HEDGECOCK ET AL.: GENETIC VARIATION IN PACIFIC SARDINES 



anal-fin-base length, and 12) caudal-peduncle 

 depth. 



Tissue Samples 



After morphological measurements were 

 made, eye, heart, hver, and skeletal (epaxial) 

 muscle tissues were dissected from each speci- 

 men for electrophoretic analyses. Tissue samples 

 were kept in plastic well-trays on ice during dis- 

 section, then covered and stored frozen at -70°C 

 for a period of several days prior to electrophor- 

 esis. The day before electrophoresis, tissue sam- 

 ples were thawed, equal volumes of 0.5 M Tris- 

 HCl, pH 7.1 buffer were added to the samples, 

 and the tissues were homogenized on ice, by 

 hand, with a ground-glass pestle. Homogenized 

 samples were then returned to the -70°C 

 freezer overnight. On the day of electrophoresis, 

 samples were allowed to thaw slowly on ice. 



Otoliths 



Sagittal otoliths were removed from speci- 

 mens and cleared overnight in a 2'7f KOH solu- 

 tion. They were then rinsed in deionized water 

 for one or more days, air dried, placed by pairs in 

 gelatin capsules with the specimen number, and 

 stored in envelopes labeled by population. Speci- 

 men identification for the Magdalena Bay sample 

 was lost. 



The age of each specimen was determined by 

 counting otolith annuH following the methods of 

 Collins and Spratt (1969). Each pan- of otoHths 

 was placed for examination under water in a 

 separate well (1 cm in diameter and painted 

 black) drilled into a strip of plexiglass. All annuli 

 were counted under a binocular dissection micro- 

 scope with incident illumination by one of us (F. 

 L. Sly). His recounts agreed with his initial 

 counts (98% consistency); in a comparison test, 

 80% of his counts were in agi'eement with those 

 of California State Department of Fish and 

 Game otolith readers. 



Electrophoretic Protocol, Genetic 

 Interpretation, and Allozyme Nomenclature 



Methods for horizontal starch-gel electrophor- 

 esis, protein assays, and genetic interpretation 

 of zymogi'ams were substantially the same as 

 those described previously (Ayala et al. 1973; 

 Tracey et al. 1975; Utter et al. 1987). The proto- 

 col used to separate and resolve 20 enzymes or 

 proteins inferred to be encoded by a total of 32 



genes is summarized in Table 1. Nomenclatures 

 for proteins, for genes inferred to encode these 

 proteins, and for alleles at these genes are de- 

 tailed by Utter et al. (1987). Proteins are re- 

 ferred to by the capitalized abbreviations given 

 in Table 1 and the coiTesponding genes by these 

 same abbreviations italicized in upper and lower 

 case. Numerical suffixes distinguish among 

 isozymes or multiple pi'oteins in order of increas- 

 ing anodal migi-ation. Alleles are symbolized by 

 italicized numerals obtained by adding or sub- 

 tracting the number of milhmeters separating 

 variants from the most common electromorphs 

 observed for each protein. Alleles encoding com- 

 mon electromorphs are arbitrarily designated 

 100. Specimens from several populations were 

 included in every electrophoretic run so that re- 

 peated comparisons of relative mobihties of their 

 allozymes were made. 



Allozyme Data Analysis 



Maximum-likelihood estimates of allelic fre- 

 quencies and observed proportions of hetero- 

 zygous genotypes at each locus scored in at least 

 two population samples were computed from 

 numbers of individuals in allehc or genotypic 

 categories and the total numbers of genomes 

 (2A^) or individuals (N) sampled, respectively. 

 Observed and expected proportions of hetero- 

 zygous genotypes at each locus were averaged 

 over loci to obtain means (Nei's [1978] unbiased 

 estimates of H„ and //p, respectively). The pro- 

 portion of genes for which any electrophoretic 

 variation was detected in a population sample 

 was defined as P; for the population sample sizes 

 used, this criterion of polymorphism is close to 

 the frequently used definition that the most com- 

 mon allele cannot exceed a frequency of 0.99 for 

 a polymorphic locus. Averaging of P and H over 

 population samples was done using angular 

 transformation of these proportional values fol- 

 lowed by back-transformation of means and con- 

 fidence limits. Owing to the nature of the re- 

 sults, no further genetic statistics were calcu- 

 lated. 



Morphometric Data Analysis 



The BMDP multivariate statistical software 

 package (Dixon 1981) was used to perform dis- 

 criminant function (P7M) and principal compo- 

 nent (P4M) analyses on log-transformed morpho- 

 metric data. Standard settings were used in the 

 discriminant analysis for tolerance (0.01), F-to- 



655 



