Pooling hair samples to increase DNA yield for PCR

Hairs are useful non-invasive sources of DNA, but the DNA yield can be very small, thus promoting genotyping errors. Using multiple hairs can counter this problem, but may introduce multiple contributors to a sample if collected remotely. With microsatellite genotyping, samples representing multiple animals are obvious if three or more alleles are detected at any locus: these samples can then be removed from any analyses. However, some multiple-individual samples may have only one or two alleles at each of the loci examined. We investigated the probability of failing to identify mixed pooled samples by simulating pooled samples (10 000 replicates) from microsatellite data from the northern and southern hairy-nosed wombats (NHNW, Lasiorhinus krefftii; SHNW, L. latifrons), species with low and high genetic diversity respectively. The majority (81.7%) of the 40 000 simulated samples had three or four alleles, so were readily identified as mixed. In the remaining 1-or-2-allele SHNW samples, forensic science software (DNAMIX) correctly identified mixed versus single-individual samples for all cases when the probability of locus failure was low (P-(LF) = 0.1), and 99% of samples when locus failure was high (P-(LF) = 0.5). For NHNW however, the probability of failing to identify a mixed sample was too high for population size estimation (0.05), even when the probability of locus failure was low. In cases such as this, pooled samples may be adequate for less demanding tasks, such as estimation of allele proportions. However, for animal populations with at least average levels of genetic variation, pooling of samples could safely be utilised for most applications.