Evaluating sources of bias in pedigree-based estimates of breeding population size

Abstract

Applications of genetic-based estimates of population size are expanding, especially for species for which traditional demographic estimation methods are intractable due to the rarity of adult encounters. Estimates of breeding population size ( N S ) are particularly amenable to genetic-based approaches as the parameter can be estimated using pedigrees reconstructed from genetic data gathered from discrete juvenile cohorts, therefore eliminating the need to sample adults in the population. However, a critical evaluation of how genotyping and sampling effort influence bias in pedigree reconstruction, and how these biases subsequently influence estimates of N S , is needed to evaluate the efficacy of the approach under a range of scenarios. We simulated a model system to understand the interactive effects of genotyping and sampling effort on error in genetic pedigrees reconstructed from the program COLONY . We then evaluated how errors in pedigree reconstruction influenced bias and precision in estimates of N S using three different rarefaction estimators. Results indicated that pedigree error can be minimal when adequate genetic data are available, such as when juvenile sample sizes are large and/or individuals are genotyped at many informative loci. However, even in cases for which data are limited, using results of the simulation analysis to understand the magnitude and sources of bias in reconstructed pedigrees can still be informative when estimating N S . We applied results of the simulation analysis to evaluate Nˆ S for a population of federally endangered Atlantic sturgeon ( Acipenser oxyrinchus oxyrinchus ) in the Delaware River, USA. Our results indicated that N S is likely to be three orders of magnitude lower compared with historic breeding population sizes, which is a considerable advancement in our understanding of the population status of Atlantic sturgeon in the Delaware River. Our analyses are broadly applicable in the design and interpretation of studies seeking to estimate N S and can help to guide conservation decisions when ecological uncertainty is high. The utility of these results is expected to grow as rapid advances in genetic technologies increase the popularity of genetic population monitoring and estimation.