FISHERY BULLETIN: VOL. 87, NO. 3, 1989 



By contrast, the substantial allozyme poly- 

 morphism in the northern anchovy allows tests 

 of both random mating and spatial homogeneity 

 of allelic frequencies. Chi-square goodness-of-f!t 

 tests detect no significant departures from the 

 genotypic proportions expected under random 

 mating, although substantial excesses of 

 heterozygotes are found at the Hbdh-2 and Lgg 

 loci. While liver-tissue degi'adation may have 

 contributed to this result for Hbdh-2 (see Ma- 

 terials and Methods), this explanation cannot 

 hold for Lgg, which was scored reliably from 

 both liver and eye zymograms. Differences in 

 allelic frequencies either among age classes or 

 between sexes can produce excess heterozygos- 

 ity, and significant interaction of sex and allehc 

 frequency is detected by fitting of log-linear 

 models to Hbdh-2 data. For Lgg, on the other 

 hand, sex and allelic frequency are independent 

 given locality. It must be remembered that the 

 chi-square test of Hardy-Weinberg-Castle 

 equihbrium has little power to detect failure of 

 its basic assumptions, notably no selection at 

 the locus and an infinite, unsubdivided popula- 

 tion (Wallace 1958; Lewontin and Cockerham 

 1959). 



The northern anchovy, again in contrast to the 

 Pacific sardine, appears to have a complex popu- 

 lation structure as evidenced by significant 

 heterogeneity of allelic frequencies at 5 of 11 

 polymorphic loci (Table 7), correlations of some 

 alleles with latitude (Fig. 4), and dependence of 

 some allelic frequencies on sex and age. This 

 heterogeneity is unexpected. All samples were 

 collected within the area occupied by the central 

 stock, which has been considered a single, ran- 

 domly mating population, primarily on the basis 

 of transferrin-allele frequencies and meristic 

 data (Vrooman et al. 1981; see MacCall et al. 

 1983). Differences among populations within this 

 area have nevertheless been described for 

 growth and age composition (Parrish et al. 1985), 

 size-adjusted otolith weight (Spratt 1972), size at 

 age (Collins 1969; Mais 1974; Mallicoate and Par- 

 rish 1981), seasonality of spawning (Mais 1974; 

 Parrish^), and migi-ation patterns (Haugen et al. 

 1969; Mais 1974), together with between-year 

 variation in many of these life history traits. 

 Similar genetic heterogeneity of anchovy stocks 

 has been described for EngranUs encrasicholis 

 (Altukhov et al. 1969a, b; Dobrovolov 1978), al- 



'Parrish, R. H. 1983. Evidence for a fall spawning an- 

 chovy stock. Paper presented at 1983 CalCOFI Confer- 



though homogeneity of allehc frequencies was 

 reported by Grant (1985b) for E. capensis. 



For loci polymorphic over the nine population 

 samples, Wright's (1978) measure of average 

 genetic variance among populations, F^t- and 

 Nei's (1978) average genetic distance D — two 

 measures that are maximized by allele replace- 

 ment among populations — are both relatively 

 small: 0.032 and 0.008, respectively. Significant 

 heterogeneity of allelic frequencies without sub- 

 stantial allele replacement may reflect popula- 

 tion subdivision and differentiation resulting 

 from ecological, rather than historical processes. 

 We will explore the causes of this paradoxical 

 genetic heterogeneity in subsequent reports 

 drawing on much larger sets of allozyme, sex, 

 age, and morphological data for northern an- 

 chovy collected between 1982 and 1985. 



Why does the Pacific sardine have low genetic 

 variation? One possibility is that this species 

 originally had levels of variation typical of 

 clupeoids, but that much of this was lost in the 

 collapse of the California sardine fishery in the 

 1950's and early 1960's. That this fishery collapse 

 did not constitute a population genetic bottle- 

 neck, however, appears likely for several 

 reasons. First, the genetically effective popula- 

 tion size during the bottleneck would have had to 

 have been very small, on the order of 10 or fewer 

 individuals, in order to account for the current 

 low level of heterozygosity (Chakraborty and 

 Nei 1977). Second, sardine populations in south- 

 ern Baja California and in the Gulf of California 

 were unaffected by the collapse of the California 

 fishery (Murphy 1969; Sokolov 1974), yet these 

 populations today show low variation also. Fin- 

 ally, by analogy, the Japanese sardine, Sar- 

 diuops melanostida, which also experienced a 

 severe fishery collapse in the 1940's but has since 

 recovered (Kondo 1980), does not have reduced 

 levels of genetic variation (Fujio and Kato 1979; 

 Table 8). 



Alternatively, a restriction in population size 

 in the more distant past might explain low var- 

 iation in the Pacific sardine. The historical 

 record of scale deposits in varved, anaerobic 

 marine sediments of the Santa Barbara Basin, 

 southern California, does show that, relative to 

 northern anchovy and Pacific hake. Pacific sar- 

 dines were always less abundant and much 

 more frequently absent (Soutar 1967; Soutar 

 and Isaacs 1969, 1974). Over the past 1,850 

 years, the Pacific sardine was abundant during 

 12 periods, each lasting from 20 to 150 years. 

 Intervals between these periods of abundance 



666 



