is Tonila, Jalisco, Mexico, and geographically corresponds to samples in clade C.
All of the phylogenetic analyses as well as Cytb genetic distances (6.58% and 7.14% respectively) indicate that clade E might represent a species distinct from O. couesi and O. mexicanus. However, assignment of specimens comprising clade E is unclear because there are 2 junior synonyms of O. couesi in Panama that could correspond to this clade. Goldman (1912) described O. gatunensis from the Canal Zone, Panama, and Bole (1937) described O. azuerensis from Paracoté, Panama. Samples examined herein are from near the type locality of O. azuerensis. However, given the close proximity of the type locality for O. gatunensis (150 km to the northeast), and because of unavailability of specimens representing gatunensis (which has priority over azuerensis), it is premature to assign specimens to either gatunensis or azuerensis. Also, azuerensis and gatunensis could be separate species; consequently, until specimens of gatunensis can be examined, it is best to refer specimens conservatively from clade E to Oryzomys species 1.
Clade F contained 2 samples from Costa Rica that formed a well-supported monophyletic clade separate from other samples of O. couesi. DNA sequences from the Cytb gene imply that the 2 specimens from Costa Rica are as distinct from the remainder of O. couesi (11.93%) as O. couesi is from O. palustris (11.30%). This also was depicted in the BEST analysis where individuals from clade F were not affiliated with either the other members of the couesi grouping (clade II) or the members of the palustris group (clade I). Two names are available for taxa from this region (O. dimidiatus and O. couesi richmondii), but neither could be examined in this study. The type localities for O. dimidiatus and O. couesi richmondii are approximately 160 km north of the locality from which specimens examined herein were collected, and it is not unreasonable to expect O. dimidiatus or O. c. richmondii to have a distribution that includes the specimens examined herein. However, morphological comparisons of specimens examined with a specimen of O. dimidiatus imply that 2 distinct taxa are represented. Further investigations are needed to resolve the morphology and genetic relationships of the Costa Rican specimens (clade F), O. c. richmondi and O. dimidiatus. Until samples of O. dimidiatus and O. c. richmondii can be examined, insufficient data are available to determine the most appropriate name for the specimens in clade F; consequently, it is referred to as Oryzomys species 2.
Examination of genetic relationships among taxa currently assigned to O. palustris and O. couesi suggest the presence of multiple taxa (Figs. 1 and 5). Geographic differentiation, genetic divergence, and cranial differences between eastern and western forms of palustris and eastern, western, and southern forms of couesi correspond to historically named taxa (Bole 1937; Goldman 1918; Merriam 1901). It is tentatively suggested that these differences collectively may represent an additional 4 species; if this assumption is correct the palustris complex would include 6 species in total.
The taxonomic history of this group is complex, and a number of approaches will be required to resolve it. A morphological approach to differences between O. texensis and O. palustris, and between O. couesi, O. mexicanus, Oryzomys species 1, and Oryzomys species 2, is necessary. Although these species appear to be difficult to distinguish morphologically, the molecular framework provided herein can allow for a more detailed morphological investigation. Additionally, increased sampling, for both molecular and morphological work, is needed throughout Alabama and Mississippi, central Mexico, Costa Rica, and Panama to better understand the species boundaries. Especially important will be molecular data from the type localities of O. m. regillus, O. m. crinitus, O. m. Iambi, O. m. bulleri, and O. m. albiventer, which are required to explore hypotheses presented by Carleton and Arroyo-Cabrales (2009). Samples from the type localities of O. gatunensis and O. dimidiatus and O. c. richardsoni and O. c. richmondii art necessary for determining the most appropriate names for Oryzomys species 1 and Oryzomys species 2. Furthermore, examinations of populations from all 6 species using faster-evolving genes or markers (i.e., control region or microsatellites) are necessary to test the validity of named phylogroups (subspecies) and the biogeographic boundaries of these groups. Finally, recent specimens, with associated tissues, from the more obscure taxa (O. penninsulae, O. nelsoni, and O. gorgasi) would be valuable additions to understanding the evolution and biogeography of this group and the genus as a whole.
We thank M. R. Mauldin, R. R. Chambers, D. D. Henson, R. N. Platt, A. O. Stallings, R. K. Baker, and L. D. Porr for laboratory assistance; M. Gaines, S. Beckmann, N. Dappen, D. Henson, M. Massanet, J. Nielsen, J. O’Connor, T. Planton, R. Prendiville, R. Rose, and R. Stevens for field assistance; and R. J. Baker for karyological equipment and expertise. In addition, we thank R. R. Chambers, L. D. Densmore, C. F. Fulhorst, D. D. Henson, R. N. Platt, R. E. Strauss, C. W. Thompson, M. Carleton, R. Timm, and 2 anonymous reviewers for comments on earlier versions of this manuscript, and M. Pinto, H. Mantilla, and M. Venegas for Spanish translations. Specimens were obtained from field collections by J. Hanson, J. Indorf, R. Timm, and the 2001 and 2004 Field Methods classes at Texas Tech University or from the following collections: Abilene Christian University Natural History Collections; Angelo State Natural History Collections; Monte L. Bean Life Science Museum, Brigham Young University; Carnegie Museum of Natural History; Collecion de Mamiferos CEAMISH, Universidad Autónoma del Estado de Morelos; Louisiana State University Natural History Museum; Museum of Southwestern Biology; Museum of Texas Tech University; Museum of Vertebrate Zoology; Royal Ontario Museum; and the University of Kansas Museum of Natural History. Financial support for fieldwork was provided by a grant from James Sowell to Texas Tech University, the Department of Biological Sciences at Texas Tech University, and the University of Miami Department of Biology Savage Fund and Kushlan Fund (to JLI). Laboratory work was funded, in part, by the National Institutes of Health (grant R01 AI-41435) to RDB.
Specimens examined.—Specimens included in this study are listed below by locality. Specific identification numbers (museum numbers or collector numbers) and GenBank accession numbers are listed in parentheses, respectively. Specimens with 1 GenBank accession number only have Cytb sequences; multiple GenBank accession numbers are separated by slashes (/) with Cytb sequences to the left, Rbp3 sequences in the middle, and Adh1-I2 sequences to the right. Karyotyped individuals are indicated with an asterisk following the museum number. Museum acronyms follow Hafner et al. (1997), and abbreviations for identification numbers are as follows: Abilene Christian University Natural History Collections (ACUNHC); Angelo State Natural History Collections (ASNHC); Monte L. Bean Life Science Museum, Brigham Young University (BYU); Carnegie Museum of Natural History (CM); Collecion de Mamiferos CEAM-ISH, Universidad Autónoma del Estado de Morelos (CMC); Louisiana State University, Museum of Natural History (LSUMZ); Louisiana State University, Museum of Natural History, tissue collection (LSUM); Museum of Southwestern Biology (MSB); Museum of Texas Tech University (TTU); Museum of Texas Tech University, tissue collection (TK; vouchers unavailable); Museum of Vertebrate Zoology (MVZ); Royal Ontario Museum (ROM); Robert M. Timm (RMT; vouchers at University of Kansas Museum of Natural History [KU]); and Jane Indorf (SBI, EVGL, SCVA, ANNER, KPSP, SJBP, LPI; no voucher specimens are available for these individuals). All localities are in the United States unless specified otherwise.
Aegialomys xanthaeolus.—ECUADOR: Guayas; Bosque Protector Cerro Blanco, Centro de Visitantes (TTU103309, EU340015/ GQ178247/GQ178261).
Holochilus sciureus.—PERU: Loreto; Iquitos, Zona Marina (TTU75634, EU074631/DQ227456/EU273418).
Melanomys chrysomelas.—NICARAGUA: Region Autonomous Atlantica Norte; El Balsamo (TTU100324, EU340017/EU648996/ EU649053).
Melanomys caliginosus.—ECUADOR: El Oro; Zaruma, Cerro Urcu (TTU102727, EU340019).
Nectomys apicalis.—PERU: Kiteni; Rio Urubamba (MVZ166700, EU340013).
Nectomys squamipes.—PARAGUAY: Paraguari; Parque Nacional Ybycui (TTU108150, EU074634/EU649004/EU273419).
Oryzomys couesi aquaticus.—MEXICO: Tamaulipas; 2 km W Ciudad Madero (ASNHC3378, EU074659/GQ178248/GQ178262); Texas; Cameron County, Brownsville, Resaca de la Palma State Park (TTU77220, DQ370034/EU649019/GQ178263; TTU77221, EU074662/GQ178249/EU273425); Texas; Hildago County, 6.4 km S, 4.8 km W Mission (ACUNHC 248, EU074665).
Oryzomys couesi couesi.—GUATEMALA: Baja Verapaz; 3.2 kmi W Puruhla (KU65179, FJ360633); El Peten; El Remate (ROM99609, FJ971266); El Peten; 10 km N Tikal (ROM99393, FJ971268; ROM99439, FJ971267); HONDURAS: Atlantida; Jardin Botanico Lancetilla (TTU103830, EU074666; TTU103837*, EU074667; TTU84450*; TTU84451*; TTU84375*; TTU84476*; TTU84477*; TTU103837*; TTU103838*); Colon; La Ceiba (TTU104083*; TTU104087*; TTU104089*); Cortes; La Guama (TTU104418*); Olancho; 4 km E Catacamas, Escuela de Sembrador (TTU84671, DQ185384; TTU84697*, DQ185383/EU273426/GQ178264; TTU84692*; TTU84694*).
Oryzomys couesi cozumelae.—MEXICO: Quintana Roo; Isla Cozumel, 30 km SE San Miguel (ASNHC1646, EU074658).
Oryzomys couesi richardsoni.—NICARAGUA: Granada (ROM112231, FJ971265; R0M112237, FJ971264); Region Atlántica Autonoma Sur; Bluefields (MVZ 140661, FJ971275); El Paraisito (TTU100593, EU074663/EU649018/EU273427; TTU100597, EU074664).
Oryzomys couesi teapensis.—MEXICO: Campeche; 10 km S Candelaria (ASNHC7144, EU074657); Chiapas; 9.5 km S Palenque (ASNHC1637, EU074660); Oaxaca; 3.6 km ENE Guichicovi (BYU32608, FJ971274); Tabasco; 6 km S, 2 km E El Triunfo (ASNHC7191, EU074656); 5.5 km E Emiliano Zapata (BYU32610, FJ971272; CMC2227, FJ971271).
Oryzomys mexicanus bulleri.—MEXICO: Nayarit; 2 km S, 36 km E San Blas (ASNHC3371, EU074651).
Oryzomys mexicanus mexicanus.—MEXICO: Colima; 8.5 km W Quesería (ASNHC3349, EU074652); Playa de Oro (TK126531, EU074645; TK126521, EU074646/EU273424/GQ178268; TK126728, GQ178245/GQ178250/GQ178269; TK126763, GQ178246/GQ178251/GQ178270); Jalisco; Chamela (TTU45123, EU074655; TTU37749, EU074653); Michoacan; Hildago, 2 km NW Presa Pucuato (TK45816, EU074654/EU273422/GQ178266); Oaxaca; Las Minas (TK93218, DQ185385; TTU82862*, DQ185386/ EU273423/GQ178267).
Oryzomys mexicanus zygomaticus.—EL SALVADOR: La Paz; 4.8 km NW San Luis Talpa (CM111499, FJ360634; CM111503, FJ360635; CM111507, FJ360636); MEXICO: Chiapas; Mapastepec, Tutuan (TTU108149, EU074648/GQ178252/GQ178270; TTU104675, EU074649/GQ178254/GQ178273; TTU104688*; TTU104636*; TTU104689*; TTU104690*); Chiapas; Ocozocoatla (TTU104634, EU074650/GQ178253/GQ178272; TTU104621*).
Oryzomys palustris coloratus.—Florida; Miami-Dade County, Everglades National Park, near Homestead, Rock Reef Pass (EVGL01, FJ974114; EVGL02, FJ974115; EVGL05, GQ148811/ EU649022/EU074639; EVGL06, EU074639/EU273432/GQ178280).
Oryzomys palustris natator.—Florida; Okeechobee County; Kissimmee Prairie Preserve State Park (KPSP01, FJ974108; KPSP02, FJ974109).
Oryzomys palustris palustris.—Alabama; Colbert County, Cherokee, Natchez Trace Parkway Mile Marker 312.4 (MSB81543, EU074636/GQ178256/GQ178277); Tallapoosa County, Horshoe Bend National Military Park (MSB81643, FJ974112; MSB81644, FJ974113); Florida; Franklin County; Apalachiola Bay, Apalachiola National Estuarine Research Reserve (ANNER01, FJ974106; ANNER02, FJ974107); Gulf County, Port St. Joe (SJBP01, FJ974110; SJBP02, FJ974111); South Carolina; Richland County, Congaree Swamp National Monument (MSB74956, EU074637/GQ178257/ GQ178278); Virginia; Norfolk County, 0.8 km E US Hwy 17 or 9.7 km from Virginia/North Carolina border (SCVA15, EU074640/ EU273433/GQ178279).
Oryzomys palustris planirostris.—Florida; Lee County, Little Pine Island (LPI01, FJ974116; LPI02, FJ974117).
Oryzomys palustris sanibeli.—Florida; Lee County, J. N. “Ding” Darling National Wildlife Refuge Sanibel Island (SBI02, EU074638/ GQ178259/GQ178281; SBI01, FJ974118; SBI03, FJ974119).
Oryzomys texensis.—MEXICO: Tamaulipas; Matamoros (ASNHC3432, FJ974120; ASNHC3439, FJ974122; ASNHC3442, FJ974121); Arkansas; Crittenden County, West Memphis (TTU82963, FJ974129); Louisiana; Cameron Parish, 13.5 km SE Rockefeller Refuge Headquarters (LSUMZ28985, FJ971263); 6 km SE Rockefeller Refuge Headquarters (LSUMZ34173, FJ971261); 13.7 km NE Rockefeller Refuge Headquarters (LSUMZ34171, FJ971255); Rockefeller Refuge (LSUM8428, FJ974123; LSUM8433, FJ974124; LSUM8436, FJ974125); East Baton Rouge Parish, Baton Rouge (LSUMZ34179); Iberville Parish, 28 km S LSU campus (LSUMZ23854, FJ971257); Livingston Parish, 4.02 km S, 4.02 km W Livingston (LSUMZ28488, FJ971259); Terrebonne Parish, 8.05 km N 3.22 W Cocodrie (LSUMZ28688, FJ971256; LSUM28689, FJ971260); Mississippi; Jefferson Davis County, 8 km SW Mount Olive (LSUMZ28494, FJ971258); Lee County, Tupelo, Natchez Trace Parkway Mile Marker 261.8 (MSB81544, EU074643); Oklahoma; Okmulgee County, 4.8 km E Dewar, Eufaula Wildlife Management Area (TTU62980, DQ37032/EU273434/GQ178274); Tennessee; Shelby County, 8.02 km N Memphis (TTU79152, FJ974126; TTU79153, FJ974127); Edward J. Meeman Biological Station (TTU79154, FJ974128); Texas; Brazoria County, Peach Point Wildlife Management Area (TTU108151, DQ370033/GQ178255/GQ178276); Calhoun County, Guadalupe Delta Wildlife Management Area (TTU75177, DQ370031); Freestone County. Richland Creek Wildlife Management Area (TTU75311, EU074642/AY163623/GQ178275); Galveston County, Texas City Virginia Point (TTU82860, EU074644; TTU82920,DQ185382/DQ207949/EU273431).
Oryzomys species 1.—PANAMA: Veraguas; Montijo, Corregimiento de Arenas, Portobel, Cerro Hoya Lower (MSB91818, EU074669/EU273428/GQ178282; MSB91812, EU074668/ EU649020/EU273429).
Oryzomys species 2.—COSTA RICA: Refugio Nacional de Vida Silvestre Mixto Maquenque (RMT4689, EU074670/EU273430/ GQ178283; RMT4695, EU074671/GQ178260/GQ178284).
. 2006. Intron 2 (Adh1 -I2) of the alcohol dehydrogenase gene: a potential nuclear DNA phylogenetic marker for mammals. Occasional Papersx, Museum of Texas Tech UniversityBulletin of the American Museum of Natural History 256:1–16.
1857. Mammals: general report upon the zoology of the several Pacific railroad routes. Vol. 8, part 1. Pp. 1–757 in Reports of explorations and surveys to ascertain the most practicable and economical route for a railroad from the Mississippi River to the Pacific Ocean. Washington, D.C., Senate Executive Document 78.
. 1976. Karyotypic variation in oryzomyine rodents (Cricetinae) with comments on chromosomal evolution in the neotropical cricetine complex. Occasional Papers of the Museum of Zoology, Louisiana State University 49:1–48.
. 2005. Superfamily Muroidea. Pp. 894–1531 in Mammal species of the world: a taxonomic and geographic reference (WilsonD. E., ReederD. M., eds.). 3rd ed. Johns Hopkins University Press, Baltimore, Maryland.
. 1992. A molecular perspective on mammalian evolution from the gene encoding interphotoreceptor retinoid binding protein with convincing evidence for bat monophyly. Molecular Phylogenetics and Evolution 1:148–160.
2006. Phylogenetic relationships of oryzomine rodents (Muroidea: Sigmodontinae): separate and combined analyses of morphological and molecular data. Bulletin of the American Museum of Natural History 296:1–149.