Molecular Systematics of the Deep-Sea Fish Genus Gonostoma (Stomiiformes: Gonostomatidae): Two Paraphyletic Clades and Resurrection of Sigmops

A recent morphological analysis demonstrated that Cyclothone, one of four currently recognized gonostomatid genera, was nested within Gonostoma, rendering the latter paraphyletic. In an attempt to resolve the interrelationships of Gonostoma and Cyclothone using molecular data, we determined the nucleotide sequences of a portion of the mitochondrially encoded 12S and 16S ribosomal RNA genes from 11 gonostomatids, including all seven valid Gonostoma species, and Diplophos taenia as an outgroup. Secondary structure-based alignment resulted in a character matrix consisting of 566 base pairs of unambiguously aligned sequences from the two genes, which were subjected to phylogenetic analyses using maximum-parsimony and maximum-likelihood methods. The resultant tree topologies from the two methods were congruent, with many nodes being strongly supported by various tree statistics. As in the morphological analysis, the molecular data indicated the monophyly of Gonostoma + Cyclothone and paraphyly of Gonostoma, although the intrarelationships of Gonostoma differed greatly from those suggested by the morphological data. The molecular data supported the existence of two basal, paraphyletic clades within Gonostoma, the first comprising two light-colored species (G. atlanticum, G. denudatum) and representing a sister clade of Cyclothone, and the second, five dark-colored species (G. ebelingi, G. elongatum, G. longipinnis, G. gracile, G. bathyphilum). Although several alternative classifications exist that reflect the molecular phylogeny, the resurrection of Sigmops Gill, 1883 for the clade of Gonostoma comprising five dark-colored species is proposed to maximize the information content in the classification.

Fig. 1.

Single most-parsimonious tree of 12 gonostomatids and an outgroup. Numbers above branches indicate bootstrap values obtained for 500 replicates. Numbers below branches indicate decay indices up to 10 steps longer than most-parsimonious tree. Taxa abbreviated as in Table 1. Size of fish illustrations roughly proportional to maximum size recorded (Mukhacheva, 1972; Badcock, 1984; Miya, 1994)

Fig. 2.

ML tree of 12 gonostomatids and an outgroup. The tree topology was fully congruent with obtained by the MP analysis (see Fig. 1). Bootstrap values for each internal branch shown as percentages above branches. All branch lengths significantly different from zero (P < 0.01). Scale indicates expected nucleotide substitutions per site. Taxa abbreviated as in Table 1

Fig. 3.

Comparisons between phylogenetic hypotheses derived from the present molecular data (left) and morphological data (right; Harold, 1998). Numbers above and below branches on left tree indicate bootstrap values for MP/ML analyses and decay indices, respectively. A single number indicates that values were the same for all analytical methods. Closed and open rectangles on right tree indicate unreversed characters and homoplasies, respectively, the numbers of them being denoted below. Taxa abbreviated as in Table 1

Fig. 4.

Relationships between the present molecular phylogeny and four possible generic classifications of Gonostoma + Cyclothone. Note that all classifications, except for that conventionally used, reflect the molecular phylogeny. “Alternative B” best maximizes the information content in the classification. For details, see text

Accepted: 24 Sept 1999