STATISTICAL PROPERTIES OF THE VARIATION AT LINKED MICROSATELLITE LOCI- IMPLICATIONS FOR THE HISTORY OF HUMAN Y-CHROMOSOME

It has recently been suggested that observed levels of variation at mi crosatellite loci can be used to infer patterns of selection in genome s and to assess demographic history. In order to evaluate the feasibil ity of these suggestions it is necessary to know something about how l evels of variation at microsatellite loci are expected to fluctuate du e simply to stochasticity in the processes of mutation and inheritance (genetic sampling). Here we use recently derived properties of the st epwise mutation model to place confidence intervals around the varianc e in repeat score that is expected at mutation-drift equilibrium and o utline a statistical test for whether an observed value differs signif icantly from expectation. We also develop confidence intervals for the time course of the buildup of variation following a complete eliminat ion of variation, such as might be caused by a selective sweep or an e xtreme population bottleneck. We apply these methods to the variation observed at human Y-specific microsatellites. Although a number of sug gested the possibility of a very recent sweep, our analyses suggest th at a sweep or extreme authors have su,, bottleneck is unlikely to have occurred anytime during the last approximately 74,000 years. To gener ate this result we use a recently estimated mutation rate for microsat ellite loci of 5.6x10(-4) along with the variation observed at autosom al microsatellite loci to estimate the human effective population size . This estimate is 18,000, implying an effective number of 4,500 Y chr omosomes. One important general conclusion to emerge from this study i s that in order to reject mutation-drift equilibrium at a set of linke d microsatellite loci it is necessary to have an unreasonably large nu mber of loci unless the observed variance is far below that expected a t mutation-drift equilibrium.