Predicting the genetic drift in small populations

Abstract

Predicting the genetic drift in small populations is considered over a short interval of time. A detailed description of the dynamic and spatial structure of the population is used, together with several measures of the change of genetic variability with time: mean inbreeding and kinship coefficients within and between groups of animals; percentages of genes contributed by the various foundation groups that made up the initial population; rate of loss of founder genes; joint distribution and random fluctuations of allelic frequencies. This approach is illustrated by means of two theoretical populations that model actual cattle breeds, and is applied to two problems: the conservation of a small population with or without freezing of semen, and the conflict between diffusion of genetic progress and random drift in a selected population conducted with or without superovulation and embryo transfer. Two main conclusions were drawn from these examples: 1. (a) Drift should not be characterized by single values (genetic effective size, or mean inbreeding coefficients), but rather by a set of criteria that have differentiated sensitivities to variations in demographic structure, mating scheme and gene diffusion. 2. (b) Quite large random variations in gene frequencies can be promoted by selection in seemingly large populations.