The sampling theory of neutral alleles in an island population of fluctuating size (Q1817532)

The demographic models that have been applied in classical studies of the genetic structure of subdivided populations often provided a weak description of natural populations because they require the unrealistic assumption that population size remains constant over time and space. An important biological question concerns the effect of local population size fluctuations on the genetic variability within, and the genetic differentiation between, semi-isolated populations. This paper further studies the mathematical properties of the continuous-generation island model of population genetics first proposed by the author and \textit{J. A. Hartigan} [Genetics 139, 429-437 (1995)] in which semi-isolated island populations are allowed to fluctuate randomly in size. The distribution of alleles on islands under a haploid birth, death and immigration (BDI) process is studied by analytical methods and it is shown that the limiting distribution of allele frequencies on each island has a Dirichlet density and is identical to the distribution obtained using a diffusion approximation for Wright's island model at equilibrium (with an appropriate change of parameters). For a sample or a population of \(N\) individuals under the BDI model, the distribution of the sizes of families of unique allele types is shown to be the compound multinomial Dirichlet, a distribution whose properties are well-known. Surprisingly, the sampling distribution of alleles does not depend on time; the populations generated by a BDI process immediately achieve a steady-state sampling distribution and therefore no ``equilibrium'' assumption is needed to apply the model. The expectation and variance, over replicate islands, of \(f_0\), the probability of IBD for random pairs of alleles, are derived from the statistical moments of this distribution.