FISHERY BULLETIN; VOL. 85, NO. 1 



quite variable among species, and four species have 

 more alleles restricted to southern localities; 2) 

 relatively few restricted alleles are found in these 

 shore fishes, further increasing the already large 

 sampling variation in the number and frequency 

 of restricted alleles (Waples 1986, in press; M. 

 Slatkin^). 



That the Channel Islands populations are no more 

 genetically isolated than those at La Jolla was some- 

 what unexpected, as La Jolla is part of the major 

 mainland metapopulation that includes much of the 

 distributional range of these shore fishes. It was 

 therefore thought that La Jolla samples would show 

 the greatest overall genetic affinity with other 

 localities. Such a pattern was reported by Haldor- 

 son (1980), who found allele frequencies in the surf- 

 perch Damalichthys vacca to be similar in a series 

 of mainland populations but distinctive at Catalina. 

 Furthermore, Tegner and Butler's (1985) study of 

 drift bottles released at the Channel Islands in- 

 dicated at most 5-10% reach the mainland, sug- 

 gesting that the amount of genetic exchange may 

 likewise be low. 



However, these findings are not inconsistent with 

 the results of the present study when two factors 

 are considered. First, Tegner and Butler's (1985) 

 study was designed to estimate the numerical im- 

 pact on local green abalone, Haliotis fulgens , popu- 

 lations of larvae derived from the Channel Islands. 

 Because relatively few H. fulgens larvae appear like- 

 ly to cross from the Channel Islands to the main- 

 land, it was concluded that the Channel Islands 

 populations cannot be expected to reseed those on 

 the mainland that are locally depleted through over- 

 fishing, pollution, destruction of habitat, etc. Al- 

 though a small percentage (say 5%) of larval 

 exchange may not exert a significant numerical im- 

 pact on a population, migration at that rate is very 

 high from the perspective of maintaining similar fre- 

 quencies of neutral alleles. In fact, the exchange of 

 only a few breeding individuals per generation is suf- 

 ficient to prevent substantial genetic divergence 

 between populations (Spieth 1974). 



Second, the Channel Islands populations might 

 well have proved to be relatively more divergent in 

 the present study if additional mainland populations 

 had been included, as was the case in Haldorson's 

 study. Nevertheless, it is noteworthy that Channel 

 Islands populations do not appear to be genetically 

 isolated to any substantial degree. They may thus 

 play a more significant role in the population struc- 



'M. Slatkin, Department of Zoology, University of California, 

 Berkeley, CA 94720. 



ture of marine species in this area than had been 

 believed. The consistently strong affinity between 

 Channel Islands and La Jolla populations suggests 

 that the Southern California Eddy may be effective 

 as a means of larval transport between mainland and 

 island localities. 



The second major point to emerge from this study 

 is that the population genetic structure of Caulola- 

 tilus princeps is very different from that of any of 

 the other species. In fact, the pattern of genetic af- 

 finity between populations of the ocean whitefish 

 is almost exactly the opposite of the pattern typical 

 of the remaining shore fishes. This result was puz- 

 zling at first, as the life history features of this 

 species are not particularly unusual. However, 

 through the aid of H. Geoffrey Moser (National 

 Marine Fisheries Service, La Jolla, CA), we obtained 

 unpublished larval capture data that shed consider- 

 able light on this problem. Figure 2 is a plot of these 

 data, collected by CalCOFI sampling programs dur- 

 ing 1955-59. In this 5-yr period no C. princeps lar- 

 vae were collected north of central Baja California, 

 Mexico (lat. 30 °N). In this respect, the larval distri- 

 bution of the ocean whitefish is similar to that 

 observed by Kramer and Smith (1973) for the 

 California yellowtail, Seriola dorsalis{= S. lalandi). 

 In contrast, larvae of the other species in this study 

 for which data are available were frequently taken 

 in the Southern California Bight during 1955-59 

 (percentage of positive collection localities north of 

 lat. 30°N: Chromis punctipinnis, 28%; G. nigricans, 

 44%; M. calif omiensis, 54%; S. pulcher, 39%; 

 Waples 1986). In a more extensive survey of larval 

 catches, Moser et al. (1986) confirmed the unusual 

 pattern for the ocean whitefish for years 1954-81 

 (only 4 of 163 larvae taken north of 30°N, and none 

 taken in the Southern California Bight), and sug- 

 gested some possible explanations for the southward 

 shift observed in this species. Thus while the 

 southern populations are near the periphery of the 

 range for most study species, it is the northern 

 populations that are far removed from the apparent 

 sources of ocean whitefish larvae. 



As we have seen, a significantly nonrandom pat- 

 tern of genetic affinity among areas or pairs of areas 

 was found when data for Caulolatihis princeps were 

 omitted. This result is not entirely unexpected, as 

 removing the most aberrant data in an analysis of 

 this nature will generally result in an improved 

 significance level of the test statistic. On the other 

 hand, such an approach seems justified in this case, 

 as the objectives of this study were to search for 

 generalized patterns of genetic differentiation and 

 to attempt to explain data for anomalous species in 



8 



