Statistical confidence for likelihood‐based paternity inference in natural populations

Measuring levels of genetic variation is an important aspect of conservation genetics. The informativeness of such measurements is related to the variability of the genetic markers used; a particular concern in species, such as bears, which are characterized by low levels of genetic variation resulting from low population densities and small effective population sizes. We describe the development of microsatellite analysis in bears and its use in assessing interpopulation differences in genetic variation in black bears from three Canadian National Parks. These markers are highly variable and allowed identification of dramatic differences in both distribution and amount of variation between populations. Low levels of variation were observed in a population from the Island of Newfoundland. The significance of interpopulation differences in variability was tested using a likelihood ratio test of estimates of theta = 4Ne mu.

Many evolutionary studies, particularly kinship studies, have been limited by the availability of segregating genetic marker loci. Microsatellites promise to alleviate these problems. Microsatellite loci are segments of DNA with very short sequence motifs repeated in tandem; their often numerous alleles differ in the number of these repeat units. They are very common in eukaryotic DNA and can be amplified by the polymerase chain reaction, which allows the use of minute or degraded DNA samples. The alleles can be scored consistently and compared unambiguously, even across different gels. Copyright © 1993. Published by Elsevier Ltd.

While genotyping wild red deer (Cervus elaphus) at microsatellite loci for paternity assignment, we found three loci (MAF65, BOVIRBP and CelJP23) with segregating nonamplifying alleles. Nonamplifying alleles were detected through mismatches between known mother-offspring pairs and by significant deviations from Hardy-Weinberg equilibria. In a wide range of molecular ecology application, and especially in parentage assignment, the possible existence of undetectable alleles must be taken into account; this may be particularly important for microsatellite data.