Incorporating uncertainty into non-invasive DNA-based mark-recapture population estimates
Abstract
Determining the number of individuals in a population is an important component of wildlife management and conservation. Non-invasive mark-recapture, where DNA is collected from sources such as hair or scat, is an increasingly popular method of obtaining a population estimate. However, non-invasive sources provide small quantities of DNA that is often degraded or contaminated and, thus, are prone to genotyping errors, which bias the resulting population estimate. In this thesis, I describe an algorithm developed to reduce the bias and incorporate the additional uncertainty into the variance of the population estimate. Typically with non-invasive DNA-based mark-recapture, each specimen is genotyped, these genotypes are converted into a single full-information capture history, and then a chosen estimator is used to return a single estimate of the population size. In contrast, the proposed algorithm, Genotyping Uncertainty Added Variance Adjustment (GUAVA), first computes the probability that a pair of specimens came from a common individual, given their observed genotypes, for all pairs of specimens. GUAVA then uses this set of probabilities to generate a distribution of capture histories. For the chosen population estimator, this results in a distribution of estimates. The mean of the distribution represents the population estimate and the variance of the distribution is incorporated into the variance of the estimate. The GUAVA algorithm is evaluated via simulation on a range of population sizes, sample sizes, marker sets, and genotyping error rates. Application of GUAVA is also demonstrated on a red wolf (Canis rufus) dataset. GUAVA has the potential to save time and money by reducing the cost per specimen, requiring fewer loci, and only a single amplification per locus, in contrast to multiple amplifications currently recommended for these studies.
Degree
Ph.D.
Advisors
Craig, Purdue University.
Subject Area
Biostatistics|Statistics|Forestry
Off-Campus Purdue Users:
To access this dissertation, please log in to our
proxy server.