RESULTS 



Table 1 summarizes our data for mean pelagic duration and geographic 

 distribution for the 115 species examined. Intraspeci f ic variation appears to 

 be low for most species. Of 57 species with N > 2, this variation ranged from 

 to 25 days, with a mode of 3 days, a mean of 5.0 days, and a standard deviation 

 of 4.97 days. Only seven species (5 broadly distributed) varied intraspecif ical ly 

 by more than 10 days, and only three, Coris gamaird and Gomphosus varius 

 (Labridae) and Rhinomuraena amboinensis (Muraemdae), varied by more than 20 

 days. Based on observations of specimens collected in other areas, la;ge 

 i ntraspecif ic variation appears to be more common in these families and also the 

 Gobiidae and Scaridae (Brothers, et al . , 1983; and Brothers, unpublished data) 

 than in the remaining families. As yet, we have no data on geographical or 

 environmental effects on intraspecif ic variability in larval duration, if any. 



Based on species mean values, there is a conspicuous threshold effect in 

 the relationship between pelagic duration and species distribution at approximately 

 45 days total pelagic duration (fig. 1). Below this threshold, duration of the 

 pelagic stage does not correlate with breadth of distribution (Spearman Rank 

 Correlation = 0.12, N = 92, p > 0.2), suggesting that for most Indo-Pacific 

 coral-reef fishes (79.3% of species examined, and all representatives in 14 of 

 the families), the length of time that pelagic eggs and/or larvae disperse is 

 not the principal determinant of the extent of species distributions. In 

 contrast, of the 23 species that have mean pelagic durations longer than 45 

 days, all but three are broadly distributed in the Indo-Pacific, with more than 

 half (14 of 23) inhabiting 25 or more biogeographic areas (as compared with 3 

 out of 92 subthreshold species; this difference is significant at p < 0.001; x 2 

 = 53.7, df = 2). The exceptions to the above are Chaetodon collare (total 

 duration 46 days), Neocirrhites armatus (51 days), and Rhinomuraena amboinesis 

 (74 days). Of these, the first two are close to the threshold value. We are 

 unable, however, to account for the limited distribution of the muraenid, R. 

 amboinesis . The two specimens examined differed markedly in apparent larval 

 duration (63 and 85 days), but both are well above the threshold and are con- 

 sistent with our data for other muraenid species. 



The data also suggest that unusually long pelagic durations are critical 

 for colonization of geographically isolated areas. The average pelagic duration 

 for Central Pacific species that also are found in the Hawaiian Islands, widely 

 considered an isolated out-lier of the Indo-Pacific faunal province (Ekman, 

 1953; Briggs, 1974), is_68.4% longer than those that are not found in Hawaii 

 (Central_Pacific only, X = 49.1 days, S.D. = 9.5, N = 91; species also in 

 Hawaii, X = 49.1 days, S.D. = 10.3; N = 15; difference significant at p < 0.001, 

 Kolmol gorov-Smirnov test for unequal sample sizes; fig. 2). The minimum observed 

 duration of Hawaiian-inclusive species, 35 days, is longer than the mean 

 durations of 59.5% of the species examined, suggesting that most Central Pacific 

 species may be incapable of colonizing Hawaiian reefs under normal circumstances. 

 The pattern for colonizing Eastern Pacific reefs by Indo-West Pacific species 

 is similar (fig. 2). Trans-Pacific species examined have a mean pelagic duration 

 25.6% longer than even Hawaiian-inclusive species, and a minimum duration (48 

 days) that is longer than mean durations of 81.0% of the species we examined. 



55 



