An empirical evaluation of effective population size estimators

Bradley J Swanson, Purdue University

Abstract

The purpose of this study was to evaluate demographic and genetic methods of estimating the effective population size of a population. Effective population size (Ne) describes how populations lose genetic variation due to random genetic drift. Reduced genetic variation increases inbreeding, which, increases the expression of deleterious traits, prevents a population from responding to environmental changes, or can initiate a mutational meltdown. Given the consequences, it is important to have accurate estimates of N e. Molecular markers allow direct examination of the genetic change, but not all molecular markers work equally well. Based on a review of the literature, microsatellites were shown to produce more intuitive estimates of local effective population size and appear to work on an ecological time scale. Allozymes produce estimates more appropriate for the species level or the effective size across evolutionary time. Several demographic methods, of varying complexity, also exist to estimate effective population size. Demographic based estimates of effective population size were compared to the actual genetic change in microsatellite allele frequencies over time in a metapopulation of kangaroo rats (Dipodomys spectabilis). The most accurate and precise demographic estimator of Ne across a single generation accounted for sex ratio, standardized variance in reproductive success, and overlapping generations. When estimating Ne across multiple generations it was most important to account for fluctuating population size. It is also possible to detect population bottlenecks based on changes in allele frequencies over time. Bottlenecked populations show an excess of heterozygosity compared to that expected at drift-mutation equilibrium as well as a shift in the modal allele frequency class. It was not possible to detect known bottlenecks using the excess heterozygosity method. When the infinite allele model was used some of the known bottlenecks were detected but a high number of false positives were also generated. The mode-shift method performed better, detecting more of the known bottlenecks and returning a lower number of false positives. The superior performance of the mode-shift method is probably related to the small initial population size of the founded populations and the small number of loci used.

Degree

Ph.D.

Advisors

Waser, Purdue University.

Subject Area

Ecology|Molecular biology

COinS