OUP user menu

The Phylogenetic Position of the Rodent Genus Typhlomys and the Geographic Origin of Muroidea

Sharon A. Jansa, Thomas C. Giarla, Burton K. Lim
DOI: http://dx.doi.org/10.1644/08-MAMM-A-318.1 1083-1094 First published online: 15 October 2009

Abstract

In this report, we provide the 1st phylogenetic assessment of the evolutionary relationships of Platacanthomyidae. This enigmatic family of rodents comprises 2 genera, Platacanthomys and Typhlomys, that are distributed disjunctly in western India and southern Asia, respectively. We analyze sequence data from the nuclear genes encoding the interphotoreceptor retinoid-binding protein (IRBP) and the growth-hormone receptor (GHR) to address the relationships of Typhlomys cinereus to other rodents, with a particular focus on testing whether its evolutionary affinities lie with glirids or muroids. Our results provide compelling evidence that Typhlomys is a muroid and that it represents the earliest split within this clade. We use the resulting phylogeny to explore the origin and evolutionary history of muroid rodents. We conclude that both Myodonta and Muroidea originated in Eurasia and that the 3 earliest divergences within Muroidea were restricted to the Eurasian supercontinent. Moreover, our analyses support the view that global muroid diversity resulted from independent radiations in separate continental regions.

Key words
  • growth-hormone receptor gene (GHR)
  • interphotoreceptor retinoid-binding protein gene (IRBP)
  • Muroidea
  • phylogeny
  • Platacanthomyidae
  • Rodentia
  • Typhlomys
  • systematics

With more than 1,500 species and a nearly global distribution, muroid rodents overwhelmingly represent the most diverse and geographically widespread subordinal clade of mammals (Wilson and Reeder 2005). Historically, muroid systematics has been controversial, and a number of different classifications have been proposed based almost entirely on varied interpretations of morphological traits (Alston 1876; Carleton and Musser 1984; Ellerman 1940, 1941, 1949; Musser and Carleton 2005; Simpson 1945; Thomas 1896; Tullberg 1899; Winge 1941; see Table 1 for representative classifications). Recent molecular phylogenetic studies of muroids have documented monophyly of the group to the extent sampled, and have identified 5 well-supported clades (Jansa and Weksler 2004; Michaux and Catzeflis 2000; Michaux et al. 2001; Steppan et al. 2004). A recent classification (Musser and Carleton 2005) recognized these 5 groups as the families Spalacidae, Calomyscidae, Nesomyidae, Cricetidae, and Muridae. However, Musser and Carleton (2005) recognize a 6th, family-level grouping within muroids—Platacanthomyidae—that has been conspicuously absent from any phylogenetic study to date.

View this table:
Table 1

Historical and current classifications of myomorph rodents depicting the varied taxonomic positions of platacanthomyids.

Alston (1876)Thomas (1896)Miller and Gidley (1918)Simpson (1945)Chaline et al. (1977)Musser and Carleton (2005)
MyomorphaMyomorphaMuroideaMyomorphaMuroideaMuroidea
MyoxidaeGliridaeMuscardinidaeMuroideaCricetidaePlatacanthomyidae
LophiomyidaeGlirinaeIschyromyidaeCricetidaeHesperomyinaeSpalacidae
MuridaePlatacanthomyinaeCricetidaeSpalacidaeSpalacinaeMyospalacinae
SminthinaeMuridaePlatacanthomyidaeRhizomyidaeMyospalacinaeRhizomyinae
HydromyinaeHydromyinaeRhizomyidaeMuridaePlatacanthomyinaeSpalacinae
PlatacanthomyinaeRhynchomyinaeSpalacidaeMurinaeNesomyidaeTachyoryctinae
GerbillinaePhloeomyinaeMuridaeDendromurinaeNesomyinaeCalomyscidae
PhloeomyinaeGerbillinaeOtomyinaeOtomyinaeNesomyidae
DendromyinaeOtomyinaePhloeomyinaeRhizomyidaeCricetomyinae
CricetinaeDendromyinaeRhynchomyinaeRhizomyinaeDelanymyinae
MurinaeMurinaeHydromyinaeTachyoryctinaeDendromurinae
ArvicolinaeLophiomyinaeGliroideaGerbillidaeMystromyinae
SiphneinaeSigmodontinaeGliridaeGerbillinaeNesomyinae
SpalacidaeNeotominaePlatacanthomyidaeTaterillinaePetromyscinae
SpalacinaeMicrotinaeDipodoideaArvicolidaeCricetidae
BathyerginaeSpalacidaeZapodidaeDendromuridaeArvicolinae
GeomyidaeRhizomyinaeDipodidaeCricetomyidaeCricetinae
GeomyinaeSpalacinaeMuridaeLophiomyinae
HeteromyinaeMurinaeNeotominae
TheridomyidaeHydromyinaeSigmodontinae
DipodidaeTylomyinae
Muridae
Leimacomyinae
Deomyinae
Gerbillinae
Murinae
Otomyinae

The family Platacanthomyidae comprises 2 extant, monotypic genera, Platacanthomys, which has a restricted distribution in highland regions of southwestern India, and Typhlomys, which is known from montane regions of southern China and northwestern Vietnam. These 2 genera are morphologically quite different from each other (Carleton and Musser 1984), but based on anatomical comparisons, there seems little doubt that they are more closely related to each other than either is to any other rodent. Since the original description of Platacanthomys (Blyth 1859), platacanthomyids have been associated taxonomically either with dormice of the family Gliridae (Ellerman 1940; Simpson 1945; Thomas 1896; Winge 1941) or with muroids (Alston 1876; Chaline et al. 1977; Miller and Gidley 1918; Musser and Carleton 2005; Table 1). In classifications that adopt the latter arrangement, platacanthomyids have been included as an independent lineage (Miller and Gidley 1918; Musser and Carleton 2005), as closer to cricetids (Chaline et al. 1977), or as having particular affinities with murines (Peters 1865). In an intriguing departure, Ellerman (1940, 1949) noted similarities between molars of Platacanthomys and the nesomyine genus Gymnuromys, thus suggesting an alliance between platacanthomyids and the native rodents of Madagascar.

To date, the systematic debate regarding platacanthomyids has centered on whether the number of cheek teeth in each jaw quadrant (4 in glirids, 3 in platacanthomyids, and 3 or fewer in muroids) or apparent similarities in molar occlusal pattern (long reentrant folds in glirids, platacanthomyids, and the muroid Gymnuromys, among others) are phylogenetically relevant characters. Additional anatomical differences between platacanthomyids and glirids, including structure of the bullae and dentary (Carleton and Musser 1984), have been used to argue that platacanthomyids should be allied with muroids and not with glirids. Despite these morphological comparisons, none of the competing hypotheses regarding platacanthomyid relationships has ever been tested, because neither genus has been included in any phylogenetic study to date. Thus, the evolutionary relationships of this enigmatic group of rodents remain effectively unknown.

Platacanthomyid specimens are rare in museum collections: only 21 specimens are recorded in the most recent online catalogs of North American collections (available at http://www.manisnet.org), and these were all collected before routine preservation of tissues for genetic work (the most recent dates from 1929). A recently collected specimen of Typhlomys cinereus included fresh tissues that provide the basis for the 1st assessment of platacanthomyid systematics using molecular data. In this report, we use sequence data from the nuclear genes encoding the interphotoreceptor retinoid-binding protein (IRBP) and the growth-hormone receptor (GHR) to address the phylogenetic relationships of Typhlomys relative to other rodents, with a particular focus on testing whether its evolutionary affinities lie with glirids or muroids. We then use the resulting phylogeny to explore the origin and evolutionary history of muroid rodents.

Materials and Methods

Taxon and gene sampling.—The voucher specimen of Typhlomys (ROM 118593, field number F47816) consists of the skin, skull, and postcranial skeleton of a young adult mule collected at Shuipu Village, Yuping Township, Libo County, Guizhou Province, People's Republic of China (25°28′59″N, 107°52′54″E), by B. K. Lim and J. L. Eger on 9 April 2007. Robert S. Voss (American Museum of Natural History) identified the specimen as T. cinereus daloushanensis based on qualitative external and craniodental characters.

In addition to this specimen, our taxon sample includes representatives of all recognized families of muroid rodents and all but 2 of the 21 recognized subfamilies (Musser and Carleton 2005). To test possible relationships between dormice of the family Gliridae and Typhlomys, we included exemplars of all glirid subfamilies, including 6 of the 9 recognized genera. To provide a more inclusive picture of rodent phylogeny, we also included representatives from the rodent families Aplodontiidae, Castoridae, Chinchillidae, Dipodidae, Geomyidae, Heteromyidae, and Sciuridae. Sequences from the lagomorph genus Lepus were used to root our rodent phytogenies.

For this study, we sequenced 1.2 kb of exon 1 from the IRBP gene and just under 1.0 kb of exon 10 of the GHR gene from T cinereus. Most of the remaining IRBP sequences were obtained during our previous work on muroid systematics (Jansa and Weksler 2004; Jansa et al. 2006), and additional sequences from this gene and the majority of GHR sequences were downloaded from GenBank (Adkins et al. 2001, 2003; DeBry 2003; DeBry and Sagel 2001; Huchon et al. 2002; Steppan et al. 2004). To maximize taxonomic overlap between the 2 gene data sets, we also sequenced the same fragment of GHR from an additional 11 species (Appendix I ). Our final data matrix of IRBP sequences consisted of 62 species coded for 1,257 base pairs (bp), and our final GHR data set consisted of 58 species coded for 954 bp. Whenever possible, we used sequences of the 2 genes from the same species, but in 17 instances we used sequences from 2 different species of the same genus, and in 4 cases we used 2 different genera from the same family. For these 21 composite taxa, monophyly of the higher-level grouping is not in question. All specimen voucher information and GenBank accession numbers are given in Appendix I.

DNA sequencing.—Whole genomic DNA was extracted from tissue samples using a DNeasy Blood and Tissue Kit (Qiagen, Inc., Valencia, California). A 1.2-kb fragment of exon 1 from the IRBP gene was amplified using primers IRBPA and IRBPB (Appendix II ). This amplification product was then used in a 2nd round of polymerase chain reaction to generate fragments of a suitable size for sequencing using IRBPA paired with IRBPF and IRBPE paired with IRBPB. In addition, a 1.0-kb portion of exon 10 from the GHR gene was amplified from genomic DNA using primers GHRF1 and GHRendAlt. This polymerase chain reaction product was reamplified using GHRF1 paired with GHR750R and GHRF50 paired with GHRendAlt. The resulting polymerase chain reaction products were purified using either a QiaQuick polymerase chain reaction purification kit (Qiagen, Inc.) or Exonuclease I and Shrimp Alkaline Phosphatase (Hanke and Wink 1994). These products were sequenced in both directions using amplification primers and dye-terminator chemistry (BigDye version 3.1 Cycle Sequencing Kit; Applied Biosystems Inc., Foster City, California) on an ABI 3730 automated sequencer (Applied Biosystems Inc.). Sequences were edited and compiled using Sequencher 4.7 (GeneCodes Inc., Ann Arbor, Michigan). Base-calling ambiguities between strands were resolved by choosing the call with the clearest trace; any heterozygous bases were given the appropriate International Union of Pure and Applied Chemistry ambiguity code.

Phylogenetic and biogeographic analyses.—Sequence alignments for GHR and IRBP were produced using CLUSTALW (Thompson et al. 1994) and manually adjusted with reference to the amino acid sequences using MacClade version 4.08 (Maddison and Maddison 2005). We performed phylogenetic analyses using maximum-parsimony, maximum-likelihood, and Bayesian inference approaches. Maximum-parsimony analyses were conducted on each gene separately and in combination using heuristic search algorithms implemented in PAUP* 4.0b10 (Swofford 2002) with all characters unordered and equally weighted. Heuristic searches were performed with 1,000 repetitions of random taxon addition followed by tree-bisection-reconnection branch swapping. Nodal support values for the maximum-parsimony analysis were estimated using nonparametric bootstrapping, with heuristic searches (10 random-addition replicates, tree-bisection-reconnection branch swapping) on 1,000 pseudoreplicate data sets.

Maximum-likelihood analyses were performed using genetic-algorithm searches as implemented in GARLI version 0.95 (Zwickl 2006), and Bayesian inference analyses were performed using MrBayes 3.1.2 (Ronquist and Huelsenbeck 2003). For each of these model-based analyses, we 1st evaluated the fit of various models of nucleotide substitution for the IRBP and GHR data sets separately and in combination using the Akaike information criterion as implemented in Modeltest 3.6 (Posada and Crandall 1998). For the maximum-likelihood tree search, we specified the best-fit model in GARLI, and allowed substitution parameters to be estimated from the data. We employed default settings for the genetic algorithm, started the run with random trees, and set the run to automatically terminate after 10,000 generations of no improvement in the log-likelihood (lnL) score. All maximum-likelihood tree searches were performed independently 3 times to ensure convergence on a stable log-likelihood score and tree. To estimate nodal support values for the maximum-likelihood analysis, we performed 1,000 bootstrap replicates under the best-fit model in GARLI using tree-search settings as above.

For Bayesian inference analyses of each gene data set, we specified the best-fit model of nucleotide substitution in MrBayes and conducted 3,000,000 generations of Metropolis-coupled Markov chain Monte Carlo using 1 cold and 3 incrementally heated chains with trees sampled every 100 generations. For the combined-gene data set, we sampled 5,000,000 generations of Metropolis-coupled Markov chain Monte Carlo as above, allowed all substitution parameters to vary independently between the 2 genes, and allowed branch lengths to be estimated proportionally for each gene. We estimated the burn-in for each run by plotting the log-likelihood score per generation using the program Tracer version 1.3 (http://tree.bio.ed.ac.uk/software/tracer/). We discarded all trees prior to stationarity and used the remaining trees to estimate tree topology, nodal posterior probability values, parameter estimates, and branch lengths.

We tested alternative phylogenetic relationships for Typhlomys using the likelihood-based Shimodaira–Hasegawa test (Shimodaira and Hasegawa 1999). Specifically, we tested the following 2 hypotheses: whether the combined-gene data set could reject a topology where Typhlomys was the sister taxon to Gliridae, and whether this data set could reject Typhlomys as part of a clade with Nesomyinae (as proposed by Ellerman 1949). To do so, we executed the combined-gene data set in GARLI and searched for the most likely tree compatible with the constraint (either hypothesis 1 or 2 above) under a GTR+I+Γ model. We then performed a Shimodaira–Hase-gawa test in PAUP* using 1,000 RELL replicates allowing uniform model parameters to be estimated during the run.

To infer the geographic origin and biogeographic history of muroid rodents, we coded each of our terminal taxa as occurring in one or more of 4 continental regions: Eurasia, Africa, North America, or South America. We then optimized ancestral areas on the combined-gene maximum-likelihood tree using the parsimony criterion for ancestral state optimization of unordered characters as implemented in Mesquite 2.5 (Maddison and Maddison 2008).

Results

Phylogenetic results.—Most of the relationships among muroids depicted in Figs. 13 have been recovered in previous studies and have been discussed elsewhere (Jansa and Weksler 2004; Michaux and Catzeflis 2000; Michaux et al. 2001; Steppan et al. 2004). Here, we focus on novel results concerning the position of Typhlomys relative to other rodents. All 3 analyses (maximum parsimony, maximum likelihood, and Bayesian inference) of the IRBP and GHR data sets recovered Typhlomys nested within Myodonta (Figs. 1 and 2; tree statistics given in Appendix III ). The strongest possible bootstrap support for monophyly of Myodonta (node D on Fig. 3) provides compelling evidence that Typhlomys belongs in this clade. Within Myodonta, analyses of each gene consistently recover Typhlomys on the branch separating dipodids from the clade including all other muroid families (Spalacidae + Calomyscidae + Nesomyidae + Cricetidae + Muridae). For the IRBP data set, parsimony bootstrap support for the nodes determining this placement ranges from fairly low to very strong (69% for Typhlomys + remaining muroids; 95% for the clade including all muroids except Typhlomys), and likelihood bootstrap support is very strong for these 2 nodes (96% and 98%, respectively). For the GHR data set, parsimony bootstrap support values are high for these 2 nodes (83% and 98%, respectively), and likelihood bootstrap values are comparably strong (89% and 100%, respectively). Bayesian analysis of each gene recovers the 2 nodes flanking the position of Typhlomys with posterior probability of 1.0.

Fig. 1

The phylogenetic tree resulting from maximum-likelihood analysis of the interphotoreceptor retinoid-binding protein gene (IRBP) data set under its best-fit model. Nodal support from a parsimony bootstrap analysis, a maximum-likelihood bootstrap analysis, and a Bayesian analysis are indicated with shaded pie diagrams at nodes. For the maximum-parsimony and maximum-likelihood bootstrap analyses (MP and ML fractions of the circle, respectively), black indicates bootstrap values ≥ 75%, gray indicates bootstrap values between 50% and 75%, and white indicates bootstrap values ≤ 50%. For the Bayesian analysis (BPP; lower right one-third of circle), black indicates posterior probability values ≥ 0.95 and white indicates posterior probability values < 0.95. The tree is rooted with Lepus (not shown).

Fig. 2

The phylogenetic tree resulting from maximum-likelihood analysis of the growth-hormone receptor gene (GHR) data set under its best-fit model. The tree is rooted with Lepus (not shown). Conventions for indicating nodal support are described in the legend to Fig. 1.

Fig. 3

The phylogenetic tree resulting from a mixed-model maximum-likelihood analysis of the combined interphotoreceptor retinoid-binding protein gene and growth-hormone receptor gene (IRBP + GHR) data set. The tree is rooted with Lepus and Chinchilla (not shown). Circles at nodes reflect Bayesian posterior probability values, where black indicates values ≥ 0.95 and white indicates values < 0.95. Letters at selected nodes indicate higher-level groupings referred to in the text. The 4 grids show support for the indicated node from each of the 9 phylogenetic analyses performed. ML = maximum-likelihood bootstrap values; MP = maximum-parsimony bootstrap values; BPP = Bayesian posterior probability values.

Analyses of the combined-gene data set also recover Typhlomys on the branch between Dipodidae and the clade including all other muroid families (Fig. 3). The 2 nodes defining this relationship receive extremely strong support from the combined-gene data set (bootstrap values ≥ 99%; Bayesian nodal posterior probability = 1.0) in all but 1 analysis: parsimony analysis of this data set recovers Typhlomys + all other muroids in 87% of bootstrap replicates.

Although bootstrap support and nodal posterior probability values provide convincing evidence for the phylogenetic position of Typhlomys, we also evaluated 2 competing hypotheses concerning relationships of this taxon using likelihood-based Shimodaira–Hasegawa tests. The topology that unites Typhlomys with glirids is significantly less likely than the best tree based on the combined-gene data set (−2ΔlnL = 91.97, P = 0.001). Similarly, Ellerman's (1949) hypothesis that platacanthomyids might be allied with the nesomyine rodents of Madagascar is soundly rejected (−2ΔlnL = 195.25, P < 0.001).

Biogeography.—Twenty-eight steps were required to optimize the 4 biogeographic areas across the combined-gene, maximum-likelihood topology. Based on this analysis, the ancestral area for the crown clade Muroidea optimizes unambiguously as Eurasia (Fig. 4). Moreover, Eurasia optimizes unambiguously as the ancestral area for the 2 nodes flanking Muroidea (Myodonta and all muroids except Typhlomys [Platacanthomyidae]). The ancestral area for spalacids optimizes unambiguously as Eurasia, as does the node linking Calomyscidae with Nesomyidae + Cricetidae + Muridae (Fig. 4). Ancestral area optimizations for the next 2 nodes along the backbone of the tree (Nesomyidae + Cricetidae + Muridae and Muridae + Cricetidae) are ambiguous and could optimize either as Eurasia or as Africa. These 2 nodes are not well supported in any analysis of the combined-gene data set, and until additional data stabilize relationships among the 5 core muroid families, we refrain from discussing these ancestral area optimizations.

Fig. 4

The simplified phylogeny from Fig. 3 with subfamilies or families of muroids shown as terminal taxa. The present-day distribution for members of each terminal taxon is shown to the right of its name. Parsimony reconstructions of ancestral areas are indicated by shaded circles; all possible reconstructions are shown for nodes with equivocal ancestral areas. The 3 asterisks indicate taxonomic differences between our phylogeny and the classification of Musser and Carleton (2005) as follows: we place Lophiomyinae in Muridae rather than Cricetidae; our concept of Murinae includes otomyines rather than Otomyinae having subfamilial status; and we place Saccostomus in Dendromurinae rather than in Cricetomyinae.

The nodes defining the family-level clades Nesomyidae, Cricetidae, and Muridae are well supported (Fig. 3), but only the ancestral state optimization for Nesomyidae is unambiguous (Africa; Fig. 4). For this level of analysis, we did not consider Madagascar and mainland Africa to be separate areas of endemism; however, it is clear from this analysis that native Malagasy muroids (Nesomyinae) dispersed from Africa to Madagascar and subsequently radiated in isolation. The ancestral area for Muridae contains 2 equally parsimonious solutions (Africa or Eurasia), whereas the ancestral area for Cricetidae could be Eurasia, Africa, or North America (Fig. 4).

Discussion

Analyses of both IRBP and GHR sequences alone and in combination provide strong support for inclusion of Typhlomys within the clade Myodonta (Muroidea + Dipodidae; Figs. 13). Within Myodonta, Typhlomys is consistently recovered on the branch between Dipodidae and the clade including all other muroid families. Of the 2 nodes defining this position (Fig. 3, nodes B and C), monophyly of Muroidea excluding Typhlomys (node B) receives strong support from all analyses of the 3 data sets (bootstrap values ≥ 95%; Bayesian posterior probability = 1.0). However, the node linking Typhlomys with Muroidea (node C) receives notably lower support (bootstrap values < 90%) from parsimony analysis of all 3 data sets (particularly the IRBP data set) and from likelihood analysis of GHR alone, despite the consistently high posterior probability values recovered across all data sets for this node.

The fact that Bayesian posterior probability values are much higher than parsimony or likelihood bootstrap values for this node is not surprising: this phenomenon has been extensively reported in the systematics literature, even if its basis remains poorly understood (Alfaro et al. 2003; Brown and Lemmon 2007; Huelsenbeck and Rannala 2004; Lemmon and Moriarty 2004). The salient question pertaining to the present example is whether a competing position for Typhlomys should be considered given the available data. Likelihood-based topology tests (Shimodaira and Hasegawa 1999) soundly reject the 2 competing hypotheses for the phylogenetic position of Typhlomys discussed earlier. Given the topology shown in Fig. 3 the 2 most plausible alternative positions for Typhlomys relative to other muroids are first, that Typhlomys falls on the branch separating Spalacidae from remaining muroids, and second, that Typhlomys is the sister taxon to Spalacidae. These 2 hypotheses also are rejected based on the results of Shimodaira–Hasegawa tests (−2ΔlnL = 32.77, P = 0.002, and −2ΔlnL = 32.77, P = 0.002, respectively).

Collectively, these results suggest that Typhlomys—and by extension the platacanthomyid lineage—is the earliest divergence event among extant muroids (Fig. 4). Although Steppan et al. (2004) gave the name “Eumuroidea” to the clade comprising Calomyscidae, Nesomyidae, Cricetidae, and Muridae, this taxonomy was not followed by Musser and Carleton (2005). Although new superfamily names, including Eumuroidea, could be proposed for all nested subclades within the Muroidea once branching patterns among the families are better understood, we see no justification for such a proliferation of superordinal names.

The basal position of platacanthomyids within Muroidea has implications for the biogeographic history of muroids. Reconstruction of ancestral areas on our muroid phylogeny (Fig. 4) suggests that both Myodonta and Muroidea originated in Eurasia, and that at least the first 3 divergences within Muroidea—those leading to platacanthomyids, spalacids, and calomyscids—occurred on the Eurasian supercontinent. Our analysis is consistent with evidence from the fossil record, which suggests that all 3 lineages originated in Eurasia and did not disperse widely from there. The earliest fossils attributable to platacanthomyids belong to the extinct genus Neocometes, which is known from early Miocene (17–18 million years ago [mya]) deposits in southern China, Thailand, and central Europe as far west as eastern Spain (Fejfar 1999; Qiu and Li 2003). An unidentified “platacanthomyid” is known from Siwalik deposits in Pakistan dated to 17 mya (Flynn 2003). Neocometes disappears from the European fossil record by the middle Miocene (11–13 mya), and the earliest fossils attributable to the modern genera Platacanthomys and Typhlomys appear in the late Miocene deposits of Yuanmou and Shihuiba in southern China (Ni and Qiu 2002; Qiu and Li 2003). No fossils of any platacanthomyid have been found from northern China, and platacanthomyids are unknown outside Eurasia. Thus, it appears that platacanthomyids were widespread throughout Eurasia in the early Miocene, went extinct in Europe in the mid-Miocene, and survive only as disjunct relictual lineages in western India (Platacanthomys) and southern mainland Asia (Typhlomys).

Our biogeographic analysis also agrees with fossil evidence suggesting that the mole-rat lineage (Spalacidae; Fig. 4) originated in Eurasia. Although species of Spalax (the sole Recent genus in Spalacinae) extend as far south as northern Africa, they are unknown from sub-Saharan Africa, and the core distribution for spalacines—both fossil and Recent forms—is in Eurasia (Savic and Nevo 1990). Similarly, both extinct and living myospalacines are unknown from outside China and Russia. The 2 Recent genera of Rhizomyinae, Cannomys and Rhizomys, are known exclusively from southern Asia with a distribution extending from Sumatra through southern China into India. Of the 4 extant lineages of Spalacidae, only Tachyoryctinae has a modern distribution outside Eurasia, occurring in eastern Africa as far south as Mt. Kilimanjaro. The extinct genus Prokanisamys from the early Miocene of Pakistan (de Bruijn et al. 1981) is the earliest fossil attributable to the more inclusive clade comprising Rhizomyinae + Tachyoryctinae and is thought to lie along the branch preceding divergence of these 2 subfamilies (Flynn 1990). Given that the earliest fossil attributable to Tachyoryctinae occurs in early Miocene deposits of Pakistan, and tachyorytines do not appear in the African fossil record until the Pliocene, we suggest that the ancestor of the African mole rats also arose in Asia and subsequently dispersed to Africa (Flynn 1990).

Uncertainty surrounding the branching order among Calomyscidae, Nesomyidae, Cricetidae, and Muridae clouds both the phylogeny and the biogeographic history for this part of the muroid evolutionary tree (Figs. 3 and 4). Despite these ambiguities, our analyses clearly support the view that global muroid diversity resulted from independent radiations in separate continental regions (Carleton and Musser 1984; Chaline and Mein 1979; Steppan et al. 2004). For example, the ancestral area for Nesomyidae unambiguously optimizes as sub-Saharan Africa, and the members of this clade apparently dispersed out of Africa only once to found the native rodents of Madagascar (Nesomyinae). All other divergences within this subfamily apparently occurred on the African continent. Although the ancestral area optimization for Muridae is ambiguous and could include either Eurasia, Africa, or a combination of the 2, murids clearly originated in the Old World and, except for worldwide introductions by humans, did not disperse from there. Similarly, Cricetidae most likely originated in either Eurasia or a more inclusive Holarctic supercontinent, with 2 of the more substantial radiations localized in either Eurasia (Cricetinae) or across the Holarctic (Arvicolinae). Although our phylogeny is uninformative regarding when and where early diversification of the primarily South American Sigmodontinae occurred, the bulk of available evidence (Steppan et al. 2004; Weksler 2003) supports a radiation in South America after arrival from North America.

Although many details of muroid phylogeny and biogeography remain to be discovered, the addition of platacanthomyids to the muroid branch of the tree-of-life means that we now have a complete family-level phylogeny for this diverse clade of mammals. The fact that platacanthomyids represent the earliest branch within Muroidea has implications not only for taxonomy and biogeography, as discussed above, but also for future studies of phenotypic character evolution and rates of diversification. Additionally, because previous studies of muroid phylogeny (Jansa and Weksler 2004; Michaux et al. 2001; Steppan et al. 2004) employed dense sampling of muroid subfamilies (mirrored in the present study), we also have a substantially complete phylogeny of muroids at this lower hierarchical level. As taxon and gene sampling improves within each of these clades, patterns of diversification are becoming increasingly clear, and the emerging picture of muroid evolution is one of episodic diversification (Jansa and Weksler 2004; Michaux and Catzeflis 2000; Steppan et al. 2004). As a result, certain parts of the muroid tree-of-life, including the 1st few splits within the clade, are well resolved and robustly supported by the available data. In contrast, other nodes, including the relationships among the principal families, may remain difficult to resolve with any certainty.

Acknowledgments

We are continually grateful to the curators and collections managers at the many museums who generously provided tissues for our ongoing studies of rodent systematics. In particular, we thank L. Heaney, B. Patterson, and W. Stanley (FMNH) and M. Carleton, J. Jacobs, and L. Gordon (USNM). In addition, we thank J. Eger, who assisted with fieldwork, R. Voss, who confirmed the identity of the voucher specimen of Typhlomys, and M. Korpela, who helped generate several of the GHR sequences for this report. Primary funding of fieldwork in China was supported by United States National Science Foundation grant DEB-030820, with secondary funding from the Royal Ontario Museum Governors’ Funds. Laboratory research was supported by funds from the University of Minnesota.

Appendix I

Specimen information

GenBank accession numbers for all sequences used in this study. Accession numbers in bold indicate sequences generated for this study. Voucher numbers for these new sequences are given after the accession number in brackets. In cases where the museum voucher number is unknown, the collector number is listed followed by the holding museum acronym. Acronyms are as follows: MNHN = Musée National d'Histoire Naturelle, Paris; CMNH = Carnegie Museum of Natural History; DM = Durban Museum; USNM = National Museum of Natural History, Washington, D.C.; FMNH = Field Museum of Natural History; UA = University of Antananarivo; ROM = Royal Ontario Museum. Taxonomy follows Wilson and Reeder (2005). IRBP = interphotoreceptor retinoid-binding protein gene; GHR = growth-hormone receptor gene; NA = not available.

Aplodontiidae

Aplodontia rufa (IRBP: AJ427238, GHR: AF332030).

Calomyscidae

Calomyscus sp. (IRBP: AY163581, GHR: AY294901).

Castoridae

Castor canadensis (IRBP: AJ427239, GHR: AF332026).

Chinchillidae

Chinchilla lanigera (IRBP: AF297280, GHR: AY701337).

Cricetidae

Arvicolinae.—Eothenomys melanogaster (IRBP: AY163583, GHR: AM392399); Microtus sikimensis (IRBP: AY163593, M. arvalis GHR: AM392386); Myodes gapperi (IRBP: AY326080, GHR: AF540623).

Cricetinae.—Cricetulus longicaudatus (IRBP: AY326082, C. migratorius GHR: AY294926); Phodopus sungorus (IRBP: AY163631, GHR: AF540640).

Lophiomyinae.—Lophiomys imhausi (IRBP: AY326090, GHR: NA).

Neotominae.—Peromyscus maniculatus (IRBP: AY163630, P. leucopus GHR: AY294927).

Sigmodontinae.—Phyllotis xanthopygus (IRBP: AY163632, P. darwini GHR: AF332023); Reithrodon auritus (IRBP: AY163634, GHR: AY294930); Rhipidomys nitela (IRBP: AY163636, R. mastacalis GHR: AY294929); Sigmodon alstoni (IRBP: AY163640, S. hispidus GHR: AF540641).

Tylomyinae.—Tylomys nudicaudus (IRBP: AY163643, GHR: AY294933).

Dipodidae

Allactaginae.—Allactaga sibirica (IRBP: AY326076, GHR: AY294897).

Dipodinae.—Dipus sagitta (IRBP: AJ427232, GHR: AM407908); Jaculus jaculus (IRBP: AM407907, GHR: AF332040).

Sicistinae.—Sicista tianshanica (IRBP: AF297288, GHR: NA).

Zapodinae.—Zapus princeps (IRBP: AF297287, Z. hudsonius GHR: AF332041).

Geomyidae

Thomomys talpoides (IRBP: AJ427234, Geomys bursarius GHR: AF332028).

Gliridae

Glirinae.—Glirulus japonicus (IRBP: AB253965, GHR: NA); Glis glis (IRBP: AB253962, GHR: AM407916).

Graphiurinae.—Graphiurus murinus (IRBP: AY303219, GHR: AF332031).

Leithiinae.—Dryomys nitedula (IRBP: AB253956, Dryomys sp. GHR: AY294896); Eliomys quercinus (IRBP: AB253958, GHR: NA); Muscardinus avellanarius (IRBP: AB253959, GHR: NA).

Heteromyidae

Perognathinae.—Chaetodipus californiens (IRBP: AY303217, Perognathus flavus GHR: AF332029).

Leporidae

Lepus crawshayi (IRBP: AJ427250, L. capensis GHR: AF332016).

Muridae

Deomyinae.—Acomys spinosissimus (IRBP: AY326074, A. ignitus GHR: AY294923); Deomys ferrugineus (IRBP: AY326084, GHR: AY294922).

Gerbillinae.—Meriones unguiculatus (IRBP: AY326095, GHR: AF332021); Tatera robusta (IRBP: AY326113, GHR: AY294920).

Murinae.—Mastomys natalensis (IRBP: AY326093, M. hildebrandtii GHR: AY294916); Mus musculus (IRBP: NM015745, GHR: BC075720); Phloeomys cumingi (IRBP: AY326103, Phloeomys sp. DQ019070); Rattus exulans (IRBP: AY326105, GHR: DQ019074); Rhabdomys pumilio (IRBP: AY326106, GHR: AY294913).

Otomyinae.—Otomys anchietae (IRBP: AY326101, Parotomys sp. GHR: AY294912).

Nesomyidae

Cricetomyinae.—Beamys hindei (IRBP: AY326077, GHR: AY294904); Cricetomys emini (IRBP: AY326081, C. gambianus GHR: AY294905); Saccostomus campestris (IRBP: AY326109, GHR: GQ272601 [MNHN-1999-438]).

Dendromurinae.—Dendromus nyikae (IRBP: AY326083, D. mesomelas GHR: AY294902); Steatomys parvus (IRBP: AY326110, GHR: GQ272602 [CMNH 98495]).

Mystromyinae.—Mystromys albicaudatus (IRBP: AY163594, GHR: GQ272600 [DM 3452]).

Nesomyinae.—Brachytarsomys albicauda (IRBP: AY326078, GHR: GQ272593 [GKC 2606 (USNM)]); Brachyuromys betsileoensis (IRBP: AY326079, GHR: GQ272594 [FMNH 151659]); Eliurus minor (IRBP: GQ272605 [SMG 11771 (FMNH)], GHR: AY294911); Gymnuromys roberti (IRBP: AY326087, GHR: GQ272595 [FMNH 159724]); Hypogeomys antimena (IRBP: AY326089, GHR: GQ272596 [SMG 7128, FMNH 154636]); Macrotarsomys bastardi (IRBP: AY326092, GHR: GQ272597 [FH 28 (UA)]); Monticolomys koopmani (IRBP: AY326096, GHR: GQ272598 [SMG 9426 (FMNH)]); Nesomys rufus (IRBP: AY326099, N. audeberti GHR: AY294910); Voalavo gymnocaudus (IRBP: AY326114, GHR: GQ272604 [FMNH 154041]).

Petromyscinae.—Petromyscus collinus (IRBP: DQ191517, P. monticularis GHR: AY294906).

Platacanthomyidae

Typhlomys cinereus (IRBP: GQ272606 [ROM 118593], GHR: GQ272603 [ROM 118593]).

Sciuridae

Sciurilluspusillus (IRBP: AY227617, Sciurus niger GHR: AF332032).

Spalacidae

Myospalacinae.—Myospalax aspalax (IRBP: AY326097, GHR: GQ272599 [MSB94335]).

Rhizomyinae.—Rhizomys pruinosus (IRBP: AY326107, GHR: AY294899).

Spalacinae.—Spalax zemni (IRBP: U48589, S. ehrenbergi GHR: AY294898).

Tachyoryctinae.—Tachyoryctes splendens (IRBP: AY326112, GHR: AY294900).

Appendix II

Primers used in polymerase chain reaction amplification and sequencing of the interphotoreceptor retinoid-binding protein gene (IRBP) and growth-hormone receptor gene (GHR).

View this table:

Primers used in polymerase chain reaction amplification and sequencing of the interphotoreceptor retinoid-binding protein gene (IRBP) and growth-hormone receptor gene (GHR).

IRBPA5′ATGGCCAAGGTCCTCTTGGATAACTACTGCTT
IRBPB5′CGCAGGTCCATGATGAGGTGCTCCGTGTCCTG
IRBPF5′CTCCACTGCCCTCCCATGTCT
IRBPE5′AACAGATGCGCAGGGCCATCGT
GHRF15′GGRAARTTRGAGGAGGRGAACACMATCTT
GHRF505′TTCTAYARYGATGACTCYTGGGT
GHR750R5′GTAAGGCTTTCTGTGGTGATRTAA
GHRendAlt5′GATTTTGTTCAGTTGGTCTGTGCTCAC

Appendix III

Tree statistics from parsimony analysis of the 3 data sets, and maximum-likelihood parameter estimates for each data set under its best-fit model (GTR+I+Γ in all 3 cases). IRBP = interphotoreceptor retinoid-binding protein gene; GHR = growth-hormone receptor gene.

View this table:

Tree statistics from parsimony analysis of the 3 data sets, and maximum-likelihood parameter estimates for each data set under its best-fit model (GTR+I+Γ in all 3 cases). IRBP = interphotoreceptor retinoid-binding protein gene; GHR = growth-hormone receptor gene.

IRBPGHRCombined
Parsimony
No. characters1,2579542,211
No. informative characters5964961,092
Tree length3,5792,3785,983
No. most-parsimonious trees171412
Consistency index0.360.450.39
Retention index0.560.660.60
Likelihood
Frequency of A0.220.280.25
Frequency of C0.300.260.28
Frequency of G0.300.210.26
Frequency of T0.180.250.21
Alpha1.171.361.19
Proportion of invariant sites0.300.170.24
Rate A< >C1.380.971.14
Rate A< >G5.434.204.65
Rate A< >T1.080.580.81
Rate C< >G0.690.910.78
Rate C< >T7.153.915.57
–Log-likelihood18,495.5613,135.5131,832.45

Footnotes

  • Associate Editor was Mark S. Hafner.

Literature Cited

View Abstract