The temporal method for estimating the effective population number from the standardized variance of allele frequency change (F) is evaluated for a model incorporating the unusual life history features of Pacific salmon (Oncorhynchus spp.). Computer simulations were used to establish the relationship between F and Nb, the effective number of breeders per year. Results indicate that the temporal method is applicable to the Pacific salmon model and is in most ways analogous to the method as applied to organisms with discrete generations. Two potential sources of bias in the Pacific salmon model not found in the discrete generation model are the irregularity of allele frequency change over short time periods and the occurrence of a time lag in estimating Nb after a rapid increase in population size. Neither of these factors, however, appears to pose a serious problem to the usefulness of the temporal method; a more important consideration is precision of the estimate Nb. In this respect, the unusual life history features of Pacific salmon confer an advantage in comparison with the discrete generation model. Because Pacific salmon populations are fragmented into several temporally isolated spawning groups, the quantity being estimated (Nb) is smaller than for a comparable population with discrete generations; therefore, the effects of genetic drift are proportionally larger in comparison with sampling error, leading to greater precision in the estimate of effective size. For a similar reason, selection has less effect on F in Pacific salmon than it does in organisms with more typical life history features. With Pacific salmon, it is also possible to sample a different spawning population of Nb individuals each year (rather than every generation, which typically is several years), and precision of Nb increases with each new sample analyzed.Analysis of electrophoretic data from 25 hatcheries of chinook salmon (O. tshawytscha) sampled in two or more years revealed a strong correlation between Nb and the number of juveniles produced by the hatchery (a rough indication of true Nb). This result supports the conclusion from the simulations that the temporal method is fairly robust to various sources of random noise and therefore can be used successfully to study a variety of hatchery and wild populations. © 1990 The American Genetic Association.
