Development and evaluation of a protocol to identify individuals of Trichosurus vulpecula with non-invasively recovered DNA

Abstract

The Australian brushtail possum (Trichosurus vulpecula) is a pervasive marsupial pest of New Zealand. Impacting on the native flora and fauna and the nation's livestock industry by its vectoring of bovine turberculosis, T. vulpecula is a priority for control and eventual eradication. Current pest control initiatives involve aerial deployment of chemical poisons, baiting and trapping. To establish the success of such control operations, estimates of possum population size pre- and post culling are required. Currently several monitoring methodologies - requiring the detection and trapping of individuals - are available to estimate indices of abundance (e.g. the residual trap-catch index). But these monitoring protocols are constrained by logistical and analytical considerations. The necessity to overcome the limitations of traditional monitoring schemes presents the opportunity to develop and evaluate the implementation of non-invasive genetic monitoring systems for possums. This thesis aimed to optimise an efficient amplification system for a panel of eight microsatellite loci that allow the identification of individual possums, characterise the occurrence of genotyping error across a range of conditions, and evaluate the use of salivary DNA retrieved from interference devices as template for amplification. Optimisation of amplification conditions for all loci in the panel was evaluated with DNA extracted from possum tissue collected at three localities in Canterbury region. Allele polymorphism was analysed by capillary electrophoresis and uorescence based detection of fragments. After optimisation, locus Tv16 was discarded from the panel due to its linkage with locus Tv27 and amplfication of unspecific fragments. Microsatellite diversity patterns of the seven remaining loci revealed moderate to high polymorphism and heterozygosity, no evidence of genetic structuring between localities across Canterbury (Fst = 0:03), and a sufficiently low overall probability of identity adjusted for siblings (PIsib) (3 x 10⁻³) to ensure a robust identification of individual possums based on their multi-locus genotype. Further exclusion of locus Tv54 was recommended based on its high PIsib (0.63-1.00) and incidence of genotyping error. Amplification of template DNA extracted from tissue was not exempt from genotyping error (mean error rate per locus (el) = 4,8% and observed error rate per multi-locus genotype (eobs) = 33.3%), these errors being associated in equal measure with stochastic causes (e.g. allele drop-out and false alleles) and systematic causes (e.g. scoring errors, sample swapping or contamination). No evidence of null alleles was detected. Six loci were successfully assembled into a multiplex PCR assay. The implementation of mutliplex PCR had no significant effects on the incidence of genotyping error or the consistency of allele size estimation compared to standard PCR, and represented a substantial reduction in labour and resources needed to obtain a genotype (92% cost reduction relative to singleplex). While 1:6 dilution of DNA extracted from tissue did not show significant effects on the amplification success and the mean genotyping error rate per locus, the use of template DNA retrieved from saliva decreased the performance of the microsatellite amplification system significantly. Only 18 of 24 samples were able to generate positive or partially positive genotypes, loci with amplicons > 200 bp being the most affected, while the mean error rate per locus increased to 45%. Altogether, these results indicate that locus characteristics (i.e. amplicon size) and quality of template DNA are crucial factors affecting the sensitivity and reliability of the protocol developed. Potential ways to improve the remote collection of DNA from saliva are recommended.