Evolution of Cytochrome b and the Molecular Systematics of Ammocrypta (Percidae: Etheostomatinae)

Investigations of phylogenetic relationships using comparative morphological characters have led to the hypothesis that Ammocrypta is the sister taxon of the Etheostoma subgenus Ioa. Subsequent molecular analyses have not recovered this proposed relationship. In this investigation, variation in complete cytochrome b sequences sampled from 27 percid species, representing all darter genera and all species of Ammocrypta, is examined in a phylogenetic context. Third codon purine transitions are saturated when all darter species examined are compared; however, saturation is not detected in any character class among species of Ammocrypta. Maximum-parsimony analysis, which excludes potentially homoplasious character classes, and maximum-likelihood analysis of cytochrome b sequences statistically reject the hypothesis that Ammocrypta is the sister taxon of the Etheostoma subgenus Ioa. Etheostoma is not recovered as monophyletic; however, the hypothesis of monophyly cannot be rejected using maximum-parsimony and maximum-likelihood methods. Morphological characters are combined with cytochrome b sequences in a total evidence analysis of relationships within Ammocrypta. A novel hypothesis of relationships is proposed, which includes a sister-taxon relationship between A. clara and A. pellucida. With regard to relationships among species of Ammocrypta, the morphological data do not significantly conflict with the cytochrome b data. Morphological characters provide additional synapomorphies supporting the hypothesis that A. beani and A. bifascia are sister taxa. The biogeographic implications of the total evidence analysis are discussed, and the continued taxonomic placement of Ammocrypta in Etheostoma is discouraged.

Fig. 1.

Alternative hypotheses of relationships among Ammocrypta species. (A) Wiley and Hagen (1997); (B) Simons (1992); C. Page (1981); and D. Williams (1975)

Fig. 2.

The accumulation of transition differences at the third codon position in cytochrome b, differences involving purines (squares) and pyrimidines (diamonds) are illustrated. (A) Comparisons in the 28 OTU percid dataset. (B) Comparisons in the 8 OTU dataset (C. asprella and Ammocrypta species)

Fig. 3.

Phylogram depicting the single most-parsimonious tree resulting from maximum parsimony analysis ignoring third codon purine (A ↔ G) substitutions. The tree length is 1955 steps, CI (excluding uninformative characters) = 0.283. Bootstrap pseudoreplicate scores (000 iterations) are listed in bold below branches and decay indices are presented in parentheses. Scale of optimized character changes at bottom of figure

Fig. 4.

Phylogram depicting the single tree recovered in maximum-likelihood analysis of all substitutions (Ln = −11,066.23). The HKY85inv+Γ model parameters were transition:transversion ratio (ti:tv) = 5.01, proportion of invariable sites = 0.56, and alpha shape parameter of the gamma distribution (α) = 1.27. Bootstrap pseudoreplicate scores (100 iterations) are listed in bold below branches. Scale of inferred substitutions per site is at bottom of figure

Fig. 5.

Single most-parsimonious tree resulting from maximum-parsimony analysis of the combined cytochrome b and morphology dataset. The tree length is 547 steps, CI (excluding uninformative characters) = 0.624. Open bars denote number of unique transitions (CI = 1.0), and circles denote unique transversions (CI = 1.0). Dark bars tally the number of nucleotide substitutions that are homoplasious along a particular internode. Unique morphological changes (CI = 1.0) are marked with a dash, and homoplasious morphological changes are indicated with a star. Bootstrap replicate scores and decay indices (in parentheses) are listed to left of the appropriate internode

Accepted: 20 Oct 1999