WAPLES and ROSENBLATT: LARVAL DRIFT IN SOUTHERN CALIFORNIA 



which involves one northern and one southern 

 population. The same five species identified in the 

 previous analysis (Chromis punctipinnis, Clino- 

 cotttis analis, P. clathratus, E. jacksoni, M. califor- 

 niensis) have the highest correlation with rankings 

 of the other species, although only for the former 

 three is r^ > 0.40. Again, results for Caulolatilus 

 princeps (r^ = -0.77) are strongly negatively cor- 

 related with those of the other species. The chi- 

 square value testing the equality of rankings for 

 pairs of localities (7.09; 5 df) is not statistically 

 significant (critical value = 11.07). 



In light of the results obtained above, the analysis 

 was repeated after deletion of Caulolatilus princeps. 

 When this was done, the r^ values for each of the 

 other species increased, to as high as 0.93 and 0.94 

 for Clinocottus analis and Chromis punctipinnis, 

 respectively. The chi-square value (11.84) indicates 

 that for the remaining species the rankings of pairs 

 of localities are significantly heterogeneous. With 

 data for Caulolatilus princeps omitted, it is even 

 more apparent that the Guadalupe-Punta Eugenia 

 comparison is the most divergent, and La Jolla- 

 Channel Islands remains the most similar pair of 

 localities (Table 4). 



DISCUSSION 



Two major points emerge from the analysis of pat- 

 terns of genetic similarity between areas. First, 

 large-scale patterns of larval dispersal for most 

 species appear to be affected in a similar way by the 

 local current regime. The recurrent patterns can be 

 summarized as follows: 1) La Jolla and the Chan- 

 nel Islands are the two areas with the greatest (and 

 Punta Eugenia and Guadalupe the two areas with 

 the lowest) overall genetic affinity with other popu- 

 lations; 2) the two northern populations share 

 similar allele frequencies, while the two southern 

 populations have much stronger genetic affinities 

 with the northern populations than with each other. 



That the southern populations are relatively iso- 

 lated genetically is not surprising, since they are at 

 the periphery of the distributional range for most 

 of the species. However, it was not expected that 

 the Punta Eugenia populations would show nearly 

 the same degree of genetic isolation as do those from 

 Guadalupe, an oceanic island with a substantial 

 endemic component in its marine flora and fauna 

 (Briggs 1974). The nature of genetic differentiation 

 of Guadalupe shore fishes is discussed more fully in 

 Waples (1986). That many marine species with 

 northern affinities are found along the coast of Baja 

 California, Mexico only in localized upwelling areas 



(Dawson 1945; Hubbs 1960) may be responsible for 

 the observed divergence of Punta Eugenia popula- 

 tions. These upwelling populations, isolated from 

 other shore fish populations by areas with water 

 temperatures up to 10°C warmer, may represent 

 largely independent reproductive units. One aspect 

 of the population structure that seems clear from 

 the results of this study is that the southern popu- 

 lations studied exchange genes much more frequent- 

 ly with northern populations than with each other. 

 Such a finding would be difficult to predict on the 

 basis of known current patterns, which are quite 

 variable and complex off the coast of Baja Califor- 

 nia (Fig. 1). 



Because the southerly flowing California Current 

 is the dominant hydrological feature in the study 

 area, it is of interest to examine the possibility that 

 the link between northern and southern populations 

 is due primarily to one-way gene flow from the 

 north. This possibility can be evaluated in terms of 

 the presence or absence of rare alleles. If gene flow 

 were unidirectional (north to south), one would ex- 

 pect most alleles present in the northern populations 

 also to appear in samples from the south. Alleles 

 originating in the southern populations, on the other 

 hand, would have no tendency to spread to the 

 north. For the 10 species combined, 50 alleles are 

 found in one or more northern populations but are 

 absent from all southern populations, while only 36 

 alleles are restricted to southern populations 

 (Waples 1986). 



These data thus do not provide evidence for gene 

 flow only from the north, as such a model would 

 predict more alleles restricted to southern popula- 

 tions. Furthermore, the average frequency of alleles 

 restricted to the southern populations (0.0098) is 

 slightly higher than the frequency of those restricted 

 to the north (0.0092); this is the opposite of the result 

 expected if unidirectional migration were "swamp- 

 ing" alleles restricted to the south. It is possible that 

 the episodic northward advection of water from the 

 south is an important source of genetic exchange 

 among populations. Such movement is known to 

 occur even in years not associated with El Nino 

 events, and organisms with southern affinities that 

 apparently have been transported into southern and 

 central California are reported on a fairly regular 

 basis (Hubbs and Schultz 1929; Hubbs 1948; Rado- 

 vich 1961; Brinton 1981). The data for restricted 

 alleles are consistent with the hypothesis that such 

 processes may be important in the overall genetic 

 structure of these shore fishes. Two factors, how- 

 ever, argue for caution regarding this conclusion: 

 1) the pattern of occurrence of restricted alleles is 



