ELSEVIER Available online at www.sciencedirect.com 'I' ScienceDirect Molecular Phylogenetics and Evolution 43 (2007) 1076-1095 MOLECULAR PHYLOGENETICS AND EVOLUTION www.elsevier.com/locate/ympev Phylogeny of the Procyonidae (Mammalia: Carnivora): Molecules, morphology and the Great American Interchange Klaus-Peter Koepfli ^'*, Matthew E. Gompper ^, Eduardo Eizirik ^, Cheuk-Chung Ho ^, Leif Linden ^, Jesus E. Maldonado ^, Robert K. Wayne ^ '^ Department of Ecology and Evolutionary Biology, University of California, Los Angeles, CA 90095-1606, USA Department of Fisheries and Wildlife Sciences, University of Missouri, Colombia, MO 65211, USA "Faculdade de Biociencias, PUCRS, Av. Ipiranga, 6681, Predio 12, Porto Alegre, RS 90619-900, Brazil "^ Smithsonian Institution, NMNHINZP?Genetic Program, 3001 Connecticut Avenue NW, Washington, DC 20008, USA Received 10 June 2006; revised 22 September 2006; accepted 2 October 2006 Available online 11 October 2006 Abstract The Procyonidae (Mammalia: Carnivora) have played a central role in resolving the controversial systematics of the giant and red pandas, but phylogenetic relationships of species within the family itself have received much less attention. Cladistic analyses of morpho- logical characters conducted during the last two decades have resulted in topologies that group ecologically and morphologically similar taxa together. Specifically, the highly arboreal and frugivorous kinkajou (Polos flavus) and olingos (Bassaricyon) define one clade, whereas the more terrestrial and omnivorous coatis (Nasua), raccoons (Procyon), and ringtails (Bassariscus) define another clade, with the similar-sized Nasua and Procyon joined as sister taxa in this latter group. These relationships, however, have not been tested with molecu- lar sequence data. We examined procyonid phylogenetics based on combined data from nine nuclear and two mitochondrial gene seg- ments totaling 6534 bp. We were able to fully resolve relationships within the family with strongly supported and congruent results from maximum parsimony, maximum likelihood, minimum evolution, and Bayesian analyses. We identified three distinct lineages within the family: a (Nasua, Bassaricyon) clade, a (Bassariscus, Procyon) clade, and a Polos lineage, the last of which is sister to the other two clades. These findings, which are in strong disagreement with prior fossil and morphology-based assessments of procyonid relationships, reem- phasize the morphological and ecological flexibility of these taxa. In particular, morphological similarities between unrelated genera pos- sibly reflect convergence associated with similar lifestyles and diets rather than ancestry. Furthermore, incongruence between the molecular supermatrix and a morphological character matrix comprised mostly of dental characters [Baskin, J.A., 2004. Bassariscus and Probassariscus (Mammalia, Carnivora, Procyonidae) from the early Barstovian (Middle Miocene). I. Vert. Paleo. 24, 709-720] may be due to non-independence among atomized dental characters that does not take into account the high developmental genetic correlation of these characters. Finally, molecular divergence dating analyses using a relaxed molecular clock approach suggest that intergeneric and intrageneric splits in the Procyonidae mostly occurred in the Miocene. The inferred divergence times for intrageneric splits for several genera whose ranges are bisected by the Panamanian Isthmus is significant because they suggest diversification well precedes the Great American Interchange, which has long been considered a primary underlying mechanism for procyonid evolution. ? 2006 Elsevier Inc. All rights reserved. Keywords: Procyonidae; Carnivora; Molecular phylogeny; Great American Interchange; Biogeography; Molecules versus morphology 1. Introduction Corresponding author. Fax: +1310 206 3987. E-mail address: klausk@lifesci.ucla.edu (K.-P. Koepfli). 1055-7903/$ - see front matter ? 2006 Elsevier Inc. All rights reserved, doi: 10.1016/j.ympev.2006.10.003 The Procyonidae is one of the 11 traditional families of the mammalian order Carnivora. The family consists of 14 extant species in six genera (Wozencraft, 2005) that are geo- graphically distributed across the Americas and includes K.-P. Koepfli et al. I Molecular Phylogenetics and Evolution 43 (2007) 1076-1095 1077 highly arboreal tropical frugivores (olingos [Bassaricyon spp.] and the prehensile-tailed kinkajou [Potos flavus]) and terres- trial to arboreal omnivores found across a diversity of habi- tats (ringtail and cacomistle [Bassariscus spp.]; coatis [Nasua spp. and Nasuella oliv?cea]; and raccoons [Procyon spp.]) (Zeveloff, 2002; Nowak, 2005). As a result of their diets and in contrast to many other species in the Carn?vora, most pro- cyonids are characterized by having hypocarnivorous denti- tions. Coat patterns among the species are diverse, with Nasua, Procyon and Bassariscus having striking tail annula- tions and masks, while in Bassaricyon and Potos such mark- ings are either reduced or lacking altogether. Systematic relationships of the extant Procyonidae have received a great deal of indirect study. As pivotal subjects in the long-running controversy over placement of giant and red pandas (Ailuropoda and Ailurus), procyonids became the perennial outgroup or ingroup (O'Brien etal., 1985, 1991; Mayr, 1986; Tagle et al., 1986; Goldman et al., 1989; Zhang and Shi, 1991; Hashimoto etal., 1993; Pecon Slattery and O'Brien, 1995). Yet these studies did not directly examine relationships within the Procyonidae and earher direct stud- ies of procyonid interrelationships, based on overall similar- ity of various anatomical features, yielded conflicting results. For example, Gill (1872) and Holhster (1916) excluded ring- tails (Bassariscus spp.) from the family, whereas McGrew (1938) adamantly included them. Pocock (1921) concluded that procyonid genera probably shared a common ancestor, but placed each genus in its own subfamily because of their morphological dissimilarites. Davis (1941) and Segafl (1943) both questioned the placement of Potos within procyonids. Story (1951) examined the pattern of the carotid arteries and their branches in Procyon lotor and other procyonid taxa and used these patterns to infer relationships among genera, with Procyon and Nasua being most similar. Recent cladistic analyses using morphological characters have been more congruent and suggest that the family is composed of two primary clades, one containing Bassariscus, Nasua and Procyon, with Nasua (plus Nasuella) and Procyon joined as sister taxa, and the other clade containing Bassari- cyon and Potos (Baskin, 1982, 1989, 1998, 2004; Decker and Wozencraft, 1991) (Fig. 1). These two major clades have been classified either as subfamilies (Procyoninae and Potosinae; Decker and Wozencraft, 1991) or tribes (Procyonini and Potosini; Baskin, 2004). Relying primarily on dental charac- ters. Baskin (1982, 1989, 1998, 2004) investigated the rela- tionships among fossil and extant genera within the family. The classification of procyonids into the Potosini and Pro- cyonini (Fig. lA) stems from Baskin's (1998, 2004) recogni- tion of three subfamilies as comprising the Procyonidae: Procyoninae (which includes the extant genera and their fos- sil relatives), Ailurinae {Ailurus and its fossil relatives), and the extinct Simocyoninae. This arrangement suggests that these three subfamilies descended from a common ancestor, a conclusion supported by other paleontological studies, except that the subfamilies are elevated to famihal rank (Ginsberg et al., 1997; Wang, 1997). However, other neonto- logical and paleontological evidence (e.g., Wolsan, 1993; but see Wang, 1997) suggests that Ailurinae and Simocyoninae are not part of the monophyletic Procyonidae, a conclusion that is firmly supported, at least for the Ailurinae (as repre- sented by Ailurus) by molecular data (Flynn et al., 2000, 2005). Decker and Wozencraft (1991) used 129 craniodental, postcranial and soft anatomical characters to infer relation- ships among the extant genera only (Fig. IB). Despite differ- ences in the composition of the respective character matrices, the congruence of relationships among extant genera in the Baskin (2004) and Decker and Wozencraft studies (Fig. 1) suggest that morphology is a reliable indicator of procyonid phylogeny. These morphology-based phylogenetic hypotheses have not, however, been tested with molecular data, although several studies have hinted at potential disagreement between molecular and morphologic data for these taxa (Sarich, 1973; Couturier and Dutrillaux, 1986). While vari- ous DNA sequence data have been generated for several procyonid genera, these have usually been employed to assess higher level relationships among carnivoran families, especially those included in the Arctoidea (Zhang and Ryder, 1993, 1994; Vrana etal., 1994; Ledje and Arnason, 1996a,b; Dragoo and Honeycutt, 1997; Flynn and Nedbal, 1998; Flynn et al., 2000, 2005; Yu et al., 2004; Dehsle and Strobeck, 2005; Sato et al., 2006). Therefore, a direct assess- ment of relationships among procyonid genera using DNA sequences has, until now, been lacking. Despite being presently restricted to the Americas, pro- cyonids first appear in Late Eocene or Late Ohgocene deposits of Western Europe, represented by fossils of the genus Pseudobassaris (Pohle, 1917; Wolsan, 1993; Wolsan and Lange-Badr?, 1996; McKenna and Bell, 1997). Other fossil taxa from European localities assigned to the Procy- onidae are also known (see McKenna and Bell, 1997), but paleontologists differ considerably over the taxonomic affinities of these taxa (e.g., Wolsan, 1993; Baskin, 2004). Procyonids first appear in the North American fossil record in the Early Miocene (Hunt, 1996; Baskin, 1998, 2004). Phylogenetic analyses based on morphological characters suggest a shared ancestry between North Amer- ican genera and the Early Miocene taxon Broiliana of Europe (Baskin, 1998, 2004). Once in North America, pro- cyonids diversified and gave rise to a number of genera, including the extant lineages. The hneage leading to Bas- saricyon and Potos (which both lack a fossil record) is believed to have diverged in the Early Miocene, based on their presumed relationship to the stem fossil genera Bas- saricyonoides and Parapotos (Baskin, 2003) (Fig. lA). This is consistent with Simpson's (1945) earher supposition of a Miocene ancestry of the kinkajou based on its highly derived morphology. Fossils of Bassariscus first appear in the early Middle Miocene, and Nasua and Procyon first appear in the Early Pliocene and Late Miocene, respec- tively (Baskin, 1982, 1998, 2004). Therefore, fossil evidence suggests that most intergeneric divergences among extant genera occurred during the Miocene epoch, most likely in southern North American and/or Central American 1078 K.-P. Koepfli et al. I Molecular Phylogenetics and Evolution 43 (2007) 1076-1095 A Baskin (2004) B Decker & Wozencraft (1991) Amphictis^ (EU - NA) Broiliana^ (EU) Bassaricyonoides^ (NA) Parapetos^ (NA) Bassaricyon Potos Bassariscus Probassariscus^ (NA) Edaphocyon^ (NA) Arctonasua^ (NA) Cyonasua^ (SA) Amphinasua^ (SA) Brachynasua^ (SA) Chapalmalania^ (SA) Protoprocyon^ (NA) Paranasua^ (NA) Nasua Procyon Potosini Procyonini ? Bassaricyon ? Potos ? Bassariscus ? Nasua ? Nasuella Potoslnae Procyon Procyoninae Fig. 1. Phylogenetic hypotheses of Procyonidae derived from morphologic cladistic analyses. (A) Phylogeny of fossil and extant genera from Baskin, 2004. ?mphictis is the outgroup. Extinct genera are denoted by a cross (f). Geographical origins of fossil taxa are indicated in parentheses. EU, Europe; NA, North America; SA, South America. (B) Phylogeny of extant genera only from Decker and Wozencraft, 1991. Proposed tribal (A) or subfamilial (B) aflBl- iations are shown at right. subtropical and/or tropical forests (Simpson, 1945; Hershko- vitz, 1972; Baskin, 1982, 1998; Webb, 1985). A well-resolved molecular phylogeny of the Procyonidae, combined with reli- able estimates of divergence times, can be used to examine temporal patterns of diversification in this group, allowing comparisons with inferences based on the fossil record. The timing of intrageneric divergences is less well under- stood, particularly with regard to species distributed in South America. This is primarily due to either the complete lack of a fossil record for some species [e.g., five species of Bassaricyon, although these may represent just one species (Poglayen-Neuwall and Poglayen-Neuwall, 1965; Nowak, 2005)] or the fragmentary record for other species (e.g.. Nasua spp. and Procyon spp.). Regarding the latter, fossils of Nasua and Procyon from South America are first recorded in Early to Late Pleistocene deposits (Berta et al., 1978; Marshall et al., 1984). Therefore, a large temporal and spatial gap exists between the time and place Nasua and Procyon first appear in North America (Late Miocene and Early Phocene, respectively [ca. 5.3 mya for this boundary]) and the time they first appear in South America (Early to Late Pleistocene, 1.5-0.1 mya). Intriguingly, however, cur- rent geographic ranges of Bassaricyon, Nasua, Procyon, and Potos are bisected by the Isthmus of Panama. The Isthmus rose above sea level ca. 3-2.5 mya in the Late Phocene, forming a landbridge between North and South America K.-P. Koepfli et al. I Molecular Phylogenetics and Evolution 43 (2007) 1076-1095 1079 that facilitated one of the major biological events of late Cenozoic history, the Great American Interchange (Mar- shall etal., 1979, 1982; Marshall, 1985; Webb, 1985, 1997; Marshall and Sempere, 1993; Coates and Obando, 1996). Ranges of putative sister species within Nasua, Procyon, and possibly Bassaricyon overlap on or meet near the Isth- mus of Panama, suggesting that current procyonid diversity and biogeography may have been influenced by formation of the Panamanian landbridge. Whether the rise of the Isth- mus of Panama played a role in intrageneric speciation events of procyonids can be tested by comparing molecular divergence times of the species involved with the well- defined geochronology of the formation and closure of the Panamanian landbridge (e.g., Marshafl et al., 1979; Mar- shall, 1985; Marshall and Sempere, 1993). Here, we present sequence data from two mitochondrial and nine nuclear gene segments to examine phylogenetics and divergence dating of the Procyonidae from a molecular perspective. Our primary aims are to: (1) critically evaluate relationships among genera within the family, (2) assess congruence between phylogenetic hypotheses generated from the present molecular study and previous cladistic morphological studies, and (3) estimate the relative diver- gence times of genera as well as sister species. The latter issue is particularly interesting in the context of better understanding the biogeographic history of the family and the potentially pivotal role that the rise of the Isthmus of Panama may have had on the radiation of procyonids. 2. Materials and methods 2.1. Taxonomic sampling Our study includes nine species in the Procyonidae, rep- resenting all genera except Nasuella (Table 1). Numerous lines of evidence suggest that Mustelidae (badgers, martens. otters, and weasels) is sister group to Procyonidae (e.g., de Jong, 1986; Flynn et al., 1988, 2000, 2005; Sato et al., 2006). Therefore, we used five species from the Mustelidae as out- groups (Table 1). Our sampling of taxa does not include the red panda (Ailurus fulgens) because recent studies based on both mitochondrial and nuclear DNA sequences have shown that this taxon diverged prior to divergence of Procyonidae and Mustelidae (Flynn and Nedbal, 1998; Flynn et al., 2000,2005; Yu et al., 2004) and most likely rep- resents a monotypic lineage within the newly defined Musteloidea (Flynn et al., 2005; Sato et al., 2006). 2.2. Laboratory methods Total genomic DNA was extracted from tissue samples using a commercial kit (QIAamp DNA Mini Kit). We used the polymerase chain reaction (PCR) to amplify segments containing exon sequences or exon plus intron sequences from nine nuclear loci and two coding regions from the mitochondrial genome (Table 2). Amplification reaction and cycling conditions for all loci, and purification of PCR products, followed Koepfli and Wayne (2003). PCR prod- ucts were cycle sequenced using PCR primers and CEQ Dye Terminator Cycle Sequencing Quick Start Kit (Beck- man Coulter). Sequencing reactions were precipitated fol- lowing the manufacturer's protocol and products were sequenced on a CEQ2000XL automated capillary sequencer (Beckman Coulter). Electropherograms were checked for accuracy and edited using Sequencher 3.1 (Gene Codes Corporation). We were unable to amplify RAG2 sequence from Bassariscus sumichrasti, and question marks were used to represent missing data for this taxon. Seven previously generated CHRNAl and two CYTb sequences were from Koepfli and Wayne (2003) and four CYTb sequences were from Koepfli and Wayne (1998) (Appendix A). Newly generated sequences were deposited Table 1 Species, common name and origin of the tissue samples for the taxa used in this study Species Common name Source locality Voucher or collector information Procyonidae Bassaricyon alleni Allen's olingo Amazonas, Peru MVZ 155219 Bassaricyon gabbii Bushy-tailed olingo Limbo, Panama R. Kays Bassariscus astutus Ringtail Arizona, USA S. Ratnayeke Bassariscus sumichrasti Cacomistle Veracruz, Mexico I. Poglayen Nasua narica White-nosed coati Barro Colorado Island, Panama M. Gompper Nasua nasua South American coati Santa Cruz, Bolivia MSB 12987 Potos flavus Kinkajou San Jose, Costa Rica LSU-MNS 28393 Procyon cancrivorus Crab-eating raccoon Presidente Hayes, Paraguay MSB 15018 Procyon lotor North American raccoon Montana, USA UAM 33819 Mustelidae Eira barbara Tayra Santa Cruz, Bolivia MSB 58756 Enhydra lutris Sea otter Mednyi Island, Russia J. Bodkin Martes americana American marten Wyoming, USA H. Henry Mustela vison American mink Texas, USA TK 29694 Taxidea taxus American badger New Mexico, USA MSB 64932 LSU-MNS, Louisiana State University, Museum of Natural Sciences; MSB, Museum of Southwestern Biology, University of New Mexico; MVZ, Museum of Vertebrate Zoology, University of California, Berkeley; TK, Museum of Texas Tech University; UAM, University of Alaska Museum, Fair- banks. 1080 K.-P. Koepfli et al. I Molecular Phylogenetics and Evolution 43 (2007) 1076-1095 Table 2 Abbreviated gene symbol, type of sequence, and forward (F) and reverse (R) primer sequences for nine nuclear and two mitochondrial gene segments amplified and sequenced in this study Gene Type of sequence Primer sequences (5' -^ 3') Reference ?DOR?3 ?POB BDNF CHRNAl COLlOAl PNOC RAGl RAG2 Exon Exon Exon Exon/intron Exon WTl CYTb NADH5 Exon Exon Exon Exon/intron Mitochondrial coding Mitochondrial coding F: ACC CCC ATG TTT GGC TGG AA R: GAT AGG GTT CAT CAT GGA GTT F: GTG CCA GGT TCA ATC AGT ATA AGT R: CCA GCA AAA TTT TCT TTT ACT TCA A F: CAT CCT TTT CCT TAC TAT GGT T R: TTC CAG TGC CTT TTG TCT ATG F: GAC CAT GAA GTC AGA CCA GGA G R: GGA GTA TGT GGT CCA TCA CCA T F: ATT CTC TCC AAA GCT TAC CC R: GCC ACT AGG AAT CCT GAG AA F: GCA TCC TTG AGT GTG AAG AGA A R: TGC CTC ATA AAC TCA CTG AAC C F: GCT TTG ATG GAC ATG GAA GAA GAC AT R: GAG CCA TCC CTC TCA ATA ATT TCA GG F: TCA TGG AGG GAA AAC ACC AAA R: TGC ACT GGA GAC AGA GAT TC F: GAG AAA CCA TAC CAG TGT GA R: GTT TTA CCT GTA TGA GTC CT F: CGA AGC TTG ATA TGA AAA ACC ATC GTT G F: GCA AGC TTC TAC CAT GAG GAC AAA TAT C F: ATA GAC AAA ATC CCA TTC CA R: TAG TTG TCA GGG TCT CCT AG R: AAC TGC AGT CAT CTC CGG TTT ACA AGA C F: GGT GCA ACT CCA AAT AAA AGT A R: AGA ATT CTA TGA TGG ATC ATG T Murphy et al. (2001) Amrine-Madsen et al. (2003) Jiang et al. (1998) Murphy et al. (2001) Lyons et al. (1997) Venta et al. (1996) Murphy et al. (2001) Teeling et al. (2000) Murphy et al. (2001) Venta et al. (1996) L14724; Irwin et al. (1991) L15162; Irwin et al. (1991) L15408; Irwin et al. (1991) H15494; Koepfli and Wayne (1998) H15915; Irwin et al. (1991) Waits et al. (1999) Gene names are: ADORA3, A3 adenosine receptor; APOB, apolipoprotein B; BDNF, brain-derived neurotrophic factor; CHRNAl, cholinergic receptor, nicotinic, a polypeptide 1 precursor; COLlOAl, collagen type X a I; PNOC, prepronociceptin; RAGl, recombination-activating protein 1; RAG2, recom- bination-activating protein 2; WTl, Wilms tumor 1; CYTb, cytochrome b; NADH5, NADH dehydrogenase subunit 5. in GenBank under the following accession numbers: DQ660167-DQ660306 (Appendix A). 2.3. Phylogenetic analyses Mitochondrial and nuclear gene sequences were manu- ally aligned and alignments for all loci were unambiguous. Three nuclear loci {APOB, CHRNAl and WTl) contained insertions and/or deletions (indels) 1-16 bp in length that necessitated introduction of gaps into alignments of these segments. We reconstructed phylogenetic trees from three different partitions and concatenations of data: (1) nine combined nuclear gene segments; (2) two combined mitochondrial gene segments; and (3) combined nuclear and mitochon- drial segments (hereafter referred to as the supermatrix). Maximum parsimony (MP), maximum likelihood (ML), and minimum evolution (ME) trees were estimated using PAUP* (SwoflFord, 2002), while Bayesian inference (BI) was conducted using MrBayes 3 (Ronquist and Huelsenbeck, 2003). Concatenations of mitochondrial and nuclear gene segments were analyzed with MP, ML, and ME methods only, whereas the supermatrix was analyzed with all four tree reconstruction methods. For MP analyses, all nucleo- tide sites were equally weighted and gaps were coded as present or absent (1 or 0), regardless of length, to utilize potential phylogenetic signal contained by indels (Barriel, 1994). For ME and ML analyses, gaps were coded as miss- ing and we used Modeltest 3.7 (Posada and Crandall, 1998) to estimate the best-fitting model and parameters of DNA evolution (ML) or distance measure (ME) for each data partition. In both MP and ML analyses, we employed heu- ristic searches with 100 replicates of a random stepwise addition and tree bisection-reconnection branch swapping (TBR) algorithm. ME analyses utilized heuristic searches with starting trees obtained via neighbor-joining and TBR branch swapping. Nodal support was assessed by bootstrap resampling, with 1000 (MP and ME) or 500 (ML) pseu- doreplicates using the same heuristic search conditions listed above (except that only five replicates of random stepwise addition were used in ML). For BI, gaps were treated as missing and the best-fitting model of DNA evo- lution selected by Modeltest 3.7 for the supermatrix was used for the likelihood function component of Bayes' rule. We used uniform default priors for model parameters, which were estimated as part of the Markov Chain Monte Carlo (MCMC) analyses. MCMC analyses were run for 3.0 X 10^ generations, with each Markov chain started from K.-P. Koepfli et al. I Molecular Phylogenetics and Evolution 43 (2007) 1076-1095 1081 a random tree and from which trees were sampled every 100 generations, using one cold and three heated chains. We used the program Tracer vl.3 (Rambaut et al., 2003) to check that chains had mixed well and to verify that conver- gence of log-likelihood scores and model parameters had reached a stationary distribution, after discarding 300,000 generations as burn-in. Two separate MCMC runs were performed to ensure consistency of stationary distributions of likelihood values and parameter estimates between runs. Mean posterior estimates and 95% highest posterior densi- ties of likelihood scores and model parameters for the two runs were calculated using Tracer vl.3. Clade credibility values were estimated after discarding the first 3000 trees (following a burn-in of 300,000 generations) from each run. 2.4. Assessing congruence between molecular and morphological phylogenies We used several approaches to assess congruence between the phylogeny based on the molecular supermatrix and those estimated with morphological data matrices of Decker and Wozencraft (1991) and Baskin (2004). First, we compared the topology derived from MP analysis (as well as topologies derived from the model-based methods) of the molecular supermatrix with the original maximum par- simony topologies from the two morphological studies (Fig. 1). Second, we evaluated levels of bootstrap support for nodes in the molecular and morphological phylogenetic hypotheses to assess whether congruence or incongruence was significant or insignificant among data sets. Bootstrap analyses of the most parsimonious trees were not per- formed in either the Decker and Wozencraft (1991) or Baskin (2004) studies. This test was conducted to rule out the possibility that any incongruence was caused by weakly supported nodes in one or both data sets, which would sug- gest an undersampling of characters as the cause of incon- gruence rather than any real character conflict per se (Hillis and Wiens, 2000). We were unable to obtain the original data matrix from the Decker and Wozencraft (1991) (D. Decker, pers. comm.) and could, therefore, not evaluate bootstrap support for nodes recovered in their recon- structed phylogeny. Congruence between the supermatrix phylogeny and the phylogeny from Decker and Wozencraft (1991) was, therefore, assessed by comparison to their pub- lished phylogeny alone (Fig. 2 in Decker and Wozencraft, 1991). For the 18 taxa (17 ingroup, 1 outgroup) in the Baskin (2004) data set, 39 of the 40 characters in the data matrix were coded as ordered as in the original study. Max- imum parsimony bootstrap analysis of this matrix was per- formed using 1000 bootstrap pseudoreplicates with branch and bound search and the "farthest" taxon addition option in PAUP* (SwofFord, 2002). We also adjusted the Baskin (2004) data matrix so that it matched our matrix with respect to generic representation by excluding all extinct genera except Amphictis, which was used to root the tree. We then used the abridged Baskin (2004) matrix to recon- struct maximum parsimony trees (with 39 ordered and one unordered characters) and to perform bootstrap analysis using the same search conditions and settings as those used with the unabridged data matrix. Third, to exclude the possibility that any incongruence between the supermatrix topology and that derived from the abridged Baskin (2004) matrix was due to the use of different outgroups (e.g., Hilhs and Wiens, 2000), we recon- structed unrooted maximum parsimony networks from the two data matrices. For these analyses, one representative species of each genus was chosen from the supermatrix to make taxon representation of the supermatrix equal to the abridged Baskin (2004) matrix (five taxa total). Exhaustive searches were performed with both data matrices sepa- rately a?ii combined using PAUP*. We used 1000 bootstrap pseudoreplicates with branch and bound search and "far- thest" taxon addition option to evaluate nodal support for unrooted trees. The combined analysis was used to infer the most parsimonious network that had the greatest congru- ence among molecular and morphological characters in the context of total evidence (Kluge, 1989). For the combined analyses, we quantified the Bremer support (Bremer, 1988) and partitioned Bremer support (PBS) of nodes in unrooted trees using TreeRot 2.0 (Sorenson, 1998). Finally, MacClade 4.0 (Maddison et al., 2000) was used to map putative synapomorphies from the Decker and Wozencraft (1991) and Baskin (2004) studies onto the molecular supermatrix topology using the simple parsi- mony criterion with accelerated (ACCTRAN) and delayed (DELTRAN) transformation in order to evaluate ancestral character states and derivation based on the molecular topology. ACCTRAN maximizes character state reversals whereas DELTRAN maximizes parallelism of character states (SwofFord and Maddison, 1987; Maddison et al., 2000). 2.5. Estimation oj divergence times Divergence time estimates were performed using the well-supported optimal topology obtained with the molecu- lar supermatrix (see Section 3), using the Relaxed Molecu- lar Clock approach developed by Thorne et al. (1998) and Kishino et al. (2001). This topology and the DNA superma- trix data set were used with the program estbranches to obtain estimates of branch lengths and a matrix of rate var- iance-covariance. The estbranches output was used to infer divergence dates with the program divtimeSb, with 1.0 x 10^ generations run after 100,000 generations of burn-in. Sev- enteen independent runs of divtimeSb were performed to assess convergence of posterior distribution estimates with varying root age priors (see below) and fossil constraint combinations. Final runs used four conservative fossil cali- brations drawn from McKenna and Bell (1997) to con- strain node times: (i) 1.8 mya as a minimum age for the origin of Procyon (applied to node F in Fig. 3); (ii) 3.6 mya as a minimum age for the origin oiNasua (node E in Fig. 3); (iii) 11.2 mya as a minimum age of crown Procyonidae (node 1082 K.-P. Koepfli et al. I Molecular Phylogenetics and Evolution 43 (2007) 1076-1095 1D0 100 73 95 100 97 100 100 83 97 90 Bassaricyon atteni Bassartyon gabbii Nasua narica Nasua nasua ? Bassariscus astutus - Bassariscus sumichrasti ? Pr?yon cancrivorous Pr?yon lotor Potos flavus ? Enhydra lutris Mustela vison Eira barbara Martes americana 100 100 97 Taxidea taxus 1' Bassaricyon atteni Bassaricyon gabbii ? Nasua narica ? Nasua nasua Bassariscus astutus ? Bassariscus sumichrasti _|? Procyon cancrivorous '? Procyon lotor ? Potos flavus Enhydra lutris Mustela vison ? Eira barbara ? Martes americana ~ Taxidea taxus ' 0.1 substitutions/site ? 0.005 substitutions/site Fig. 2. (A) Maximum likeliliood phylogram based on a combined analysis of mitochondrial CYTb and NADH5 gene segments. (B) Maximum likelihood phylogram based on combined analysis of nine nuclear gene segments (ADORA3, APOB, BDNF, CHRNAl, COLlOAl, PNOC, RAGl, RAG2, and WTl). Numbers above internodes, from top to bottom, are bootstrap support values from MP, ML, and ME analyses, respectively. Asterisks above internodes indicate that bootstrap support was 100% in all analyses while dashes indicate bootstrap support <50%. See Table 3 for MP tree statistics and ML model parameters and scores. Branch lengths are proportional to number of substitutions per site (scale bars). H), based on a fossil assigned to Bassariscus, and allowed here to be a stem taxon in that lineage (see Section 4); and (iv) 37 mya as an upper bound for the age of crown Procy- onidae (node H), based on the putative stem taxa Pseudo- bassaris and Amphictis occurring in the Late Eocene and Early Oligoc?ne, respectively. We set the mean of the prior distribution for the ingroup root age at 24 mya (based on the Early Miocene taxon Broiliana, proposed to be the ear- hest fossil procyonine?McKenna and Bell (1997); Baskin (1982,1989, 1998, 2004)) and tested the impact of this prior value on the posterior distribution of nodal ages by varying it in both directions in seven independent runs. 3. Results 3.1. Phylogenetic estimates The length of the supermatrix alignment was 6534 bp (including gaps). The 11 gene segments contributed the fol- lowing number of nucleotide sites to the supermatrix: AD0RA3 (376), APOB (935), BDNF (548), CHRNAl (376), COLlOAl (326), PNOC (287), RAGl (1071), RAG2 (468), WTl (712), CYTb (1140), NADH5 (295). When gaps were coded following Barriel (1994), the length of the ahgn- ment was reduced to 6491 bp. The concatenated mitochon- drial data set (CYTb and NADH5) contained 677 variable characters, of which 525 were parsimony-informative, while the concatenated nuclear data set (with coded gaps) con- tained 743 variable characters, of which 426 were parsi- mony-informative (Table 3). Best-fitting models of DNA substitution were identified by Akaike Information Criteria (AIC) in Modeltest 3.7 (Posada and Crandall, 1998; Posada and Buckley, 2004) for the three data partitions (Table 3). The general time reversible model (Tavare, 1986; Yang, 1994), with among-site rate variation approximated via a gamma distribution and a proportion of invariable sites (GTR+G+I), was selected as the best-fitting model of DNA substitution for the supermatrix partition. This model was employed in the ML, ME (as the distance correction mea- sure) and BI supermatrix analyses. Topologies estimated from MP, ML, and ME analyses of concatenated nuclear data and from ML and ME analyses of the concatenated mitochondrial data were identical (Fig. 2). In the MP analysis of concatenated mitochondrial data. Potos was resolved as the sister lineage to the Bassariscus plus Procyon clade, but this relationship had <50% bootstrap support (results not shown). Otherwise, topology of the MP tree from the mitochondrial data was congruent with that based on the concatenated nuclear data. Relationships among procyonids received 100% bootstrap support by all three methods of analysis for the concatenated nuclear data (Fig. 2B). All but two nodes also had high bootstrap support in most analyses of the concatenated mitochondrial data, although MP bootstrap values tended to be lower than val- ues from ML and ME (Fig. 2A). There were no significantly conflicting nodes between the concatenated mitochondrial and nuclear gene partitions, using a value of ^70% as the criterion for significant bootstrap conflict/support (e.g., Hillis and Bull, 1993; Flynn and Nedbal, 1998). Therefore, we com- bined these partitions into a single supermatrix. Parsimony, likelihood, minimum evolution and Bayesian analyses of the supermatrix yielded phylogenies with identi- cal weU-resolved topologies (Fig. 3). This topology matched trees estimated from separate analyses of concatenated mito- chondrial and nuclear gene partitions (Fig. 2). Rooting the tree using mustehd outgroups, the kinkajou (Potos) is resolved as the sister lineage to a clade that is divided into two subgroups: one containing olingos (Bassaricyon) and coatis (Nasua), and the other containing the ringtail-cacomis- tle (Bassariscus) and raccoons (Procyon). Nodal support for K.-P. Koepfli et al. I Molecular Phylogenetics and Evolution 43 (2007) 1076-1095 1083 99 1.0 |? Bassaricyon alleni A Bassaricyon gabbii Nasua narica Nasua nasua 9S 99 100 1.0 Bassariscus asMus Bassariscus sumichrasti Procyon cammorus Procyon lotor Potos TIavus 100 _^ 100 1.0 Enhydra lutris Mustela vison 99 100 100 1.0 Eira barbara Martes americana Taxidea taxus 0.01 substitutions/site Fig. 3. Maximum likelihood phylogram of procyonids based on combined analysis of 11 gene segments (the supermatrix). Numbers above internodes are, from top to bottom, bootstrap support values from MP, ML, and ME analyses, and posterior probability values from BI, respectively. Asterisks above internodes indicate that bootstrap support and posterior probability were 100% or 1.0, respectively. Posterior probability values from BI based on 50% majority-rule consensus of 27,000 trees after discarding a burn-in of 3000 trees. Branch lengths are proportional to the number of substitutions per site (scale bar). Nodes used in molecular divergence dating analyses are lettered (A-H) and refer to Table 4. See Table 3 for MP tree statistics and ML model parameters and scores and Appendix A for estimated likelihood scores and parameter values from the two BI runs. Illustrations of representative species of procyonid genera (Eisenberg, 1989; Eisenberg and Redford, 1999; used with permission) are shown at right and are, from top to bottoiu: Bassaricyon gabbii, Nasua nasua, Bassariscus sumichrasti, Procyon lotor, and Potos flavus (not shown to scale). relationships within the supermatrix tree is consistently high across all analyses with >95% MP, ML, and ME bootstrap values and 1.0 Bayesian clade credibility values (Fig. 3). Several informative indels also support this pattern of relationships. Two deletions (of 1 and 11 bp, respectively) in the CHRNAl intron and a 1 bp insertion in the WTl intron define synapomorphies for the Bassaricyon plus Nasua clade, while a 2bp deletion in the WTl intron is a synapomorphy for the Bassariscus plus Procyon clade. All other indels are either autapomorphic for a particular procyonid genus or species or represent synapomorphies that are found in the ingroup relative to the outgroup. Furthermore, a curious fea- ture of the CYTb gene in Bassaricyon and Nasua also pro- vides support for uniting these two genera. As first noted by Ledje and Arnason (1996a), the CYTb gene in nearly all Car- n?vora studied is terminated by an AGA stop codon at codon 380, but in Bassaricyon, this codon is occupied by GGG, which codes for a glycine residue. We confirmed this finding by sequencing multiple individuals of Bassaricyon (data not shown). In Nasua, we found that codon 380 codes for either a lysine residue (AAA in N. narica and AAG in N. nasua) or a stop codon (AGA) in some individuals of A^. narica. For Bas- saricyon, Nasua nasua, and some individuals of Nasua narica, the stop codon is instead TAA (based on the mammahan mitochondrial genetic code) at position 382, indicating that the CYTb gene is 6bp longer in these two genera relative to other procyonids and Carn?vora that have been sequenced for this gene. 3.2. Comparison of molecular and morphological phylogenies Phylogenies based on parsimony analyses of two mor- phological data matrices (Decker and Wozencraft, 1991; Baskin, 2004) are largely congruent with one another, but 1084 K.-P. Koepfli et al. I Molecular Phylogenetics and Evolution 43 (2007) 1076-1095 Table 3 Sequence characteristics, descriptive statistics from parsimony analyses, and models and substitution parameters selected by AIC in Modeltest 3.7 for the combined mitochondrial, nuclear, and supermatrix partitions Mitochondrial (2 gene segments) Nuclear (9 gene segments) Supermatrix (11 gene segments) No. of nucleotides 1435 No. variable 677 No. parsimony-informative 525 No. of trees 1 Tree length 1814 Retention index 0.434 Model TrN+r+I -Ini 8949.1117 TTA 0.3444 "c 0.3362 ^G 0.0826 "T 0.2368 / 0.4379 r 0.8903 i-AC 1.0000 ??AG 30.4364 ??AT 1.0000 rca 1.0000 rci 21.4421 rr.T 1.0000 5099/5056 743 426 1 956 0.878 SYM+r 12853.5684 0.2500 0.2500 0.2500 0.2500 0.1727 1.3003 5.2732 0.5575 1.3408 6.5775 1.0000 6534/6491 1580 1056 1 2869 0.623 GTR+r+I 22697.1962 0.2690 0.2598 0.2234 0.2478 0.5764 0.6139 2.3962 8.3456 1.3060 1.0253 19.1958 1.0000 No. of nucleotides, alignment length with gaps included in the alignment/alignment length with gaps coded according to Barriel (1994). TT^, K^, KQ, and re^, empirical base frequencies; /, proportion of invariable sites; F, gamma shape parameter; r^c^ ''AG^ ''AT' ''CG' ''CT' ^i^'l ''GT' i"?te of substitution for specified nucleotides. Model names: TrN, Tamura-Nei (Tamura and Nei, 1993); SYM, symmetrical model (Zharkikh, 1994); GTR, general time reversible model (Tavare, 1986). are highly incongruent with phylogenies derived from three molecular data partitions (compare Fig. 1 with Figs. 2 and 3). In both morphological trees, Nasua is sister to Procyon and Bassaricyon is sister to Potos. Bassariscus is resolved as sister to the ({Nasua, Nasuella) Procyon) clade in the Decker and Wozencraft (1991) tree and in the original Baskin (2004) tree that includes both fossil and extant taxa. Bootstrap analysis of the MP tree using the unabridged Baskin (2004) matrix reveals that clades containing Bassaricyon + Potos and Nasua + Procyon are strongly sup- ported (87 and 92% bootstrap values, respectively. Fig. 4A). However, sister-group relationships of these genera with respect to fossil genera within each clade are only weakly supported (Fig. 4A). In the analysis of the abridged Baskin (2004) matrix, Bassariscus is resolved as the sister lineage to the {Bassaricyon, Potos) and {Nasua, Procyon) clades (Fig. 4B). Thus, for the abridged matrix, placement of Bass- ariscus (and character states that define this taxon) is sensi- tive to whether fossil taxa are included (Baskin, 2004; Fig. 4A) or excluded (present study. Fig. 4B). Nonetheless, bootstrap support values for nodes from the abridged Baskin (2004) matrix are all relatively high, with 88 and 100% for the {Bassaricyon, Potos) clade and {Nasua, Pro- cyon) clades, respectively (Fig. 4B). Unrooted parsimony analyses, with taxon sampling restricted to representatives of five extant genera, also resulted in highly incongruent topologies between the molecular supermatrix and morphological matrix from Baskin (2004) (Figs. 5A and B). Discordance between unrooted networks was significant, with nearly 100% MP bootstrap support for completely different relationships. For example, while the molecular supermatrix joined Nasua with Bassaricyon with 100% bootstrap support, the Baskin (2004) matrix joined Nasua with Procyon, also with 100% bootstrap support. When we combined the two data sets into a single matrix, the unrooted parsimony network topology was identical to the one recovered by analysis of the supermatrix alone, also with high boot- strap support (Fig. 5C). Bremer support values for the two internal nodes were high (Fig. 5C). PBS analysis showed that the supermatrix contributed a substantial value to the total Bremer support at the two nodes, while the morphological matrix contributed either high nega- tive support {Bassaricyon, Nasua) or minimal positive support {Bassariscus, Procyon). The combined unrooted analysis suggests that phylogenetic signal derived from morphological data either conflicts with or provides httle support for relationships based upon the molecular supermatrix. Putative synapomorphies that define the {Bassaricyon, Potos) and {Nasua, Procyon) clades in the two morphologi- cal studies are all resolved as convergences (homoplasies) when they are traced onto the molecular topology using MacClade 4.0 (Maddison et al., 2000) with both accelerated (ACCTRAN) and delayed (DELTRAN) transformation (Appendix A). The {Bassaricyon, Potos) clade is supported by nine and four putative synapomorphies in Decker and Wozencraft (1991) and Baskin (2004), respectively (Appen- dix A). The {Nasua, Procyon) clade is supported by five syn- apomorphies in Decker and Wozencraft (1991) (which also includes Nasuella as part of this clade) and 10 synapomor- phies in Baskin (2004). K.-P. Koepfli et al. I Molecular Phylogenetics and Evolution 43 (2007) 1076-1095 1085 88 92 100 Amphictis^ Broiliana^ Bassaricyonoides^ Parapetos^ Bassaricyon Potos Bassariscus Probassariscus^ Edaphocyoii^ Arctonasua^ Cyonasua^ Amphinasua^ Brachynasua^ Chapalmalania^ Protoprocyon^ Paranasua^ Nasua Procyon Bassaricyon Potos Nasua Procyon Bassariscus A Bassaricyon Bassariscus Fig. 4. (A) Single most-parsimonious cladogram of extant and fossil gen- era of procyonids reconstructed from analysis of 40 morphological char- acters from Baskin (2004). Tree length (TL) = 100, consistency index, excluding uninformative characters (CI) = 0.6022, and retention index (RI) = 0.7673. Arrows indicate nodes that are strongly conflicting with those of the supermatrix topology (Fig. 3). (B) Single most-parsimonious cladogram of extant genera only based on analysis of 40 morphological characters from Baskin (2004). TL = 68, CI, excluding uninformative characters = 0.8182, and RI = 0.7561. For both analyses, 39 characters were ordered and one was unordered, as in Baskin (2004) and both trees were rooted using Amphictis (not shown in B). Numbers above internodes are bootstrap support values, using 1000 pseudoreplicates for each analysis. Original morphological trees (Fig. 1) also define the clade {Bassariscus {Nasua, Procyon)), although this clade is not recovered in our analysis of the abridged Baskin (2004) matrix (Fig. 4B). This clade is supported by nine and two putative synapomorphies in the Decker and Wozencraft (1991) and Baskin (2004) studies, respectively (Appendix A). When these synapomorphies are traced onto the molec- ular topology, their evolutionary transformations are differently resolved under ACCTRAN versus DELTRAN transformations. Under ACCTRAN, derived character states are resolved as symplesiomorphies among the three 100 100 Nasua B Bassaricyon Potos Procyon Nasua 98 100 Potos C Bassaricyon Bassariscus Procyon Bassariscus 39 100 [46 61 100 r?? Nasua Potos Procyon Fig. 5. Unrooted network of extant procyonid genera based on parsimony analysis of (A) the molecular supermatrix (6491 characters; Length (L)=1135, CI, excluding uninformative characters = 0.7071, and RI = 0.5246), (B) 40 (39 ordered and one unordered) morphological char- acters from Baskin (2004) (L = 46, CI, excluding uninformative characters = 0.8611, and RI = 0.8000), and (C) combined molecular and morphological data (L = 1199, CI, excluding uninformative characters = 0.6883, and RI = 0.4909). In all three networks, numbers above internodes are bootstrap support percentages, using 1000 pseudore- plicates. In (C), numbers above boxes are Bremer support values and numbers inside of boxes are partition Bremer support values from the molecular supermatrix (above internode) and from the morphological matrix (below internode). genera, with a secondary loss inferred for Bassaricyon (because Bassaricyon is grouped with Potos in morphologi- cal trees whereas it is grouped with Nasua in molecular trees). The same characters are resolved as convergences between Nasua versus the {Bassariscus, Procyon) clade with DELTRAN. Because the original data matrix from Decker and Wozencraft (1991) was unavailable, we were unable to determine if other characters and their states in this matrix, under parsimony, were consistent with our molecular topology. We were able to explore this with Baskin (2004) matrix, however. Of the 40 characters included in his matrix, only one was consistent with our molecular topol- ogy: character 28, which defines relative placement of para- conid, protoconid and metaconid cusps on the ml trigonid. There are four states for this character (see Appendix 1 in Baskin, 2004): O = paraconid, protoconid and metaconid more or less equally spaced; 1 = paraconid and metaconid close together; 2 = paraconid and metaconid adjacent; and 1086 K.-P. Koepfli et al. I Molecular Phylogenetics and Evolution 43 (2007) 1076-1095 Table 4 Representative results of molecular divergence dating analyses using a Bayesian relaxed molecular clock method Node Age (mya) Constraint set 1" Constraint set 1" Root prior = 24 mya Root prior = 30 mya Constraint set 2^ Root prior = 24 mya A B C D E F G H 2.5 (1.2^.7) 7.0(3.7-12.1) 9.1 (4.9-15.4) 5.0 (2.6-8.7) 11.8(6.4-19.7) 11.4(6.2-19.0) 18.3 (10.3-29.7) 21.6(12.1-34.8) 2.8 (1.3-5.0) 8.0 (4.1-12.9) 10.3 (5.4-16.1) 5.7 (2.8-9.2) 13.3 (7.0-20.6) 12.8 (6.8-19.8) 20.7(11.1-30.9) 24.4(13.3-36.0) 2.8 (1.5^.8) 7.8 (4.8-12.3) 10.1 (6.5-15.7) 5.5 (3.3-8.9) 13.0(8.5-19.9) 12.6 (8.2-19.3) 20.2 (14.0-30.1) 24.0(16.7-35.2) Point estimate of age is indicated for each node (labeled as in Fig. 3), with 95% credibility intervals (CI) shown in parentheses. Root priors indicate the mean of the prior distribution of root-to-tip age used in each run of the analysis. " Constraint set 1 : conservative set of four fossil constraints (minima for nodes E, F and H, and maximum for node H), as described in Section 2. *" Constraint set 2: more restrictive set of five fossil constraints: (i-iii) minima for nodes E and F, and maximum for node H kept as in set 1; (iv) minimum of 11.2 mya based on the earliest putative Bassariscus fossil moved from node HtonodeG, thus making a less conservative assumption about its phyloge- netic placement; and (v) new minimum (16.4 mya) added for node H, based on the young end of the Early Miocene interval, from which the fossil taxon Broiliana is used as a calibration for crown Procyonidae. 3 = paraconid blade-like, anteriorly placed. State 3 is inferred to be plesiomorphic, as this is the state coded for Amphictis (the outgroup). State 1 unites the {Bassariscus, Procyon) clade whereas state 2 unites the (Bassaricyon, Nasua) clade. However, character 28 is coded as a "?" for Potos, which renders interpretation of the state transforma- tions for this character equivocal for the other two clades in the molecular tree. 3.3. Divergence time estimates We estimated divergence times of eight nodes of the molecular topology of the ingroup using the Bayesian relaxed molecular clock method (Thorne et al., 1998; Kish- ino et al., 2001). Posterior distributions of nodal ages were very consistent among multiple runs, and were usually robust to changes in fossil constraint configurations and moderate alterations of prior root ages. Major changes (e.g., halving the value) in the root age prior had an impact on the posterior point estimates, although credibility inter- vals remained largely overlapping. Prior ages and nodal constraints that are consistent with conservative assump- tions about the procyonid fossil record were used in the final runs, whose results are indicated in Table 4. Divergence time estimates suggest that the genera origi- nated and diversified during the Miocene epoch, 23.8- 5.3 mya (Fig. 6). The spht between Potos and the remaining procyonids (node H) is estimated to have occurred 21.6- 24 mya (95% CI = 12.1-36.0 mya), depending on the con- straint set and root prior used. Estimated divergence between the {Bassaricyon, Nasua) and (Bassariscus, Pro- cyon) clades (node G) is 18.3-20.7 mya (CI=10.3- 30.9 mya). Interestingly, the split between Bassaricyon and Nasua (node E, 11.8-13.3 mya, CI = 6.4-20.6) and that between Bassariscus and Procyon (node F, 11.4?12.8 mya, CI = 6.2-19.8) occurred nearly simultaneously at the end of the Middle Miocene. The divergence between sister species within Nasua, Bassariscus, and Procyon (nodes B, C, and D, respectively) occurred in the Late Miocene (Table 4), whereas divergence between the two putative species of Bassaricyon (node A) is estimated at 2.5-2.8 mya (CI = 1.2- 5.0 mya), which coincides with the Late Pliocene. 4. Discussion 4.1. Incongruence between molecular- and morphology-based procyonid phylogenies Phylogenetic analyses of the supermatrix using four different methods of tree reconstruction (MP, ML, ME and BI) all strongly support the existence of three lin- eages: a {Bassaricyon, Nasua) clade, a {Bassariscus, Pro- cyon) clade, and a Potos hneage (Figs. 2 and 3). This molecular phylogeny challenges recent views of the evolu- tionary relationships within the Procyonidae based on cladistic analyses of morphological data (Decker and Wozencraft, 1991; Baskin, 2004). In fact, molecular and morphological data sets recover strongly supported topologies that are incompatible with one another. For example, whereas our supermatrix shows that olingos {Bassaricyon) and coatis {Nasua) are each other's closest relative, the morphological matrices strongly support a grouping of olingos with kinkajous {Potos) and a group- ing of coatis with raccoons {Procyon). Bassariscus is usu- ally characterized as having a primitive dental morphology that distinguishes it from fossil and recent taxa with more derived hypocarnivorous dentitions (Baskin, 1982, 1989, 2004). Cladistic analyses of morphol- ogy lend support to this characterization, as Bassariscus has been suggested as the earliest diverging lineage within the Procyonini or Procyoninae (Fig. 1). Our molecular phylogeny, however, shows that Bassariscus is the sister group to Procyon, suggesting that the dentition of Bass- ariscus may not be so primitive after all. K.-P. Koepfli et al. I Molecular Phylogenetics and Evolution 43 (2007) 1076-1095 1087 Oligoc?ne Miocene PliocenePleJ 35 30 25 20 15 10 5 0 I MYA Bassaric^onoiaes^ Parapetos^ I 1 1 Potos Bassaricyon Nasua Bassariscus Procyt on Probassariscusf Edaphocyori^ Arctonasua^ Bractrynasua^ (SA? ChapaSmalarta^ (SA) CyonasuaT (SA) /mp/wjast/a? (SA) Protcprocyor^ Paranasu?t I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I MYA 35 30 25 20 15 10 5 0 Fig. 6. Temporal ranges of extant and fossil New World procyonid genera. Extinct genera are denoted with a cross (f). Black lines indicate temporal range based on fossil evidence (McKenna and Bell, 1997; Baskin, 1982, 1998, 2003, 2004). Gray lines appended to extant genera indicate temporal range esti- mates based on analyses of the molecular supermatrix using a Bayesian relaxed molecular clock (see Table 4). Bassaricyon and Potos lack a fossil record and are, therefore, considered recent (McKenna and Bell, 1997). Arrow under temporal range o? Procyon indicates earliest age based on McKenna and Bell (1997). Vertical gray lines represent mean Bayesian posterior divergence time estimates and red boxes are 95% credibility intervals for estimated diver- gence times. (SA) refers to genera endemic to South America. Pie., Pleistocene. To assess which tree estimate is more hkely to be correct, we first determined whether the incongruence between data sets is real (i.e., different biological and/or evolutionary pro- cesses have affected the data sets in ways such that they record different historic events) or is the result of an artifact related to sampling or methodology (Wendel and Doyle, 1998; Hillis and Wiens, 2000). In comparing our superma- trix with Baskin's (2004) matrix, we attempted to account for such possible sources of incongruence. First, incongru- ence is not due to weak support of relationships in one data set versus the other, which would suggest an undersampling of characters or high levels of homoplasy in one or the other data set. Bootstrap analyses of parsimony trees reconstructed from both data sets indicate that conflicting nodes are strongly supported by each data set (Figs. 3 and 4A). Furthermore, when all fossil taxa of the ingroup are removed from the Baskin (2004) matrix, bootstrap values increase and nodes remain significantly conflicting with those of the supermatrix tree (Fig. 4B). Second, both data sets were analyzed with the same optimality criterion (MP), and, therefore, incongruence is not a result of use of differ- ent methods of phylogeny reconstruction and thus different assumptions about character evolution. Third, incongru- ence could result from use of different outgroups or from use of different numbers of taxa in each data set. We accounted for these possible artifacts by performing unrooted MP and bootstrap analyses on the supermatrix and the abridged Baskin (2004) matrix. Relationships in unrooted trees from the two data sets, each recovered by exhaustive searches, remain significantly incongruent with one another (Figs. 5A and B). Thus, the cause of incongru- ence between molecular and morphological data sets does not arise from choice of phylogeny reconstruction method, use of different outgroups, or weak nodal support in one data set versus the other. When putative synapomorphies from the morphological data sets are mapped onto the molecular topology, many are resolved as homoplasies due to convergence or reversals of character states. These results suggest that incongruence between molecular and morphological trees may partly arise from misleading effects of adaptive convergence in morphological characters. Wiens et al. (2003) proposed sev- eral criteria for detecting morphological convergence among unrelated taxa resulting in an inaccurate estimate of phylogeny relative to an independent (e.g., molecular) data set. Criteria include: (1) strong morphological support for clades containing taxa that share a similar selective environment; (2) characters that define clades in the morphology-based tree are associated with the shared selective environment and are thus likely adaptive for that K.-P. Koepfli et al. I Molecular Phylogenetics and Evolution 43 (2007) 1076-1095 environment; and (3) a molecular topology based on a mul- tilocus data set shows that species with similar selective environments are not monophyletic. Importantly in this context, the multilocus component of the molecular data set would support the notion that included segments are not subject to identical selective pressures, as they partici- pate in a variety of cellular and metabolic processes assumed to be unrelated. Our results are consistent with an interpretation that ecological adaptation to similar habitats and diets in unre- lated taxa has confounded cladistic analyses of cranial, den- tal and postcranial characters in the Procyonidae. For example, kinkajous and olingos are both arboreal frugi- vores and share many ecological and behavioral traits (Ford and Hoffmann, 1988; Kays, 1999, 2000). Adaptations to arboreality include a short rostrum and forwardly directed orbits for stereoscopic vision, short limbs with sharply curved claws, furred soles, and a long tail (prehen- sile in the kinkajou). Adaptations to frugivory in both taxa include bunodont dentition with a reduced number of cusps on molars and greater development of masticatory musculature in the mediolateral plane compared to the sag- ittal plane (Story, 1951; Decker and Wozencraft, 1991). Indeed, a reduced number of molar cusps is one character state that unites Bassaricyon + Potos in the Decker and Wozencraft (1991) phylogeny and the dental characters uniting these same taxa in Baskin (2004) reflect bunodonty. Ecological similarities between coatis and raccoons may have also resulted in convergent or parallel evolution of morphology in these taxa. Both genera are omnivorous, with highly tuberculate molars and longer rostra relative to kinkajous and olingos. Both coatis and raccoons inhabit forested habitats, forage on the ground and in the trees, and are good climbers and use their forepaws to manipulate prey (Lotze and Anderson, 1979; Gompper, 1995; Gomp- per and Decker, 1998). As with kinkajous and olingos, putative cranial and dental synapomorphies that group coatis and raccoons (Decker and Wozencraft, 1991; Baskin, 2004) are shown to be homoplasies when mapped onto the molecular phylogeny (under both ACCTRAN and DEL- TRAN transformation). Arguments with regards to phylogenetically misleading effects of convergence assume, however, that morphological characters used in procyonid phylogeny reconstruction are actually adaptive for a particular habitat or lifestyle (crite- rion 2 in Wiens et al., 2003). This is difficult to demonstrate in general, but especially so for characters that have been established through the atomization of a trait for phyloge- netic purposes. The dental characters in the Baskin (2004) matrix may indeed reflect adaptation to differing diets among procyonid genera, as tooth morphology is strongly correlated with diet in many mammals, including the Car- n?vora (e.g.. Van Valkenburgh and Koepfli, 1993; Friscia et al., in press). Moreover, high bootstrap support for con- flicting nodes in the Baskin (2004) tree do unite taxa that tend to share a similar diet, for example Bassaricyon and Potos (criterion 1 in Wiens et al., 2003), suggesting that the source of incongruence between the supermatrix phylogeny and the Baskin (2004) phylogeny is not due to randomly distributed phylogenetic signal in the Baskin (2004) matrix (Wenzel and Siddafl, 1999; Wiens et al., 2003; Gaubert et al., 2006). This is also confirmed by the high CI and RI values of the unabridged and abridged analyses of the Baskin (2004) matrix (Fig. 4). The adaptive value of some putative synapomorphies found in Decker and Wozencraft (1991) is somewhat more difficult to evaluate because they include cranial, dental, postcranial and soft anatomical characters. Nine characters unite Bassaricyon + Potos, and while some may be associ- ated with adaptation to arboreality (e.g., sharply angled acromion process) and a particular diet (<3 cusps on molars), the functional significance is not clear for others (e.g., short and rounded external pinnae and short and stout malleus). Similarly, nine characters unite Bassariscus + Nasua + Nasuella + Procyon into the "Procy- oninae" in the Decker and Wozencraft (1991) phylogeny (Fig. IB). These are resolved as plesiomorphies or homo- plasies depending on the method of character state optimi- zation used. Three are dental characters, four are cranial, one is postcranial, and one is a soft anatomical character (presence of banded tail rings). Such coat patterns, how- ever, although less distinct, are present in Bassaricyon (Nowak, 2005), suggesting that this character is plesiomor- phic for all taxa except Potos. Nonetheless, comparative analyses suggest that tail rings, coupled with tail movement, may function in communication and thus be adaptive for species inhabiting forest habitats (Ortolani, 1999). Deter- mining the adaptive value, if any, of the characters that define the "Procyoninae" is made more difficult by the fact that Bassariscus, Nasua and Procyon inhabit a diversity of environments and have extremely varied diets across their ranges. PhylogeneticaUy misleading effects of convergence may be exacerbated if morphological characters are not independent, especially if characters are developmentally, geneticaUy, and/ or functionaUy correlated (Felsenstein, 1988; Shaffer et al., 1991; Emerson and Hastings, 1998; Wiens et al., 2003). Non- independence of characters may apply to the Baskin (2004) data set in particular. Thirty of the 40 characters (75%) in the Baskin (2004) data matrix are dental characters, and 25 of these characters are based on just four teeth (P4, Ml, ml, and m2) that have been atomized into four (m2), six (P4, ml) and nine (Ml) separate characters. Moreover, in several cases, a particular dental character is subdivided into two separate characters (e.g., characters 24 and 25 deal with the presence or absence and the position, if present, of the Ml metaconule, respectively; Baskin, 2004). While such atomiza- tion of dental morphology increases the power of cladistic analyses (more characters = potentially more resolving power), such a strategy is problematic because empirical studies demonstrate that teeth are developmentally and func- tionally integrated structures (e.g., Gingerich and Winkler, 1979; Polly, 1998; Jernvall and Jung, 2000). For example, Kangas et al. (2004) have shown that small modifications in K.-P. Koepfli et al. I Molecular Phylogenetics and Evolution 43 (2007) 1076-1095 1089 the expression o? a single gene in the enamel knot of develop- ing mouse teeth can result in correlated state changes for multiple dental characters (number of teeth as well as cusp shape and position). Furthermore, cusp number and shape have been shown to be tightly integrated through a shared developmental genetic program (Jernvall et al., 2000; Jernvall and Jung, 2000). Individual molar cusps develop iteratively from secondary enamel knots through activation of the same set of genes (Jernvall et al., 2000), which can be deactivated and reactivated, causing character states to appear and dis- appear over evolutionary time (Jernvall and Jung, 2000). Moreover, this shared developmental information is likely propagated through both upper and lower jaws, thereby fur- ther limiting the independence of serially homologous struc- tures such as teeth (e.g.. Van Valen, 1994). Such pleiotropic effects on tooth morphology challenge the method of atomiz- ing a tooth into separate characters and assuming that they are independently heritable and acted upon separately by natural selection. An example of potentially correlated (and thus non- independent) characters in the Baskin (2004) matrix are the four characters that unite Bassaricyon + Potos: two from the P4 (absence of the metacone blade and presence of the parastyle) and two from the ml molar: (a very reduced or absent paraconid and an unbasined talonid). Not only are individual cusps of these two teeth likely to be developmen- tally correlated (where no single gene codes for each indi- vidual cusp; Jernvall and Jung, 2000), but these two teeth and their associated cusps form a functionally correlated complex, the carnassial. Changes to one P4 cusp will proba- bly result in a concomitant evolutionary change in mor- phology of the ml, to maintain functional occlusion between the two teeth and efficient food processing capabil- ity. Such atomization of non-independent characters has the potential to inflate homoplasy and result in incorrect yet well-supported phylogenies, especially if selection acts on an entire trait such as a molar. Although nucleotide sites can also be affected by non- independence, particularly within genes (e.g., Cummings et al., 1995), our supermatrix topology is supported by multiple fines of independent evidence. First, tfie superma- trix topology is recovered in separate phylogenetic analy- ses of concatenated nuclear and mitocfiondrial sequence data (Fig. 2). Because loci are derived from different geno- mic compartments (nuclear versus mitocfiondrial) and the nuclear gene segments are derived from unfinked genes and, tfierefore, represent markers witfi presumably inde- pendent evolutionary fiistories, tfiese results indicate tfiat tfie gene trees derived from our sampled loci are tracing tfie same evolutionary fiistory. Moreover, patterns of rela- tionsfiip are not due to pfiylogenetic signal arising from relatively few loci. For example, separate analyses of tfie 11 gene segments using MP and ME witfi bootstrapping sfiows tfiat tfie {Bassaricyon, Nasua) and {Bassariscus, Procyon) clades are recovered by 10 and 9 of tfie gene seg- ments, respectively, witfi fiigfi bootstrap support (Table 5). It is unlikely tfiat all segments would be biased by natural selection in an identical or correlated fasfiion, as tfiey participate in a variety of unrelated cellular and molecular processes. Second, tfie {Bassaricyon, Nasua) and (Bassariscus, Pro- cyon) clades are each supported by informative indels and the (Bassaricyon, Nasua) clade is further supported by shar- ing a CYTb gene structure that is two codons longer than in other procyonid taxa. Third, the finding that Procyon and Bassariscus are sister taxa is supported by immunological distance analyses (Sarich, 1973) and affinity o? Bassaricyon and Nasua is supported by shared inversions on chromo- somes 2 and 6 in these taxa (Couturier and Dutrillaux, 1986). Fourth, statistical tests (Kishino and Hasegawa, 1989) based on ML using PAUP* strongly favor the molec- ular topology of procyonid genera over the morphology- based topologies (?InL molecular topology = 22697.196 versus ?InL morphological topologies = 22990.308; differ- ence in ?In L = 293.112, P<0.05). Finally, even when the molecular supermatrix plus 40 morphological characters from Baskin (2004) are analyzed simultaneously, the unrooted network topology is identical to the one derived from analysis of the supermatrix alone (Fig. 5C), with high positive PBS values contributed by the supermatrix. This Table 5 Bootstrap values derived from separate maximum parsimony (MP) and minimum evolution (ME) analyses of the 11 gene segments for three procyonid clades {Bassaricyon, Nasua) MP ME 87 75 72 87 91 94 100 99 96 94 89 90 99 99 97 98 99 96 69 99 <50 59 (Bassariscus, Procyon) ((Bassaricyon, Nasua) (Bassariscus, Procyon)) MP ME MP ME ADORA3 APOB BDNF CHRNAl COLlOAl PNOC RAGl RAG2 WTl CYTb NADH5 91 65 87 70 82 97 84 100 99 <50 90 88 65 78 75 85 98 66 100 100 <50 90 <50 69 83 55 <50 62 93 <50 <50 99 92 <50 55 79 56 a 61 97 54 <50 In the ME analysis of RAGl, /"O?O? was joined as the sister taxon to the (Bassariscus, Procyon) clade with a bootstrap value of 68%. 1090 K.-P. Koepfli et al. I Molecular Phylogenetics and Evolution 43 (2007) 1076-1095 further suggests that the molecular supermatrix contains a greater number of independent characters and thus robust support for relationships relative to Baskin's (2004) mor- phological matrix (see Kluge, 1989). Our results showing incongruence between molecular and morphological estimates of procyonid phylogeny are mirrored in phylogenetic studies of other families of the Carn?vora. For example, in Viverridae (genets and civets), morphological similarity in craniodental features between African and Asian linsangs (Poiana and Prionodon, respec- tively) has traditionally led researchers to consider these genera to be closely related. These taxa share a similar spot- ted coat pattern, both are hypercarnivorous and adapted to an arboreal lifestyle in heavily forested habitats (Nowak, 2005). Initial molecular studies suggested that this morpho- logical similarity actually represents convergent evolution, as Asian linsangs {Prionodon) form the sister group to the cat family (Felidae) while the African linsangs (Poiana) form the sister group to the genets (Gaubert and Veron, 2003). As with procyonids, morphological characters thought to be rehable indicators of phylogeny (e.g., denti- tion and basicranial characters) were found to be highly homoplasious (Gaubert and Veron, 2003). Subsequent study, however, suggested that incongruence between molecular and morphological estimates of viverrid phylog- eny was caused by the overall lack of structuring phyloge- netic signal in morphological data rather than adaptive convergence (Gaubert et al., 2006). These authors noted the importance of having an independent molecular phylogeny as a reference for deducing true phylogenetic signal from morphological data. This approach, using a multigene molecular phylogeny to evaluate the veracity of morpho- logical characters, has also been implemented for other taxa, particularly primates (Fleagle and McGraw, 1999, 2002; Collard and Wood, 2000; Strait and Grine, 2004). 4.2. Dating procyonid divergence and implications for the Great American Interchange The molecular divergence times and fossil data, taken together, suggest that the Procyonidae is a product of a Mio- cene radiation. Divergence time estimates suggest that extant genera all originated and diversified during the Miocene epoch, which coincides with the first appearance of nearly all extinct procyonid genera (Fig. 6). Furthermore, these esti- mates are remarkably consistent with certain aspects of the procyonid fossil record, such as that for Bassariscus. Although our molecular estimate suggests a younger age for the origination o? Bassariscus, compared to the minimum age of the earhest fossil for this taxon {ca. 15.9mya, McKenna and Bell, 1997; Baskin, 2004), the 95% CI overlaps with the earlier age suggested by the fossil record (Fig. 6). For other taxa, our results support suggested gaps in the procyonid fos- sil record (Bassaricyon and Potos lack a fossil record alto- gether), with mean divergence times and 95% CIs indicating much earlier times of origin. In particular, the genetic dis- tances separating Potos from the other taxa are consistently large for both mitochondrial and nuclear gene segments, indicating that Potos represents an ancient lineage that has been separated from other procyonid genera since the Early Miocene, and perhaps the Oligoc?ne. Molecular dating results also show that Bassaricyon vs. Nasua and Bassariscus vs. Procyon diverged contemporane- ously near the end of the Middle Miocene, ll-12mya (Fig. 6). Such correlated patterns of divergence between inde- pendent lineages is suggestive of the influence of major regional environmental processes on diversification (e.g., Del- sue et al., 2004). The poor fossil record of procyonids in gen- eral, and in the Neotropics in particular, precludes our ability to make meaningful inferences from intergeneric divergence times because the geographic location in North or Central America where diversification took place is uncertain. Also, although we were unable to sample Nasuella, a taxon found at high elevations in the Andes Mountains of northern South America, this taxon may be important in further understand- ing the context of procyonid diversification, particularly the influence of northern Andean uplift, which has been associ- ated with the diversification of hneages within the Xenarthra (Delsuc et al., 2004). However, the mean divergence time esti- mate of 11-12 mya for the two pairs of genera coincides with a period of major global cooling, along with formation of the Antarctic ice sheet and a short-term but significant drop in global sea level (Haq et al., 1987; Zachos et al., 2001; Miller et al., 2005). The change in climate is associated with major vegetational changes across the globe, including the Neo- tropics, as forested environments were reduced and became mixed with expanding savanna-like environments (Potts and Behrensmeyer, 1992). These environmental events likely played a role in diversification of other clades of mammals (e.g.. Mercer and Roth, 2003; Delsuc et al., 2004; Johnson et al., 2006), and, in this case, could have influenced connec- tivity of ancestral procyonid taxa among suitable areas in the New World prior to the formation of the Panamanian land bridge. Indeed, fossils of mammalian herbivores collected from the Panama Canal region with affinities to taxa in North America suggest that there was a broad connection between Central and North America during the Miocene (Whitmore and Stewart, 1965). Timings of intrageneric splits range from Late Miocene to Late Pliocene (Table 4). Among these, the divergence time between the two putative species of Bassaricyon is the youngest, dated at ca. 2.5-2.8 mya (95% CI = 1.2-5.0 mya). This result should be treated with caution, however, given the uncertainty about the actual number of species included in Bassaricyon (see Section 5). The oldest intrageneric diver- gence occurs between Bassariscus astutus and B. sumichr- asti, with a mean divergence time of ca. 9.1-10.3 mya in the Late Miocene (95% CI = 4.9-16.1 mya). This resuh lends additional support to earlier behavioral research that has shown that these two species display large differences in morphology, reproductive physiology and behavior, modes of communication, and ecology (Poglayen-Neuwall and Toweill, 1988; Poglayen-Neuwall and Poglayen-Neuwall, 1995). With regards to ecology, for example, B. astutus K.-P. Koepfli et al. I Molecular Phylogenetics and Evolution 43 (2007) 1076-1095 1091 inhabits dry tropical to semi-arid and montane woodland environments, is mainly terrestrial, and tends towards car- nivory, whereas B. sumichrasti inhabits lowland tropical forests, is almost completely arboreal, and has a diet high in fruit and insects (Coates-Estrada and Estrada, 1986; Pog- layen-Neuwall and Toweill, 1988 Poglayen-Neuwall and Poglayen-Neuwall, 1995). Even more interesting are the divergence times within Nasua and Procyon because of their possible relevance to understanding faunal dynamics of the Great American Interchange (GAI). Procyonids figure prominently in the GAI, which involved reciprocal exchange of different lin- eages of formerly endemic mammals between North and South America due to formation of the Panamanian land bridge, ca. 3-2.5 mya (Marshall et al., 1979; Marshall, 1985, 1988; Webb, 1997). Fossil evidence suggests that procyo- nids dispersed from North America into South America on two different occasions, once before and once after the completion of the Panamanian land bridge (Webb, 1985). Procyonids are first recorded in South America in Late Miocene deposits from Argentina, represented by the fossil genus Cyonasua, which is morphologically similar and closely related to the North American extinct Arctonasua (Simpson, 1980; Baskin, 1982, 2004; Fig. lA). Cyonasua is thought to have dispersed into South America via island hopping or rafting during the Late Miocene (Simpson's waif dispersal), when the inter-American seaway consisted of a large archipelago during a time when global sea level was relatively low (Haq et al., 1987; Coates and Obando, 1996; Coates, 1997; Miller et al., 2005). Cyonasua gave rise to a number of descendants, including the bear-sized Cha- palmalania, but these South American endemic procyonids are thought to have gone extinct by the end of the Pliocene (Marshall, 1985). Procyonids, represented by the extant genera, are believed to have entered South America for a second time with other mammalian taxa as part of the "legions of the north" during the GAI ca. 2-3 mya, after the land bridge was completed (Simpson, 1980; Marshall, 1988; Marshall et al., 1979, 1982; Webb, 1985). Fossils o? Nasua nasua and Procyon cancrivorus from South America are first recorded in Late Pleistocene deposits (Berta et al., 1978). Therefore, Cyonasua and its descendants are not considered to be progenitors to modern procyonid taxa that now inhabit South America, based on the temporal gap between occurrence of the two lineages in South Amer- ica (Marshall, 1985; Webb, 1985) and on phylogenetic con- siderations (Baskin, 2004). However, divergence times within Nasua and Procyon, coupled with consideration of present day configuration of the geographic ranges of sister species within these genera, suggest that these taxa may have entered South America at the same time as Cyonasua also via island hopping (as diver- gence time estimates suggest that the temporal gap separat- ing these genera is now removed?Fig. 6). Estimates indicate that N. narica and N. nasua diverged ca. 7-8 mya, during the Late Miocene, whereas P. cancrivorus and P. lotor diverged ca. 5-5.7 mya, during the Early Phocene. These divergence times well predate the final emplacement of the Panamanian land bridge, ca. 3-2.5 mya (Marshall etal., 1979; Marshall, 1985; Coates and Obando, 1996). The modern presence of North and South American sister taxa oi Procyon and Nasua may be accounted for by pre-land bridge dispersal to South America (>3mya), followed by isolation, drift and then sec- ondary contact once the land bridge was completed. Infer- ring the mode of speciation or other historical processes from the configuration of modern day geographic ranges is obvi- ously fraught with many problems (e.g., Losos and Glor, 2003). However, this hypothesis might eventually be tested through discovery of older fossils of Nasua and Procyon in Neotropical fossil localities (MacFadden, 2006). 5. Conclusions The molecular phylogeny of the Procyonidae pre- sented here highlights the morphological and ecological flexibility of these taxa. The available fossil record sug- gests that most procyonid genera, including extant gen- era, originated in southern North America and/or Central America (Simpson, 1945; Hershkovitz, 1972; Baskin, 1982, 1998; Webb, 1985). Although there is no evidence regarding the exact mode of speciation, diversi- fication within such a limited geographic area would likely be driven by competition for resources, leading to ecological and morphological divergence of taxa (Dar- win, 1859; Schl?ter, 2000), and may help explain why multiple procyonid genera can coexist within the same neotropical community (e.g., Bisbal, 1986; Kays, 2000; M?rquez and Fari?a, 2003). Our findings suggest that procyonid phylogenetic rela- tionships based on morphological evidence are problem- atic, and further, that phylogenetic hypotheses relating fossil and living taxa based primarily on dental evidence are also open to question because of extensive homoplasy asso- ciated with this type of evidence. This should not be taken as an indictment of all morphological characters, or even all those used in previous procyonid studies. Rather, careful study of dental and skeletal anatomy has shown that mor- phological features can accord with molecular estimates of phylogeny for a group (e.g., papionine primates; Fleagle and McGraw, 2002). Therefore, a thorough r??valuation of the morphology of procyonids in light of our molecular results would seem to be a fruitful avenue of research. Furthermore, understanding genetic bases of morphogene- sis and ontogeny of cranial and dental characters can help to elucidate correlation or independence among morpho- logical features and perhaps lead to better justifications for the atomization of characters or suites of characters (Lovejoy et al., 1999; Hlusko, 2004). For example, the study by Popwics (1998) on ontogeny of carnassial and postcar- nassial teeth in ferrets (Mustela putorius) showed that while early stages of tooth ontogeny were similar to other carni- vores, later stages of ontogeny resulted in species-specific dental morphologies. Such elegant studies, combined with an understanding of developmental genetics of 1092 K.-P. Koepfli et al. I Molecular Phylogenetics and Evolution 43 (2007) 1076-1095 morphogenesis, provide a way to possibly dififerentiate homology from homoplasy (Lieberman, 2000) and thereby enhance the phylogenetic signal obtained from morpholog- ical traits. Understanding the genetic basis of phenotypic traits offers the best hope for deducing the historical forces that have influenced morphological evolution as well as improving our ability to reconstruct phylogenies that relate both fossil and living taxa. Our results also suggest that earher classifications of the Procyonidae (Decker and Wozencraft, 1991; McK- enna and Bell, 1997; Baskin, 2004) need to be revised. The molecular phylogeny indicates that subfamilies or tribes defined on the basis of cladistic morphological analyses do not accurately reflect genealogical history of the fam- ily. The molecular phylogeny is consistent with the classi- fication proposed by Trouessart (1904), in which Potos is placed as the sole member of the Potosinae and all other genera are placed within Procyoninae. However, beyond these issues is the greater concern over uncertainty surrounding the exact number of species that comprise the Procyonidae (Glatston, 1994). First, it is still unknown how many species of olingo exist. Some consider the five described species to represent only one or two spe- cies at most. Second, multiple populations of Procyon, and one Nasua, are endemic to Caribbean islands. While most of these may be recent introductions, others have a long his- tory of isolation (Pons et al., 1999; Cuar?n et al., 2002; Hel- gen and Wilson, 2002, 2003, 2005; McFadden, 2004). Such examples highlight how little we actually know about the Procyonidae, and call for the need for further molecular sys- tematic studies to not only help us close these gaps in knowl- edge, but also to help establish a solid foundation for the conservation of procyonid biodiversity (e.g., Avise, 1989). Acknowledgments We gratefully thank the individuals and institutions listed in Table 1 for providing tissue samples that made this study possible. We thank Jennifer Leonard and Kelly Sainz for outstanding laboratory assistance and Eli GefFen for help in the initial phase of this project. We thank Jon Baskin for providing valuable insights regard- ing procyonid fossil calibration dates. We thank John Pol- linger for scanning the illustrations included in Fig. 3 and Benny Gee for the final composition of Fig. 3. We thank Peter Adam, Carolyne Bardeleben, Karen Kapheim, Gra- ham Slater, Blaire Van Valkenburgh, and three anony- mous reviewers for providing constructive comments and criticisms that improved the manuscript. This study was supported by research funds from the U.S. National Sci- ence Foundation. Appendix A. Supplementary data Supplementary data associated with this article can be found, in the online version, at doi:10.1016/j.ympev. 2006.10.003. References Amrine-Madsen, H., Koepfli, K.-P., Wayne, R.K., Springer, M.S., 2003. A new phylogenetic marker, apolipoprotein B, provides compelling evidence for eutherian relationships. Mol. Phylogenet. Evol. 28, 225-240. Avise, J.C., 1989. A role for molecular genetics in the recognition and con- servation of endangered species. Trends Ecol. Evol. 4, 279-281. Barriel, V., 1994. Molecular phylogenies and how to code insertion/dele- tion events. Life Sei. 317, 693-701. Baskin, J.A., 1982. Tertiary Procyoninae (Mammalia: Carn?vora) of North America. J. Vert. Paleo. 2, 71-93. Baskin, J.A., 1989. Comments on New World Tertiary Procyonidae (Mammaha; Carn?vora). J. Vert. Paleo. 9, 110-117. Baskin, J.A., 1998. Procyonidae. In: Janis, CM., Scott, K.M., Jacobs, L.L. (Eds.), Evolution of Tertiary Mammals of North America, vol. 1. Cam- bridge University Press, Cambridge, pp. 144?151. Baskin, J.A., 2003. New procyonines from the Hemingfordian and barsto- vian of the gulf coast and Nevada, including the first fossil record of the Potosini. Buh. Am. Mus. Nat. Hist. 279, 125-146. Baskin, J.A., 2004. Bassariscus and Probassariscus (Mammalia, Carn?vora, Procyonidae) from the early Barstovian (Middle Miocene). J. Vert. Paleo. 24, 709-720. Berta, A., Marshall, L.G., 1978. South American Carn?vora. In: Westphal, F. (Ed.), Fossilium catalogus I: Animalia, W. Junk, The Hague, pp. 1^8. Bisbal, F.J., 1986. Food habits of some neotropical carnivores in Venezu- ela (Mammalia, Carn?vora). Mammalia 50, 329-339. Bremer, K., 1988. The limits of amino acid sequence data in angiosperm phylogenetic reconstruction. Evolution 42, 795-803. Coates, A.G., 1997. The forging of Central America. In: Coates, A.G. (Ed.), Central America: A Natural and Cultural History. Yale University Press, New Haven, CT, pp. 1-37. Coates, A.G., Obando, J.A., 1996. The geological evolution of the Central American isthmus. In: Jackson, J.B.C., Budd, A.F., Coates, A.G. (Eds.), Evolution and Evironment in Tropical America. University of Chicago Press, Chicago, IL, pp. 21-56. Coates-Estrada, R., Estrada, A., 1986. Fruiting and frugivores at a stran- gler fig in the tropical rain forest of Los Tuxtlas, Mexico. J. Trop. Ecol. 2, 349-357. Collard, M., Wood, B., 2000. How reliable are human phylogenetic hypotheses? Proc. Nati. Acd. Sei. USA 97, 5003-5006. Couturier, J., Dutrillaux, B., 1986. Evolution chromosomique chez les Car- nivores. Mammalia 50, 124-162. Cuar?n, A., Morales-Martinez, M., McFadden, K.W., Valenzuela, D., Gompper, M.E., 2002. The status of dwarf carnivores on Cozumel Island, Mexico. Biodivers. Conserv. 13, 317-331. Cummings, M.P., Otto, S.P., Wakeley, J., 1995. Sampling properties of DNA sequence data in phylogenetic analysis. Mol. Biol. Evol. 12, 814-822. Darwin, C.R., 1859. On the Origin of Species by Means of Natural Selec- tion. John Murray, London. Davis, D.D., 1941. The arteries in the forearm of carnivores. Field Mus. Nat. Hist. Zool. Ser. 27, 137-227. Decker, D.M., Wozencraft, W.C., 1991. Phylogenetic analysis of recent procyonid genera. J. Mammal. 72, 42-55. de Jong, W.W., 1986. Protein sequence evidence for monophyly of the carnivore families Procyonidae and Mustelidae. Mol. Biol. Evol. 3, 276-281. Delisle, I., Strobeck, C, 2005. A phylogeny of the Caniformia (order Car- n?vora) based on 12 complete protein-coding mitochondrial genes. Mol. Phylogenet. Evol. 37, 192-201. Delsuc, F., Vizcai'no, S.F., Douzery, E.J.P., 2004. Influence of Tertiary paleoenvironmental changes on the diversification of South American mammals: a relaxed molecular clock study with xenarthrans. BMC Evol. Biol. 4, 11-24. Dragoo, J.W., Honeycutt, R.L., 1997. Systematics of mustelid-like carni- vores. J. Mammal. 78, 426^43. K.-P. Koepfli et al. I Molecular Phylogenetics and Evolution 43 (2007) 1076-1095 1093 Eisenberg, J.F., 1989. Mammals of the Neotropics: The Northern Neo- tropics (Vol. 1). University of Chicago Press, Chicago, IL. Eisenberg, J.F., Redford, K.H., 1999. Mammals of the Neotropics: The Central Neotropics (vol. 3). University of Chicago Press, Chicago, IL. Emerson, S.B., Hastings, P.A., 1998. Morphological correlations in evolu- tion: consequences for phylogenetic analysis. Q. Rev. Biol. 73, 141-162. Felsenstein, J., 1988. Phylogenies and quantitative characters. Ann. Rev. Ecol. Syst. 19,445^71. Fleagle, J.G., McGraw, W.S., 1999. Skeletal and dental morphology sup- ports diphyletic origin of baboons and mandrills. Proc. Nati. Acad. Sei. USA 96, 1157-1161. Fleagle, J.G., McGraw, W.S., 2002. Skeletal and dental morphology of Afri- can papionins: unmasking a cryptic clade. J. Hum. Evol. 42, 252-267. Flynn, J.J., Neff, N.A., Tedford, R.H., 1988. Phylogeny of the Carn?vora. In Benton, M.J. (Ed.), The Phylogeny and Classification of the Tetrapods, vol. 2; Mammals. Systematics Association Special Volume 35B, Clar- endon Press, Oxford, pp. 73-116. Flynn, J.J., Nedbal, M.A., 1998. Phylogeny of the Carn?vora (Mammalia): congruence vs incompatibility among multiple data sets. Mol. Phyloge- net. Evol. 9, 414^26. Flynn, J.J., Nedbal, M.A., Dragoo, J.W., Honeycutt, R.L., 2000. Whence the red panda? Mol. Phylogenet. Evol. 17, 190-199. Flynn, J.J., Finarelli, J.A., Zehr, S., Hsu, J., Nedbal, M.A., 2005. Molecular phylogeny of the Carn?vora (Mammalia): assessing the impact of increased sampling on resolving enigmatic relationships. Syst. Biol. 54, 317-337. Ford, L.S., Ho?'mann, R.S., 1988. Potos flavus. Mammal. Species 321, 1-9. Friscia, A.R., Van Valkenburgh, Biknevicius, A., in press. An ecomorpho- logical analysis of extant small carnivorans. J. Zool. Lond. Gaubert, P., Veron, G., 2003. Exliaustive sample set among Viverridae reveals the sister-group to felids: the linsangs as a case of extreme morphological convergence within FeUfomiia. Proc. R. Soc. Lond. B 270,2523-2530. Gaubert, P., Wozencraft, W.C., Cordeiro-Estrela, P., Veron, G., 2006. Mosaics of convergences and noise in morphological phylogenies: what's in a viverrid-like carnivoran? Syst. Biol. 54, 865-894. Gill, T., 1872. Arrangement of the families of mammals. Smithsonian Misc. Coll. 11, 1-98. Gingerich, P.D., Winkler, D.A., 1979. Patterns of variation and correlation in the dentition of the red fox, Vulpes vulpes. J. Mammal. 60, 691-704. Ginsberg, L., Morales, J., Soria, D., Herraez, E., 1997. Discovery of an ancestor of the lesser panda in the Middle Miocene of Madrid (Spain). C. R. Acad. Sei. Paris, Earth Planet. Sei. 325, 447^51. Glatston, A.R., 1994. The red panda, olingos, coatis, raccoons, and their relatives. lUCN/SSC Mustelid, Viverrid, and Procyonid Specialist Group, International Union for Conservation of Nature and Natural Resources, Gland, Switzerland. Goldman, D., Girl, P.R., O'Brien, S.J., 1989. Molecular genetic-distance estimates among the Ursidae as indicated by one- and two-dimensional protein electrophoresis. Evolution 43, 282-295. Gompper, M.E., 1995. Nasua narica. Mammal. Sp. 487, 1-10. Gompper, M.E., Decker, D.M., 1998. Nasua nasua. Mammal. Sp. 580, 1-9. Haq, B.U., Hardenbol, J., Vail, P.R., 1987. Chronology of fluctuating sea levels since the Triassic. Science 235, 1156-1167. Hashimoto, T., Otaka, E., Adachi, J., Mizuta, K., Hasegawa, M., 1993. The giant panda is closer to a bear, judged by a- and b-hemoglobin sequences. J. Mol. Evol. 36, 282-289. Helgen, K.M., Wilson, D.E., 2002. The history of the raccoons of the West Indies. J. Barbados Mus. Hist. Soc. 48, 1-11. Helgen, K.M., Wilson, D.E., 2003. Taxonomic status and conservation rel- evance of the raccoons {Procyon spp.) of the West Indies. J. Zool. 259, 69-76. Helgen, K.M., Wilson, D.E., 2005. A systematic and Zoogeographie over- view of the raccoons of Mexico and Central America. In: S?nchez-Cor- dero, v., Medellin, R.A. (Eds.), Contribuciones Mastozool?gicas en Homenaje a Bernado Villa. Instituto de Biolog?a e Instituto de Eco- logia, UNAM, Mexico City, pp. 221-236. Hershkovitz, P., 1972. The Recent mammals of the Neotropical region: a Zoogeographie and ecological review. In: Keast, A., Erk, F.C., Glass, B. (Eds.), Evolution, Mammals and Southern Continents. State Univer- sity of New York Press, Albany, NY, pp. 311^31. Hillis, D.M., Bull, J.J., 1993. An empirical test of bootstrapping as a method for assessing confidence in phylogenetic analysis. Syst. Biol. 42, 182-192. Hillis, D.M., Wiens, J.J., 2000. Molecules versus morphology in systemat- ics: conflicts, artifacts, and misconceptions. In: Wiens, J.J. (Ed.), Phylo- genetic Analysis of Morphological Data. Smithsonian Institution Press, Washington, DC, pp. 1-19. Hlusko, L.J., 2004. Integrating the genotype and phenotype in hominid paleontology. Proc. Nati. Acad. Sei. USA 101, 2653-2657. Hollister, N., 1916. The genera and subgenera of raccoons and their allies. Proc. U.S. Nati. Mus. 49, 143-150. Hvmt Jr., R.M., 1996. Biogeography of the Order Carn?vora. In: Gittleman, J.L. (Ed.), Carnivore Behavior, Ecology and Evolution, vol. 2. Cornell University Press, Ithaca, NY, pp. 485-541. Irwin, D.M., Kocher, T.D., Wilson, A.C., 1991. Evolution of the cyto- chrome b gene of mammals. J. Mol. Evol. 32, 128-144. Jernvall, J., Ker?nen, S.V.E., Thesleff, I., 2000. Evolutionary modification of development in mammalian teeth:quantifying gene expression pat- terns and topography. Proc. Nati. Acad. Sei. USA 97, 14444-14448. Jernvall, J., Jung, H.-S., 2000. Genotype, phenotype, and developmental biology of molar tooth characters. Yearbook Phys. Anthro. 43, 171- 190. Jiang, Z., Pri?t, C, Galibert, F., 1998. Traced orthologous amplified sequence tags (TOASTs) and mammalian comparative maps. Mam- mal. Genome 9, 577-587. Johnson, W.E., Eizirik, E., Pecon-Slattery, J., Murphy, W.J., Antunes, A., Teeling, E., O'Brien, S.J., 2006. The Late Miocene radiation of the mod- ern Felidae: a genetic assessment. Science 311, 73-77. Kangas, A.T., Evans, A.R., Thesleff, I., Jernvall, J., 2004. Nonindependence of mammahan dental characters. Nature 432, 211-214. Kays, R.W., 1999. Food preferences of kinkajous (Potos flavus) a frugivo- rous carnivore. J. Mammal. 80, 589-599. Kays, R.W., 2000. The behavior and ecology of olingos (Bassaricyon gab- bii) and their competition with kinkajous (Potos flavus) in central Pan- ama. Mammalia 64, 1-10. Kishino, H., Hasegawa, M., 1989. Evaluation of the maximum likelihood estimate of the evolutionary tree topologies from DNA sequence data, and the branching order in Hominoidea. J. Mol. Evol. 29, 170-179. Kishino, H., Thorne, J.L., Brvmo, W.J., 2001. Performance of a divergence time estimation method under a probabilistic model of rate evolution. Mol. Biol. Evol. 18,352-361. Kluge, A.G., 1989. A concern for evidence and a phylogenetic hypothesis of relationships among Epicrates (Boidae, Serpentes). Syst. Zool. 38, 7- 25. Koepfli, K.-P., Wayne, R.K., 1998. Phylogenetic relationships of otters (Carn?vora: Mustelidae) based on mitochondrial cytochrome b sequences. J. Zool. Lond. 246, 401^16. Koepfli, K.-P., Wayne, R.K., 2003. Type I STS markers are more informa- tive than cytochrome b in phylogenetic reconstruction of the Musteli- dae (Mammalia: Carn?vora). Syst. Biol. 52, 571-593. Ledje, C, Arnason, U., 1996a. Phylogenetic analyses of complete cyto- chrome b genes of the order Carn?vora with particular emphasis on the Caniformia. J. Mol. Evol. 42, 135-144. Ledje, C, Arnason, U., 1996b. Phylogenetic relationships with caniform carnivores based on analyses of the mitochondrial 12S rRNA gene. J. Mol. Evol. 42, 641-649. Lieberman, D.E., 2000. Ontogeny, homology and phylogeny in the homi- nid craniofacial skeleton: the problem of the browridge. In: O'Higgins, P., Cohn, M.J. (Eds.), Development, Growth and Evolution. Academic Press, San Diego, CA, pp. 85-122. Losos, J.B., Glor, R.E., 2003. Phylogenetic comparative methods and the geography of speciation. Trends Ecol. Evol. 18, 220-227. Lotze, J.-H., Anderson, S., 1979. Procyon lotor. Mammal. Sp. 119, 1-8. Lovejoy, CO., Cohn, M.J., White, T.D., 1999. Morphological analysis of the mammalian postcranium: a developmental perspective. Proc. Nati. Acad. Sei. USA 96, 13247-13252. 1094 K.-P. Koepfli et al. I Molecular Phylogenetics and Evolution 43 (2007) 1076-1095 Lyons, L.A., Laughlin, T.F., Copeland, N.G., Jenkins, N.A., Womack, J.E., O'Brien, S.J., 1997. Comparative anchor tagged sequences (CATS) for integrative mapping of mammalian genomes. Nat. Genet. 15, 47-56. MacFadden, B.J., 2006. Extinct mammalian biodiversity of the ancient New World tropics. Trends Ecol. Evol. 21, 157-165. Maddison, D.R., Maddison, W.P., 2000. MacClade 4: analysis of phylog- eny and character evolution. Version 4.0. Sinauer Associates, Sunder- land, MA. M?rquez, A., Fari?a, R.A., 2003. Dental morphology and diet in canids and procyonids from Uruguay. Mammalia 67, 567-573. Marshall, L.G., 1985. Geochronology and land-mammal biochronology of the transamerican faunal interchange. In: Stehli, F.G., Webb, S.D. (Eds.), The Great American Interchange. Plenum Press, NY, pp. 49-85. Marshall, L.G., 1988. Land mammals and the great American interchange. Amer. Sei. 76, 380-388. Marshall, L.G., Butler, R.F., Drake, R.E., Curtis, G.H., Tedford, R.H., 1979. Calibration of the Great American Interchange. Science 204, 272-279. Marshall, L.G., Webb, S.D., Sepkoski Jr., J.J., Raup, D.M., 1982. Mamma- lian evolution and the great American interchange. Science 215, 1351- 1357. Marshall, L.G., Berta, A., Hoffstetter, R., Pascual, R., Reig, O.A., Bombin, M., Mones, A., 1984. Mammals and stratigraphy: geochronology of the continental mammal-bearing Quaternary of South America. Palaeo- vertebrata, Mem. 1984, Ext. 1-76. Marshall, L.G., Sempere, T., 1993. Evolution of the Neotropical Cenozoic land mammal fauna in its geochronologic, stratigraphie, and tectonic context. In: Goldblatt, P. (Ed.), Biological Relationships Between Africa and South America. Yale University Press, New Haven, CT, pp. 329-392. Mayr, E., 1986. Uncertainty in science: is the giant panda a bear or a rac- coon? Nature 323, 769-771. McFadden, K.W., 2004. The ecology, evolution, and natural history of the dwarf carnivores of Cozumel Island, Mexico. Ph.D. Dissertation, Columbia University. McGrew, P.O., 1938. Dental morphology of the Procyonidae with a description of Cynarctoides, gen. nov. Field Mus. Nat. Hist., Geol. Ser. 22, 323-339. McKenna, M.C., Bell, S.K., 1997. Classification of Mammals Above the Species Level. Columbia University Press, New York, NY. Mercer, J.M., Roth, V.L., 2003. The effects of Cenozoic global change on squirrel phylogeny. Science 299, 1568-1572. Miller, K.G., Kominz, M.A., Browning, J.V., Wright, J.D., Mountain, G.S., Katz, M.E., Sugarman, P.J., Cramer, B.S., Christie-Blick, N., Pekar, S.F., 2005. The Phanerozoic record of global sea-level change. Science 310, 1293-1298. Murphy, W.J., Eizirik, E., Johnson, W.E., Zhang, Y.-P., Ryder, O.A., O'Brien, S.J., 2001. Molecular phylogenetics and the origin of placental mammals. Nature 409, 614-618. Nowak, R.N., 2005. Walker's Carnivores of the World. Johns Hopkins University Press, Baltimore, MD. O'Brien, S.J., Nash, W.G., Wildt, D.E., Bush, M.E., Benveniste, R.E., 1985. A molecular solution to the riddle of the giant panda. Nature 317, 140-144. O'Brien, S.J., Benveniste, R.E., Nash, W.G., Martenson, J.S., Eichelberger, M.A., Wildt, D.E., Bush, M., Wayne, R.K., Goldman, D., 1991. Molecu- lar biology and evolutionary theory: the giant panda's closest relatives. In: Warren, L., Koprowski, H. (Eds.), New Perspectives on Evolution, vol. 4. Wiley-Liss, NY, pp. 225-250. Ortolani, A., 1999. Spots, stripes, tail tips and dark eyes: predicting the function of carnivore colour patterns using the comparative method. Biol. J. Linn. Soc. 67, 433^76. Pecon Slattery, J., O'Brien, S.J., 1995. Molecular phylogeny of the red panda (Ailurusfulgens). J. Hered. 86,413^22. Pocock, R., 1921. On the external characters and classification of the Procyonidae. Proc. Zool. Soc. Lond. 27, 389^22. Poglayen-Neuwall, I., Poglayen-Neuwall, I., 1965. Gefangenschaftsbeo- bachtungen an makib?ren (Bassaricyon Allen 1876). Z. S?ugetierk. 30, 321-366. Poglayen-Neuwall, I., Poglayen-Neuwall, I., 1995. Observations on the ethology and biology of the Central American cacomixtle, Bassariscus sumichrasti (de Saussure, 1860) in captivity, with notes on its ecology. Zool. Garten N.F. 65, 11^9. Poglayen-Neuwall, I., Toweill, D.E., 1988. Bassariscus astutus. Mammal. Sp. 327, 1-8. Pohle, H., 1917. Pseudobassaris riggsi, gen. nov., spec. nov. fur Amphictis spec. Riggs. Sitzungsberichte der Gesellschaft Naturforschender Fre- unde zu Berlin 1917,403^11. Polly, P.D., 1998. Variability, selection, and constraints: development and evolution in viverravid (Carn?vora, Mammalia) molar morphology. Paleobiology 24, 409^29. Pons, J.-M., Volobouev, V., Ducroz, J.-F., Tillier, A., Reudet, D., 1999. Is the Guadeloupean raccoon (Procyon minor) really an endemic species? New insights from molecular and chromosomal analyses. J. Zool. Syst. Evol. Res. 37, 101-108. Popwics, T.E., 1998. Ontogeny of postcanine tooth form in the ferret, Mus- tek putorius (Carn?vora: Mammalia), and the evolution of dental diversity within the Mustelidae. J. Morph. 237, 69-90. Posada, D., Crandall, K.A., 1998. MODELTEST: testing the model of DNA substitution. Bioinformatics 14, 817-818. Posada, D., Buckley, T.R., 2004. Model selection and model averaging in phylogenetics: advantages of the AIC and Bayesian approaches over likelihood ratio tests. Syst. Biol. 53, 793-808. Potts, R., Behrensmeyer, A.K., 1992. Late cenozoic terrestrial ecosystems. In: Behrensmeyer, A.K., Damuth, J.D., DiMichele, W.A., Potts, R., Sues, H.-D., Wing, S.L. (Eds.), Terrestrial Ecosystems Through Time: Evolutionary Paleoecology of Terrestrial Plants and Animals. Univer- sity of Chicago Press, Chicago, IL, pp. 419-541. Rambaut, A., Drummond, A.J., 2003. Tracer vl.3, http://evolve.zoo.ox.ac.uk/. Ronquist, F., Huelsenbeck, J.P., 2003. MrBayes 3: Bayesian phylogenetic inference vmder mixed models. Bioinformatics 19, 1572-1574. Sarich, V.M., 1973. The giant panda is a bear. Nature 245, 218-220. Sato, J.J., Wolsan, M., Suzuki, H., Hosoda, T., Yamaguchi, Y., Hiyama, K., Kobayashi, M., Minami, S., 2006. Evidence from nuclear DNA sequences sheds light on the phylogenetic relationships of Pinnipedia: single origin with affinity to Musteloidea. Zool. Sei. 23, 125-146. Schl?ter, D., 2000. The Ecology of Adaptive Radiation. Oxford Series in Ecology and Evolution. Oxford University Press, Oxford. Segall, W., 1943. The auditory region of the arctoid carnivores. Field Mus. Nat. Hist. Zool. Ser. 29, 33-59. Shaffer, H.B., Clark, J.M., Kraus, F., 1991. When molecules and mor- phology clash: a phylogenetic analysis of the North American ambystomatid salamanders (Caudata: Ambystomatidae). Syst. Zool. 40, 284-303. Simpson, G.G., 1945. The principles of classification and a classification of mammals. Buff. Am. Mus. Nat. Hist. 85, 1-350. Simpson, G.G., 1980. Splendid Isolation: The Curious History of South American Mammals. Yale University Press, New Havan, CT. Sorenson, M.D., 1998. TreeRot, version 2. University of Michigan, Ann Arbor, MI. Story, H.E., 1951. The carotid arteries in the Procyonidae. Field Zool. 32, 477-557. Strait, D.S., Grine, F.E., 2004. Inferring hominoid and early hominid phy- logeny using craniodental characters: the role of fossil taxa. J. Hum. Evol. 47, 399^52. Swofford, D.L., 2002. PAUP*. Phylogenetic Analysis Using Parsimony (*and Other Methods), Version 4.0. Sinauer Associates, Sunderland, MA. Swofford, D.L., Maddison, W.P., 1987. Reconstructing ancestral character states under Wagner parsimony. Math. Biosci. 87, 199-229. Tagle, D.A., Miyamoto, M.M., Goodman, M., Hofmann, O., Braunitzer, G., G?ltenboth, R., Jalanka, H., 1986. Hemoglobin of pandas: phyloge- netic relationships of carnivores as ascertained with protein sequence data. Naturwissenschaften 73, 512-514. Tamura, K., Nei, M., 1993. Estimation of the number of nucleotide substi- tutions in the control region of mitochondrial DNA in humans and chimpanzees. Mol. Biol. Evol. 10, 512-526. K.-P. Koepfli et al. I Molecular Phylogenetics and Evolution 43 (2007) 1076-1095 1095 Tavare, S., 1986. Some probabilistic and statistical problems on the analy- sis of DNA sequences. Lee. Math. Sei. 17, 57-86. Teeling, B.C., Scally, M., Kao, D.J., Romagnoli, M.L., Springer, M.S., 2000. Molecular evidence regarding the origin of echolocation and flight in bats. Nature 403, 188-192. Thorne, J.L., Kishino, H., Painter, I.S., 1998. Estimating the rate of evolution of the rate of molecular evolution. Mol. Biol. Evol. 15, 1647-1657. Trouessart, E.-L., 1904. Catalogus Mammalium: tam viventium quam fos- silium. R. Friedl?nder, Berlin. Van Valen, L.M., 1994. Serial homology: the crests and cusps of mamma- han teeth. Acta Palaeon. Pol. 38, 145-158. Van Valkenburgh, B., Koepfli, K.-P., 1993. Cranial and dental adaptations to pr?dation in canids. Symp. Zool. Soc. Lond. 65, 15-37. Venta, P.J., Brouillette, J.A., Yuzbasiyan-Gurkan, V., Brewer, G.J., 1996. Gene-specific universal mammalian sequence-tagged sites: application to the canine genome. Biochem. Genet. 34, 321-341. Vrana, P.B., Milkovitch, M.C., Powell, J.R., Wheeler, W.C, 1994. Higher level relationships of the arctoid Carn?vora based on sequence data and "total evidence". Mol. Phylogenet. Evol. 3, 47-58. Waits, L.P., Sullivan, J., O'Brien, S.J., Ward, R.H., 1999. Rapid radia- tion events in the Family Ursidae indicated by likelihood phyloge- netic estimation from multiple fragments of mtDNA. Mol. Phylogenet. Evol. 13, 82-92. Wang, X., 1997. New cranial material of Simocyon from China, and its implications for phylogenetic relationship to the red panda (Ailurus). J. Vert. Paleo. 17, 184-198. Webb, S.D., 1985. Late Cenozoic mammal dispersals between the Ameri- cas. In: Stehli, F.G., Webb, S.D. (Eds.), The Great American Inter- change. Plenum Press, NY, pp. 357-386. Webb, S.D., 1997. The great American faunal interchange. In: Coates, A.G. (Ed.), Central America: A Natural and Cultural History. Yale Univer- sity Press, New Haven, CT, pp. 97-122. Wendel, J.F., Doyle, J.J., 1998. Phylogenetic incongruence: window into genome history and molecular evolution. In: Soltis, P.S., Soltis, D.E., Doyle, J.J. (Eds.), Molecular Systematics of Plants II: DNA Sequenc- ing. Kluwer, Dordrecht, The Netherlands, pp. 265-296. Wenzel, J.W., Siddall, M.E., 1999. Noise. Cladistics 15, 51-64. Whitmore Jr., F.C., Stewart, R.H., 1965. Miocene mammals and Central American seaways. Science 148, 180-185. Wiens, J.J., Chippindale, P.T., Hillis, D.M., 2003. When are phylogenetic analyses misled by convergence? A case study in Texas cave salaman- ders. Syst. Biol. 52, 501-514. Wolsan, M., 1993. Phylogeny and classification of early European Mustel- ida (Mammalia: Carn?vora). Acta Theriol. 38, 345-384. Wolsan, M., Lange-Badr?, B., 1996. An arctomorph carnivoran skull from the Phosphorites du Quercy and the origin of procyonids. Acta Palaeo. Pol. 41, 277-298. Wozencraft, W.C, 2005. Order Carn?vora. In: Wilson, D.E., Reeder, D.M. (Eds.), Mammal Species of the World: A Taxonomic and Geo- graphic Reference. The Johns Hopkins University Press, Baltimore, pp. 279-348. Yang, Z., 1994. Estimating the pattern of nucleotide substitution. J. Mol. Evol. 39, 105-111. Yu, L., Li, Q.-W., Ryder, O.A., Zhang, Y.-P., 2004. Phylogenetic relation- ships within mammalian order Carn?vora indicated by sequences of two nuclear genes. Mol. Phylogenet. Evol. 33, 694?705. Zachos, J., Pagani, M., Sloan, L., Thomas, E., Billups, K., 2001. Trends, rhythms, and aberrations in global climate 65 Ma to present. Science 292, 686-693. Zevelolf, S.I., 2002. Raccoons: A Natural History. Smithsonian Institution Press, Washington, DC. Zhang, Y.-p., Ryder, O.A., 1993. Mitochondrial DNA sequence evolution in the Arctoidea. Proc. Nati. Acad. Sei. USA 90, 9557-9561. Zhang, Y.-p., Ryder, O.A., 1994. Phylogenetic relationships of bears (the Ursidae) inferred from mitochondrial DNA sequences. Mol. Phyloge- net. Evol 3, 351-359. Zhang, Y.-p., Shi, L.-M., 1991. The riddle of the giant panda. Nature 352, 573. Zharkikh, A., 1994. Estimation of evolutionary distances between nucleo- tide sequences. J. Mol. Evol. 39, 315-329.