Abstract

Phylogenies of closely related animal species are often inferred using mitochondrial DNA (mtDNA) gene sequences. The accuracy of mtDNA gene trees is compromised through hybridization that leads to introgression of mitochondrial genomes. Using DNA sequences from 6 single-copy nuclear genes and 2 regions of the mitochondrial genome, we investigated the temporal and geographic signature of mitochondrial and nuclear introgression in the Etheostoma spectabile darter clade. Phylogenetic analyses of the nuclear genes result in the monophyly of the E. spectabile clade; however, with respect to sampled specimens of 5 species (Etheostoma fragi, Etheostoma uniporum, Etheostoma pulchellum, Etheostoma burri, and E. spectabile), the mitochondrial phylogeny is inconsistent with E. spectabile clade monophyly. Etheostoma uniporum and E. fragi are both fixed for heterospecific mitochondrial genomes. Limited nuclear introgression is restricted to E. uniporum. Our analyses show that the pattern of introgression is consistently asymmetric, with movement of heterospecific mitochondrial haplotypes and nuclear alleles into E. spectabile clade species; introgressive hybridization spans broad temporal scales; and introgression is restricted to species and populations in the Ozarks. The introgressed mitochondrial genome observed in E. fragi has an obscure phylogenetic placement among darters, an ancient age, and is possibly a mitochondrial fossil from an Etheostoma species that has subsequently gone extinct. These results indicate that introgression, both ancient and more contemporaneous, characterizes the history of diversification in the E. spectabile species clade and may be relatively common among clades comprising the species-rich North American freshwater fauna.

A pattern emerging from several recent phylogenetic investigations among closely related species is that sampled DNA sequences from genes with independent evolutionary histories often do not converge onto a similar inferred species tree (Shaw 2002; Bachtrog et al. 2006). One perspective that is gaining wider acceptance is one championed by Maddison (1997, p. 523), where “all of the gene trees are part of the species tree, which can be visualized like a fuzzy statistical distribution, a cloud of gene histories.” Maddison (1997) suggested that phylogenetic analysis of multiple loci should be undertaken in an explicit coalescent framework, because it is expected that some gene trees will deviate from the underlying species tree due to stochastic variation in the coalescent process. However, it is important to consider that comparing and highlighting incongruence between gene trees can be biologically meaningful, ensuring that processes other than incomplete lineage sorting, such as hybridization, are considered and not lost in the “cloud of gene histories.”

Typically, phylogenetic relationships of animal species with recent diversification are investigated with gene trees inferred from mitochondrial DNA (mtDNA) gene sequences (Avise 1986). Mitochondrial genes are thought to have a better chance of tracking the species tree due to a higher mutation rate relative to nuclear genes, and ancestral alleles shared between incipient species will sort faster due to a smaller effective population size as a consequence of uniparental inheritance (Pamilo and Nei 1988; Moore 1995). However, it is important to consider that the mitochondrial genome is genetically a single locus and not necessarily a representative of the multitude of evolutionary histories of the unlinked genes in the nuclear genome. There are several mechanisms that will lead to incongruence between gene trees and the species tree, most notably incomplete lineage sorting of ancestral polymorphisms and introgression resulting from interspecific hybridization (Maddison 1997; Funk and Omland 2003). This prospect is particularly troubling for phylogenies of closely related species based only on cytoplasmic genome sequence data, because there is a relatively high frequency of heterospecific mitochondrial and chloroplast lineages that have crossed species boundaries in many plant and animal clades (Chan and Levin 2005).

Despite the possibility of being misled by using a single locus, the fact remains that mitochondrial gene sequences have an appreciably higher probability of resolving relationships among closely related animal species than DNA sequences from a single nuclear gene and perhaps substantial sets of nuclear genes (Moore 1995; Hudson and Coyne 2002). Given the need for phylogenetic trees among closely related species for comparative studies, detecting mtDNA introgression is a very important aspect of studying the phylogenetic history of these clades (Funk and Omland 2003; Grant et al. 2005). A notable pattern in the form of biased introgression of maternally inherited genomes can aid phylogeneticists in this endeavor. There are several mechanisms that independently, and in combination, could bias introgression of the cytoplasmic genome: sexual selection and asymmetric reproductive barriers (Chan and Levin 2005), demographic effects (Rieseberg et al. 1996b), differences in the magnitude of selection on particular genes (Funk and Omland 2003), and cytonuclear compatibilities (Rieseberg et al. 1996a). This biased cytoplasmic introgression manifests itself without introgression of alleles from the nuclear genome. Examples include Hawaiian crickets (Shaw 2002), the Drosophila yakuba species group (Bachtrog et al. 2006), Arctic and Brook charr (Doiron et al. 2002), mormyrid fishes (Sullivan et al. 2004), European cyprinid fishes (Freyhof et al. 2005), and lamprologine cichlid fishes (Schelly et al. 2006). In all these cases, phylogenetic analysis of mtDNA sequences would reflect the heterospecific origin of the mitochondrial genomes, resulting in an incorrect conclusion with regard to the inferred species phylogeny. The method that has shown the most promise to detect mitochondrial introgression involves a direct assessment of incongruence between mitochondrial and nuclear gene trees, as well as differences between molecular phylogenies and morphological characters (Shaw 2002; Bachtrog et al. 2006). Because of the uniparental inheritance of the mitochondrial genome, this straightforward phylogenetic approach can also identify the directionality of introgression, and given the phylogenetic pattern coupled with molecular branch lengths, it may also provide information on the relative timing of introgressive hybridization events.

Our focus on mtDNA introgression is the result of ongoing investigations in the phylogenetic relationships of darters, a clade of approximately 225 freshwater teleost fish species endemic to eastern North America. In particular, published studies have found substantial incongruence between the mtDNA gene tree and a single nuclear gene tree that is consistent with extensive mtDNA introgression in several species of the Etheostoma spectabile species clade (Lang and Mayden 2007; Ray et al. 2008), collectively called the orangethroat darters. This clade contains 8 recognized species (Distler 1968; Ceas and Page 1997; Ceas and Burr 2002; Nelson et al. 2004), 2 subspecies that we regard here as species, and at least 4 undescribed species. Breeding males of the E. spectabile clade are brilliantly colored and superficially resemble the rainbow darter, Etheostoma caeruleum, differing only in breeding pigment pattern, pectoral fin ray counts, and infraorbital canal characteristics (Trautman 1930; Page 1983). Reflecting the overall similarity in color pattern, E. caeruleum and species of the E. spectabile species clade have traditionally been classified in the darter subclade Oligocephalus (Page 1981, 1983; Bailey and Etnier 1988). Species in the E. spectabile clade and E. caeruleum co-occur throughout their respective distributions in the Eastern and Interior Highlands of eastern North America (Fig. 1), and putative F1 hybrid specimens have been documented (Distler 1968). In addition, natural hybridization between Etheostoma pulchellum with Etheostoma radiosum and Etheostoma whipplei has been documented (Linder 1955, 1958; Branson and Campbell 1969; Echelle et al. 1974; Mendelson 2003a).

FIGURE 1.

Geographic distribution of Etheostoma spectabile clade species. The geographic distributions of species in the Eastern Highlands are colored grey and include Etheostoma bison, Etheostoma tecumsehi, Etheostoma lawrencei, Etheostoma kantuckeense, Mamequit darter, Sheltowee darter, and Ihiyo darter. Region of ambiguous species designation between E. spectabile and Etheostoma pulchellum is indicated. Approximate sampling localities are shown with solid circles. Inset map shows the geographic distribution of Etheostoma caeruleum.

FIGURE 1.

Geographic distribution of Etheostoma spectabile clade species. The geographic distributions of species in the Eastern Highlands are colored grey and include Etheostoma bison, Etheostoma tecumsehi, Etheostoma lawrencei, Etheostoma kantuckeense, Mamequit darter, Sheltowee darter, and Ihiyo darter. Region of ambiguous species designation between E. spectabile and Etheostoma pulchellum is indicated. Approximate sampling localities are shown with solid circles. Inset map shows the geographic distribution of Etheostoma caeruleum.

Phylogenetic analyses of mitochondrial and nuclear genes indicate that Oligocephalus is not monophyletic and the E. spectabile species clade is distantly related to E. caeruleum (Lang and Mayden 2007). The gene trees presented in Lang and Mayden (2007) also reveal that although the E. spectabile species clade was monophyletic for the sampled nuclear gene, the species complex was not monophyletic in the mitochondrial gene tree. In particular, mtDNA haplotypes sampled from Etheostoma uniporum were closely related to those sampled from E. caeruleum, and mtDNA haplotypes sampled from Etheostoma fragi were not closely related to any of the sampled darter species. The similarity of mtDNA haplotypes observed in E. uniporum and E. caeruleum was explored in Ray et al. (2008), where they discovered phylogenetic patterns consistent with hybridization, followed by introgression of E. caeruleum mtDNA haplotypes into E. uniporum. The discrepancy between the mitochondrial and the nuclear gene trees observed within the E. spectabile species clade appears consistent with multiple instances of heterospecific mtDNA introgression.

Our research goals are to investigate the extent of introgression of mitochondrial genomes and nuclear alleles in the E. spectabile clade through comparisons of gene trees inferred from mtDNA and multiple nuclear genes. We are interested in determining if extensive mtDNA introgression has occurred without nuclear introgression, as seen in other animal clades (Powell 1983; Spolsky and Uzzell 1984; Bachtrog et al. 2006). Incongruence between inferred mitochondrial and nuclear gene trees would suggest that mtDNA introgression has been a feature of the diversification of the E. spectabile species clade, whereas limited incongruence among the phylogenies inferred from the nuclear genes would indicate a lack of introgression in the nuclear genome. We intend to focus on both the temporal and the geographic contexts of introgression, asking 2 main questions. First, what is the relative timing of introgression events as determined through comparison of gene trees? Second, given that species of the E. spectabile complex co-occur with E. caeruleum in 2 areas with rich aquatic biodiversity (Fig. 1), are hybrids and genomes with introgressed alleles found throughout their overlapping geographic distributions, or is introgression constrained to particular geographic regions? Our research strategy is unique among published studies involving teleost fishes in that it uses 6 single-copy nuclear genes to investigate mtDNA introgression and estimate phylogenetic relationships within a clade of closely related species.

MATERIALS AND METHODS

Specimen Sampling, DNA Sequencing, and Alignment

Etheostoma spectabile clade species were collected throughout their geographic distributions using a seine net (Fig. 1; Supplementary Table A1, http://www.oxfordjournals.org/our_journals/sysbio/). On collection, specimens were anesthetized using MS-222, and a tissue biopsy was taken from the right pectoral fin and cataloged in the Yale Fish Tissue Collection. The specimens were fixed in formalin and later transferred to 70% ethanol and deposited in the fish collection at the Yale Peabody Museum of Natural History (YPM) or the University of Tennessee Research Collection of Fishes. Dr. Patrick Ceas (St. Olaf College) provided additional specimens of E. spectabile and E. pulchellum. The darter species E. caeruleum, E. radiosum, and E. whipplei were sampled to ensure inclusion of potential parental species involved in hybridization with E. spectabile clade species. To explore the phylogenetic relationships of the E. spectabile clade with respect to the entire darter clade, representative species from all the major darter lineages were sampled. Sequences of the mtDNA gene cytochrome b (cytb) collected from specimens sampled across the geographic distribution of E. caeruleum were downloaded from GenBank (Ray et al. 2006) to investigate the geographic context of introgression. The non-darter percids, Sander vitreus and Perca flavescens, were included as outgroups in all phylogenetic analyses.

Nucleic acids were isolated with standard phenol-chloroform extraction and ethanol precipitation methods or with a Qiagen DNAeasy Tissue Extraction Kit (QIAGEN, Valencia, CA) following the manufacturer's protocol. The mitochondrial genes cytb and NADH subunit 2 (ND2) were amplified using polymerase chain reaction (PCR), with primers and cycling conditions reported in Near et al. (2000) and Kocher et al. (1995). Nuclear genes were amplified using PCR, with primers and cycling conditions reported in previous studies; S7 ribosomal protein intron 1 (Chow and Hazama 1998), RAG1 exon 3 (Lopez et al. 2004), mixed lineage leukemia (MLL) (Dettaï and Lecointre 2005), Tmo-4C4 (Streelman and Karl 1997), and myoglobin (Mb) (Grove et al. 2004). Primers for the KELCH locus were provided by Leos Kral (KELCH-F GCAAGAAACCAGGCTAACA and KELCH-R GCATGAAATGCCGACTGT). After initial sequencing, specific primers were designed for PCR and sequencing of MLL (MLL-F2 TTGGCCTCACRCCTTTCTACGGGG and MLL-R2 CTGAGSGACAGTTGGGCCTC) and Mb (MVO-F2b AGCTGTTCCCCAAGTTTGCCG and MVO-R2 GCATCGAGTCCTGCCTTCTC). All 6 nuclear genes were sequenced for a subset of our specimens that included 46 individuals sampled from 27 darter species. Individuals from multiple populations of E. fragi, E. uniporum, E. caeruleum, and putative hybrid specimens were sampled. Thermal cycling for PCR amplification of the nuclear genes included an initial denaturation step of 94°C for 3 min, followed by 30 cycles of denaturing at 94°C (30 s); primer annealing at 57.4°C (30 s) for RAG1 exon 3 and MLL, and 55°C for Mb, Tmo-4C4, and KELCH; and extension at 72°C (1.5 min), followed by a final incubation period of 72°C for 5 min.

Amplified DNA products from PCR were cleaned prior to direct DNA sequencing using enzymatic purification with exonuclease 1 and shrimp alkaline phosphatase, incubated at 37°C for 15 min, followed by 80°C for 15 min to inactivate the enzymes, or using a Qiagen Qiaquick PCR Purification Kit (QIAGEN). Cleaned PCR products were used as templates for Big Dye (Applied Biosystems, Foster City, CA) cycle sequencing, and sequencing reactions were read using an ABI 3100 automated sequencer at the Molecular Systematics and Conservation Genetics Laboratory (Department of Ecology and Evolutionary Biology, Yale University, New Haven, CT). Primers were designed for sequencing the RAG1 locus (RAG1F2 GATCTTTCAGCCCCTGCAYNCCC and RAG1R2 GCTCAAAGGCCTTTGACTGRC). Contiguous sequences were built using Sequencher (GeneCodes, Ann Arbor, MI). The cytb and ND2 sequences were aligned by eye using the text editor in PAUP* v. 4.01 (Swofford 2002). All nuclear genes were aligned in ClustalX 1.8 (Thompson et al. 1997), followed by minor adjustments using the data editor in MacClade 4.0 (Maddison and Maddison 2000). Some individuals contained heterozygous genotypes for the sampled nuclear loci, and any heterozygous base pair positions were coded using standard degeneracy codes.

Phylogenetic Analyses and Species Tree Estimation

Partitioned Bayesian analyses were used to generate phylogenetic trees from the mitochondrial and nuclear gene alignments (Larget and Simon 1999). Phylogenies were estimated using several combinations of the sampled mitochondrial and nuclear genes: cytb for all sampled darter specimens including the supplemental E. caeruleum sequences sampled from GenBank (Ray et al. 2006), the combined cytb and ND2 mitochondrial gene alignment for E. spectabile complex species and other major darter lineages, each nuclear gene separately, and all nuclear genes combined into a single data matrix. Character partitions used in the partitioned Bayesian analyses were designated based on intron and exon regions and codon positions in coding regions for each gene. The partitioning scheme used in this study is similar to that used in other studies of darter phylogeny that have used both mitochondrial and nuclear gene sequences (Page et al. 2003; Lang and Mayden 2007; Keck and Near 2008; Piller et al. 2008). The models of molecular evolution for each character partition used in the Bayesian analyses were selected using the Akaike Information Criterion as executed in the computer program Modeltest v. 3.06 (Posada and Crandall 1998). Partitioned Bayesian analyses were run in MrBayes v. 3.1.2 (Ronquist and Huelsenbeck 2003) for 20 million generations, and parameter convergence was verified with Tracer v1.4 (Rambaut and Drummond 2003). Four chains were run simultaneously in each analysis. The burn-in period for the Bayesian analyses was 1 × 106 generations, and the parameters and trees generated before the burn-in period were discarded.

Given the potential for introgression in the E. spectabile clade and resulting incongruence between different gene phylogenies, it is important to explore different strategies to estimate the species tree from the individual gene trees. There are differing opinions regarding the merits of different methods to estimate a species tree, given data sampled from multiple genes. These methods include construction of a consensus tree from the individual gene trees (Jennings and Edwards 2005) and estimation of the species tree from a concatenated multiple-gene data set (Nylander et al. 2004; Rokas and Carroll 2005; Kubatko and Degnan 2007). Two methods have been developed to incorporate assumptions of coalescent models in estimating the optimal species tree: one minimizes the number of coalescent events across multiple gene trees (Maddison 1997; Maddison and Knowles 2006; Carstens and Knowles 2007), and the second uses a Bayesian strategy to estimate the species tree (Edwards et al. 2007; Liu and Pearl 2007). These methods assume that the incongruence among gene trees is due to incomplete lineage sorting; thus, our species tree analyses focused on the nuclear gene trees to remove any incongruence caused by mitochondrial introgression. We performed a partitioned Bayesian analysis that combined the data from the 6 sampled nuclear genes as discussed above. Mesquite v. 2.01 (Maddison and Maddison 2007) was used to estimate the species tree, given assumptions of lineage sorting and minimization of deep coalescences (Maddison 1997). A gene tree was estimated from each of the 6 nuclear loci using maximum likelihood trees estimated using the computer program RAxML (Stamatakis 2006), and these trees were used in Mesquite along with an association matrix that related each of the 46 sampled gene sequences to the 27 sampled species. A heuristic tree search within the Mesquite coalescent module using deep coalescences with multiple loci criterion and subtree pruning and regrafting branch swapping was used to find optimal species trees that minimized deep coalescence events. When multiple optimal trees were returned, a 50% majority rule consensus tree was generated using the computer program Phyutility (Smith and Dunn 2008).

A Bayesian method implemented in the computer program BEST v. 2.0 was also used to estimate the species tree (Liu and Pearl 2007). We incorporated multiple alleles of the 6 nuclear genes from the E. spectabile clade species Etheostoma burri, E. fragi, E. pulchellum, E. spectabile, and E. uniporum. Multiple alleles of the nuclear genes were also sampled from E. caeruleum. The optimal substitution models presented in Table 1 were for each locus. The prior distribution of theta (θ) was an inverse gamma distribution with α = 3 and β = 0.03 (Liu et al. 2008). Chains were run for 1 × 108 generations, and the frequency of sampling was every 1000 generations. Convergence of the chains was assessed by examining the log-likelihood values using the computer program Tracer v. 1.4 (Rambaut and Drummond 2003), and the first 70 000 trees sampled before reaching stationarity were discarded as burn-in. The estimated posterior distribution of species trees was determined with the “sumt” command in BEST that produced a 50% majority rule consensus tree.

TABLE 1.

Selected optimal molecular evolutionary models for character partitions identified for each gene

Gene Length (bp) Model of first codon Model of second codon Model of third codon Model of intron 
Cytb 1140 TIMef+I+G GTR+I+G GTR+I+G NA 
ND2 1047 TrN+I+G GTR+I+G GTR+I+G NA 
S7 intron 1 543 NA NA NA TVM+I 
Tmo4C4 473 F81+I F81 SYMd +G NA 
Mb 409 TrNef TVM HKY TVM 
KELCH 754 K81uf+I TrN+I F81+I HKY 
MLL 807 Exon 25: TrN TrN+I TrNef TIM 
  Exon 26: HKY TrN Trn+I  
RAG1 exon 3 1377 TVM+I TVM+I TVM+G NA 
Gene Length (bp) Model of first codon Model of second codon Model of third codon Model of intron 
Cytb 1140 TIMef+I+G GTR+I+G GTR+I+G NA 
ND2 1047 TrN+I+G GTR+I+G GTR+I+G NA 
S7 intron 1 543 NA NA NA TVM+I 
Tmo4C4 473 F81+I F81 SYMd +G NA 
Mb 409 TrNef TVM HKY TVM 
KELCH 754 K81uf+I TrN+I F81+I HKY 
MLL 807 Exon 25: TrN TrN+I TrNef TIM 
  Exon 26: HKY TrN Trn+I  
RAG1 exon 3 1377 TVM+I TVM+I TVM+G NA 

Abbreviation: NA = the gene does not contain the specified region.

Phylogenetic Congruence of Gene Trees

The significance of constraining the gene trees that did not support the monophyly of the E. spectabile clade and E. uniporum specimens was explored using Bayes factors that quantified the support of alternative hypotheses given the observed data (Suchard et al. 2005). A Bayes factor is defined as the ratio of marginal likelihoods, or the likelihood of data under a particular model after parameter estimation, from 2 competing hypotheses. The Bayes factor can be interpreted as the success of each hypothesis at predicting the data (Kass and Raftery 1995). In this case, the null hypotheses were the optimal gene trees resulting from the unconstrained Bayesian analyses and were compared against the alternative hypotheses or the phylogenies that resulted from the constrained searches to enforce monophyly of the E. spectabile species clade and the sampled E. uniporum specimens. Each sampled gene that did not result in optimal gene trees that support the monophyly of the E. spectabile clade and E. uniporum was analyzed in MrBayes with the appropriate monophyly constraint. The Markov chain Monte Carlo (MCMC) parameters were identical to the unconstrained analyses discussed above. The marginal log-likelihoods were retrieved via the “sump” function in MrBayes and interpretations were based on the guidelines presented by Kass and Raftery (1995).

Relative Timing of Introgression

The relative timing among inferred mtDNA introgression events was estimated using an uncorrelated lognormal (UCLN) model of rate variation executed in the computer program BEAST v. 1.4.7 (Drummond et al. 2006; Drummond and Rambaut 2007). A uniform prior with a mean of 100 arbitrary time units with upper and lower bounds at 99 and 101 time units was set for the most recent common ancestor (MRCA) of the entire darter clade on the mtDNA gene tree. BEAST was run for 4 × 107 generations with a Yule speciation prior and the optimal partitioned molecular evolution models used in the MrBayes analyses. We used the computer program Tracer v. 1.4 to determine convergence and measure the effective sample size of estimated parameters and relative ages. The ages of the MRCA for nodes in the mtDNA gene tree characterized by introgression were tracked and the distribution of relative age estimates that included the 95% highest posterior density (HPD) was reported.

RESULTS

Alignment and Optimal Molecular Evolutionary Models

The initial alignment of the cytb mtDNA contained sequences from 273 individuals that included all E. spectabile clade species and the E. caeruleum sequences reported in Ray et al. (2006). The alignment consisted of 1140 base pairs with no insertions or deletions. Replicated haplotypes were removed, leaving 170 individuals in the cytb alignment. All new DNA sequences generated for this study were submitted to GenBank (FJ381003–FJ381428), and all data matrices and resulting trees were submitted to TreeBASE (http://www.treebase.org; S2262). Following the protocols outlined in Brandley et al. (2005), the optimal molecular evolutionary models for each gene were chosen by comparison of Bayes factors for each partitioning scheme. These models as well as the length of the alignments for all sampled genes are shown in Table 1.

Phylogenetic Analyses of mtDNA Sequences

The E. spectabile clade was not monophyletic in the phylogenies resulting from the Bayesian analyses of the cytb and combined cytb and ND2 mtDNA alignments, and the standard deviation of split frequencies was less than 0.05, indicating convergence with regard to the tree topology (Figs. 2 and 3). A core clade containing specimens of all E. spectabile clade species, except E. uniporum and E. fragi, was resolved and supported with significant posterior probabilities in the Bayesian inferred mtDNA phylogenies (≥95% posterior probability). However, sampled specimens of 5 E. spectabile clade species (E. fragi, E. uniporum, E. pulchellum, E. burri, and E. spectabile) were distributed among 7 different clades in the mtDNA phylogenies (Fig. 2). Most notably, all the haplotypes sampled from E. uniporum and E. fragi were distantly related to the E. spectabile clade. As reported in Ray et al. (2008), all the E. uniporum mtDNA haplotypes were phylogenetically nested within those sampled from E. caeruleum. Most of the sampled E. uniporum haplotypes form a distinct clade of specimens sampled throughout the limited geographic distribution of the species (Fig. 1), but the mtDNA haplotypes of several E. uniporum specimens were more closely related to E. caeruleum sampled from the same river drainage (Fig. 2). Average uncorrected pairwise distances of cytb and ND2 between E. uniporum haplotypes and those from the core E. spectabile species clade in the Bayesian phylogeny ranged between 17% and 19%, and were as high as 7% between specimens sampled from the widespread E. uniporum clade and E. caeruleum. Our new mtDNA phylogenies also showed that mtDNA haplotypes of E. fragi did not appear closely related to any sampled darter lineage (Figs. 2 and 3). Average uncorrected pairwise distances at cytb and ND2 between E. fragi and specimens in the core E. spectabile clade were 20% and the minimum pairwise distance between E. fragi and E. caeruleum mtDNA gene sequences was 15%.

FIGURE 2.

Phylogeny resulting from a partitioned Bayesian analysis of mitochondrial cytbgene sequences; the phylogenetic tree is connected at the arrowheads. The red block and red branches highlight the Etheostoma spectabileclade species, and the blue box highlights the Etheostoma caeruleumclade. Asterisks identify nodes supported by Bayesian posterior probabilities ≥ 95%.

FIGURE 2.

Phylogeny resulting from a partitioned Bayesian analysis of mitochondrial cytbgene sequences; the phylogenetic tree is connected at the arrowheads. The red block and red branches highlight the Etheostoma spectabileclade species, and the blue box highlights the Etheostoma caeruleumclade. Asterisks identify nodes supported by Bayesian posterior probabilities ≥ 95%.

FIGURE 3.

A phylogeny of 44 individuals focusing on putative introgressed mitochondrial genomes resulting from the partitioned Bayesian analysis of 2 mitochondrial genes, cytband ND2. The red and blue boxes highlight the Etheostoma spectabileand Etheostoma caeruleumclades, respectively. Red branches indicate E. spectabileclade species. Asterisks identify nodes supported by Bayesian posterior probabilities ≥ 95%.

FIGURE 3.

A phylogeny of 44 individuals focusing on putative introgressed mitochondrial genomes resulting from the partitioned Bayesian analysis of 2 mitochondrial genes, cytband ND2. The red and blue boxes highlight the Etheostoma spectabileand Etheostoma caeruleumclades, respectively. Red branches indicate E. spectabileclade species. Asterisks identify nodes supported by Bayesian posterior probabilities ≥ 95%.

Sampled mtDNA haplotypes from E. spectabile, E. pulchellum, and E. burri were not monophyletic in the phylogenetic analyses (Figs. 2 and 3). Haplotypes sampled from these species either were distributed in the E. spectabile species clade or in the clade of E. caeruleum haplotypes or were closely related to E. whipplei (Figs. 2 and 3). These populations of E. spectabile and E. burri are polymorphic with regard to their mitochondrial genomes because they include individuals with E. spectabile-like mitochondrial haplotypes and those that exhibit E. caeruleum-like mitochondrial haplotypes. In 2 instances, all sampled individuals from a population were fixed for haplotypes that were similar to other either E. caeruleum or E. whipplei. Three specimens of E. spectabile sampled from Doe Run (St. Francis River System, St. Francis County, MO) had mtDNA haplotypes typical of E. caeruleum found in the same river system. The 3 E. pulchellum specimens sampled from Mill Branch (Arkansas River Drainage, White County, AR) contained haplotypes that were closely related to those found in specimens of E. whipplei collected at the same locality (Figs. 2 and 3). There were 2 examples where individuals of E. spectabile and E. burri sampled from the same location possessed 2 divergent mtDNA haplotypes that were either phylogenetically nested in the core E. spectabile species clade or most closely related to E. caeruleum haplotypes sampled from the same river drainages.

Phylogenetic Analyses of Nuclear Gene DNA Sequences

In stark contrast to the phylogenies resulting from the mtDNA genes, Bayesian phylogenetic analyses for each of the 6 sampled nuclear genes provided substantial support for monophyly, or monophyletic to the exclusion of a single specimen sampled from the E. spectabile species clade (Figs. 4 and 5). The standard deviation of split frequencies was less than 0.05, indicating convergence with regard to the tree topology. The alleles sampled at the 6 nuclear gene loci for E. uniporum and E. fragi were monophyletic and related to a clade comprising the remaining E. spectabile species or were each nested in a clade containing all other E. spectabile clade species. Bayesian analyses for 3 of the 6 nuclear genes (KELCH, MLL, and Mb) resulted in phylogenies that contained a clade of all specimens sampled from 1011.8the E. spectabile species complex supported with significant Bayesian posterior probabilities (Fig. 4). Bayesian phylogenies inferred from the S7 ribosomal protein intron 1 and RAG1 exon 3 supported near monophyly of the E. spectabile clade because in each of these 2 gene trees, a single E. uniporum specimen had an allele that was more closely related to the alleles sampled from E. caeruleum (Fig. 5a,b). Bayesian analysis of the Tmo-4C4 locus resulted in a phylogeny that lacked resolution with regard to interpreting the monophyly of the E. spectabile clade (Fig. 5c).

FIGURE 4.

Phylogenies resultingfrom partitioned Bayesian analyses of the nuclear genes (a) KELCH, (b) MLL, and (c) Mb. The gray box highlights the Etheostoma spectabileclade species. Asterisks identify nodes supported by Bayesian posterior probabilities ≥ 95%.

FIGURE 4.

Phylogenies resultingfrom partitioned Bayesian analyses of the nuclear genes (a) KELCH, (b) MLL, and (c) Mb. The gray box highlights the Etheostoma spectabileclade species. Asterisks identify nodes supported by Bayesian posterior probabilities ≥ 95%.

FIGURE 5.

Partitioned Bayesian analyses of the nuclear genes (a) ribosomal protein S7 intron 1 (S7), (b) RAG1 exon 3 (RAG1), and (c) Tmo-4C4 (Tmo). The gray box highlights the Etheostoma spectabile clade species. Asterisks identify nodes supported by Bayesian posterior probabilities ≥ 95%.

FIGURE 5.

Partitioned Bayesian analyses of the nuclear genes (a) ribosomal protein S7 intron 1 (S7), (b) RAG1 exon 3 (RAG1), and (c) Tmo-4C4 (Tmo). The gray box highlights the Etheostoma spectabile clade species. Asterisks identify nodes supported by Bayesian posterior probabilities ≥ 95%.

The E. spectabile species clade was monophyletic and supported with a significant posterior probability in a Bayesian phylogenetic analysis of the concatenated nuclear genes, and the standard deviation of split frequencies was less than 0.05, indicating convergence with regard to the tree topology (Fig. 6). E. fragi and E. uniporum were most closely related to one another, and all other sampled E. spectabile species were resolved as a clade with significant Bayesian posterior support. Only 2 interspecific nodes in the phylogeny were supported with a Bayesian posterior probability less than 1.00, and the phylogeny presented well-resolved relationships among the major darter lineages (Fig. 6).

FIGURE 6.

Phylogeny resulting from a partitioned Bayesian analysis of the concatenated 6 nuclear genes. The gray box encloses the Etheostoma spectabile clade species. Asterisks identify nodes supported by Bayesian posterior probabilities ≥ 95%.

FIGURE 6.

Phylogeny resulting from a partitioned Bayesian analysis of the concatenated 6 nuclear genes. The gray box encloses the Etheostoma spectabile clade species. Asterisks identify nodes supported by Bayesian posterior probabilities ≥ 95%.

The E. spectabile species clade was also monophyletic in the species analysis that minimized deep coalescences (Fig. 7a). A total of 6 optimal species trees, differing only in outgroup relationships and in the relationship between Etheostoma vitreum and the clade containing the species Etheostoma microperca, Etheostoma fonticola, and Etheostoma proeliare were combined in a majority rule consensus tree. The E. spectabile species clade was found in each of the 6 species trees, as was the sister species relationship of E. uniporum and E. fragi.

FIGURE 7.

Results from species tree analyses. (a) A 50% majority rule consensus tree resulting from a minimizing deep coalescences analysis. A gray box encloses Etheostoma spectabile clade species. Values at nodes represent consensus support. (b) A 50% majority rule consensus tree resulting from a BEST analysis. A gray box encloses E. spectabile clade species. Values at nodes represent posterior probability support.

FIGURE 7.

Results from species tree analyses. (a) A 50% majority rule consensus tree resulting from a minimizing deep coalescences analysis. A gray box encloses Etheostoma spectabile clade species. Values at nodes represent consensus support. (b) A 50% majority rule consensus tree resulting from a BEST analysis. A gray box encloses E. spectabile clade species. Values at nodes represent posterior probability support.

Similar to the concatenated data Bayesian analysis and the parsimony minimizing deep coalescence species tree analysis, the BEST species tree analysis resulted in a monophyletic E. spectabile species clade that was supported with a significant Bayesian posterior probability (Fig. 7b). Etheostoma caeruleum was not closely related to the E. spectabile species clade, but was more closely related to Etheostoma luteovinctum, E. radiosum, and E. whipplei.

Phylogenetic Congruence of Gene Trees

The Bayesian phylogenies inferred from the mitochondrial and nuclear genes differ substantially with regard to the relationships of the E. spectabile species clade (Figs. 23456). Comparing the optimal unconstrained mtDNA gene trees against the constrained mtDNA gene trees that reflect the monophyly of the E. spectabile species clade resulted in very high Bayes factors, providing very strong support for the optimal mtDNA gene trees and rejecting monophyly of the E. spectabile clade. The Bayes factor calculated for the contrast between the optimal S7 gene tree and the E. spectabile species clade constrained S7 gene tree was moderate and interpreted as strong support for the optimal S7 gene tree. Comparing the optimal gene trees for RAG1 and Tmo-4C4 with the constrained gene trees reflecting monophyly of the E. spectabile clade resulted in Bayes factors that are interpreted as positive support for the optimal unconstrained Bayesian gene trees (Table 2). Interestingly, although multiple levels of support exist for the optimal unconstrained phylogenetic hypotheses, even the lowest Bayes factor for RAG1 provides significant support for introgression in the E. spectabile species complex.

TABLE 2.

Comparison of alternative phylogenetic hypotheses using Bayes factors

Gene Constrained group H0: unconstrained HA: constrained monophyly Bayes factor = 2log B10  =  2(H0HABayes factor interpretation 
All taxa mtDNA Etheostoma spectabile − 15 982.70 − 17 043.39 2121.38 Very strong 
 group monophyly     
Subsample mtDNA Etheostoma spectabile − 24 001.02 − 25 053.67 2105.30 Very strong 
 group monophyly     
Subsample S7 Etheostoma spectabile − 3241.29 3274.07 65.56 Strong 
 group monophyly     
 Etheostoma uniporum group − 3241.29 − 3300.85 119.12 Strong 
 monophyly     
Subsample RAG1 Etheostoma spectabile − 4280.20 − 4284.55 8.70 Positive 
exon 3 group monophyly     
 Etheostoma uniporum group − 4280.20 − 4285.01 9.62 Positive 
 monophyly     
Subsample Tmo-4C4 Etheostoma spectabile − 1722.98 − 1728.76 11.56 Positive 
 roup monophyly     
Gene Constrained group H0: unconstrained HA: constrained monophyly Bayes factor = 2log B10  =  2(H0HABayes factor interpretation 
All taxa mtDNA Etheostoma spectabile − 15 982.70 − 17 043.39 2121.38 Very strong 
 group monophyly     
Subsample mtDNA Etheostoma spectabile − 24 001.02 − 25 053.67 2105.30 Very strong 
 group monophyly     
Subsample S7 Etheostoma spectabile − 3241.29 3274.07 65.56 Strong 
 group monophyly     
 Etheostoma uniporum group − 3241.29 − 3300.85 119.12 Strong 
 monophyly     
Subsample RAG1 Etheostoma spectabile − 4280.20 − 4284.55 8.70 Positive 
exon 3 group monophyly     
 Etheostoma uniporum group − 4280.20 − 4285.01 9.62 Positive 
 monophyly     
Subsample Tmo-4C4 Etheostoma spectabile − 1722.98 − 1728.76 11.56 Positive 
 roup monophyly     

Relative Timing of mtDNA Introgression

Estimation of relative divergence times in the mtDNA cytb gene tree using the UCLN model in BEAST revealed that introgressed mtDNA lineages have differing ages of origination (Fig. 8). The MRCA of the entire darter clade was calibrated with an age of 100 relative time units, and the UCLN mean age estimate for this node was 93.8 (95% HPD: 86.0–100.5). The estimated relative age for the mtDNA haplotype lineage observed in E. fragi was 68.4 (95% HPD: 59.9–76.7) relative time units. The widespread E. uniporum clade that is phylogenetically nested within sampled E. caeruleum haplotypes was much younger, with an estimated relative age of the stem lineage at 11.8 (95% HPD: 9.5–14.2) time units and a crown node age of 9.3 (95% HPD: 6.7–11.9) time units (Fig. 8).

FIGURE 8.

Histograms of the posterior density of relative ages for the MRCA of each identified mtDNA introgression event, as estimated using BEAST, with the same taxon sampling as the mitochondrial gene tree presented in Figure 2. Circles identify the node age presented in the histogram, and the color of the marked node (gray or black) matches that of the histogram when multiple nodes are dated in a single graph. Thick branches in the phylogenies mark heterospecific haplotype lineages found in Etheostoma spectabile clade species. The mean age is reported as well as the 95% upper and lower HPD. The first histogram estimates the relative age of the MRCA of the entire darter clade.

FIGURE 8.

Histograms of the posterior density of relative ages for the MRCA of each identified mtDNA introgression event, as estimated using BEAST, with the same taxon sampling as the mitochondrial gene tree presented in Figure 2. Circles identify the node age presented in the histogram, and the color of the marked node (gray or black) matches that of the histogram when multiple nodes are dated in a single graph. Thick branches in the phylogenies mark heterospecific haplotype lineages found in Etheostoma spectabile clade species. The mean age is reported as well as the 95% upper and lower HPD. The first histogram estimates the relative age of the MRCA of the entire darter clade.

The E. uniporum mtDNA haplotypes that were phylogenetically nested within E. caeruleum haplotypes sampled from the same river drainages ranged in estimated relative age from 9.9 (95% HPD: 7.9–11.9) to 1.0 (95% HPD: 0.4–1.7) relative time units. The relative age of the haplotype lineage observed in E. pulchellum specimens that are closely related to E. whipplei was estimated as 2.5 (95% HPD: 1.4–3.6) time units, and the estimated relative ages of mtDNA haplotypes more closely related to E. caeruleum, but found in specimens of E. burri and E. spectabile, ranged between 0.5 (95% HPD: 0.1–1.0) and 1.8 (95% HPD: 1.1–2.5) relative time units (Fig. 8).

DISCUSSION

Through comparisons of phylogenies inferred from mitochondrial and nuclear genes, we have been able to reveal a complex pattern of repeated mitochondrial introgression, with little introgression of nuclear alleles in the E. spectabile species clade. Our results were intimated in previous studies that had a more limited sampling of specimens and nuclear genes. Lang and Mayden (2007) presented a phylogeny based on mtDNA sequences of the ND2 gene that showed extensive nonmonophyly of the E. spectabile clade, but the group was monophyletic in the phylogeny inferred from the nuclear S7 ribosomal protein intron 1. In the mtDNA phylogeny, E. uniporum was more closely related to E. caeruleum, and E. fragi was nested in an inclusive and monophyletic Etheostoma but not closely related to the E. spectabile clade or E. caeruleum. A more intensive sampling revealed that all E. uniporum mtDNA haplotypes were phylogenetically nested in E. caeruleum and never closely related to other species in the E. spectabile clade (Ray et al. 2008).

Because phylogenetic incongruence among genes can result from both introgression and ancestral polymorphism (Funk and Omland 2003), much effort has been aimed at distinguishing between these 2 processes (Sang and Zhong 2000; Holder et al. 2001; Nielsen and Wakeley 2001; Buckley et al. 2006). Although our analyses did not explicitly test for ancestral polymorphism through coalescent-based simulations (Knowles and Maddison 2002; Knowles 2004), the fact that mtDNA haplotypes sampled from E. uniporum and E. fragi were more closely related to distantly related Etheostoma lineages argues against lineage sorting events being responsible for this phylogenetic pattern. Supporting this conclusion is the striking and well-supported monophyly of the E. spectabile species clade resulting from our phylogenetic analyses of nuclear genes, both the single-gene and the combined-gene analyses (Figs. 456) and the species trees inferred from the 6 nuclear genes using 2 different methods (Fig. 7).

There are multiple instances of introgressive hybridization that have occurred at different temporal periods in the evolutionary history of the E. spectabile species clade. The E. uniporum and E. fragi mitochondrial genomes have been completely replaced with those from other Etheostoma species and lineages. In addition, the nuclear gene trees suggest rare and limited introgression of E. caeruleum alleles into the nuclear genome of E. uniporum. The introgression of E. caeruleum mtDNA across E. uniporum species boundaries has occurred at a minimum during 2 distinct temporal periods in the history of the species. The 3 oldest mean UCLN estimates for the ages of the introgressed E. caeruleum haplotypes in E. uniporum are similar and range between 11.9 and 7.8 relative time units, and all 3 age estimates exhibit overlapping 95% HPD estimates (Fig. 8). These oldest introgressed mtDNA lineages include the geographically widespread E. uniporum clade as well as the haplotypes (E. uniporum M, O, and P) that are closely related to E. caeruleum haplotypes sampled from the Current River (Figs. 2 and 7). One sampled E. uniporum haplotype (T) was well nested in a clade of E. caeruleum sampled from the Strawberry, Spring, Gasconade, and Eleven Point rivers (Fig. 2). The mean UCLN age estimate for this haplotype was 1.0 (95% HPD: 1.7–0.4) relative time units, and its 95% HPD did not overlap with the lineage age estimate of any other heterospecific mtDNA haplotype observed in E. uniporum. This indicates a separate and more recent secondary introgression of E. caeruleum mtDNA in E. uniporum (Fig. 8). Repeated hybridization between E. caeruleum and E. uniporum has resulted in the recurrent introgression of E. caeruleum mitochondrial genomes into E. uniporum and has also resulted in limited introgression of alleles in the nuclear genome. The S7 ribosomal intron 1 gene tree shows that the alleles sampled from E. uniporum C are more closely related to those sampled from E. caeruleum and E. luteovinctum (Fig. 5a), and alleles sampled from E. uniporum M for the RAG1 exon 3 locus are more closely related to those sampled from E. caeruleum (Fig. 5b). Alleles sampled from these specimens for KELCH, MLL, and Mb were all more closely related to other sampled E. uniporum specimens and nested in a monophyletic E. spectabile species clade (Fig. 4). Evidence of 2 separate instances of nuclear introgression in E. uniporum could indicate high introgression pressure on E. uniporum from E. caeruleum. And although conclusions about the timing of nuclear introgression are difficult, given this cursory examination of the nuclear genome, our analyses highlight the importance of using multiple independent nuclear markers. The origin of the introgressed mitochondrial genome found in E. fragi presents a problem that is much less obvious than the E. caeruleum introgressed mitochondrial genomes in E. uniporum. Our mitochondrial gene tree (Fig. 3) and the one presented in Lang and Mayden (2007) indicate that the mitochondrial genome in E. fragi is not closely related to mtDNA haplotypes observed in any extant Etheostoma species or lineage. The mean UCLN estimated age for the node that relates the introgressed mtDNA haplotypes in E. fragi to other darters in the mtDNA gene trees (Figs. 2 and 3) is quite old at 68.4 (95% HPD: 59.9–76.7) relative time units (Fig. 8). In contrast to the divergent phylogenetic placement inferred from the mtDNA haplotypes, gene trees estimated from each of the 6 nuclear genes show that E. fragi is more closely related to other species in the E. spectabile species clade (Figs. 4 and 5). Phylogenies resulting from analysis of the concatenated nuclear genes and the species trees estimated from minimizing deep coalescence and the Bayesian method implemented in BEST also result in monophyly of the E. spectabile clade and support the hypothesis that the geographically adjacent E. fragi and E. uniporum are sister species (Figs. 1, 6, and 7). A preliminary phylogenetic analysis of cytb sampled from 222 of 225 darter species results in a phylogenetic nesting of the introgressed E. fragi mitochondrial genome in Etheostoma but is not closely related to any extant darter species. Given the obscure phylogenetic relationships and the ancient relative age, we hypothesize that the mitochondrial genome found in E. fragi represents an mtDNA fossil, meaning that the mtDNA haplotypes found in E. fragi have originated from an Etheostoma species that has subsequently gone extinct. A similar mechanism of recent extinction was proposed to explain orphan allozyme alleles present in the unisexual hybrid teleost fish species Menidia clarkhubbsi (Echelle et al. 1989), and mitochondrial lineages of extinct species have been reported in tetraploid Hyla tree frog species (Holloway et al. 2006).

The patterns of mtDNA introgression observed in E. uniporum and E. fragi raises questions regarding the occurrence and frequency of mtDNA introgression in other E. spectabile clade species. The mtDNA gene trees reveal that individuals of E. spectabile, E. burri, and E. pulchellum have heterospecific mtDNA haplotypes (Figs. 2 and 3) but do not have introgressed nuclear alleles (Fig. 6). The introgressed mtDNA haplotypes observed in E. spectabile and E. burri are phylogenetically nested in the E. caeruleum clade and are very similar to haplotypes of E. caeruleum sampled from the same river drainages. The heterospecific mtDNA haplotypes observed in E. pulchellum are phylogenetically closely related to the haplotypes sampled from E. whipplei, a syntopic congener (Figs. 2 and 3). We are confident that the sampled E. whipplei mtDNA haplotypes are not introgressed from another Etheostoma species because E. whipplei and E. radiosum are resolved as sister species in both the mitochondrial and the nuclear gene phylogenies (Figs. 2, 3, 6, and 7). This phylogenetic relationship is consistent with previous molecular phylogenetic analyses (Lang and Mayden 2007) and traditional taxonomic arrangements based on external morphology and male nuptial coloration (Hubbs and Black 1941; Moore and Rigney 1952). The UCLN relative age estimates for the introgressed mtDNA haplotypes in E. spectabile, E. burri, and E. pulchellum are very young (Fig. 8), indicating a fairly contemporaneous set of introgression events when compared with the patterns of mtDNA introgression observed in E. uniporum and E. fragi.

The patterns of mitochondrial introgression observed in the E. spectabile species clade, both contemporaneous and ancient, are consistently asymmetric. Heterospecific mtDNA haplotypes are only found in E. spectabile clade species, whereas E. spectabile clade mtDNA haplotypes are never introgressed in other species such as E. caeruleum. Asymmetric mitochondrial introgression is not uncommon and is observed in a diverse array of animal lineages (Good et al. 2003; McGuire et al. 2007; Takami et al. 2007). The asymmetric pattern of mitochondrial introgression observed in the E. spectabile clade could have resulted from asymmetric behavioral reproductive isolation, exemplified if heterospecific spawning between E. spectabile males and E. caeruleum/E. whipplei females occurs more frequently than the reciprocal heterospecific spawning event. This is the pattern observed in a laboratory behavioral study that demonstrated a significantly higher frequency of interspecific spawning between E. radiosum females and E. pulchellum males than between E. pulchellum females and E. radiosum males (Mendelson 2003a) that was facilitated through sneak fertilization of E. pulchellum males. Putative F1 hybrids between these 2 species have been reported (Branson and Campbell 1969); however, it is not known if E. pulchellum populations sympatric with E. radiosum contain introgressed E. radiosum mtDNA haplotypes.

Another mechanism that could result in the asymmetric introgression observed in the E. spectabile species clade is postzygotic isolation asymmetry, where reciprocal hybrid crosses yield different degrees of offspring viability or fitness (Tiffin et al. 2001; Bolnick and Near 2005; Turelli and Moyle 2007; Bolnick et al. 2008). One explanation for isolation asymmetry is the presence of genetic incompatibilities involving uniparentally inherited genetic factors such as mitochondrial genomes (Turelli and Moyle 2007). A study of the freshwater teleost clade Centrarchidae demonstrated that species with accelerated rates of mitochondrial nucleotide substitution tend to be the worse maternal parent in experimental interspecific crosses, supporting the hypothesis that increased mitonuclear incompatibilities can result in isolation asymmetry (Bolnick et al. 2008). Interesting patterns have also emerged through experimental laboratory crosses performed on darter species that exhibit a wide range of divergence times (Hubbs and Strawn 1957; Hubbs 1958, 1959, 1967, 1971; Mendelson 2003b). Experimental hybrids from 2 sets of interspecific crosses, E. caeruleum females × E. zonale males, and E. luteovinctum females × E. binotatum males, exhibit high viability that is similar to control conspecific crosses, but in reciprocal heterospecific crosses, hybrid offspring of E. zonale or E. binotatum females underwent abnormal development patterns and had significantly reduced viability (Mendelson et al. 2006, 2007). There are no experimental data for reciprocal crosses between E. spectabile clade species and E. caeruleum; however, experiments should investigate if patterns of isolation asymmetry are concordant with expectations, given the asymmetry of mitochondrial introgression observed in our gene trees and previous studies by Lang and Mayden (2007) and Ray et al. (2008).

Selection on specific combinations of cytonuclear genotypes may result in the biased fixation of 1 parental mtDNA haplotype that would exhibit an asymmetric phylogenetic pattern (Seehausen 2004; Doiron et al. 2002). For example, wild brook charr populations, Salvelinus fontinalis, that inhabit colder, high-altitude lakes as compared with populations in lower-elevation warmer habitats, are fixed with mitochondrial haplotypes of arctic charr, Salvelinus alpinus, suggesting a selective advantage to introgressed brook charr populations (Doiron et al. 2002). Whether E. caeruleum mtDNA haplotypes confer a selective advantage onto E. spectabile clade species and a selective sweep is responsible for the complete replacement of the mitochondrial genome observed in E. uniporum requires further investigation.

The phylogenetic perspective on mtDNA introgression in the E. spectabile species clade provided by our analysis of mitochondrial and nuclear gene trees reveals a very interesting biogeographic pattern. Introgression of heterospecific mtDNA into E. spectabile clade species occurs only in those species distributed in the Ozarks and is not observed in species distributed in the Eastern Highlands. Two of the E. spectabile clade species (E. spectabile and E. pulchellum) are widely distributed in the Great Plains, Edwards Plateau, northern Ozarks, upper Mississippi River Basin, and northern tributaries of the Ohio River Drainage that were glaciated during the Pleistocene (Fig. 1). All remaining species have more restricted geographic distributions in the Ozark Mountains and Eastern Highlands, which are areas characterized by a high degree of diversity and endemism of freshwater teleost fish species and were relatively immune to the effects of Pleistocene glaciation (Wiley and Mayden 1985; Near et al. 2001; Near and Keck 2005; Soltis et al. 2006). Investigations of reported E. spectabile × E. caeruleum hybrids from northern tributaries of the Ohio River Drainage using allozyme analyses did not find evidence of hybridization between these 2 species (Martin and Richmond 1973; McLeod et al. 1980). The propensity for introgression of mtDNA in Ozark species of the E. spectabile clade is illustrated by a pattern of introgression from several sources; E. uniporum is fixed for mtDNA haplotypes that originated from E. caeruleum, whereas E. fragi is fixed for mtDNA haplotypes that originated from an Etheostoma lineage that is presumably extinct. Even the more contemporaneous introgression of mtDNA haplotypes from E. whipplei and E. caeruleum into E. pulchellum, E. spectabile, and E. burri is found only in Ozark populations and never in species from the Eastern Highlands. One potential explanation for this pattern lies in the historical biogeography of the 2 lineages. Haplotype and morphological diversity of E. caeruleum is significantly higher in the Ozarks (Knapp 1964; Ray et al. 2006) and may be associated with the long-term stability of this region (Pflieger 1971). The phylogenetically basal position of E. spectabile species distributed in the Ozarks compared with that distributed in the Eastern Highlands suggests an Ozark origin for the clade, followed by colonization of the Eastern Highlands and previously glaciated areas. Thus, the Ozark region has been an area where E. spectabile species and E. caeruleum have co-occurred for a longer period of time, providing greater opportunities for hybridization and the observed introgression.

Introgressive hybridization events, both ancient and contemporaneous, have likely played a role in the diversification of the E. spectabile clade. Our nuclear analyses support the monophyly of the E. spectabile clade, a fact that would have been obscured if our inferences were based solely on the mtDNA phylogeny. This study highlights the necessity of nuclear gene phylogenies when investigating the evolutionary history of closely related species with a history of hybridization. Not only does our comparison of mitochondrial and nuclear gene trees uncover the presence of mitochondrial introgression in nearly one-third of the species in the E. spectabile clade, but it also refines the pattern, illustrating that introgressive hybridization spans temporal scales, yet remains restricted to species and populations in the Ozarks. Comparisons of demographic parameters, the evolution of pre- and postzygotic isolating mechanisms, differences in habitat utilization, and different histories of paleogeographic disturbance between the Ozarks and the Eastern Highlands are required to explore potential mechanisms resulting in the patterns of asymmetric mitochondrial introgression that characterize diversification of species in the E. spectabile clade.

SUPPLEMENTARY MATERIAL

Supplementary material can be found at http://www.oxfordjournals.org/our_journals/sysbio/.

FUNDING

The National Science Foundation (DEB-0716155) and the YPM supported this research.

This study was completed as a partial fulfillment for C.M.B.’s master of science degree, awarded by the Department of Ecology and Evolutionary Biology at the University of Tennessee. Our work on orangethroat darters was greatly facilitated by advice, discussions, and assistance in the field from P. A. Ceas, B. H. Bauer, R. C. Carlson, G. R. Dinkins, D. A. Etnier, R. C. Harrington, K. L. Heckman, P. R. Hollingsworth Jr., C. D. Hulsey, B. P. Keck, and W. C. Starnes. Gregory Watkins-Cowell provided collection assistance at the Peabody Museum of Natural History, and Jennifer Parris assisted with the University of Tennessee Research Collection of Fishes. B. H. Bauer, C. D. Hulsey, J. A. McGuire, and D. C. Tank provided comments on earlier versions of our manuscript.

References

Avise
JC
Mitochondrial DNA and the evolutionary genetics of higher animals
Philos. Trans. R. Soc. B
 , 
1986
, vol. 
312
 (pg. 
325
-
342
)
Bachtrog
D
Thornton
K
Clark
A
Andolfatto
P
Extensive introgression of mitochondrial DNA relative to nuclear genes in the Drosophila yakuba species group
Evolution
 , 
2006
, vol. 
60
 (pg. 
292
-
302
)
Bailey
RM
Etnier
DA
Comments on the subgenera of darters (Percidae) with descriptions of two new species of Etheostoma (Ulocentra) from southeastern United States
Misc. Pub. Mus. Zool. Univ. Mich
 , 
1988
, vol. 
175
 (pg. 
1
-
48
)
Bolnick
DI
Near
TJ
Tempo of hybrid inviability in centrarchid fishes (Teleostei: Centrarchidae)
Evolution
 , 
2005
, vol. 
59
 (pg. 
1754
-
1767
)
Bolnick
DI
Turelli
M
López-Fernández
H
Wainwright
PC
Near
TJ
Accelerated mitochondrial evolution and “Darwin's corollary”: asymmetric viability of reciprocal F1 hybrids in centrarchid fishes
Genetics
 , 
2008
, vol. 
178
 (pg. 
1037
-
1048
)
Brandley
MC
Schmitz
A
Reeder
TW
Partitioned Bayesian analyses, partition choice, and the phylogenetic relationships of scincid lizards
Syst. Biol.
 , 
2005
, vol. 
54
 (pg. 
373
-
390
)
Branson
BA
Campbell
JB
Hybridization in the darters Etheostoma spectabile and Etheostoma radiosum cyanorum
Copeia
 , 
1969
, vol. 
1969
 (pg. 
70
-
75
)
Buckley
TR
Cordeiro
M
Marshall
DC
Simon
C
Differentiating between hypotheses of lineage sorting and introgression in New Zealand alpine cicadas (Maoricicada dugdale)
Syst. Biol.
 , 
2006
, vol. 
55
 (pg. 
411
-
425
)
Carstens
BC
Knowles
LL
Estimating species phylogeny from gene-tree probabilities despite incomplete lineage sorting: an example from Melanoplus grasshoppers
Syst. Biol.
 , 
2007
, vol. 
56
 (pg. 
400
-
411
)
Ceas
PA
Burr
BM
Etheostoma lawrencei, a new species of darter in the E. spectabile species complex (Percidae: subgenus Oligocephalus), from Kentucky and Tennessee. Ichthyol. Explor
Freshwaters
 , 
2002
, vol. 
13
 (pg. 
203
-
216
)
Ceas
PA
Page
LM
Systematic studies of the Etheostoma spectabile complex (Percidae; subgenus Oligocephalus), with descriptions of four new species
Copeia
 , 
1997
, vol. 
1997
 (pg. 
496
-
522
)
Chan
KMA
Levin
SA
Leaky prezygotic isolation and porous genomes: rapid introgression of maternally inherited DNA
Evolution
 , 
2005
, vol. 
59
 (pg. 
720
-
729
)
Chow
S
Hazama
K
Universal PCR primers for S7 ribosomal protein gene introns in fish
Mol. Ecol
 , 
1998
, vol. 
7
 (pg. 
1255
-
1256
)
Dettaï
A
Lecointre
G
Further support for the clades obtained by multiple molecular phylogenies in the acanthomorph bush
C.R. Biol.
 , 
2005
, vol. 
328
 (pg. 
647
-
689
)
Distler
DA
Distribution and variation of Etheostoma spectabile (Agassiz) (Percidae, Teleostei)
Univ. Kan. Sci. Bull.
 , 
1968
, vol. 
48
 (pg. 
143
-
208
)
Doiron
S
Bernatchez
L
Blier
PU
A comparative mitogenomic analysis of the potential adaptive value of arctic charr mtDNA introgression in brook charr populations (Salvelinus fontinalis Mitchill)
Mol. Biol. Evol.
 , 
2002
, vol. 
19
 (pg. 
1902
-
1909
)
Drummond
AJ
Ho
SYW
Phillips
MJ
Rambaut
A
Relaxed phylogenetics and dating with confidence. PLoS
Biol.
 , 
2006
, vol. 
4
 (pg. 
699
-
710
)
Drummond
AJ
Rambaut
A
BEAST: Bayesian evolutionary analysis by sampling trees
BMC Evol. Biol.
 , 
2007
, vol. 
7
 pg. 
214
 
Echelle
AA
Dowling
TE
Moritz
CC
Brown
WM
Mitochondrial-DNA diversity and the origin of the Menidia clarkhubbsi complex of unisexual fishes (Atherinidae)
Evolution
 , 
1989
, vol. 
43
 (pg. 
984
-
993
)
Echelle
AA
Schenk
JR
Hill
LG
Etheostoma spectabile-E. radiosum hybridization in Blue River, Oklahoma
Am. Midl. Nat
 , 
1974
, vol. 
91
 (pg. 
182
-
194
)
Edwards
SV
Liu
L
Pearl
DK
High-resolution species trees without concatenation
Proc. Natl. Acad. Sci. USA
 , 
2007
, vol. 
104
 (pg. 
5936
-
5941
)
Freyhof
J
Lieckfeldt
D
Pitra
C
Ludwig
A
Molecules and morphology: evidence for introgression of mitochondrial DNA in Dalmatian cyprinids
Mol. Phylogenet. Evol.
 , 
2005
, vol. 
37
 (pg. 
347
-
354
)
Funk
DJ
Omland
KE
Species-level paraphyly and polyphyly: frequency, causes, and consequences, with insights from animal mitochondrial DNA
Annu. Rev. Ecol. Syst
 , 
2003
, vol. 
34
 (pg. 
397
-
423
)
Good
JM
Demboski
JR
Nagorsen
DW
Sullivan
J
Phylogeography and introgressive hybridization: chipmunks (genus Tamias) in the northern Rocky Mountains
Evolution
 , 
2003
, vol. 
57
 (pg. 
1900
-
1916
)
Grant
PR
Grant
BR
Petren
K
Hybridization in the recent past
Am. Nat
 , 
2005
, vol. 
166
 (pg. 
56
-
67
)
Grove
TJ
Hendrickson
JW
Sidell
BD
Two species of Antarctic icefishes (genus Champsocephalus) share a common genetic lesion leading to the loss of myoglobin expression
Polar Biol.
 , 
2004
, vol. 
27
 (pg. 
579
-
585
)
Holder
MT
Anderson
JA
Holloway
AK
Difficulties in detecting hybridization
Syst. Biol.
 , 
2001
, vol. 
50
 (pg. 
978
-
982
)
Holloway
AK
Cannatella
DC
Gerhardt
HC
Hillis
DM
Polyploids with different origins and ancestors form a single sexual polyploid species
Am. Nat
 , 
2006
, vol. 
167
 (pg. 
E88
-
E101
)
Hubbs
C
Fertility of F1 hybrids between the percid fishes, Etheostoma spectabile and E. lepidum
Copeia
 , 
1958
, vol. 
1958
 (pg. 
57
-
59
)
Hubbs
C
Laboratory hybrid combinations among etheostomatine fishes
Tex. J. Sci.
 , 
1959
, vol. 
11
 (pg. 
49
-
56
)
Hubbs
C
Geographic variations in survival of hybrids between etheostomine fishes
Bull. Tex. Mem. Mus
 , 
1967
, vol. 
13
 (pg. 
1
-
72
)
Hubbs
C
Survival of intergroup percid hybrids
Jap. J. Ichthyol
 , 
1971
, vol. 
18
 (pg. 
65
-
75
)
Hubbs
C
Strawn
K
Survival of F1 hybrids between fishes of the subfamily Etheostomatinae
J. Exp. Zool
 , 
1957
, vol. 
134
 (pg. 
31
-
60
)
Hubbs
CL
Black
JD
The subspecies of the American percid fish, Poecilichthys whipplii
Occas. Papers Mus. Zool. Univ. Mich
 , 
1941
, vol. 
429
 (pg. 
1
-
28
)
Hudson
RR
Coyne
JA
Mathematical consequences of the genealogical species concept
Evolution
 , 
2002
, vol. 
56
 (pg. 
1557
-
1565
)
Jennings
WB
Edwards
SV
Speciational history of Australian grass finches (Poephila) inferred from thirty gene trees
Evolution
 , 
2005
, vol. 
59
 (pg. 
2033
-
2047
)
Kass
RE
Raftery
AE
Bayes factors
J. Am. Stat. Assoc.
 , 
1995
, vol. 
90
 (pg. 
773
-
795
)
Keck
BP
Near
TJ
Assessing phylogenetic resolution among mitochondrial, nuclear, and morphological datasets in Nothonotus darters (Teleostei: Percidae)
Mol. Phylogenet. Evol.
 , 
2008
, vol. 
46
 (pg. 
708
-
720
)
Knapp
LW
Systematic studies of the rainbow darter, Etheostoma caeruleum (Storer), and the subgenus Hadropterus (Pisces, Percidae) [PhD dissertation]
1964
Ithaca (NY)
Cornell University
(pg. 
1
-
207
)
Knowles
LL
The burgeoning field of statistical phylogeography
J. Evol. Biol.
 , 
2004
, vol. 
17
 (pg. 
1
-
10
)
Knowles
LL
Maddison
WP
Statistical phylogeography
Mol. Ecol
 , 
2002
, vol. 
11
 (pg. 
2623
-
2635
)
Kocher
TD
Conroy
JA
McKaye
KR
Stauffer
JR
Lockwood
SF
Evolution of NADH dehydrogenase subunit 2 in east African cichlid fish
Mol. Phylogenet. Evol.
 , 
1995
, vol. 
4
 (pg. 
420
-
432
)
Kubatko
LS
Degnan
JH
Inconsistency of phylogenetic estimates from concatenated data under coalescence
Syst. Biol.
 , 
2007
, vol. 
56
 (pg. 
17
-
24
)
Lang
NJ
Mayden
RL
Systematics of the subgenus Oligocephalus (Teleostei: Percidae: Etheostoma) with complete subgeneric sampling of the genus Etheostoma
Mol. Phylogenet. Evol.
 , 
2007
, vol. 
43
 (pg. 
605
-
615
)
Larget
B
Simon
DL
Markov chain Monte Carlo algorithms for the Bayesian analysis of phylogenetic trees
Mol. Biol. Evol.
 , 
1999
, vol. 
16
 (pg. 
750
-
759
)
Linder
AD
The fishes of Blue River in Oklahoma with descriptions of two new percid hybrid combinations
Am. Midl. Nat
 , 
1955
, vol. 
54
 (pg. 
173
-
191
)
Linder
AD
Behavior and hybridization of two species of Etheostoma (Percidae)
Trans. Kan. Acad. Sci.
 , 
1958
, vol. 
61
 (pg. 
195
-
212
)
Liu
L
Pearl
DK
Species trees from gene trees: reconstructing Bayesian posterior distributions of a species phylogeny using estimated gene tree distributions
Syst. Biol.
 , 
2007
, vol. 
56
 (pg. 
504
-
514
)
Liu
L
Pearl
DK
Brumfield
RT
Edwards
SV
Estimating species trees using multiple-allele DNA sequence data
Evolution
 , 
2008
, vol. 
62
 (pg. 
2080
-
2091
)
Lopez
JA
Chen
WJ
Orti
G
Esociform phylogeny
Copeia
 , 
2004
, vol. 
2004
 (pg. 
449
-
464
)
Maddison
DR
Maddison
WP
MacClade 4: analysis of phylogeny and character evolution, version 4.0
2000
Sunderland (MA)
Sinauer
Maddison
WP
Gene trees in species trees
Syst. Biol.
 , 
1997
, vol. 
46
 (pg. 
523
-
536
)
Maddison
WP
Knowles
LL
Inferring phylogeny despite incomplete lineage sorting
Syst. Biol.
 , 
2006
, vol. 
55
 (pg. 
21
-
30
)
Maddison
WP
Maddison
DR
Mesquite: a modular system for evolutionary analysis. Version 2.01.
2007
 
Martin
FD
Richmond
RC
Analysis of five enzyme-gene loci in four etheostomid species (Percidae: Pisces) in an area of possible introgression
J. Fish. Biol.
 , 
1973
, vol. 
5
 (pg. 
511
-
517
)
McGuire
JA
Linkem
CW
Koo
MS
Hutchison
DW
Lappin
AK
Orange
DI
Lemos-Espinal
J
Riddle
BR
Jaeger
JR
Mitochondrial introgression and incomplete lineage sorting through space and time: phylogenetics of crotaphytid lizards
Evolution
 , 
2007
, vol. 
61
 (pg. 
2879
-
2897
)
McLeod
MJ
Wynes
DL
Guttman
SI
Lack of biochemical evidence for hybridization between two species of darters. Comp
Biochem. Phys. B
 , 
1980
, vol. 
67
 (pg. 
323
-
325
)
Mendelson
TC
Evidence of intermediate and asymmetrical behavioral isolation between orangethroat and orangebelly darters (Teleostei: Percidae)
Am. Midl. Nat
 , 
2003
, vol. 
150
 (pg. 
343
-
347
)
Mendelson
TC
Sexual isolation evolves faster than hybrid inviability in a diverse and sexually dimorphic genus of fish (Percidae: Etheostoma)
Evolution
 , 
2003
, vol. 
57
 (pg. 
317
-
327
)
Mendelson
TC
Imhoff
VE
Iovine
MK
Analysis of early embryogenesis in rainbow and banded darters (Percidae: Etheostoma) reveals asymmetric postmating barrier
Environ. Biol. Fish
 , 
2006
, vol. 
76
 (pg. 
351
-
360
)
Mendelson
TC
Imhoff
VE
Venditti
JJ
The accumulation of reproductive barriers during speciation: postmating barriers in two behaviorally isolated species of darters (Percidae: Etheostoma)
Evolution
 , 
2007
, vol. 
61
 (pg. 
2596
-
2606
)
Moore
GA
Rigney
CC
Taxonomic status of the percid fish Poecilichthys radiosus in Oklahoma and Arkansas, with descriptions of new subspecies
Copeia
 , 
1952
, vol. 
1952
 (pg. 
7
-
15
)
Moore
WS
Inferring phylogenies from mtDNA variation: mitochondrial-gene trees versus nuclear-gene trees
Evolution
 , 
1995
, vol. 
49
 (pg. 
718
-
726
)
Near
TJ
Keck
BP
Dispersal, vicariance, and timing of diversification in Nothonotus darters
Mol. Ecol
 , 
2005
, vol. 
14
 (pg. 
3485
-
3496
)
Near
TJ
Page
LM
Mayden
RL
Intraspecific phylogeography of Percina evides (Percidae: Etheostomatinae): an additional test of the Central Highlands pre-Pleistocene vicariance hypothesis
Mol. Ecol
 , 
2001
, vol. 
10
 (pg. 
2235
-
2240
)
Near
TJ
Porterfield
JC
Page
LM
Evolution of cytochrome b and the molecular systematics of Ammocrypta (Percidae: Etheostomatinae)
Copeia
 , 
2000
, vol. 
2000
 (pg. 
701
-
711
)
Nelson
JS
Crossman
EJ
Espinoza-Perez
H
Findley
LT
Gilbert
CR
Lea
RN
Williams
JD
Common and scientific names of fishes from the United States, Canada
Mexico. Am. Fish. Soc. Spec. Pub
 , 
2004
, vol. 
29
 (pg. 
1
-
386
)
Nielsen
R
Wakeley
J
Distinguishing migration from isolation: a Markov chain Monte Carlo approach
Genetics
 , 
2001
, vol. 
158
 (pg. 
885
-
896
)
Nylander
JAA
Ronquist
F
Huelsenbeck
JP
Nieves-Aldrey
JL
Bayesian phylogenetic analysis of combined data
Syst. Biol.
 , 
2004
, vol. 
53
 (pg. 
47
-
67
)
Page
LM
The genera and subgenera of darters (Percidae, Etheostomatini)
Occas. Pap. Mus. Nat. Hist. Univ. Kan
 , 
1981
, vol. 
90
 (pg. 
1
-
69
)
Page
LM
Handbook of darters
 , 
1983
Neptune City (NJ)
T.F.H. Publications
(pg. 
1
-
271
)
Page
LM
Hardman
M
Near
TJ
Phylogenetic relationships of barcheek darters (Percidae: Etheostoma, subgenus Catonotus) with descriptions of two new species
Copeia
 , 
2003
, vol. 
2003
 (pg. 
512
-
530
)
Pamilo
P
Nei
M
Relationships between gene trees and species trees
Mol. Biol. Evol.
 , 
1988
, vol. 
5
 (pg. 
568
-
583
)
Pflieger
WL
A distributional study of Missouri fishes
Univ. Kan. Publ. Mus. Nat. Hist
 , 
1971
, vol. 
20
 (pg. 
225
-
570
)
Piller
KR
Bart
HL
Hurley
DL
Phylogeography of the greenside darter complex, Etheostoma blennioides (Teleostomi: Percidae): a wide-ranging polytypic taxon
Mol. Phylogenet. Evol.
 , 
2008
, vol. 
46
 (pg. 
974
-
985
)
Posada
D
Crandall
KA
MODELTEST: testing the model of DNA substitution
Bioinformatics
 , 
1998
, vol. 
14
 (pg. 
817
-
818
)
Powell
JR
Interspecific cytoplasmic gene flow in the absence of nuclear gene flow: evidence from Drosophila
Proc. Nat. Acad. Sci. USA
 , 
1983
, vol. 
80
 (pg. 
492
-
495
)
Rambaut
A
Drummond
AJ
Tracer, MCMC trace analysis package. Version 1.4.
2003
 
Ray
JM
Lang
NJ
Wood
RM
Mayden
RL
History repeated: recent and historical mitochondrial introgression between the current darter Etheostoma uniporum and rainbow darer Etheostoma caeruleum (Teleostei: Percidae)
J. Fish. Biol.
 , 
2008
, vol. 
72
 (pg. 
418
-
434
)
Ray
JM
Wood
RM
Simons
AM
Phylogeography and post-glacial colonization patterns of the rainbow darter, Etheostoma caeruleum (Teleostei: Percidae)
J. Biogeog
 , 
2006
, vol. 
33
 (pg. 
1550
-
1558
)
Rieseberg
LH
Sinervo
B
Linder
CR
Ungerer
MC
Arias
DM
Role of gene interactions in hybrid speciation: evidence from ancient and experimental hybrids
Science
 , 
1996
, vol. 
272
 (pg. 
741
-
745
)
Rieseberg
LH
Whitton
J
Linder
CR
Molecular marker incongruence in plant hybrid zones and phylogenetic trees
Acta Bot. Neerl
 , 
1996
, vol. 
45
 (pg. 
243
-
262
)
Rokas
A
Carroll
SB
More genes or more taxa? The relative contribution of gene number and taxon number to phylogenetic accuracy
Mol. Biol. Evol.
 , 
2005
, vol. 
22
 (pg. 
1337
-
1344
)
Ronquist
F
Huelsenbeck
JP
MrBayes 3: Bayesian phylogenetic inference under mixed models
Bioinformatics
 , 
2003
, vol. 
19
 (pg. 
1572
-
1574
)
Sang
T
Zhong
Y
Testing hybridization hypotheses based on incongruent gene trees
Syst. Biol.
 , 
2000
, vol. 
49
 (pg. 
422
-
434
)
Schelly
R
Salzburger
W
Koblmuller
S
Duftner
N
Sturmbauer
C
Phylogenetic relationships of the lamprologine cichlid genus Lepidiolamprologus (Teleostei: Perciformes) based on mitochondrial and nuclear sequences, suggesting introgressive hybridization
Mol. Phylogenet. Evol.
 , 
2006
, vol. 
38
 (pg. 
426
-
438
)
Seehausen
O
Hybridization and adaptive radiation
Trends Ecol. Evol.
 , 
2004
, vol. 
19
 (pg. 
198
-
207
)
Shaw
KL
Conflict between nuclear and mitochondrial DNA phylogenies of a recent species radiation: what mtDNA reveals and conceals about modes of speciation in Hawaiian cricket
Proc. Natl. Acad. Sci. USA
 , 
2002
, vol. 
99
 (pg. 
16122
-
16127
)
Smith
SA
Dunn
CW
Phyutility: a phyloinformatics tool for trees, alignments and molecular data
Bioinformatics
 , 
2008
, vol. 
24
 (pg. 
715
-
716
)
Soltis
DE
Morris
AB
McLachlan
JS
Manos
PS
Soltis
PS
Comparative phylogeography of unglaciated eastern
North America. Mol. Ecol
 , 
2006
, vol. 
15
 (pg. 
4261
-
4293
)
Spolsky
C
Uzzell
T
Natural interspecies transfer of mitochondrial DNA in amphibians
Proc. Natl. Acad. Sci. USA
 , 
1984
, vol. 
81
 (pg. 
5802
-
5805
)
Stamatakis
A
RAxML-VI-HPC: maximum likelihood-based phylogenetic analyses with thousands of taxa and mixed models
Bioinformatics
 , 
2006
, vol. 
22
 (pg. 
2688
-
2690
)
Streelman
JT
Karl
SA
Reconstructing labroid evolution with single-copy nuclear DNA
Proc. R. Soc. B
 , 
1997
, vol. 
264
 (pg. 
1011
-
1020
)
Suchard
MA
Weiss
RE
Sinsheimer
JS
Models for estimating Bayes factors with applications to phylogeny and tests of monophyly
Biometrics
 , 
2005
, vol. 
61
 (pg. 
665
-
673
)
Sullivan
JP
Lavoue
S
Arnegard
ME
Hopkins
CD
AFLPs resolve phylogeny and reveal mitochondrial introgression within a species flock of African electric fish (Mormyroidea: Teleostei)
Evolution
 , 
2004
, vol. 
58
 (pg. 
825
-
841
)
Swofford
DL
PAUP*: phylogenetic analysis using parsimony (*and other methods)
Version 4
 , 
2002
Sunderland (MA)
Sinauer Associates
Takami
Y
Nagata
N
Sasabe
M
Sota
T
Asymmetry in reproductive isolation and its effect on directional mitochondrial introgression in the parapatric ground beetles Carabus yamato and C. albrechti
Popul. Ecol
 , 
2007
, vol. 
49
 (pg. 
337
-
346
)
Thompson
JD
Gibson
TJ
Plewniak
F
Jeanmougin
F
Higgins
DG
The ClustalX windows interface: flexible strategies for multiple sequence alignment aided by quality analysis tools
Nucleic Acids. Res.
 , 
1997
, vol. 
22
 (pg. 
4673
-
4680
)
Tiffin
P
Olson
MS
Moyle
LC
Asymmetrical crossing barriers in angiosperms
Proc. R. Soc. B
 , 
2001
, vol. 
268
 (pg. 
861
-
867
)
Trautman
MB
The specific distinctiveness of Poecilichthys coeruleus Storer and Poecilichthys spectabilis Agassiz
Copeia
 , 
1930
, vol. 
1930
 (pg. 
12
-
13
)
Turelli
M
Moyle
LC
Asymmetric postmating isolation: Darwin's corollary to Haldane's rule
Genetics
 , 
2007
, vol. 
176
 (pg. 
1059
-
1088
)
Wiley
EO
Mayden
RL
Species and speciation in phylogenetic systematics, with examples from the North American fish fauna
Ann. Mo. Bot. Gard
 , 
1985
, vol. 
72
 (pg. 
596
-
635
)

Author notes

Associate Editor: Mark S. Hafner

Supplementary data