Phylogeography of two moray eels indicates high dispersal throughout the indo-pacific

Abstract

Reef fishes disperse primarily as oceanic "pelagic" larvae, and debate continues over the extent of this dispersal, with recent evidence for geographically restricted (closed) populations in some species. In contrast, moray eels have the longest pelagic larval stages among reef fishes, possibly providing opportunities to disperse over great distances. We test this prediction by measuring mitochondrial DNA (mtDNA) and nuclear DNA variation in 2 species of moray eels, Gymnothorax undulatus (N = 165) and G. flavimarginatus (N = 124), sampled at 14-15 locations across the Indo-Pacific. The mtDNA data comprise 632 bp of cytochrome b and 596 bp of cytochrome oxidase I. Nuclear markers include 2 recombination-activating loci (421 bp of RAG-1 and 754 bp of RAG-2). Analyses of molecular variance and Mantel tests indicate little or no genetic differentiation, and no isolation by distance, across 22 000 km of the Indo-Pacific. We estimate that mitochondrial genetic variation coalesces within the past about 2.3 million years (My) for G. flavimarginatus and within the past about 5.9 My for G. undulatus. Permutation tests of geographic distance on the mitochondrial haplotype networks indicate recent range expansions for some younger haplotypes (estimated within approximately 600 000 years) and episodic fragmentation of populations at times of low sea level. Our results support the predictions that the extended larval durations of moray eels enable ocean-wide genetic continuity of populations. This is the first phylogeographic survey of the moray eels, and morays are the first reef fishes known to be genetically homogeneous across the entire Indo-Pacific.

Figure 1

Map of sampling localities and sketch of generalized moray leptocephalus larvae (bottom right). Each sampling locality is indicated by a number in parentheses that corresponds to localities in Figure 2 and Tables 1–3. Localities are categorized as Indian Ocean (black shading), western/central Pacific Ocean (open circles), or eastern Pacific Ocean (gray shading) to denote the 3 major subdivisions of the Indo-Pacific. See Table 1 for sample size at each locality. Dotted lines represent the biogeographic barriers of the Sunda Shelf and the Eastern Pacific Barrier.

Figure 2

Haplotype networks for mtDNA (left) and for RAG-1 and RAG-2 for both Gymnothorax flavimarginatus and G. undulatus. Squares represent inferred ancestral haplotypes; circles, additional haplotypes; and small black-filled circles, missing haplotypes needed to connect observed haplotypes. The size of the circle or square is proportional to the number of individuals sampled that share that haplotype. For mitochondrial haplotypes, shading denotes oceanic region as in Figure 1: Indian Ocean (black fill), western/central Pacific Ocean (open circles), and eastern Pacific Ocean (gray fill). Localities are listed for each nuclear haplotype as numbered in Figure 1 and Tables 1–3, increasing from west to east. For mitochondrial haplotype networks, groups A–E show regions for which hypotheses of panmixia could be rejected by NCPA in favor of either restricted gene flow or contiguous range expansion (Table 4).

Figure 3

Mismatch distributions for mitochondrial haplotypes (1228 bp, black line) superimposed on the expected distribution calculated for the assumption of a demographically expanding population. Observed distributions do not differ significantly from expectations either in magnitude or in raggedness (see Results). Peak values of approximately 17 (Gymnothorax flavimarginatus) and 10 (G. undulatus) differences between haplotypes are consistent with an interpretation that some geographically isolated haplotype lineages formed during transitory geographic fragmentation of populations approximately 0.3–0.6 My ago and have persisted through subsequent merging of populations.