Darwin first recognized the importance of episodic intercontinental dispersal in the establishment of worldwide biotic diversity. Faunal exchange across the Bering Land Bridge is a major example of such dispersal. Here, we demonstrate with mitochondrial DNA evidence that three independent dispersal events from Asia to North America are the source for almost all lizard taxa found in continental eastern North America. Two other dispersal events across Beringia account for observed diversity among North American ranid frogs, one of the most species-rich groups of frogs in eastern North America. The contribution of faunal elements from Asia via dispersal across Beringia is a dominant theme in the historical assembly of the eastern North American herpetofauna.
Episodes of intercontinental dispersal have had a dramatic impact on current large-scale floral and faunal distribution patterns. Intercontinental dispersal occurs primarily when tectonic plates collide (Macey et al. 2000; Bossuyt & Milinkovitch 2001), when decreasing sea-levels expose land bridges (Wallace 1876; Darwin 1883), or when climatic changes produce dispersal corridors in otherwise inhospitable areas. A faunal exchange between Africa and Asia was induced by the plate-tectonic closing of the Tethys Sea, and a major exchange between North America and South America occurred following formation of a land bridge by the combined influence of plate tectonics and a decrease in sea-level (Brown & Lomolino 1998).
Climatic fluctuations producing episodic connections between Eurasian and North American faunal and floral elements have been invoked recently to explain disjunct distributions in bacteria (Maekawa et al. 2005), flowering plants (Wen 1999), aphids (von Dohlen et al. 2002), alligators and cryptobranchid salamanders (Pough et al. 2004) and even humans (Brown & Lomolino 1998). Particularly important for herpetofaunal exchanges was a Mid-Miocene connection of eastern Eurasia and eastern North America by a continuous temperate deciduous forest; this connection provided habitats for amphibian and reptilian groups otherwise excluded from the high latitudes at which Eurasia and North America are in close proximity.
Several eastern North American lizard and frog groups may have entered this region from Asia during the Miocene. We present molecular phylogenetic evidence of intercontinental faunal exchanges for ranid frogs and for two genera of scincid lizards, which share disjunct distributions in eastern North America and eastern Asia. The broader phylogenetic contexts of both groups place their origins in the Old World (Bossuyt & Milinkovitch 2001; Brandley et al. 2005), indicating that the eastern North American species descend from Eurasian ancestors.
2. Material and methods
The scincid genera, Scincella and Eumeces, are placed in separate subfamilies, Lygosominae and Scincinae, respectively. No previous phylogenetic work has considered all Northern Hemisphere genera in this family. In the Lygosominae, we sample Scincella lateralis from eastern North America and compare it to three species of Scincella and Sphenomorphus indicus from eastern Asia. In the Scincinae, we sample each species of Eumeces (including Neoseps reynoldsi) occurring east of the Mississippi River, several western North American Eumeces and two eastern Asian species (Eumeces capito and Eumeces quadrilineatus). Central American Eumeces are not part of the North American radiation (Brandley et al. 2005). Based on morphological characters, Griffith et al. (2000) suggest that North American scincines form a monophyletic group. We sample all scincine genera occurring in western Asia and most from North Africa (Chalcides, Eurylepis, Novoeumeces, Ophiomorus and Scincus). To root the phylogenetic hypothesis of the Scincidae, members of two separate families are included, Eremias grammica from the Lacertidae and Platysaurus capensis from the Cordylidae.
Ranidae is one of the largest frog families. To investigate the affinities of North American taxa, members of all monophyletic North American species groups are sampled (R. catesbeiana, R. sylvatica, R. warszewitschii, R. areolata, R. pipiens and all taxa in the R. boylii species group; Hillis & Davis 1986; Macey et al. 2001). These taxa are compared to a wide range of ranids occurring in Asia and Europe, and rooted with a pipid (Xenopus laevis), bufonid (Bufo andrewsi) and rhacophorid (Polypedates leucomystax).
We sequenced a segment of the mitochondrial genome containing part of nad1, trnI, trnQ, trnM, nad2, trnW, trnA, trnN, trnC, trnY and part of cox1. Sequencing primers for scincid lizards are from Macey et al. (1997) and primers for ranid frogs are from Macey et al. (2001).
DNA sequences were aligned manually. Positions encoding proteins were translated to amino acids using MacClade v. 4.03 (Maddison & Maddison 2001) for confirmation of alignment. Alignments of sequences encoding tRNAs were based on secondary structural models (reviewed in Macey & Verma 1997). Regions with extensive length variation were deemed unalignable and excluded from phylogenetic analyses.
The alignment for scincid lizards is 1933 positions of which 101 were excluded from analyses. Of the remaining 1832 positions, 949 are parsimony informative. The alignment for ranid frogs is 2126 positions of which 61 were excluded from analyses. Of the remaining 2065 positions, 951 are parsimony informative. These sequence alignments are available in electronic supplementary material, appendices A and B.
Phylogenetic trees were inferred by parsimony using Paup* beta v. 4.0b8 (Swofford 2001) using default settings. Bootstrap resampling (Felsenstein 1985) was applied to assess support for individual nodes using 500 replicates. Decay indices (=‘branch support’ of Bremer 1994) were calculated for all internal branches using searches that retained suboptimal nodes. All searches were heuristic using 100 random additions per replicate or search (see appendix C in electronic supplementary material for maximum-likelihood methods).
Museum voucher numbers and locality data for all newly reported sequences are deposited in GenBank files AY607272–AY607316, DQ005638 and DQ471440. A few of these sequences have been extended from other studies (Macey et al. 1997, 1998). Previously reported sequences used here are: X. laevis, M10217 (Roe et al. 1985); R. catesbeiana, R. sylvatica, R. temporaria, R. boylii, R. pretiosa, R. aurora, R. cascadae and R. muscosa, AF314016–AF314023 and AF314029, respectively (Macey et al. 2001).
Phylogenetic relationships among the Scincidae and Ranidae are presented in figure 1 (see appendix C in electronic supplementary material for maximum-likelihood results, which are similar). North American scincines do not form a monophyletic group. The lygosomine Scincella lateralis groups with East Asian Scincella and Sphenomorphus with strong support (bootstrap 98%, decay index 14) to the exclusion of West Asian Ablepharus and Asymblepharus, which form a monophyletic group (bootstrap 100%, decay index 27). North American members of the scincine genus Eumeces appear monophyletic with weak support (bootstrap less than 50, decay index 2), yet they are strongly grouped with East Asian Eumeces (E. capito and E. quadrilineatus, bootstrap 100, decay index 30).
North American ranid frogs also do not form a monophyletic group. Eastern North American ranids (R. catesbeiana, R. sylvatica, R. areolata and R. pipiens) and R. warszewitschii from Costa Rica together form a clade (bootstrap 74, decay index 10). Western North American ranids form a separate monophyletic group (bootstrap 99, decay index 11), the R. boylii species group (Macey et al. 2001). Among Eurasian R. temporaria and R. japonicus species groups form a clade (bootstrap 100, decay index 12) that is the sister taxon to the western North American R. boylii species group (bootstrap 87, decay index 10). All other ranids sampled are phylogenetically outside a group comprising eastern North America, western North America and the R. temporaria–R. japonicus clade.
Our results identify several intercontinental dispersals that contributed to the current distributions of the lizards and frogs examined here. Among lizards, North American Eumeces group with East Asian Eumeces to the exclusion of all other Northern Hemisphere scincid lizards, indicating a dispersal event across Beringia from Asia (figure 2). The North American species of Scincella is nested within Asian taxa indicating an additional dispersal event from Asia to North America.
Among ranid frogs, two clades of North American species are identified: one corresponding to the R. boylii group of western North America and another consisting of eastern North American taxa plus the single Neotropical species R. warszewitschii. The North American Rana do not form a monophyletic group. The R. boylii species group is the sister taxon to the R. temporaria–R. japonicus clade, and two equally parsimonious interpretations can explain dispersals across Beringia. One hypothesis is that separate dispersals across Beringia from Asia established the eastern and western North American clades, with the eastern North American dispersal preceding the one that established the R. boylii group in western North America. Alternatively, a single dispersal event from Asia to North America may have occurred followed by vicariant separation of eastern and western North American forms and a subsequent dispersal from western North America back to Asia to establish the R. temporaria–R. japonicus clade.
There are 14 native continental species of lizards in eastern North America (east of the Mississippi River; Conant & Collins 1998; table 1). Here, we have determined that ancestral lineages of the six Eumeces species and of the single Scincella species are both of Asian origin. Macey et al. (1999) showed that the ancestral lineage of four eastern North American Ophisaurus species is of Asian origin. The three remaining lizard taxa occurring in continental eastern North America, from separate taxonomic groups, are derived from the neotropics: Anolis (West Indies; Jackman et al. 1999), Cnemidophorus (Reeder et al. 2002) and Sceloporus (Wiens & Reeder 1997). Hence, 11 of the 14 eastern North American lizard species descend from ancestral lineages representing three dispersal events across Beringia. The origins of the 38 frog taxa in eastern North America (Conant & Collins 1998) are less well understood than lizards. All 14 Rana species appear to descend from a single dispersal event from Asia because our sampling includes members of all eastern North American species groups. The 17 hylid frogs occurring east of the Mississippi River are descended from a neotropical ancestor, and the North American taxa gave rise to the Eurasian species through dispersal across Beringia (Faivovich et al. 2005). Phylogenetic data are not yet available to evaluate geographic origins of the other frog groups in eastern North America.
Eighty per cent of lizard species and at least a third of frog taxa occurring east of the Mississippi River are derived from four independent dispersal events across Beringia. The contribution of Asian faunal elements to North America is a dominant theme in the assembly of the eastern North American fauna.
We thank the National Science Foundation (DEB-9726064; fellowships to J.A.S.), Ruth Rand for the Eumeces laticeps sample and Steve Gotte for the Eumeces multivirgatus sample. We thank Jonathan J. Fong for assistance with handling DNA sequences and GenBank submissions.