Crawford 2005

MOLECULAR
PHYLOGENETICS
AND
EVOLUTION
Molecular Phylogenetics and Evolution 35 (2005) 536–555
www.elsevier.com/locate/ympev
Cenozoic biogeography and evolution in direct-developing frogs

of Central America (Leptodactylidae: Eleutherodactylus)
as inferred from a phylogenetic analysis of nuclear
and mitochondrial genes
Andrew J. Crawford a,*, Eric N. Smith b
a
Naos Labs, Smithsonian Tropical Research Institute, Apartado 2072, Balboa, Ancón, Panama
b
Department of Biology, The University of Texas at Arlington, Arlington, TX 76019, USA
Received 16 May 2004; revised 18 February 2005

Available online 7 April 2005
Communicated by A. Larson
Abstract
We report the first phylogenetic analysis of DNA sequence data for the Central American component of the genus Eleuthero-
dactylus (Anura: Leptodactylidae: Eleutherodactylinae), one of the most ubiquitous, diverse, and abundant components of the Neo-
tropical amphibian fauna. We obtained DNA sequence data from 55 specimens representing 45 species. Sampling was focused on
Central America, but also included Bolivia, Brazil, Jamaica, and the USA. We sequenced 1460 contiguous base pairs (bp) of the
mitochondrial genome containing ND2 and five neighboring tRNA genes, plus 1300 bp of the c-myc nuclear gene. The resulting
phylogenetic inferences were broadly concordant between data sets and among analytical methods. The subgenus Craugastor is
monophyletic and its initial radiation was potentially rapid and adaptive. Within Craugastor, the earliest splits separate three north-
ern Central American species groups, milesi, augusti, and alfredi, from a clade comprising the rest of Craugastor. Within the latter
clade, the rhodopis group as formerly recognized comprises three deeply divergent clades that do not form a monophyletic group; we
therefore restrict the content of the rhodopis group to one of two northern clades, and use new names for the other northern (mex-
icanus group) and one southern clade (bransfordii group). The new rhodopis and bransfordii groups together form the sister taxon to
a clade comprising the biporcatus, fitzingeri, mexicanus, and rugulosus groups. We used a Bayesian MCMC approach together with
geological and biogeographic assumptions to estimate divergence times from the combined DNA sequence data. Our results cor-
roborated three independent dispersal events for the origins of Central American Eleutherodactylus: (1) an ancestor of Craugastor
entered northern Central America from South American in the early Paleocene, (2) an ancestor of the subgenus Syrrhophus entered
northern Central America from the Caribbean at the end of the Eocene, and (3) a wave of independent dispersal events from South
America coincided with formation of the Isthmus of Panama during the Pliocene. We elevate the subgenus Craugastor to the genus
rank.
Ó 2005 Elsevier Inc. All rights reserved.
Keywords: Chortis block; Craugastor; Eleutherodactylinae; Paleogeography; Mesoamerica; Molecular phylogenetics; Proto-Antilles; Systematics
1. Introduction
*
Corresponding author. Present address: Smithsonian Tropical
Research Institute, attn.: Andrew J. Crawford—Naos Labs, Unit
With over 700 Neotropical species, Eleutherodacty-
0948, APO, AA 34002-0948, USA. Fax: +011 507 212 8790. lus ranks as the most species-rich of all vertebrate gen-
E-mail address: andrew@dna.ac (A.J. Crawford). era, and the number of species continues to climb
1055-7903/$ - see front matter Ó 2005 Elsevier Inc. All rights reserved.
doi:10.1016/j.ympev.2005.03.006
A.J. Crawford, E.N. Smith / Molecular Phylogenetics and Evolution 35 (2005) 536–555 537
(AmphibiaWeb, 2005; Campbell, 1999; Duellman, be polyphyletic. Although taxonomic sampling was lim-
1999a; Frost, 2004; Hedges, 1999; Lynch and Duell- ited, their optimal reconstructions suggested that the sis-
man, 1997). For example, during a recent 4-year peri- ter group to Craugastor could be Brachycephalus from the
od, 1999–2002, Eleutherodactylus taxa were newly southern Atlantic rainforests of Brazil, and the Andean
described or resurrected at a mean rate of one every genus Phrynopus could be the sister group to either the
month (e.g., Campbell and Savage, 2000; Duellman subgenus Euhyas or to the subgenus Eleutherodactylus.
and Pramuk, 1999; Lynch, 2001a; Savage and Myers, However, placement of these two South American genera
2002). Half of all the worldÕs frog species are Neotrop- relative to Eleutherodactylus was not statistically signifi-
ical, and a quarter of these are Eleutherodactylus cant in either a Bayesian or a parsimony framework.
(AmphibiaWeb, 2005; Duellman, 1999b). Many frog Unfortunately, these other genera are not included in this
communities are dominated by Eleutherodactylus, both study, so we are unable to evaluate these hypotheses.
in terms of diversity and abundance (Lieberman, 1986; The following biogeographic scenario for the origins
Scott, 1976), especially those in the highlands (Doan of Eleutherodactylus in Central America follows Duell-
and Arriaga, 2002; Hofer and Bersier, 2001). This man (2001, pp. 807 and 809) and is based on a mix of
genus possesses an unusual, though not unique, mode data and informed speculation. The three lineages of
of development among anurans in which the tadpole Central American Eleutherodactylus are thought to have
stage has been lost, eggs are laid terrestrially, and the arrived there through two (Savage, 1982) or possibly
young hatch out as tiny froglets (Callery et al., 2001; three (Hedges, 1989) separate dispersal events. The
Elinson et al., 1990). Because of its direct development, genus Eleutherodactylus (including Craugastor) first en-
Eleutherodactylus has provided a model system for tered Central America from South America in the late
studies of the evolution of development (Elinson and Cretaceous to early Paleocene via an hypothesized pro-
Ninomiya, 2003; Hanken et al., 2001). We find it to to-Antillean land bridge (Savage, 1966, 2002) formed at
be no small tragedy, therefore, that this tremendous the leading edge of the Caribbean tectonic plate as it
group of frogs is so poorly known regarding its biogeo- moved east between North and South America (Burke,
graphic origins, phylogenetic relationships, taxonomic 1988). Among these South American frogs came the
groupings, or in some cases even basic field ancestor of Craugastor, according to this model. With
identification. the break-up of the land bridge, the ancestral Euhyas
The goals of this study are to estimate phylogenetic came to occupy the area that would become Cuba
relationships among Central American Eleutherodacty- (Hedges, 1989; Hedges et al., 1992). These West Indian
lus, estimate levels of genetic divergence among the ma- Eleutherodactylus diverged from their mainland counter-
jor clades, and use these results to elucidate the parts an estimated 77–63 million years ago (mya)
biogeographic origins of these frogs. Although there (Hedges, 1996), coinciding with the land-bridge model.
are still points of contention among students of the Second, Syrrhophus is thought to have originated in
genus, we may summarize our current state of knowl- northern Mesoamerica 40–30 mya through the dispersal
edge as follows. Eleutherodactylus almost certainly orig- from the Greater Antilles by an ancestral member of the
inated in South America because this continent is home Euhyas lineage (Hass and Hedges, 1991; Hedges, 1989).
to all species of all other genera in the subfamily Eleut- Third, with the gradual formation of a new land bridge
herodactylinae and most species in all other subfamilies (Coates et al., 2004) and the reconnection of Central and
of Leptodactylidae (Duellman, 1999a; Lynch, 1971). South America just prior to 3 mya (Coates and Obando,
Not surprisingly then, the bulk of species diversity of 1996), species of the subgenus Eleutherodactylus entered
Eleutherodactylus falls within the primarily South Amer- lower Central America (Savage, 2002; Vanzolini and
ican subgenus of the same name. Of the approximately Heyer, 1985) and today extend as far north as Honduras
400 species in this group, 16 occur in Central America. (McCranie and Wilson, 2002).
The genus Eleutherodactylus contains four other subgen- Eleutherodactylus are notorious for their high pheno-
era (Hedges, 1989). The subgenus Craugastor ranges typic variability within populations (Savage and Emer-
from the southwestern United States to northwestern son, 1970) and scant morphological divergence among
South America (Lynch, 1986) and contains just over species (Campbell and Savage, 2000; Savage, 1981).
100 species. Two subgenera, Euhyas and Pelorius, are Few morphological characters are informative for parsi-
found in the Greater Antilles and collectively contain mony analysis of interspecific relationships within this
fewer than 100 species. The subgenus Syrrhophus in- diverse genus (Lynch, 1986, 2000; Lynch and Duellman,
cludes two dozen species and ranges from Texas to Be- 1997). Besides the three samples included in the study by
lize and Guatemala. Within Central America, Darst and Cannatella (2004), the only previous molecu-
therefore, one may find Eleutherodactylus from three lar phylogenetic studies of Central American Eleuthero-
subgenera. dactylus used allozyme data to investigate relationships
Darst and Cannatella (2004) found molecular phylo- within species groups (Miyamoto, 1983, 1984, 1986),
genetic evidence that the genus Eleutherodactylus may and the conclusions we feel were hampered by
538 A.J. Crawford, E.N. Smith / Molecular Phylogenetics and Evolution 35 (2005) 536–555
inadequate sampling within species or within popula- but still confined to Central America, each one ranging
tions. Therefore, further DNA sequence data are crucial from Mexico to central Panama: the rugulosus (33 spp.),
to advancing our understanding of this group. gollmeri (10 spp.), and rhodopis (16 spp.) groups. All
Based on the above taxonomy and biogeographic sce- nine species groups (sensu Savage, 2002) except the
nario, we investigate the following questions. Does the bufoniformis group were represented by at least one
subgenus Craugastor represent a monophyletic group? specimen in the present study (Table 1). The wide-rang-
Was the initial radiation of the ancestral Craugastor ing gollmeri and rhodopis groups were investigated here
lineage rapid or adaptive? Are Syrrhophus and Euhyas in detail.
sister taxa, as suggested by allozyme, immunological,
and some morphological data (Hedges, 1989; Hass and 2.2. Taxonomic sampling
Hedges, 1991)? Do the members of the subgenus Eleut-
herodactylus in lower Central America represent a single We obtained DNA sequence data from 55 frogs rep-
invasion with subsequent speciation, or did speciation resenting 45 species, including two outgroup taxa, four
pre-date dispersal? We also look in detail at two wide- undescribed taxa and one unidentified specimen from
spread Central American species groups within Crauga- Brazil (Table 1). Four of five subgenera of Eleuthero-
stor, the rhodopis and gollmeri groups, to test their dactylus are represented among the samples, as are eight
monophyly and to investigate their geographic origins. of nine species groups (sensu Savage, 2002) within the
HedgesÕ hypothesis represents an unusual biogeographic subgenus Craugastor. Mitochondrial DNA (mtDNA)
scenario, a continental radiation having been derived sequences were obtained from 43 samples, and single-
from an island ancestor, which has been greeted with copy nuclear DNA (scnDNA) sequences were obtained
some skepticism (e.g., Duellman, 2001, p. 807). Here from 40 samples. Thus, both loci were sequenced for a
we use mitochondrial and nuclear DNA sequence data subset of 28 samples. All samples are supported by cor-
to investigate these phylogenetic and biogeographic responding voucher specimens available in public re-
questions. Finally, we investigate absolute times of orig- search collections, and all genetic data are available in
ination of species groups within Craugastor under the GenBank (Table 1).
proto-Antilles model and compare these divergence
times with the geological history of Central America. 2.3. Laboratory techniques
Genomic DNA was extracted from liver and/or thigh

2. Materials and methods muscle tissues using either standard phenol–chloroform
methods or the Qiagen QIAamp tissue kit. Mitochon-
2.1. The study group drial DNA fragments were PCR amplified using the
primers L4437 and H5934 (Macey et al., 1997a). Cycle
The approximately 100 species of Craugastor are di- sequencing utilized D-Rhodamine dye-terminator chem-
vided currently into nine (Savage, 2002) or 11 (Lynch, istry, and products were analyzed on an ABI Prism 377
2000) species groups, and most species are found only automated DNA sequencer (Applied Biosystems). Both
in Central America. Most of these species groups are heavy and light strands were completely sequenced using
phenetic, lacking synapomorphies. We can divide Sa- many internal primers (Table 2). Sequences were aligned
vageÕs nine groups into three sets according to their geo- using Sequencher 3.0 (Gene Codes) and by eye with the
graphic ranges [a concise overview of Eleutherodactylus same program and with GeneDoc (Nicholas and Nicho-
taxonomy and Craugastor distributions is found in las, 1997). Homology of DNA characters was inferred
Duellman (2001, pp. 806–811)]. Species in the three from the nucleotide sequences, the inferred amino acid
northern groups, augusti (formerly known as the genus sequences, and transfer RNA (tRNA) secondary struc-
Hylactophryne, 2 species), alfredi (13 spp.), and milesi ture models (Kumazawa and Nishida, 1993; Macey et
(12 spp.) groups reach as far south as Mexico, Guate- al., 1997c). These protocols yielded a mtDNA fragment
mala, and Honduras, respectively. The two uniquely of approximately 1460 base pairs (bp) containing the
southern groups are the fitzingeri (13 spp.; Savage following five complete and two partial genes, plus a
et al., 2004) and bufoniformis (2 spp.) groups, and the stem–loop structure, listed here in order from the 50 to
northern extent of their ranges reaches just over the Nic- the 30 end: a fragment of the tRNAMET gene, the com-
aragua–Honduras border in the former group but stops plete NADH dehydrogenase subunit 2 (ND2) gene,
in southeastern Costa Rica in the latter group. One tRNATRP, tRNAALA, tRNAASN, the origin of light
small group, the biporcatus group (6 spp.), is largely strand replication (OL), tRNACYS, tRNATYR, and a
southern, but with a disjunct species in Guatemala. fragment of the cytochrome oxidase I (COI) gene. This
The preceding three are the only groups within Crauga- mtDNA fragment is referred to as the ND2-WANCY
stor that have representatives in northwestern South region. For those samples that deviated from this stan-
America. The final three species groups are widespread dard gene order, the duplicated genes, rearranged genes,
Table 1
Institutional voucher numbers, locality information, and GenBank accession numbers for sampled taxa (FMNH = Field Museum of Natural
History, Chicago; UTA = University of Texas at Arlington; SIUC = Southern Illinois University at Carbondale; MVZ = Museum of Vertebrate
Zoology, University of California, Berkeley; CH = Cı́rculo Herpetológico de Panamá; USNM = National Museum of Natural History, Washington,
DC; SMF = Senckenberg Museum, Frankfurt am Main, Germany; ZUFRJ = Departamento de Zoologia, Universidade Federal do Rio de Janeiro,
Brazil)
Species and taxonomya Institutional Collection localityb Geographic GenBank Accession No.
voucher # coordinates
mtDNA c-myc
Subfamily Leptodactylinae
Genus Leptodactylus
Species group melanonotus
L. melanonotus UTA A-53817 San Marcos, GT 14°520 N, 091°540 W AY273099 Y337266
Species group fuscus
L. labialis UTA A-48666 Puerto Barrios, Izabal, GT 15°360 N, 088°430 W AY273100 None
Subfamily Eleutherodactylinae
Genus Eleutherodactylus
Subgenus Eleutherodactylus
Species group unistrigatus
E. ridens FMNH 257746 Las Cruces, Puntarenas, CR 08°470 N, 082°590 W AY273101 AY211306
E. museosus SIUC H-06970 El Copé, Coclé, PA 08°400 N, 080°360 W AY273103 None
E. pardalis FMNH 257675 Fortuna, Chiriquı́, PA 09°450 N, 082°130 W AY273102 AY211305
Species group conspicillatus
E. gaigei SIUC H-06965 El Copé, Coclé, PA 08°400 N, 080°360 W None AY211290
E. sp. (Brazil) ZUFRJ 7861 Pernambuco, BR 07°160 S, 035°230 W None AY211305
Species group discoidalis
E. ibischi UTA A-55243 Sud Cinti, Chuquisaca, BO 20°510 S, 064°190 W None AY211288
Subgenus Euhyas
Species group gossei
E. pantoni USNM 327872 St. Andrew Parish, JM 18°050 N, 076°430 W AY273104 AY211282
Subgenus Syrrhophus
Species group pipilans
E. pipilans UTA A-51050 Nenton, Huehuetenango, GT 15°460 N, 091°510 W AY273105 None
Species group nitidus
E. dilatus UTA A-55248 Omiltemi, Guerrero, MX 17°330 N, 099°410 W None AY337268
Subgenus Craugastor
Species group milesi
E. trachydermus UTA A-48500 Livingston, Izabal, GT 15°430 N, 089°140 W AY273106 AY211300
E. daryi UTA A-55251 Xucaneb, Alta Verapaz, GT 15°390 N, 089°470 W AY273107 AY211316
Species group augusti
E. augusti 1 UTA A-54930 Chilpancingo, Guerrero, MX 17°370 N, 099°050 W AY273108 AY211289
E. augusti 2 MVZ 226839 Roswell, New Mexico, US 33°220 N, 104°150 W AY273109 None
Species group alfredi
E. bocourti UTA A-55235 Purulha, Baja Verapaz, GT 15°150 N, 090°100 W AY273110 AY211301
E. xucanebi UTA A-51369 Huehuetenango, GT 15°520 N, 091°140 W None AY211298
Species group biporcatus
E. megacephalus FMNH 257714 La Selva, Heredia, CR 10°250 N, 084°020 W AY273111 AY211296
Species group rugulosus
E. ranoides MVZ 207277 Volcán Cacao, Guanacaste, CR 10°550 N, 085°270 W AY273112 AY211287
Species group fitzingeri
E. crassidigitus 1 FMNH 257676 Fortuna, Chiriquı́, PA 09°450 N, 082°130 W AY273113 None
E. crassidigitus 2 FMNH 257693 Nusagandı́, San Blas, PA 09°200 N, 078°590 W AY273114 None
E. crassidigitus 3 FMNH 257695 Nusagandı́, San Blas, PA 09°200 N, 078°590 W None AY337269
E. fitzingeri FMNH 257745 Las Cruces, Puntarenas, CR 08°470 N, 082°590 W AY273117 AY211297
E. longirostris CH 4735 Cana, Darién, PA 07°450 N, 077°410 W AY273116 None
E. talamancae FMNH 257694 Nusagandı́, San Blas, PA 09°200 N, 078°590 W None AY337270
E. tabasarae SIUC H-06964 El Copé, Coclé, PA 08°400 N, 080°360 W AY273115 None
Species group gollmeri
E. chac UTA A-55261 Livinatón, Izabal, GT 15°430 N, 089°140 W AY273130 None
E. gollmeri 1 FMNH 257561 Fortuna, Chiriquı́, PA 09°450 N, 082°130 W AY273124 AY211279
E. gollmeri 2 FMNH 257696 Nusagandı́, San Blas, PA 09°200 N, 078°590 W AY273123 None
E. laticeps UTA A-55249 El Volcán, Alta Verapaz, GT 15°280 N, 089°520 W AY273129 AY337267
E. lineatus UTA A-55250 El Volcán, Alta Verapaz, GT 15°280 N, 089°520 W AY273126 None
E. mimus FMNH 257605 La Selva, Heredia, CR 10°250 N, 084°020 W AY273125 AY211281
E. noblei 1 FMNH 257616 El Copé, Coclé, PA 08°400 N, 080°360 W AY273127 AY211280
E. noblei 2 USNM 534195 Quebrada Machin, Colón, HN 15°190 N, 085°170 W AY273128 AY211285
(continued on next page)
Table 1 (continued)
Species and taxonomya Institutional Collection localityb Geographic GenBank Accession No.
voucher # coordinates
mtDNA c-myc
Species group mexicanus (species group rhodopis prior to this analysis)
E. mexicanus UTA A-55233 Guelatao, Oaxaca, MX 17°260 N, 096°290 W AY273118 AY211312
E. omiltemanus UTA A-55240 Xochipala, Guerrero, MX 17°400 N, 99°490 W None AY337271
E. pygmaeus 1 UTA A-55241 Nueva Dehli, Guerrero, MX 17°260 N, 100°110 W None AY211309
E. pygmaeus 2 UTA A-55246 Rı́o Salado, Oaxaca, MX 16°120 N, 097°060 W AY273119 AY211313
E. saltator UTA A-55239 Nueva Dehli, Guerrero, MX 17°290 N, 100°120 W AY273122 AY211311
E. sartori UTA A-51105 La Fraternidad, San Marcos, GT 14°560 N, 091°530 W AY273121 AY211308
E. sp. nov. A UTA A-55247 Miahuatlán, Oaxaca, MX 16°100 N, 097°000 W AY273120 AY211310
Species group rhodopis
E. rhodopis 1 UTA A-55245 Yaxchilán, El Petén, GT 17°110 N, 091°060 W AY273133 AY211315
E. rhodopis 2 UTA A-55231 Jacaltepec, Oaxaca, MX 17°520 N, 096°140 W AY273131 AY211294
E. rhodopis 3 UTA A-54811 Chimalapa, Oaxaca, MX 16°450 N, 094°450 W AY273132 None
E. loki UTA A-54820 Los Tuxtlas, Veracruz, MX 18°220 N, 095°060 W AY273134 None
Species group bransfordii (species group rhodopis prior to this analysis)
E. bransfordii FMNH 257700 Nusagandı́, San Blas, PA 09°200 N, 078°590 W AY273140 AY211304
E. lauraster SMF 79760 Selva Negra, Matagalpa, NI 12°000 N, 085°550 W AY273138 None
E. persimilis FMNH 257566 CATIE, Cartago, CR 09°540 N, 083°390 W AY273141 AY211299
E. podiciferus 1 FMNH 257653 Las Cruces, Puntarenas, CR 08°470 N, 082°590 W AY273135 AY211319
E. podiciferus 2 FMH 257595 Tapantı́, Cartago, CR 09°450 N, 083°470 W None AY211307
E. polyptychus FMNH 257555 Isla Colón, Bocas del Toro, PA 09°230 N, 082°170 W AY273139 AY211322
E. stejnegerianus 1 FMNH 257803 Rincón de Osa, Puntarenas, CR 08°420 N, 083°310 W AY273137 None
E. stejnegerianus 2 FMNH 257801 Rincón de Osa, Puntarenas, CR 08°420 N, 083°310 W None AY211320
E. sp. nov. B FMNH 257689 Fortuna, Chiriquı́, PA 09°450 N, 082°130 W None AY211318
E. sp. nov. C FMNH 257760 Rı́o Claro, Puntarenas, CR 08°450 N, 083°030 W AY273136 AY211303
Numerals following conspecific names match numbers used in Figs. 1–3.
a
Taxonomy follows Lynch and Duellman (1997) and Frost (2004), except where noted.
b
Countries abbreviated by their ISO 3166 two-letter codes: BO, Bolivia; BR, Brazil; CR, Costa Rica; GT, Guatemala; HN, Honduras; JM,
Jamaica; MX, Mexico; NI, Nicaragua; PA, Panama; and US, United States.
and pseudogenes were replaced with ambiguities (NÕs) 2.4. Phylogenetic inference
for purposes of this study. Description and analysis of
these unusual features will be discussed elsewhere. Because all models of DNA sequence evolution used
A fragment of the cellular myelocytomatosis (c- in this study assume that nucleotide frequencies are con-
myc) gene was amplified using the primers cmyc1U stant over time and among lineages, we first tested the
and cmyc3L (Table 2). PCR products were sequenced mitochondrial and nuclear data sets for significant
directly and in both directions. The c-myc genes of departure from this assumption of stationarity using a
three problematic samples (Eleutherodactylus laticeps, v2 test implemented in PAUP* 4.0b10 (Swofford,
Leptodactylus melanonotus, and Syrrhophus dilatus) 1998). To evaluate potential phylogenetic heterogeneity
were sequenced using a protocol differing from that between the two data sets we used the incongruence
used for the mitochondrial fragment. PCR products length difference (ILD) permutation test (Farris et al.,
were obtained on a gradient thermal cycler at anneal- 1995) with 6800 random partitions of the combined data
ing temperatures of 54, 55.6, and 57.5 °C. PCR prod- set as implemented in PAUP*.
ucts were cloned into plasmids (Topo TA, Invitrogen) Prior to likelihood-based phylogenetic analyses we
that were isolated using the QIAprep Spin Miniprep identified a model of DNA sequence evolution that re-
Kit (Qiagen). Sequencing reactions were prepared quired a minimum number of parameters yet was ade-
using SequiTherm EXCEL II DNA Sequencing Kit- quate to explain the data. This evaluation was
LC (Epicentre Technologies) and the products were conducted on neighbor-joining (NJ) trees (Saitou and
analyzed with a LI-COR 4200 Dual Laser Long-read Nei, 1987) based on JC69 distances (Jukes and Cantor,
automated sequencer. Genic structure of the c-myc 1969) using Modeltest version 3.06 (Posada and Crand-
fragment was determined from consensus splice sites all, 1998). The model and the parameter estimates were
and by alignment with an mRNA sequence from chosen by AkaikeÕs minimum theoretical information
Xenopus (King, 1991). These protocols yielded approx- criterion, or AIC (Akaike, 1974), and were then used
imately 1340 bp sequences that included roughly in maximum likelihood (ML) analyses (Felsenstein,
540 bp of exon 2, 500 bp of intron 2, and 300 bp of 1981). These results were also used to choose the appro-
exon 3. priate model for the Bayesian analyses.
Table 2 Table 2 (continued)

Primers used in this study for amplifying and sequencing the
Nucleotide sequence (50 –30 )
mitochondrial ND2-WANCY gene region and intron 2 plus partial
flanking exons of the nuclear gene, c-myc cmyc4Lb (exon 3) TTGGCTGCGGTATATCKTTTYTC
cmyc7L (intron 2) AATGCATACAAGTTAGTAAT
Nucleotide sequence (50 –30 ) cmyc7La (intron 2) AAGGCATACGAGTTAGTAAT
mtDNA primer (lab name) cmyc7Lg (intron 2) AATACATATGCGTTAGTAAT
L4437 (METf.6)a AAGCTTTCGGGCCCATACC cmyc7Lv (intron 2) AATGCATACAAGTTAGTAAA
H5934 (COIr.1)a AGRGTGCCAATGTCTTTGTGRTT cmyc8L (intron 2) GCGTCGCTGCCCTAAACTAYC
H4996 (ND2r.564s)b AGTATGCTAAGAGTTTTC cmyc5U (exon 3) TATACCGCATCCARGAAAA
H4980 (ND2r.6)a ATTTTTCGTAGTTGGGTTTGRTT
The first two primers listed for each genome were used for amplifying
H4980 (ND2r.B) AATTTTCGAATTTGTGTTTGGTT
and sequencing. The other primers were used for sequencing only.
H4980 (ND2r.D) ATTTTCCGAACTTGTGTTTGATT Mitochondrial primers are designated by their 30 ends corresponding
H4980 (ND2r.C) AGTTTACGAATTTGAGTTTGGTT
to the homologous position in the human genome (Anderson et al.,
H4980 (ND2r.U1) ATTTTTCGGGTTTGAGTTTGATT
1981). In parentheses are indicated the laboratory names of mtDNA
H4980 (ND2r.G1) ATTTTTCGAATTTGTGTTTGATT
primers and the priming location of c-myc primers.
H4980 (ND2r.L) ATTTTTCGGACTTGTGTTTGGTT a
Indicates primers previously published in Macey et al. (1997a).
H4980 (ND2r.M) AGTTTACGAACTTGTGTTTGGTT b
Indicates primers previously published in Macey et al. (1997b).
H4980 (ND2r.H) ATTTTTCGTAGCTGTTTTTGGTT c
Indicates an unpublished primer from J. R. Macey.
H4980 (ND2r.G2) ATTTTTCGAATCTGTGTTTGATT d
Indicates primers previously published in Crawford (2003a).
H4980 (ND2r.G3) ATTTTTCGAAGTTGTGTTTGGTT
L4811 (ND2f.7) GGCATTGCCCCMTTYCACTTCTG
L4882 (ND2f.15)a TGACAAAAACTAGCACC Bayesian phylogenetic analyses (Rannala and Yang,
L4882 (E-ND2f.15) TGACAAAAACTTCCACC 1996) were conducted on separate and combined data
L4976 (ND2f.0533) AATTGTAGGCGGCTGAGGAGGC sets using MrBayes version 3.0b4 (Huelsenbeck and
L5038 (ND2f.590s) GCTCACCTTGGCTGAAT
L5002 (ND2f.5)b AACCAAACCCAACTACGAAAAAT
Ronquist, 2001). Four Monte Carlo Markov chains
L5002 (ND2f.C) AACCAAACTCAAATTCGTAAAGT (MCMC; Yang and Rannala, 1997), one cold and three
L5002 (E-ND2f.5) AATCAAACTCAACTACGAAAACT heated, were run simultaneously for one million genera-
L5002 (ND2f.B) AACCAAACTCAAGTTCGAAAAAT tions (Metropolis-coupled MCMC). Trees were sampled
L5002 (ND2f.U1) AATCAAACTCAAACCCGAAAAAT every 100 generations. Burn-in was evaluated by exami-
L5002 (ND2f.k) AACCAAACACAAATTCGAAAAKT
L5002 (ND2f.M1) AACCAAACACAACTTCGAAAGAT
nation of the plateau in log-likelihood over generations.
L5002 (ND2f.G1) AATCAAACACAAATTCGAAAAAT Burn-in occurred in fewer than 60,000 generations in all
L5002 (ND2f.U4) AACCAAACTCAAGTCCGAAAACT cases, so the first 1000 trees were always excluded, leav-
L5002 (ND2f.L) AACCAAACACAAACACGAAAAAT ing 9000 trees for estimating the marginal posterior dis-
L5002 (ND2f.M3) AACCAAAAACAGCTACGAAAAAT tribution of topologies and parameter values. Default
L5002 (ND2f.M2) AATCAAACACAACTTCGAAAAAT
L5002 (ND2f.G2) AATCAAACACAGATTCGAAAAAT
priors and conditions were used in all cases.
H5465 (ND2r.1413) GGCGAGAAAGAGTGAGAGA Using PAUP* 4.0b10, maximum parsimony (MP)
H5687 (ANSr.2)c GCGTTTAGCTGTTAACTAAA (Camin and Sokal, 1965) and weighted parsimony
H5687 (E-ANSr.2) GTTCTTAGCTGTTAATTAAG (WP) analyses were conducted on mtDNA, scnDNA,
H5686 (ANSr.A) GTATTTAGCTGTTAATTAA and combined data sets, whereas ML analyses were
H5686 (ANSr.B) GTTTTTAGCTGTTAACTAA
H5686 (ANSr.C) GTTTTTAGCTGTTACCTAA
limited to the separate mt- and scnDNA data sets.
H5575 (ALAr.2) CGCAAGTCTTACAGAAAC MP reconstructions employed Fitch parsimony (Fitch,
H5586 (ALAr.1) GGTTAGTGTCCCGCAAGT 1971) with gap sites excluded, while WP reconstruc-
L5551 (TRPf.5)b GACCAAAGGCCTTCAAAGCC tions utilized a tri-level weighting scheme (Benabib
L5551 (TRPf.A) AACCCTGAGCCTTCAAAGCT et al., 1997; Flores-Villela et al., 2000) with gaps
L5551 (E-TRPf.5) AACCCTGGGCCTTCAAAGCC
L5551 (TRPf.C) AACCCCGAGCCTTCAAAGCT
coded as a fifth base. Tri-level weighting incorporates
L5551 (TRPf.D) GACCAAAAGCCTTCAAAGCT three different levels of information on the structure
L5603 (ALAf.1) AAGACTTGCGGGACACTAACC and inferred function of nucleotide substitutions. Un-
L5603 (ALAf.B) AAGACCTGCAGGATATTAACC der this WP scheme, transitions have a weight of 1,
c-myc primer (location) transversions are weighted 2, and any nucleotide sub-
cmyc1Ud (exon 2) GAGGACATCTGGAARAARTT stitution that is inferred to cause an amino-acid sub-
cmyc3Ld (exon 3) GTCTTCCTCTTGTCRTTCTCYTC stitution is weighted +1 more. MP and WP
cmyc5L (exon 2) ATGGGTGGYGTTTCCATRTT employed ACCTRAN optimization of character state
cmyc5Lp (exon 2) ATGGGCGGCGTGTCCATATT
cmyc3Ud (exon 2) TCTTTCCTTACCCGTTGAATGATRC
changes. Tree searching was conducted with 500 ran-
cmyc3Up (exon 2) TTCCCTTACCCGTTGAATGA dom addition sequence replicates, while ML searches
cmyc4U (exon 2) TATGGAAACRCCACCCATCAG were initiated from a NJ tree. All searches used the
cmyc6Ld (intron 2) CAAAAGCCAGMCATTGGAAGATAA tree-bisection-regrafting method of proposing new
cmyc6Lg (intron 2) CCAGCCATCGAAAGATAA topologies. MP and WP bootstrap analyses (Felsen-
cmyc6U (intron 2) CGGCACGCTTTCTAAGAA
cmyc4La (exon 3) CTTGGATGCGGTATATCKTTT
stein, 1985) involved 2000 pseudoreplicates with 2–10
random addition sequences each.
We employed likelihood ratio tests (LRTs) to evalu- runs. We also conducted an additional analysis using
ate whether either data set, ND2-WANCY or c-myc, just the ND2-WANCY data and assuming the Bayesian
was significantly unlikely under the assumption of con- consensus topology based on just those data.
stancy of rates of molecular evolution (Felsenstein,
1981). Significance of the LRTs was evaluated assuming 2.6. Paleogeography
that the expectation of twice the absolute value of the
difference in support (Ln) under the clock versus the Our effort in phylogeny reconstruction and diver-
non-clock model was v2n2 distributed, where n equals gence time inference presented in this paper would be
number of sequences in the data set (Felsenstein, 1981). incomplete without a biogeographic context. Herein
For testing the relative support of alternative topolo- we present a revised consensus model of the complex
gies we used the paired-sites test of Shimodaira and paleogeography of Central America and the Caribbean
Hasegawa (1999, SH test) as implemented in PAUP*. from 75 to 18 mya. In our paleogeographic summary
This test uses bootstrap resampling and corrects critical we focused on land configuration (subaerial), instead
values for multiple comparisons. For a given topological of plate-tectonics, and tried to incorporate information
test, we constrained only the node/s in question and per- on elevation where possible. For northern Central
formed a new ML search, as above. The magnitude of America we based most of our paleogeography on
the difference in likelihood support for the ML tree Smith (2001) and references therein. Kuenzi et al.
(H1) and the constrained tree (H0) was evaluated by (1979) and Rogers (2000) provided additional insights
RELL sampling with 1000 bootstrap replicates (Kishino on the elevational configuration of this region through-
and Hasegawa, 1989). out the Cenozoic. For the Antillean region we followed
Iturralde-Vinent and MacPhee (1999). Regarding the
2.5. Estimating divergence times southern Central American region, we have incorpo-
rated ideas from Lloyd (1963), Coates and Obando
Rather than assume an externally calibrated rate of (1996), and Orvis and Horn (2000). However, the diver-
molecular evolution to estimate divergence times, we as- sity of mammalian fossils of strictly North American
sumed the proto-Antillean biogeographic model and affinity collected from a mid-Miocene formation in cen-
then estimated divergence times and evolutionary rates. tral Panama suggests that continuous land stretched
We then compared these results with a paleogeological from North America to Panama 19–16 mya (Fer-
model and previously published information on rates rusquı́a-Villafranca, 1975; MacFadden, 2005; Whitmore
of molecular evolution. We employed a MCMC ap- and Stewart, 1965). For northern South America we fol-
proach to estimate posterior probability distributions lowed Hoorn (1993), Dı́az de Gamero (1996), Marshall
of absolute divergence times, as developed by Thorne and Lundberg (1996), Vergara (1997), Iturralde-Vinent
et al. (1998; Kishino et al., 2001) and implemented for and MacPhee (1999), Audemard and Audemard
multigenic data (Thorne and Kishino, 2002) in ThorneÕs (2002), and Donato et al. (2003). We present the paleo-
software package ‘‘multidistribute’’ version 05/Aug/03 geographic reconstruction in Fig. 5.
(ftp://statgen.ncsu.edu/pub/thorne/). Divergence times
were estimated using all sequences of both genes, assum-
ing the Bayesian consensus topology based on the com- 3. Results
bined data (see Section 3). Some assumptions about
divergence times are required to facilitate the decoupling 3.1. DNA sequence alignments
of rate from time. Therefore, we temporally constrained
two nodes on the tree, one basal and one nested. For our IIn addition to gapped sites, the following gene re-
basal constraint we began with the proto-Antilles model gions were removed from all samples in the analysis
and restricted the origin of Craugastor to the interval due to ambiguities in the alignment. In the ND2-
80–60 mya (Savage, 1966, 1982), corresponding to a WANCY region we removed 119 consensus sites from
likely time interval for the hypothesized formation and the 30 end of ND2 following a completely conserved
break-up of a land connection between Nuclear Central tryptophan amino acid site and continuing through
America and northern South America (Burke, 1988; to the 2 bases shared by ND2 and tRNATRP. This por-
Iturralde-Vinent and MacPhee, 1999). For our nested tion of ND2 was highly variable in both inferred ami-
constraint, we bounded the divergence time of E. bran- no acid sequences and in length. Prior to analyses, we
sfordii and E. polyptychus to the interval 13–7 mya based also removed from the alignment 7 nucleotide sites of
on Crawford (2003a), whose temporal results were based the tRNATRP D-loop and 8 of the T-loop, 13 sites of
on a recalibration of rates estimated from toads (Macey the OL loop structure, 8 of tRNACYS T-loop, and in
et al., 1998). The MCMC analysis used default protocols, tRNATYR we removed 7 sites of the T-loop, 10 of
and multiple analyses were run to check for convergence the D-loop and 5 nucleotide sites between the T-stem
of the posterior distributions from independent MCMC and AC-stem. Thus, the mtDNA analysis consisted
of 1288 sites, of which 318 were constant and 854 were based on the c-myc data. Insofar as paralinear distances
parsimony-informative. In the c-myc alignment we re- are resistant to biases caused by heterogeneity of nucle-
moved 148 aligned sites from the 50 end of intron 2 otide frequencies, non-stationarity does not appear to
due to the prevalence of gaps, and 217 sites from have hindered substantially our estimate of relationships
around the intron 2/exon 3 boundary due to a stretch among the deeper nodes. No evidence of incongruence
of 8–23 pyrimidines in intron 2 followed by 13–26 tri- between data sets was found by the ILD test (p = 0.989).
nucleotide (GAR)n repeats in phase with exon 3. Thus, For likelihood-based analyses of the mtDNA multi-
the c-myc data set included 905 sites, of which 630 gene fragment, the general time reversible model (Ta-
were constant and 156 were parsimony-informative. varé, 1986) was found by the AIC to be the best fit
Our alignment is available from TreeBASE at www.tree model of sequence evolution, with the distribution of
base.org (study accession number S1249, matrix acces- rates of evolution among sites described by the gamma
sion number M2177), and includes DNA sequence anno- density shape parameter, a (Yang, 1994), and a propor-
tation and a complete list of all sites excluded from the tion of invariable sites, I (Hasegawa et al., 1987), aka,
analyses. Complete annotation of tRNA secondary the GTR + C + I model. A limited case of the preceding
structures is available from the authors. model was selected by the AIC for the c-myc data, i.e.,
equal base frequencies and equal transition rates (aka,
3.2. Stationarity, compatibility, and parameter estimation TVMef + C + I). As this model is a special case of the
GTR but not the HKY model, the former was used in
While the c-myc data show no signs of departure the Bayesian analysis of the c-myc data. For both data
from equal nucleotide frequencies among taxa (p = 1), sets, the models and parameter values obtained by the
the mtDNA data set reject strongly the hypothesis of AIC using Modeltest were adopted for ML topology
stationarity among samples whether we applied the v2 searches and branch-length estimation. Parameter val-
test to all the data, to the ingroup, or to just the subge- ues estimated for both genes by Modeltest and MrBayes
nus Craugastor (p < 0.000001). To examine heuristically are presented in Table 3. Because of the differences in
the effects of this departure from stationarity, we com- parameter estimates, no ML analysis was conducted
pared NJ trees based on LogDet distances (Lake, on the combined data set. A Bayesian analysis of the
1994; Lockhart et al., 1994) versus an JC69 NJ tree, a combined data was conducted using MrBayes version
BioNJ (Gascuel, 1997) tree using TrN93 distances 3.0b4, in which all parameters except one were estimated
(Tamura and Nei, 1993), and our MP, ML, and Bayes- independently in each data partition. Because the esti-
ian trees (see below). On only one basal node (i.e., mates of the C-distribution shape parameter, a, derived
among subgenera or among species groups) does the from Modeltest were similar in both data sets, only one a
LogDet mtDNA tree disagree with the other NJ trees, parameter was estimated across the combined data set.
the ML tree, or the Bayesian consensus tree based on
the mtDNA data. Only the LogDet tree recovers a sis- 3.3. Phylogenetic reconstructions
ter-group relationship between the gollmeri group and
the rhodopis + bransfordii clade. This relationship, how- Inferred phylogenies and nodal support based on
ever, does not appear in any of our reconstructions MP, WP, and Bayesian analyses of ND2-WANCY and
Table 3
Parameter estimates obtained from Modeltest and MrBayes for the models of evolution assumed in ML and Bayesian phylogenetic analyses of
separate mitochondrial and nuclear data sets
ND2-WANCY c-myc
Modeltest MrBayes Modeltest MrBayes
pA 0.3458 0.3435 (0.3282–0.3599) 0.25 0.2479 (0.2240–0.2743)
pC 0.3248 0.3260 (0.3123–0.3403) 0.25 0.2592 (0.2348–0.2851)
pG 0.0813 0.0823 (0.0766–0.0885) 0.25 0.2543 (0.2296–0.2800)
pT 0.2481 0.2482 (0.2359–0.2602) 0.25 0.2387 (0.2143–0.2652)
a 0.6682 0.6685 (0.5957–0.7448) 0.6664 0.6491 (0.4421–0.9628)
I 0.1654 0.1633 (0.1300–0.1942) 0.4555 0.4235 (0.3149–0.5182)
A-C 0.5132 0.5573 (0.4259–0.6949) 1.4156 1.6271 (1.0529–2.4675)
A-G 6.5057 7.0618 (5.7369–8.8227) 2.9437 2.9734 (1.9555–4.2380)
A-T 0.5896 0.6589 (0.4917–0.8656) 0.3860 0.5175 (0.2789–0.8640)
C-G 0.6885 0.7457 (0.5395–1.0274) 0.6427 0.7281 (0.4125–1.1518)
C-T 3.1106 3.4121 (2.6978–4.2332) 2.9437 3.8876 (2.6921–5.4432)
Using the Akaike information criterion, the following models were selected for ML analyses: GTR + C + I for the ND2-WANCY data and
TVMef + C + I (i.e., equal transition rates and equal base frequencies) for the c-myc data. MrBayes analyses assumed a GTR + C + I model for both
data sets. pN, estimated nucleotide frequencies. Substitution rates are relative to G-T = 1.
Fig. 1. Gene trees for the ND2-WANCY mitochondrial genes (left) and c-myc gene fragment (right) resulting from Bayesian MCMC phylogenetic
analyses. Support for each node is indicated by three numbers. The upper or first, lighter-colored number shows Bayesian marginal posterior
probabilities. The lower or trailing paired black numbers show the percent support obtained from parsimony bootstrap analyses, in both unweighted
(before the slash) and weighted (after the slash) analyses. Bootstrap values <50% are indicated with a hyphen. Those branches inferred from
parsimony bootstrap analyses that conflicted with the Bayesian topologies are indicated by the dotted lines. Maximum-likelihood topologies were
identical to the Bayesian consensus trees shown here, except for the position of (E. trachydermus, E. daryi) in the c-myc tree (see Section 3).
c-myc data separately (Fig. 1) and combined (Fig. 2) are this issue. In the combined data set, however, Bayesian
concordant in their support for the monophyly of spe- inference gives strong support (0.98 mpp; Fig. 2) to a sis-
cies groups, but show some discrepancies in their estima- ter-group relationship between Craugastor and all other
tion of relationships among species groups and among sampled Eleutherodactylus, while parsimony methods
subgenera. These discrepancies tend to involve clades give ambiguous results (64% MP bss, 58% WP bss;
not strongly supported by one or both data sets or opti- Fig. 2).
mality criteria, with one notable exception deep in the Bayesian consensus topologies for the ND2-
phylogeny. Bayesian and MP analyses of separate WANCY and c-myc data sets (Fig. 1) are identical
mtDNA and c-myc data sets recover a sister-group rela- to those of their respective ML trees (not shown),
tionship between the subgenus Euhyas and the subgenus with one small exception, again deep in the phylogeny.
Syrrhophus, as represented by Eleutherodactylus pan- Unlike the c-myc Bayesian consensus tree, the c-myc
toni + E. pipilans or E. dilatus, respectively. The discrep- ML tree places the milesi group (E. trachydermus + E.
ancy lies in the placement of this clade. Based on the daryi) as the sister lineage to the rest of the Crauga-
mtDNA data, this clade could be either the sister group stor taxa, as found in the mtDNA and combined data
to Craugastor, as weakly suggested by Bayesian infer- estimates (see Fig. 2). However, the branch length is
ence (0.68 marginal posterior probability, or mpp; Fig. non-significant for this internode on the ML c-myc
1), or the sister group to the subgenus Eleutherodactylus tree.
clade, as supported by WP analysis (91% bootstrap sup- Parsimony analyses among data sets yield the follow-
port, or bss; Fig. 1). The c-myc data are equivocal on ing hypotheses of relationships. Two MP trees of length
Fig. 2. Bayesian phylogenetic inference of relationships among all sampled taxa based on combined mtDNA and c-myc data sets. Those taxa
represented in both data sets are indicated with a circle at the tip of their respective terminal branches. See Fig. 1 for explanation of the three nodal-
support values and the black dotted lines. The origin of each sample is indicated by two-letter ISO 3166 country codes to the right of each taxon
name. Previously recognized subgenera are indicated by vertical bars. We recommend here that Craugastor be recognized as a separate genus. Species
groups are indicated by brackets and follow taxonomic changes recommended in this paper.
7009 were inferred from the mtDNA data, differing constancy using a LRT (p < 0.001 in both cases).
only within the fitzingeri group. The homoplasy index Therefore, we employ the MCMC approach to Bayes-
excluding uninformative characters (HIeuc) is 0.7468, ian estimation of the ‘‘rate of evolution of the rate of
and the rescaled consistency index (RC) is 0.1238. The molecular evolution’’ (Thorne and Kishino, 2002;
MP trees place the subgenus Craugastor as the sister Thorne et al., 1998) using the combined mitochondrial
lineage to all other ingroup taxa. The WP tree infers plus nuclear data sets, and assuming the combined data
the same relationships among subgenera. The shortest Bayesian tree (Fig. 2) and the proto-Antilles model
WP tree differs from the MP trees only among taxa with- (Fig. 3). The rates of evolution in the two genes show
in the gollmeri group. With weighting, the relative a similar degree of departure from clock-like behavior.
homoplasy is reduced slightly: HIeuc = 0.7032 and For the ND2-WANCY gene region, m = 0.1700
RC = 0.1724. In the MP analysis of the c-myc data, (0.0901–0.2944), and m = 0.2498 (0.0808–0.5364) for c-
the 500 random-addition sequence replicates recovered myc, where m = 0 implies a perfect molecular clock.
4600 shortest trees with length 597 steps. All of these However, the direction or magnitude of rate evolution
trees put E. pantoni + E. dilatus as the sister group to among branches is not significantly correlated between
all other ingroup samples, suggesting the following rela- genes (rank correlation coefficient of 0.4093,
tionships among subgenera: (((Euhyas, Syrrhophus) p = 0.135). The higher estimate and 95% confidence
Eleutherodactylus) Craugastor). Unlike the Bayesian interval for the nuclear gene probably reflect the much
c-myc consensus tree (Fig. 1), however, all the MP smaller absolute number of mutations observed in the
c-myc trees place the milesi group (E. trachydermus + E. c-myc data.
daryi) as the sister lineage to all other Craugastor species Absolute rates of divergence vary widely among
groups (in agreement with the mtDNA and combined- branches for both genes, but the central tendencies are
data Bayesian results). MP analysis of the combined roughly consistent with previous studies (e.g., Macey
data sets and all taxa, including those with missing data, et al., 2001). Among the 105 branches, the median esti-
resulted in 10,000 shortest trees of length 7609, all of mated rate of divergence for ND2-WANCY is 0.8410%
which agree with the mtDNA rather than the c-myc per million years (my) (SD = 0.4958%). These rates are
MP topologies in their relationships among subgenera per lineage, and the multidistribute software package
and species groups. The 156 parsimony-informative sites (Thorne and Kishino, 2002) employs the F84 + C model
in the c-myc data set are evidently overwhelmed by the of DNA sequence evolution. Similarly for the c-myc
854 contained in the ND2-WANCY data. gene fragment, the median is 0.0380%/my
The most fundamental difference among the phyloge- (SD = 0.0249%). To estimate the relative rate of evolu-
netic results involves the placement of the E. pan- tion in the mitochondrial versus nuclear genes, we plot-
toni + E. pipilans clade, representing the subgenera ted the estimated mtDNA divergence versus scnDNA
Euhyas and Syrrhophus in our analysis. On the one divergence for each branch of the phylogeny. The
hand, the combined data analyses, as well as the parsi- squared PearsonÕs correlation coefficient (r2) between
mony analyses of the mtDNA data, all support Crauga- rates is 0.0360 and the slope is 17.521 with the regression
stor as the sister group to all other sampled line forced through the origin (Fig. 4). This relative rate
Eleutherodactylus, especially in the Bayesian combined is similar to a previous result of 16 based on silent-site
analysis (Fig. 2) and the WP mtDNA analysis (Fig. 1). divergences estimated within species groups (Crawford,
Bayesian and ML analyses of the mtDNA data places 2003b).
the E. pantoni + E. pipilans clade as the sister group to Given the basal and nested node constraints, the
subgenus Craugastor, albeit with a low mpp of 0.68 Bayesian MCMC estimation of divergence times pro-
(Fig. 1). Therefore, using the SH test, we asked whether vided the following results based on simultaneous anal-
the mtDNA significantly reject the hypothesis of Crau- ysis of the mtDNA and c-myc data (Fig. 3). The
gastor as the sister group to all other sampled Eleuther- Caribbean subgenus Euhyas (represented here by E.
odactylus. We found no significant difference in pantoni) shared a common ancestor with the subgenus
likelihood support between the ML tree (28254.8239) Syrrhophus (E. pipilans and E. dilatus) 46–24 mya. The
and one constrained to contain a clade of all non-Crau- width of this 95% credibility interval probably reflects
gastor ingroup taxa (28256.4542; p = 0.409). We adopt the missing data among Syrrhophus samples (Table
the combined data Bayesian topology (Fig. 2) as our sin- 1), but the mean estimate of 35 mya matches well the
gle tree used in the estimates of divergence rates and 40–30 mya interval estimated previously from immuno-
times, below. logical and allozyme data (Hass and Hedges, 1991;
Hedges, 1989). Within Craugastor, most species groups
3.4. Evolutionary rates and divergence times arose even earlier, principally during the Eocene. With-
in the well-sampled species groups (gollmeri, bransfor-
Both the mitochondrial and nuclear DNA sequence dii, mexicanus, and fitzingeri), group members share a
data are significantly inconsistent with a model of rate most recent common ancestor (MRCA) that probably
Fig. 3. Absolute divergence times of each node estimated using the Bayesian MCMC method of Thorne and Kishino (2002), based on the topology
shown in Fig. 2, incorporating both mtDNA and c-myc sequence data, and assuming the proto-Antillean land-bridge model. For each node, the
horizontal line indicates the point estimate and the vertical line indicates the extent of the central 95% of the posterior distribution of divergence
times. The two pairs of horizontal heavy black lines indicate temporal constraints placed on the divergence times of two nodes prior to the analysis.
The vertical bar on the right indicates geological periods and epochs where Q, Quaternary and P, Pliocene.
lived during the Oligocene or possibly the Early 4. Discussion

Miocene. Species of Eleutherodactylus, too, can be
quite old. According to this analysis, five of the eight 4.1. The proto-Antilles assumption
sets of conspecific samples show >95% probability of
having diverged prior to the Pliocene (Fig. 3). Again, Estimating absolute divergence time requires an
these figures are based on the proto-Antilles model. assumption of either rates or dates. Because our data
We find that using wider priors and lowering the min- preclude the application of a rate calibration or the
imum age of the bransfordii–polyptychus divergence to assumption of an incontrovertible vicariance event, we
5 mya has no appreciable effect on our estimates of are evaluating a controversial biogeographic scenario.
divergence times. Removing the nuclear data from We assume the proto-Antilles model and then explore
the analysis results in an increase of 10% in the mean the implications of this model in terms of divergence
age of species groups. dates and rates. Under this model, land continuity or
4.2. The origin of the subgenera of Eleutherodactylus
Our Bayesian phylogenetic analysis of the combined

data set provides significant support for the novel
hypothesis that the subgenus Craugastor is the sister
lineage to all other Eleutherodactylus. Single-gene and
combined-data analyses support a sister-group relation-
ship between the subgenera Syrrhophus and Euhyas.
Although statistically significant, these results should
be considered tentative due to limited sampling of this
enormous genus and subfamily. This study lacks sam-
ples of additional eleutherodactyline genera, such as
Phrynopus or Brachycephalus, and, like Darst and Cann-
atella (2004), this study includes only six species from
among the roughly 400 species of the subgenus Eleuther-
odactylus. Keeping in mind, therefore, that all phyloge-
netic hypotheses are subject to further scrutiny by the
addition of more taxa and more characters, we discuss
briefly the biogeographic implications in light of the sta-
Fig. 4. Relative rates of DNA sequence evolution in the mtDNA
fragment versus the c-myc gene fragment. Each point represents one tistical support we find for some relationships among
branch on the phylogeny in Fig. 3 and shows the rate of evolution at subgenera.
each gene on that branch, as estimated by the Bayesian MCMC The hypothesis that Craugastor could share a sister-
approach of Thorne and Kishino (2002). Dashed line equals the linear group relationship with the rest of Eleutherodactylus
regression forced through the origin. Points in the solid circle represent
suggests that the genus sensu lato could have arisen in
branches corresponding to E. longirostris, E. sp. nov. A, E. pygmaeus
2, and the MRCA of the latter two taxa. Points in the dashed circle northwestern South America (Fig. 5). Our divergence-
represent E. pantoni, the MCRA of E. pantoni and (E. pipilans + E. time analysis under the proto-Antilles model suggests
dilatus), and the MCRA of all non-Craugastor taxa. Note that some that the common ancestor of the Greater Antillean sub-
taxa were missing data for one gene (Table 1). The median ratio of genera (including Syrrhophus) split from a lineage ances-
mtDNA to c-myc divergence rates was 25.616 (SD = 14.134).
tral to the mainland group containing E. ibischi 66–45
mya (Fig. 3). This calibration overlaps with HedgesÕ
proximity between northwestern South America and (1996) estimate of 77–63 mya for the origin of Eleuther-
Central America allowed faunal exchange between the odactylus in the West Indies. The ancestors of the sub-
two areas during the end of the Cretaceous and the genera Syrrhophus and Euhyas then diverged 46–25
beginning of the Paleocene (e.g., Guyer and Savage, mya, agreeing with a previous estimate of 37 mya
1987, 1992; Rosen, 1975, 1985; Savage, 1982). Some (Hedges et al., 1992) and the fossil record (Holman,
paleotectonic reconstructions of the Caribbean region 1968; Iturralde-Vinent and MacPhee, 1996).
refer to a proto-Antilles island chain in the area now
occupied by lower Central America (e.g., Burke, 1988; 4.3. Are the frogs really that old?
Pindell et al., 1988; Ross and Scotese, 1988), yet the evi-
dence for a land connection between the two continents Independent estimation of the time of origin of Cra-
would have long since disappeared due to geologic pro- ugastor is difficult in the absence of early fossils repre-
cesses, insomuch as ‘‘the geography of the proto-Antilles senting this group. In our Bayesian MCMC estimation
probably has nothing to do with the geography of the of divergence times we constrained its origin to occur
existing islands [Greater Antilles]’’ (Iturralde-Vinent 80–60 mya, based on the proto-Antilles model (see
and MacPhee, 1999). However, evidence for low sea above). If this divergence time was a substantial overes-
levels during the Maastrichtian (71–65 mya) have timation, then our resulting rates of evolution should err
been identified globally (Hallam, 1984; Haq et al., on the slow side. To the contrary, our rates appear to be
1987; Vail and Hardenbol, 1979) and in northwest South on the high side of previous estimates of mtDNA diver-
America in particular (Vergara, 1997). These recorded gence. Our resulting mean rate estimate of 0.8410% per
periods of low sea levels coincide with the timing of lineage per million years is higher than previous esti-
the proto-Antilles and make a land-pass between Cen- mates for poikilotherms, which range from 0.57% to
tral America and South America more likely. Unfortu- 0.69% (reviewed in Macey et al., 2001). The absolute le-
nately, we cannot demonstrate its existence with the vel of DNA sequence divergence observed is also partic-
data at hand. We can, however, explore the implications ularly high. The ND2-WANCY divergences between
of this model for the ages of clades and the rates of Craugastor and Eleutherodactylus subgenera are quite
molecular evolution. high at 34% uncorrected distance, or an estimated
41% between Craugastor and Eleutherodactylus. Look-

ing at the c-myc data, Crawford (2003b) assumed these
same two subgenera diverged 72 mya to estimate diver-
gence rates at silent sites. Once again the estimated rates
were not low. Instead they were quite similar to esti-
mates from other nuclear genes in other frog lineages.
Thus, the proto-Antilles assumption results in reason-
able rates of molecular evolution.
4.4. The origin and diversification of Craugastor, the first

invasion
The Cretaceous–Paleocene proto-Antillean land

bridge or island-arc model can account for the present
distribution of Eleutherodactylus subgenera (Hedges,
1989; Savage, 1982), and our molecular phylogenetic re-
sults appear largely concordant with this proposal.
Monophyly of the subgenus Craugastor is well sup-
ported by Bayesian and WP analyses of the mitochon-
drial and combined data sets. Therefore, we infer
parsimoniously that Craugastor originated from a single
dispersal northward. This invasion would have repre-
sented the first occurrence of a direct-developing frog
in Neotropical Laurasia, setting the stage for a potential
adaptive radiation if this reproductive mode represented
a key innovation in that environmental context (e.g.,
Vences et al., 2002). Under this hypothesis one predicts
rapid cladogenesis relative to divergence, resulting in
short internodes and the potential for a hard polytomy
(Jackman et al., 1999; Schluter, 2000). Although most
basal nodes uniting species groups in Craugastor were re-
solved, particularly using Bayesian inference, these inter-
nodes are distinctly shorter than the subtended branches
that lead to the basal node of each species group (Fig. 2).
Further support for a rapid evolutionary radiation
comes from the observation that these short, basal inter-
nodes occur both in the phylogeny inferred from fast-
evolving mtDNA sequences and in the phylogeny
inferred from the slowly evolving nuclear gene (Fig. 1).
Ecological and morphological diversification of Cra-
ugastor appears to have taken place early in cladogene-
sis, concomitant with the origination of species groups
(cf., Harmon et al., 2003). Body size and habitat prefer-
ences vary among groups but are largely constant within
groups (Savage, 2002). For example, almost all members
Fig. 5. Cenozoic biogeography of Eleutherodactylus in Central America. of the milesi and rugulosus groups are riparian, the latter
This figure shows a geographic and temporal hypothesis for the origin of reaching larger adult body sizes. Frogs of the gollmeri,
four subgenera and the major species groups within the subgenus rhodopis, and bransfordii groups are all forest floor spe-
Craugastor. This hypothesis is based on the present phylogenetic results cies, with gollmeri taxa attaining much larger body sizes.
(Fig. 2) and the proto-Antillean land-bridge model (Fig. 3). The top
panel provides a geographic key to the other panels. Circles represent
Species in the fitzingeri group are moderately sized and
areas where phylogeny and geography coincide in time. Smoking inhabit low vegetation. Speciation within groups, there-
volcano icons represent hypothesized islands of magmatic origin. fore, did not involve obvious ecological displacement.
Consistent with this conclusion, previous ecological
138% using the ML model selected by the AIC. At the studies have found that niche partitioning is scarcely ob-
amino acid level the aligned portion of the inferred servable among sympatric Eleutherodactylus (Lieber-
ND2 protein sequence shows an average difference of man, 1986; Toft, 1980a,b).
Under the proto-Antilles hypothesis, the ancestor of 1995). Also, as recently as the Pleistocene Syrrhophus
extant Craugastor began its potentially rapid diversifica- occupied a wider range in North America, as evidenced
tion around the Paleocene–Eocene transition on the by fossils collected from 300 km north of its present
Chortis block (modern day southern Guatemala, El Sal- range (Lynch, 1964; Tihen, 1960). Therefore, the ances-
vador, Honduras, and northern Nicaragua) and adja- tral Syrrhophus lineage could have dispersed from the
cent southern Mexico now north of the Balsas basin Antilles into Florida and expanded west along the
(Fig. 5). This hypothesis we base on the observation that southern United States, subsequently becoming con-
within Craugastor the species groups with ancestral fined to central Texas and southwards.
branches positioned closest to the root node (milesi Regardless of the route, dispersal between the Antil-
and augusti + alfredi) are northern Central American les and the mainland may have been much easier during
endemics. The gollmeri group lineage diverged next, also the Oligocene and Miocene than it would appear to be
presumably originating in the north, although it con- today. There is evidence that the dry land in the Carib-
tains both southern and northern taxa (cf., Savage, bean may have been more extensive due to historical
1987). Subsequently, multiple dispersal events took sea-level fluctuations, e.g., the Pleistocene (Olson and
place from north to south, dividing Chortis block Pregill, 1982) and the Oligocene (Iturralde-Vinent and
(northern) from lower Central American species groups: MacPhee, 1999). Periods of low sea level occurred
the rhodopis–bransfordii split and the divergence of the approximately 30, 15, and 5 mya (Haq et al., 1987; Vail
mexicanus group from its sister group (Figs. 3 and 5). and Hardenbol, 1979). The low sea level at about 30
Based on the proto-Antilles model, we infer that these mya may have facilitated the dispersal of the Syrrhophus
cladogenic events took place around the Eocene–Oligo- ancestor into the mainland (Fig. 5), as well as the wide-
cene boundary. Some paleogeological reconstructions spread dispersal of Eleutherodactylus in the West Indies,
posit that the highlands of lower Central America were through GAARlandia.
an island at this time (Iturralde-Vinent and MacPhee,
1999). The geographic origins of the biporcatus and rug- 4.6. The third invasion
ulosus groups are unclear, because both groups contain
northern and southern species, though we predict they Our data suggest that the third independent origin of
are northern as well. Eleutherodactylus in Central America involved a wave of
species rather than a single dispersal event. The four
4.5. The origin of Syrrhophus, the second invasion lower Central American taxa sampled from the subge-
nus Eleutherodactylus apparently diverged from one an-
We find little doubt that the subgenera Syrrhophus other long before the Pliocene (Fig. 5) when the
and Euhyas are sister taxa, as first suggested by Hedges Panamanian land bridge formed (Coates and Obando,
(1989; also Hass and Hedges, 1991). All Bayesian and 1996). Assuming that these frogs, like many other South
MP bootstrap analyses showed strong support for this American lineages, reached Central America recently via
union, though WP analyses involving c-myc data were this land bridge (Savage, 2002; Vanzolini and Heyer,
equivocal (Figs. 1 and 2). In addition to the various data 1985; Webb and Rancy, 1996), each of our four Pana-
sets of Hedges (1989; also Hass and Hedges, 1991), sam- manian and Costa Rican species in this subgenus repre-
ples from these two clades share an unusual structural sents an independent dispersal event. In addition to our
feature of their mtDNA molecules. Eleutherodactylus divergence time estimates, we note that the c-myc data
pantoni and E. pipilans share a 2 bp deletion correspond- support a paraphyletic relationship of the lower Central
ing to the AA-acceptor bases of tRNATRP and tRNAALA. American samples of the subgenus Eleutherodactylus
Because these two genes are encoded on opposite with respect to the Brazilian sample. Therefore, the
strands, their corresponding tRNAs are necessarily topology alone suggests a minimum of two dispersal
cleaved from opposite transcripts, and both genes can events.
therefore remain functional.
The Antillean ancestor of Syrrhophus could have dis- 4.7. An alternative hypothesis concerning rates and dates
persed to Mexico via the Yucatan Peninsula, but fossil
evidence supports a possible alternative hypothesis. Despite the concordance between previously pub-
Early Miocene Eleutherodactylus fossils from Florida lished rates of evolution and the rates obtained here
(Holman, 1968) suggest that the genus might have based on the proto-Antilles model, we still cannot reject
crossed from the Caribbean into eastern North America. the alternative hypothesis that rates of molecular evolu-
However, these fossils consist of only two ilial bones and tion in Eleutherodactylus, particularly in their mtDNA,
their taxonomy is controversial (Lynch, 1971). Two could be higher than those estimated previously in tem-
more recent and less controversial pieces of fossil evi- perate-zone amphibians, reptiles, and tropical marine
dence support the presence of Eleutherodactylus in late fishes (reviewed in Macey et al., 2001). Although no evi-
Pleistocene southern Florida (Emslie and Morgan, dence has yet been found to support the hypothesis that
tropical organisms in general have higher rates of molec- not endorse at this time. Within the Eleutherodactylus
ular evolution than temperate-zone species (Bromham subgenus, the probable paraphyly of the Central Amer-
and Cardillo, 2003), Eleutherodactylus are unusual in ican taxa relative to the Brazilian sample suggests that
two respects. First, among the taxa studied here we the Central American species do not form a monophy-
uncovered five independent mtDNA gene duplications letic group relative to other South American endemics,
and rearrangements (details to be presented in a fol- in which case the cruentus group of Savage (2002) would
low-up study), and the rate of rearrangement could be not be valid (Lynch and Duellman, 1997).
positively correlated with rates of nucleotide substitu- Within Craugastor, we found that the southern mem-
tion (e.g., Shao et al., 2003). Second, Eleutherodactylus bers of the former rhodopis group are quite diverged
are unusual relative to other frogs in their wide diversity from the northern species. Although the name podicife-
of karyotypes (DeWeese, 1976), although high rates of rus predates bransfordii, we propose the name ‘‘bransfor-
chromosomal rearrangements need not imply high rates dii group’’ for the southern clade, as this name is already
of DNA sequence evolution. used commonly among workers in the field. Among the
One alternative hypothesis might be that Craugastor northern species, we propose the name ‘‘mexicanus
is half as old but has a rate of molecular evolution twice group’’ for the other former rhodopis-group clade that
as high as found in previous studies of poikilotherms. is distantly related to the rhodopis + bransfordii clade
This ‘‘young Craugastor’’ hypothesis has a disadvantage (Fig. 2). We found no support for recognizing Crauga-
in requiring that the ancestral Craugastor rafted to Cen- stor daryi in the gollmeri group (Lynch, 2000) or for rec-
tral America during the late Eocene. Trans-oceanic dis- ognizing the omiltemanus group of Ford and Savage
persal by amphibians is an unlikely phenomenon (1984). Under our organizational scheme, C. omiltem-
(Darwin, 1859), but it does happen (Hedges et al., anus would now be a member of the mexicanus group
1992; Vences et al., 2003, 2004). One advantage of a (cf., Lynch, 2000). We concur with LynchÕs (2000) place-
younger Craugastor, however, is that the southern por- ment of C. bocourti in the alfredi group. We currently
tion of its biogeographic history would be more easily have no molecular data to evaluate the validity or phy-
reconciled with the relatively young age of lower Central logenetic positions of the species assigned by Lynch
America (Coates et al., 2004; Whitmore and Stewart, (2000) to the andi and bufoniformis groups of Crauga-
1965). Such alternative possibilities highlight our need stor, so we are unable either to support or to reject these
for more independent rate calibrations for Eleuthero- hypotheses.
dactylus, and for more amphibians in general. We suspect that the unassigned Craugastor taxon
Eleutherodactylus uno will be related to the alfredi group
4.8. Taxonomic implications because it inhabits cloud-forest environments in south-
ern Mexico and has the fifth toe longer than the third,
The genus Craugastor is clearly monophyletic and has expanded and notched pads on the fingers, and lacks
a long history separate from the other subgenera. We tympanic sexual dimorphism. In summary, we propose
propose that the taxonomic rank of the subgenus Crau- to use a node-based definition of the new genus Crauga-
gastor be elevated to the status of a genus. This taxo- stor and define it as that crown clade containing the fol-
nomic change is also supported by the finding that the lowing taxa and their MRCA: C. augusti, C. bocourti
sister group of Craugastor might not even be an Eleut- (alfredi group), C. bransfordii, C. daryi (milesi group),
herodactylus (Darst and Cannatella, 2004), but rather C. fitzingeri, C. gollmeri, C. megacephalus (biporcatus
Brachycephalus or another eleutherodactyline genus. group), C. mexicanus, C. rhodopis, and C. ranoides (rug-
Monophyly of Craugastor is supported by a morpholog- ulosus group). If the members of the bufoniformis and
ical synapomorphy involving the jaw musculature fraudator groups were found to share this same MRCA,
(Lynch, 1986, 2000, 2001b), although the same character then they too would be placed in the genus Craugastor.
state appears in the E. fraudator group from Bolivia (De
La Riva and Lynch, 1997; Köhler, 2000; Lynch and
McDiarmid, 1987). The phylogenetic position of the 5. Conclusions
fraudator group is unclear at this time.
Other subgenera could probably be elevated, but we Craugastor has differentiated enormously, and this
leave such changes to future studies of Eleutherodacty- diversity seems to be associated with the old age of the
lus, leptodactylid, and hyloid systematics. We refrain group and the complex geologic history of Central
from recognizing Euhyas or Syrrhophus as full genera America during the Cenozoic that promoted numerous
at this time because our sampling from within these taxa dispersal and vicariance events. We find that ecological
was minimal. Our data suggest that Hylactophryne diversity among species groups within Craugastor coin-
should not be recognized as a genus (e.g., Darst and cides with short internodes in both the mitochondrial
Cannatella, 2004) unless both the alfredi and milesi and nuclear phylogenies, suggestive of an early adaptive
groups are elevated to the generic level, which we do radiation. Ecological and geographical correlates of spe-
ciation within groups warrant further investigation. We Rice, R. Rojas, and G. Keller for help collecting in the
hope that this study has set the genealogical stage for fu- field. Field assistance in Guatemala was provide to
ture studies of evolution within species groups. E.N.S. by M. Acevedo, R. Garcia, C.L. Guirola, J.
For improved studies of Eleutherodactylus evolution Monzón, and R. Schiele. For field assistance in Mexico,
we need to incorporate additional historical events that E.N.S. is indebt to K. Castañeda and the late J. Cama-
may be used as internal rate-calibration points for rillo-Rangel. The following individuals and institutions
molecular evolution. These events would decrease our generously donated further samples to this project: K.
dependence on a priori biogeographical hypotheses or Lips and the SIUC, G. Köhler and the SMF, A. Carna-
estimates of molecular evolutionary rates derived from val, C. Cicernos and D. Wake of the MVZ, S.B. Hedges
other taxa. Expanded taxonomic sampling of Eleuthero- and J.R. McCranie via S. Gotte and G. Zug of the
dactylus (apart from Craugastor) and additional South USNM, J. Campbell and M. Harvey at UTA, R. Ibáñez
American genera (Darst and Cannatella, 2004) are re- and C. Jaramillo of the CHP.
quired before a clear and inclusive picture of evolution Most DNA sequence data were collected in the lab-
and diversification of these frogs can be achieved. With- oratory of M. Kreitman at the University of Chicago,
in Craugastor, the high-elevation alfredi group of north- with additional data collected in the R. Fleischer Lab
ern Central America appears to be particularly old, yet at the National Zoological Park and the P. Chippin-
we have sampled very little of its potential diversity. In dale Lab at UT Arlington. For lab advice and help
the south, molecular data are lacking from the bufonifor- we thank P. Andolfatto, E. Stahl, S.-C. Tsaur, J. Glad-
mis group, from LynchÕs (2000) andi group contained stone, J.R. Macey, J. Dumbacher, and T. Castoe. For
within the fitzingeri group of Savage et al. (2004), and analytical advice starting way back in time, we thank
from within the rugulosus + biporcatus clade. Finally, A. Driskell, J. Walsh, G. Reeves, B. Quenouille, T.
to understand CraugastorÕs place among the eleuthero- Castoe, V. Gowri-Shankar, and J. Thorne. For helpful
dactylines, more molecular data are needed from these comments on the manuscript we thank T. Castoe and
additional genera, as well as northern Andean Eleuther- C. Dick. The final product was improved significantly
odactylus and the fraudator group of Bolivia. We hope by an anonymous reviewer, and by the insight and
that our analysis here contributes positively to our sharp eye of A. Larson.
understanding of relationships of at least one twig of For field and laboratory work A.J.C. was variously
the eleutherodactyline tree and promotes further re- supported by an OTS graduate research fellowship from
search on these remarkably diverse and locally abun- Peace Frogs Inc., STRI-OTS Mellon predoctoral fellow-
dant animals. ship, a Sigma Xi Grant-in-Research, a Gaige award
from the ASIH, a Grant-in-Aid from SICB, the CEB
and the Park Fund at the University of Chicago, the
Acknowledgments Biology of Small Populations RTG Postdoctoral Fel-
lowship from the UMCP and the NZP, and an NSF
We very gratefully acknowledge the following indi- International Programs Postdoctoral Fellowship. This
viduals, land owners, governmental personnel and insti- paper is based in part upon work supported by the Na-
tutions for permission to collect, export, and import tional Science Foundation under grants DEB-9705277
samples. Permits to collect in and export sample from and DEB-0102383 to J. Campbell, and a grant from
Costa Rica were graciously provided to A.J.C. by the the Wildlife Conservation Society to E.N.S.
MINAE and J. Guevara, with the support of F. Bolaños
of the Univerdidad de Costa Rica. Permission to collect
locally was kindly granted by CATIE, the OTS, L.D.
References
Gomez, B. Young, R. Matlock, and the Osa Pulcha wo-
menÕs collective. In Panama, permits were kindly Akaike, H., 1974. A new look at the statistical model identification.
granted to A.J.C. by ANAM and PEMASKY (Comar- IEEE Trans. Automatic Control 19, 716–723.
ca de Kuna Yala) and obtained with the help of O. Aro- AmphibiaWeb: Information on amphibian biology and conservation
semena and M. Leone at STRI. Local permission was [web application]. 2005. Berkeley, California: AmphibiaWeb.
kindly granted by IRHE. In Mexico, the necessary per- Available from: <http://amphibiaweb.org/>.
Anderson, S., Bankier, A.T., Barrell, B.G., de Bruijn, M.H.L.,
mits were provided by SEMARNAT. Permits for con- Coulson, A.R., Drouin, J., Eperon, I.C., Nierlich, D.P., Roe,
ducting research in Guatemala were granted to J.A. B.A., Sanger, F., Schreier, P.H., Smith, A.J.H., Staden, R., Young,
Campbell and E.N.S. by the CONAP through the help I.G., 1981. Sequence and organization of the human mitochondrial
of E. Dı́az de Gordı́llo, O.F. Lara, and M. Garcı́a. Sup- genome. Nature 290, 457–465.
Audemard, F.E., Audemard, F.A., 2002. Structure of the Mérida
port for Guatemalan research was facilitated through
Andes, Venezuela: relations with the South America–Caribbean
M. Dix from the Universidad del Valle de Guatemala geodynamic interaction. Tectonophysics 345, 299–327.
and O. Lara from the Universidad de San Carlos de Benabib, M., Kjer, K.M., Sites Jr., J.W., 1997. Mitochondrial DNA
Guatemala. A.J.C. thanks K. Lips, C. Jaramillo, A. sequence-based phylogeny and the evolution of viviparity in the
Sceloporus scalaris group (Reptilia, Squamata). Evolution 51, Elinson, R.P., del Pino, E.M., Townsend, D.S., Cuesta, F.C.,
1262–1275. Eichhorn, P., 1990. A practical guide to the developmental biology
Bromham, L., Cardillo, M., 2003. Testing the link between the of terrestrial breeding frogs. Biol. Bull. 179, 163–177.
latitudinal gradient in species richness and rates of molecular Elinson, R.P., Ninomiya, H., 2003. Parallel microtubules and other
evolution. J. Evol. Biol. 16, 200–207. conserved elements of dorsal axial specification in the direct
Burke, K., 1988. Tectonic evolution of the Caribbean. Ann. Rev. Earth developing frog, Eleutherodactylus coqui. Dev. Genes Evol. 213,
Planet. Sci. 16, 201–230. 28–34.
Callery, M.C., Fang, H., Elinson, R.P., 2001. Frogs without polliwogs: Emslie, S.D., Morgan, G.S., 1995. Taphonomy of a late Pleistocene
evolution of anuran direct development. BioEssays 23, 233–241. carnivore den, Dade County, Florida. In: Steadman, D.W., Mead,
Camin, J.H., Sokal, R.R., 1965. A method for deducing branching J.I. (Eds.), Late Quaternary Environments and Deep History: A
sequences in phylogeny. Evolution 19, 311–326. Tribute to Paul S. Martin. The Mammoth Site of Hot Springs,
Campbell, J.A., 1999. Distribution patterns of amphibians in Middle South Dakota, Scientific Papers 3, pp. 65–83.
America. In: Duellman, W.E. (Ed.), Patterns of Distribution of Farris, J.S., Källersjö, M., Kluge, A.G., Bult, C., 1995. Testing
Amphibians: A Global Perspective. Johns Hopkins University significance of incongruence. Cladistics 10, 315–319.
Press, Baltimore, MD, pp. 111–210. Felsenstein, J., 1981. Evolutionary trees from DNA sequences: a
Campbell, J.A., Savage, J.M., 2000. Taxonomic reconsideration of maximum likelihood approach. J. Mol. Evol. 17, 368–376.
Middle American frogs of the Eleutherodactylus rugulosus group Felsenstein, J., 1985. Confidence limits on phylogenies: an approach
(Anura: Leptodactylidae): a reconnaissance of subtle nuances using the bootstrap. Evolution 39, 783–791.
among frogs. Herp. Monographs 14, 186–292. Ferrusquı́a-Villafranca, I., 1975. Mamı́feros Miocenicos de México:
Coates, A.G., Obando, J.A., 1996. The geologic evolution of the Contribución al conocimiento de la paleozoografı́a del continente.
Central American Isthmus. In: Jackson, J.B.C., Budd, A.F., Rev. Inst. Geol. U.N.A.M. 1, 12–18.
Coates, A.G. (Eds.), Evolution and Environment in Tropical Fitch, W.M., 1971. Toward defining the course of evolution:
America. University of Chicago Press, Chicago, pp. 21–56. minimal change for a specific tree topology. Syst. Zool. 20,
Coates, A.G., Collins, L.S., Aubry, M.-P., Berggren, W.A., 2004. The 406–416.
Geology of the Darien, Panama, and the late Miocene–Pliocene Flores-Villela, O., Kjer, K.M., Benabib, M., Sites Jr., J.W., 2000.
collision of the Panama arc with northwestern South America. Multiple data sets, congruence, and hypothesis testing for the
GSA Bull. 116, 1327–1344. phylogeny of basal groups of the lizard genus Sceloporus (Squa-
Crawford, A.J., 2003a. Huge populations and old species of Costa mata, Phrynosomatidae). Syst. Biol. 49, 713–739.
Rican and Panamanian dirt frogs inferred from mitochondrial and Ford, L.S., Savage, JM., 1984. A new frog of the genus Eleuthero-
nuclear gene sequences. Mol. Ecol. 12, 2525–2540. dactylus (Leptodactylidae) from Guatemala. Occas. Pap. Mus.
Crawford, A.J., 2003b. Relative rates of nucleotide substitution in Nat. Hist. Univ. Kansas 110, 1–9.
frogs. J. Mol. Evol. 57, 636–641. Frost, D.R., 2004. Amphibian Species of the World: an Online
Darst, C.R., Cannatella, D.C., 2004. Novel relationships among hyloid Reference. Version 3.0 (22 August, 2004). Electronic Database
frogs inferred from 12S and 16S mitochondrial DNA sequences. accessible at http://research.amnh.org/herpetology/amphibia/in-
Mol. Phylogenet. Evol. 31, 462–475. dex.html. American Museum of Natural History, New York, USA.
Darwin, C., 1859. On the Origin of Species (A Facsimile of the First Gascuel, O., 1997. BIONJ: an improved version of the NJ algorithm
Edition). Harvard University Press, Cambridge, MA. based on a simple model of sequence data. Mol. Biol. Evol. 14,
De La Riva, I., Lynch, J.D., 1997. New species of Eleutherodactylus from 685–695.
Bolivia (Amphibia: Leptodactylidae). Copeia 1997 (1), 151–157. Guyer, D., Savage, J., 1987. Cladistic relationships among anoles
DeWeese, J.E., 1976. The karyotype of Middle American frogs of the (Sauria: Iguanidae). Syst. Zool. 35, 509–531.
genus Eleutherodactylus (Anura: Leptodactylidae): a case study of Guyer, D., Savage, J., 1992. Anole systematics revisited. Syst. Zool. 41,
the significance of the karylogic method. Ph.D. dissertation. 89–110.
Graduate School of U.S.C., Los Angeles, CA. Hallam, A., 1984. Pre-quaternary sea-level changes. Ann. Rev. Earth
Dı́az de Gamero, M.L., 1996. The changing course of the Orinoco Planet. Sci. 12, 205–243.
River during the Neogene: a review. Palaeogeogr. Palaeoclimatol. Hanken, J., Carl, T.F., Richardson, M.K., Olsson L, Schlosser, G.,
Palaeoecol. 123, 385–402. Osabutey, C.K., Klymkowsky, M.W., 2001. Limb development in
Doan, T.M., Arriaga, W.A., 2002. Microgeographic variation in a ‘‘Nonmodel’’ vertebrate, the direct-developing frog Eleuthero-
species composition of the herpetofaunal communities of Tambo- dactylus coqui. J. Exp. Zool. B 291, 375–388.
pata region, Peru. Biotropica 34, 101–117. Haq, B.U., Hardenbol, J., Vail, P.R., 1987. Chronology of fluctuating
Donato, M., Posadas, P., Miranda-Esquivel, D.R., Ortiz Jaurequizar, sea levels since the Triassic. Science 235, 1156–1167.
E., Cladera, G., 2003. Historical biogeography of the Andes region: Harmon, L.J., Schulte II, J.A., Larson, A., Losos, J.B., 2003. Tempo
evidence from Listroderina (Coleoptera: Curculionidae: Rhytirrh- and mode of evolutionary radiation in iguanian lizards. Science
inini) in the context of the South American geobiotic scenario. Biol. 301, 961–964.
J. Linn. Soc. 80, 339–352. Hasegawa, M., Kishino, H., Yano, T., 1987. ManÕs place in Homi-
Duellman, W.E., 1999a. Distribution patterns of amphibians in South noidea as inferred from molecular clocks of DNA. J. Mol. Evol. 26,
America. In: Duellman, W.E. (Ed.), Patterns of Distribution of 132–147.
Amphibians: A Global Perspective. Johns Hopkins University Hass, C.A., Hedges, S.B., 1991. Albumin evolution in the West Indian
Press, Baltimore, MD, pp. 255–328. frogs of the genus Eleutherodactylus (Leptodactylidae): Carribean
Duellman, W.E., 1999b. Global distribution of amphibians: patterns, biogeography and a calibration of the albumin immunological
conservation, and future challenges. In: Duellman, W.E. (Ed.), clock. J. Zool. Lond. 225, 413–426.
Patterns of Distribution of Amphibians: A Global Perspective. Hedges, S.B., 1989. Evolution and biogeography of West Indian frogs
Johns Hopkins University Press, Baltimore, MD, pp. 1–30. of the genus Eleutherodactylus: slow-evolving loci and the major
Duellman, W.E., 2001. The hylid frogs of Middle America, vol. 2. groups. In: Woods, C.A. (Ed.), Biogeography of the West Indies:
Contrib. Herp. 18, 695–1158. Past Present and Future. Sandhill Crane Press, Gainesville, FL, pp.
Duellman, W.E., Pramuk, J.B., 1999. Frogs of the genus Eleuthero- 305–370.
dactylus Anura: Leptodactylidae. in the Andes of northern Peru. Hedges, S.B., 1996. The origin of West Indian amphibians and reptiles.
Sci. Pap. Nat. Hist. Mus. Univ. Kansas 13, 1–78. In: Powell, R., Henderson, R.W. (Eds.), Contributions to West
Indian Herpetology: A Tribute to Albert Schwatz, Contrib. Herp., Lynch, J.D., 1971. Evolutionary relationships, osteology, and zooge-
vol. 12. SSAR, Ithaca, NY, pp. 95–128. ography of leptodactyloid frogs. Univ. Kansas Mus. Nat. Hist.
Hedges, S.B., 1999. Distribution patterns of amphibians in the West Misc. Pub. 53, 1–238.
Indies. In: Duellman, W.E. (Ed.), Patterns of Distribution of Lynch, J.D., 1986. The definition of the Middle American clade of
Amphibians: A Global Perspective. Johns Hopkins University Eleutherodactylus based on jaw musculature (Amphibia: Lepto-
Press, Baltimore, MD, pp. 211–254. dactylidae). Herpetologica 42, 248–258.
Hedges, S.B., Hass, C.A., Maxson, L.R., 1992. Caribbean biogeogra- Lynch, J.D., 2000. The relationships of an ensemble of Guatemalan
phy: molecular evidence for dispersal in West Indian terrestrial and Mexican frogs Eleutherodactylus: Leptodactylidae: Amphibia.
vertebrates. Proc. Natl. Acad. Sci. USA 89, 1909–1913. Rev. Acad. Colomb. Cienc. 24 (90), 67–94.
Hofer, U., Bersier, L.-F., 2001. Herpetofaunal diversity and abun- Lynch, J.D, 2001a. A small amphibian fauna from a previously
dance in tropical upland forests of Cameroon and Panama. unexplored páramo of the Cordillera Occidental in western
Biotropica 33, 142–152. Colombia. J. Herp. 35, 226–231.
Holman, J.A., 1968. Additional Miocene anurans from Florida. Quart. Lynch, J.D, 2001b. Four osteological synapomorphies within Eleut-
J. Florida Acad. Sci. 30, 121–140. herodactylus (Amphibia: Leptodactylidae) and their bearing on
Hoorn, C., 1993. Marine incursions and the influence of Andean subgeneric classifications. Rev. Acad. Colomb. Cienc. 25 (94), 127–
tectonics on the Miocene depositional history of northwestern 136.
Amazonia: results of a palynostratigraphic study. Palaeogeogr. Lynch, J.D., Duellman, W.E., 1997. Frogs of the genus Eleuthero-
Palaeoclimatol. Palaeoecol. 105, 267–309. dactylus in western Ecuador: systematics, ecology, and biogeogra-
Huelsenbeck, J.P., Ronquist, F., 2001. MRBAYES: Bayesian inference phy. Univ. Kansas Nat. Hist. Mus. Spec. Pub. 23, 1–236.
of phylogenetic trees. Bioinformatics 8, 754–755. Lynch, J.D., McDiarmid, R.W., 1987. Two new species of Eleuther-
Iturralde-Vinent, M.A., MacPhee, R.D.E., 1996. Age and paleo- odactylus (Amphibia: Anura: Leptodactylidae) from Bolivia. Proc.
geographical origin of Dominican amber. Science 273, 1850– Biol. Soc. Washington 100, 337–346.
1852. Macey, J.R., Larson, A., Ananjeva, N.B., Fang, Z., Papenfuss, T.J.,
Iturralde-Vinent, M.A., MacPhee, R.D.E., 1999. Paleogeography of 1997a. Two novel gene orders and the role of light-stand
the Caribbean region: implications for Cenozoic biogeography. replication in rearrangement of the vertebrate mitochondrial
Bull. Am. Mus. Nat. Hist. 238, 1–95. genome. Mol. Biol. Evol. 14, 91–104.
Jackman, T.R., Larson, A., De Queiroz, K., Losos, J.B., 1999. Macey, J.R., Larson, A., Ananjeva, N.B., Fang, Z., Papenfuss, T.J.,
Phylogenetic relationships and tempo of early diversification in 1997b. Replication slippage may cause parallel evolution in the
Anolis lizards. Syst. Biol. 48, 254–285. secondary structures of mitochondrial transfer RNAs. Mol. Biol.
Jukes, T.H., Cantor, C.R., 1969. Evolution of protein molecules. In: Evol. 14, 31–39.
Munro, H.N. (Ed.), Mammalian Protein Metabolism. Academic Macey, J.R., Larson, A., Ananjeva, N.B., Papenfuss, T.J., 1997c.
Press, New York, pp. 21–123. Evolutionary shifts in three major structural features of the
King, M.W., 1991. Developmentally regulated alternative splicing in mitochondrial genome among iguanian lizards. J. Mol. Evol. 44,
the Xenopus laevis c-Myc gene creates an intron-1 containing c-Myc 660–674.
RNA present only in post-midblastula embryos. Nucleic Acids Macey, J.R., Schulte II, J.A., Larson, A., Fang, Z., Wang, Y.,
Res. 19, 5777–5783. Tuniyev, B.S., Papenfuss, T.J., 1998. Phylogenetic relationships of
Kishino, H., Hasegawa, M., 1989. Evaluation of the maximum toads in the Bufo bufo species group from the eastern escarpment of
likelihood estimate of the evolutionary tree topologies from the Tibetan Plateau: a case of vicariance and dispersal. Mol.
DNA sequence data, and the branching order in Hominoidea. J. Phylogenet. Evol. 9, 80–87.
Mol. Evol. 29, 170–179. Macey, J.R., Strasburg, J.L., Brisson, J.A., Vredenburg, V.T.,
Kishino, H., Thorne, J.L., Bruno, W.J., 2001. Performance of a Jennings, M., Larson, A., 2001. Molecular phylogenetics of western
divergence time estimation method under a probabilistic model of North American frogs of the Rana boylii species group. Mol.
rate evolution. Mol. Biol. Evol. 18, 352–361. Phylogenet. Evol. 19, 131–143.
Köhler, J., 2000. New species of Eleutherodactylus (Anura: Leptodac- MacFadden, B.J., 2005. Middle Miocene land mammals from the
tylidae) from cloud forest of Bolivia. Copeia 2000, 516–520. Cucaracha Formation (Hemingfordian-Barstovian) of Panama. J.
Kuenzi, W.D., Horst, O.H., McGehee, R.V., 1979. Effect of volcanic Vert. Paleontology (in press).
activity on fluvial-deltaic sedimentation in a modern arc-trench Marshall, L.G., Lundberg, J.G., 1996. Miocene deposits of the
gap, southwestern Guatemala. Amazon Basin foreland. Technical comments. Science 273, 123–
Kumazawa, Y., Nishida, M., 1993. Sequence evolution of mitochon- 124.
drial tRNA genes and deep-branch animal phylogenetics. J. Mol. McCranie, J.R., Wilson, L.D., 2002. The amphibians of Honduras.
Evol. 37, 380–398. Contrib. Herp. 19, 1–625.
Lake, J.A., 1994. Reconstructing evolutionary trees from DNA and Miyamoto, M.M., 1983. Frogs of the Eleutherodactylus rugulosus
protein sequences: paralinear distances. Proc. Natl. Acad. Sci. USA group: a cladistic study of allozyme, morphological, and karyolog-
91, 1455–1459. ical data. Syst. Zool. 32, 109–124.
Lieberman, S.S., 1986. Ecology of the leaf litter herpetofauna of a Miyamoto, M.M., 1984. Central American frogs allied to Eleuthero-
Neotropical rain forest: La Selva, Costa Rica. Acta Zool. Mex. dactylus cruentus: allozyme and morphological data. J. Herp. 18,
(n.s.) 15, 1–72. 256–263.
Lloyd, J.J., 1963. Tectonic history of the South Central-American Miyamoto, M.M., 1986. Phylogenetic relationships and systematics of
orogen. In: Childs, O.E., Beebe, B.W. (Eds.), Backbone of the the Eleutherodactylus fitzingeri group (Anura, Leptodactylidae).
Americas, a Symposium. American Association of Petroleum Copeia 1986, 503–511.
Geologists, Tulsa, Oklahoma, pp. 88–100. Nicholas, K.B., Nicholas, H.B., 1997. GeneDoc: a tool for editing and
Lockhart, P.J., Steel, M.A., Hendy, M.D., Penny, D., 1994. Recov- annotating multiple sequence alignments. Distributed by the
ering evolutionary trees under a more realistic model of sequence author at: www.cris.com/~Ketchup/genedoc.shtml.
evolution. Mol. Biol. Evol. 11, 605–612. Olson, S.L., Pregill, G.K., 1982. Fossil vertebrates from the Bahamas.
Lynch, J.D., 1964. Additional hylid and leptodactylid remains from Smithsonian Contrib. Paleobiol. 48, 1–7.
the Pleistocene of Texas and Florida. Herpetologica 20, 141– Orvis, K.H., Horn, S.P., 2000. Quaternary glaciers and climate on
142. Cerro Chirripó, Costa Rica. Q. Res. 54, 24–37.
Pindell, J.L., Cande, S.C., Pitman III, W.C., Rowley, D.B., Dewey, genus Sceloporus (Squamata: Phrynosomatidae). The University of
J.F., Labrecque, J., Haxby, W., 1988. A plate-kinematic framework Texas at Arlington, Arlington, Texas, Dissertation, pp. 356.
for models of Caribbean evolution. Tectonophysics 155, 121–138. Swofford, D.L., 1998. PAUP*. Phylogenetic Analysis Using Parsi-
Posada, D., Crandall, K.A., 1998. Modeltest: testing the model of mony * and Other Methods., Version 4b10. Sinauer Associates,
DNA substitution. Bioinformatics 14, 817–818. Sunderland, MA.
Rannala, B., Yang, Z., 1996. Probability distribution of molecular Tamura, K., Nei, M., 1993. Estimation of the number of nucleotide
evolutionary trees: a new method of phylogenetic inference. J. Mol. substitutions in the control region of mitochondrial DNA in
Evol. 43, 304–311. humans and chimpanzees. Mol. Biol. Evol. 10, 512–526.
Rogers, D., 2000. Slab break-off mechanism for the late Neogene uplift Tavaré, S., 1986. Some probabilistic and statistical problems on the
of Honduras, Northern Central America. Available from: <http:// analysis of DNA sequences. Lec. Math. Life Sci. 17, 57–86.
www.ig.utexas.edu/research/projects/honduras/agu2000/ Thorne, J.L., Kishino, H., 2002. Divergence time and evolutionary rate
Rogers2000.html>. estimation with multilocus data. Syst. Biol. 51, 689–702.
Rosen, D.E., 1975. A vicariance model of Caribbean biogeography. Thorne, J.L., Kishino, H., Painter, I.S., 1998. Estimating the rate of
Syst. Zool. 24, 431–464. evolution of the rate of molecular evolution. Mol. Biol. Evol. 15,
Rosen, D.E., 1985. Geological hierarchies and biogeographical con- 1647–1657.
gruence in the Caribbean. Ann. Missouri Bot. Garden 72, 636–659. Tihen, J.A., 1960. Notes on Late Cenozoic hylid and leptodactylid
Ross, M.I., Scotese, C.R., 1988. A hierarchical model of the Gulf of frogs from Kansas, Oklahoma, and Texas. Southwestern Natural-
Mexico and Caribbean region. Tectonophysics 155, 139–168. ist 5, 66–70.
Saitou, N., Nei, M., 1987. The neighbor-joining method: a new method Toft, C.A., 1980a. Seasonal variation in populations of Panamanian
for reconstructing phylogenetic trees. Mol. Biol. Evol. 4, 406–425. litter frogs and their prey: a comparison of wetter and drier sites.
Savage, J.M., 1966. The origins and history of the Central American Oecologia 47, 34–38.
herpetofauna. Copeia 1966, 719–766. Toft, C.A., 1980b. Feeding ecology of thirteen syntopic species of
Savage, J.M., 1981. The systematic status of Central American frogs anurans in a seasonal tropical environment. Oecologia 45, 131–141.
confused with Eleutherodactylus cruentus. Proc. Biol. Soc. Wash. Vail, P.R., Hardenbol, J., 1979. Sea-level changes during the Tertiary.
94, 413–420. Oceanus 22, 71–79.
Savage, J.M., 1982. The enigma of the Central American herpetofa- Vanzolini, P.E., Heyer, W.R., 1985. The American herpetofauna and
una: dispersals or vicariance?. Ann. Missouri Bot. Gard. 69 464– the interchange. In: Stehli, F.G., Webb, S.D. (Eds.), The Great
547. American Biotic Interchange. Plenum Press, New York, pp. 475–
Savage, J.M., 1987. Systematics and distribution of the Mexican and 487.
Central American rainfrogs of the Eleutherodactylus gollmeri group Vences, M., Andreone, F., Glaw, F., Kosuch, J., Meyer, A., Schaefer,
(Amphibia: Leptodactylidae). Fieldiana: Zoology 33, 1–57. H.-C., Veith, M., 2002. Exploring the potential of life-history key
Savage, J.M., 2002. The Amphibians and Reptiles of Costa Rica. innovation: brook breeding in the radiation of the Malagasy tree
University of Chicago Press, Chicago, IL. frog genus Boophis. Mol. Ecol. 11, 1453–1463.
Savage, J.M., Emerson, S.B., 1970. Central American frogs allied to Vences, M., Kosuch, J., Rödel, M.-O., Lötters, S., Channing, A.,
Eleutherodactylus bransfordii (Cope): a problem of polymorphism. Glaw, F., Böhme, W., 2004. Phylogeography of Ptychadena
Copeia 1970, 623–644. mascareniensis suggests transoceanic dispersal in a widespread
Savage, J.M., Myers, C.W., 2002. Frogs of the Eleutherodactylus African-Malagasy frog lineage. J. Biogeogr. 31, 593–601.
biporcatus group (Leptodactylidae) of Central America and north- Vences, M., Vieites, D.R., Glaw, F., Brinkmann, H., Kosuch, J., Veith,
ern South America, including rediscovered, resurrected, and new M., Meyer, A., 2003. Multiple overseas dispersal in amphibians.
taxa. Am. Mus. Novitates 3357, 1–48. Proc. Roy. Soc. Lond. B 270, 2435–2442.
Savage, J.M., Hollingsworth, B.D., Lips, K.R., Jaslow, A.P., 2004. A Vergara, S.L., 1997. Stratigraphy, foraminiferal assemblages and
new species of rainfrog (genus Eleutherodactylus) from the Serranı́a paleoenvironments in the late Cretaceous of the upper Magdalena
de Tabasará, west-central Panama and reanalysis of the fitzingeri valley, Colombia (Part I). J. South Am. Earth Sci. 10, 111–
species group. Herpetologica 60, 519–529. 132.
Scott Jr., N.J., 1976. The abundance and diversity of the herpetofaunas Webb, S.D., Rancy, A., 1996. Late Cenozoic evolution of the
of tropical forest litter. Biotrópica 8, 41–58. Neotropical mammal fauna. In: Jackson, J.B.C., Budd, A.F.,
Schluter, D., 2000. The Ecology of Adaptive Radiation. Oxford Coates, A.G. (Eds.), Evolution and Environment in Tropical
University Press, New York. America. University of Chicago Press, Chicago, pp. 335–358.
Shao, R., Dowton, M., Murrell, A., Barker, S.C., 2003. Rates of gene Whitmore Jr., F.C., Stewart, R.H., 1965. Miocene mammals and
rearrangement and nucleotide substitution are correlated in the Central American seaways. Science 148, 180–185.
mitochondrial genomes of insects. Mol. Biol. Evol. 20, 1612–1619. Yang, Z., 1994. Maximum likelihood phylogenetic estimation from
Shimodaira, H., Hasegawa, M., 1999. Multiple comparisons of log- DNA sequences with variable rates over sites: approximate
likelihoods with applications to phylogenetic inference. Mol. Biol. methods. J. Mol. Evol. 39, 306–324.
Evol. 16, 1114–1116. Yang, Z., Rannala, B., 1997. Bayesian phylogenetic inference using
Smith, E.N., 2001. Species boundaries and evolutionary patterns of DNA sequences: a Markov chain Monte Carlo method. Mol. Biol.
speciation among the malachite lizards (formosus group) of the Evol. 14, 717–724.

Crawford 2005

Uploaded by

Copyright:

Available Formats

Crawford 2005

Uploaded by

Document Information

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Crawford 2005

Uploaded by

Copyright:

Available Formats

MOLECULAR

Cenozoic biogeography and evolution in direct-developing frogs

Received 16 May 2004; revised 18 February 2005

Genomic DNA was extracted from liver and/or thigh

Table 2 Table 2 (continued)

lived during the Oligocene or possibly the Early 4. Discussion

4.2. The origin of the subgenera of Eleutherodactylus

Our Bayesian phylogenetic analysis of the combined

41% between Craugastor and Eleutherodactylus. Look-

4.4. The origin and diversiﬁcation of Craugastor, the ﬁrst

The Cretaceous–Paleocene proto-Antillean land

You might also like