Abstract
Differentiation in morphometric traits among experimental populations of the housefly subjected to an experimental bottleneck was examined for replicate lines founded with one, four or 16 pairs of flies. Differentiation among lines within a bottleneck size was significantly greater than predicted by drift in relation to the additive genetic variation for these traits within the founding population. Two models of nonadditive genetic variance were investigated to interpret these results, one involving dominance of allelic effects within loci and another incorporating multiplicative epistasis. Both models generated more variation among lines as a direct result of sampling during the bottleneck than predicted by a model with additive gene action. The pattern of differentiation among our experimental lines in relation to these models conformed more to the model incorporating epistasis. Nevertheless, it may be difficult to distinguish differentiation among lines occurring during a bottleneck as a result of nonadditive gene action from that caused by diversifying selection among lines after the bottleneck.