## Info

We can explore the consequences of variance in family size with some examples. Figure 3.17 shows three hypothetical distributions of family size. The first is an ideal Poisson distribution where the mean family size is equal to the variance in family size. This is the standard assumption used in the Wright-Fisher model of genetic drift. The next distribution is an example of highly skewed family sizes where a relatively small proportion of the population contributes most of the progeny. The final distribution shows family size variation that is less than expected for a Poisson distribution. If the Poisson distribution is used as the standard, the other populations with differing distributions of family size show more or less genetic drift, respectively, due to modification of the bottleneck-like effect of unequal family sizes. These distributions also illustrate that the effective population size based on variance in family size has the unique quality that Ne can actually be larger than N if the variance in family size is less than the mean of family size. This stands in contrast to population size fluctuations through time and unequal breeding sex ratios, where Ne can only be less than or equal to N but not greater than N.

These family size distributions are not just theoretical entities. Many annual plants show variation in reproductive success that exceeds the mean, demonstrating that variation in family size contributes to overall rates of genetic drift (Heywood 1986). In one study of salmon, the large variance in reproductive success among anadromous males had a greater impact on the effective population size than breeding sex ratio (Jones & Hutchings 2002). In contrast, Poisson-distributed male reproductive success has been observed in laboratory populations of D. melanogaster, partly supporting the effective population size assumption behind many genetic experiments that have used fruit flies reared in the laboratory (Joshi et al. 1999).

In the last section of the chapter, we compared allele frequencies over time in 107 Drosophila populations to allele frequencies expected from the Markov chain model (Fig. 3.11). The fly populations, founded each generation with eight female and eight male flies, experienced a faster rate of fixation or loss than expected for an effective population size of16. In fact, the fly populations reached fixation and loss at a rate comparable to a population with an effective size of about 10 or 11 (this can be seen using PopGene.S2's Markov chain module as in Interact box 3.2). The concepts in this section that distinguish between census and effective population sizes can be used to

## Post a comment