systems. Electrophoretically detectable allele fre- 

 quencies can be used to estimate levels of migration 

 if it can be assumed that these frequencies reflect 

 a balance between the opposing forces of migration 

 (gene flow) and random divergence of allele fre- 

 quencies (genetic drift). The main difficulty with this 

 approach is that other forces, notably natural selec- 

 tion and historical contact, can influence allele fre- 

 quencies, and the relative importance of these forces 

 in natural populations has proved extremely diffi- 

 cult to evaluate directly. 



The present study differs from most previous ones 

 in an important way: rather than concentrate on one 

 or two species, we sampled 10 marine shore fishes 

 from the same suite of island and mainland local- 

 ities in southern California and Baja California, Mex- 

 ico. Substantial differences between species in 

 fecundity, length of larval life, and other life history 

 features allowed us to test the hypothesis that 

 species with low dispersal capability should show 

 greater genetic differences between populations 

 than do species that are better dispersers. As dis- 

 cussed by Waples (in press), the statistically signif- 

 icant negative correlation between dispersal capabil- 

 ity and levels of genetic differentiation in these 

 shore fishes is consistent with expectations based 

 on an equilibrium model involving gene flow and 

 genetic drift. Scenarios invoking natural selection 

 and/or historical (nonequilibrium) perturbations of 

 migration patterns could be hypothesized to explain 

 these results, but there is no a priori reason to ex- 

 pect the observed correlation to result from selec- 

 tion or historical factors. The test discussed by 

 Waples (in press) does not exclude the possibility of 

 selection at individual gene loci, but does suggest 

 that such forces have not been strong enough to 

 disturb the overall patterns of genetic differentia- 

 tion due to gene flow that are of interest here. 



In this paper we extend the analysis of these data 

 to address two questions regarding larval dispersal 

 that can only be understood by considering data for 

 a number of species simultaneously: 1) Are there 

 consistent patterns (across species) of genetic 

 similarity among localities that suggest common 

 avenues of larval transport? 2) If such patterns do 

 exist, can results for those species that are excep- 

 tions to the pattern be understood in terms of dif- 

 ferent behavioral or life history features that might 

 cause their larvae to be affected differently by the 

 current regime? The question of the frequency of 

 successful long distance dispersal in these shore 

 fishes and some of the problems associated with 

 estimating this frequency will be discussed in a later 

 paper. 



FISHERY BULLETIN: VOL. 85, NO. 1 



• MATERIALS AND METHODS 



Experimental Design 



Collections were made at six sites in four major 

 areas: La Jolla, CA; the California Channel Islands 

 (San Nicolas Island and Santa Catalina Island); Isla 

 de Guadalupe, Mexico; and near Punta Eugenia, 

 Mexico (Cabo Thurloe and Islas de San Benito; see 

 Figure 1). Two sites were used in the Channel 

 Islands and near Punta Eugenia because not all 

 species could be collected at a single locality. The 

 area of study describes a quadrilateral roughly 600 

 km long and 100-300 km wide that encompasses 

 almost the entire area south of Point Conception 

 governed by the California Current System. Fur- 

 thermore, few of the species studied occur north of 

 Point Conception in any numbers, and central Baja 

 California, Mexico, is at or near the southern dis- 

 tributional limit for most of these species as well. 

 The study areas thus cover a major portion of the 

 normal range for these species, and this sampling 

 pattern should have been able to detect significant 

 population subdivision if it exists. 



The study sites were also chosen in such a way 

 that the genetic affinities of populations in certain 

 areas could be evaluated. Mainland populations are 

 represented by samples taken at La Jolla. The 

 California Channel Islands harbor large populations 

 of many marine organisms, and it is important to 

 assess the degree to which these populations are in- 

 dependent of those from the mainland. Guadalupe 

 is a small, oceanic island of volcanic origin sur- 

 rounded by deep (>3,000 m) water. It is remote 

 enough (275 km west of the central Baja California 

 coast) that genetic differentiation of shore fishes 

 might be expected. Collections in the vicinity of Pun- 

 ta Eugenia were made to serve as controls for 

 evaluating the extent of differentiation at Guada- 

 lupe and to estimate the relative importance of east- 

 west larval drift in this area. 



Well-developed oceanic currents serve as poten- 

 tial transport mechanisms for pelagic larvae in the 

 study area. The California Current brings relative- 

 ly cold, low salinity water from high latitudes toward 

 the Equator; its principal characteristics have been 

 known for some time (Reid et al. 1958; Hickey 1979). 

 The California Current is most strongly developed 

 north of Point Conception; further south, nearshore 

 flow becomes somewhat variable because of the 

 eastward jut of the coastline and the complicating 

 effects of the Channel Islands (Fig. 1). Between 

 about lat. 30° and 33 °N, the current shifts toward 

 the east, and a portion of the water is deflected 



