Phylogenetic Relationships of Barcheek Darters (Percidae: Etheostoma, Subgenus Catonotus) with Descriptions of Two New Species

Etheostoma virgatum has been treated as a species occupying three widely separated regions of the Cumberland River drainage in Kentucky and Tennessee. To test the hypothesis that the three widely disjunct populations of E. virgatum are monophyletic, DNA sequence data from mitochondrial and nuclear loci were gathered on E. virgatum and other species of Catonotus including all species of barcheek darters. Morphological data were analyzed from populations throughout the range of E. virgatum. The three widely separated populations of E. virgatum, although morphologically similar, do not form a monophyletic group in phylogenetic analyses of molecular data. Consistent with this result, two of the populations are described as new species. These three species had been identified as E. virgatum because of the shared presence of bold dark stripes along the side of the body, a feature not found in the other four species of barcheeks. It is unclear whether the presence of bold stripes represents retention of a pleisiomorphic trait (lost in other barcheeks) or whether the condition arose independently in these three species.

Fig. 1.

Known localities for Etheostoma virgatum (circles), Etheostoma basilare (triangles), and Etheostoma derivativum (squares). Type locality for each species is circled. All three species have been referred to as E. virgatum

Fig. 2.

(A) Etheostoma virgatum, 63-mm SL breeding male, Clear Creek, Rockcastle County, Kentucky, 18 April 2000. Note white spots on pectoral fin that are hypothesized to be egg mimics (Porter et al., 2002). (B) Etheostoma basilare, 63-mm SL breeding male, Duke Creek (Caney Fork system), Cannon County, Tennessee, 18 April 2000. Inset: Etheostoma basilare, 56-mm SL breeding male, Duke Creek, Cannon County, Tennessee, 3 April 1992. Note egglike spots on cheek that are hypothesized to be egg mimics (Page, 2000). (C) Etheostoma derivativum, 49-mm SL breeding male, West Fork Stones River, Rutherford County, Tennessee, 22 April 2002. Photo A by M. H. Sabaj, inset by K. S. Cummings, B and C by L. M. Page

Fig. 3.

Maximum-parsimony phylograms for each gene partition. Values above nodes are nonparametric bootstrap percentages (1000 pseudoreplicates). Cytb tree length = 985, CI = 0.58, RI = 0.66; ND2 tree length = 1061, CI = 0.59, RI = 0.66; ND4 tree length = 1079, CI = 0.64, RI = 0.59; S7 intron tree length = 118, CI = 0.86, RI = 0.86

Fig. 6.

Maximum-likelihood (ML) tree obtained using Bayesian Markov chain Monte Carlo sampling. ML score = 12,558.16. Rate parameters used in site-specific rates model were, cytb first codon 0.270559, cytb second codon 0.078838, cytb third codon 2.942669, ND2 first codon 0.535517, ND2 second codon 0.154986, ND2 third codon 3.164224, and S7 intron 1 0.226306. Numbers at nodes represent posterior probabilities of the clade among sampled tree topologies. (2) = two populations sampled

Fig. 5.

Maximum-parsimony and likelihood phylograms for the mtDNA and all data combined. Values above nodes are nonparametric bootstrap percentages (1000 pseudoreplicates). MtDNA model = GTR+I+Γ, ln L = 16360.15; S7 intron + mtDNA model = GTR+I+Γ, ln L = 17890.43. See Table 2 for model parameters of each dataset

Fig. 4.

Maximum-likelihood phylograms for each gene partition. Cytb model = GTR+I+Γ, ln L = 5644.30; ND2 model = GTR+I+Γ, ln L = 5780.03; ND4 model = GTR+I+Γ, ln L = 6351.31; S7 intron model = HKY+Γ, ln L = 1450.63. See Table 2 for model parameters of each gene partition

Received: 13 Aug 2002

Accepted: 21 Mar 2003