The influence of variation in female fecundity on effective population size

Understanding the relationship between effective population size (Ne) and the number of adults in a population (N) is important for predicting genetic change in small populations. In general, Ne is expected to be close to N/2, i.e. in the range N/4-3N/4, provided that the powerful effect of population bottlenecks on reducing Ne is factored out (using the harmonic mean of N). However, some very low published estimates of Ne/N(<; 0.1) raise the possibility that other factors acting to reduce Ne have been underestimated. Here one such factor, variation in female fecundity, is investigated. Its effect on Ne depends on the standardized variance in fecundity (per breeding season), a measure that is generally independent of mean fecundity. Empirical estimates of this standardized variance from 16 animal studies yielded an average value of 0.44, and a maximum value less than 1.5. To investigate the effect of such values, three kinds of fecundity variation were modelled: random (seasonal): individual; and age-related. Fixed individual differences among females reduce Ne the most. However, to reduce Ne to N/10, the resulting standardized variance must usually be 10 or more. Random differences need to be even larger to achieve the same reduction. One possible mechanism, the random loss of whole families, requires very high family mortality (90% or more). The third model, fecundity that increases linearly with age, is ineffective at causing a marked decrease in Ne. Given the finding that very unusual conditions are required to reduce Ne below Ne/10, low estimates of Ne/N need to be examined critically: the lowest published ratio, for a natural population of oysters, was found to be questionable because of possible immigration into the population by cultivated oysters.