EFFECTS OF SPATIAL DISTRIBUTIONS OF INDIVIDUALS ON MODELS OF ISOLATION-BY-DISTANCE.

Abstract

Effective population size is one of the fundamental parameters in many population genetic models. It provides a common currency to compare populations by reference to an analytically tractable ideally behaving population. Different values of this parameter can have very significant effects on rates and modes of evolution. Sewall Wright's shifting balance theory stresses the importance of drift interacting with selection and dispersal in the process of evolution. For this process to work requires effective deme sizes of no more than a few hundred. Errors of only one order of magnitude can seriously distort our view of the mechanisms of evolution underlying a population structure. Methods exist for dealing with some of the obvious departures from the ideal population. Natural populations seldom conform to other assumptions about population structure made to calculate effective deme size. One such assumption is that individuals are uniformly distributed over area. Much work shows that this is often far from the case. Computer simulations were used to investigate the effects of different spatial distributions of individuals, in combination with various dispersal regimes, on the scale and degree of genetic differentiation. The model consists of diploid individuals arranged according to a spatial distribution with various dispersal regimes imposed upon them. Generations are discrete and the model is allowed to run for 120 to 200 generations. Wright's F-statistics are used as one measure of genetic differentiation. F-statistics do well at reflecting the overall level of differentiation but do not give any idea of spatial structure. Spatial autocorrelation techniques are used to examine the spatial scale and temporal continuity of gene frequency differentiation. Significant effects of the spatial distribution of individuals are found that are not visible through F-statistics. Stable features on the gene frequency surfaces are found to be much larger than the calculated neighborhood sizes. Very different scales of structure can result from different distributions of individuals even though they result in similar estimates of effective deme size. I conclude that it is necessary to have detailed information on a population's structure to be able to predict the effects of genetic drift on the scale of genetic differentiation.