Wildlife forensics: "supervised" assignment testing can complicate the association of suspect cases to source populations

Abstract

Forensic science techniques are an important component of investigations for wildlife-related offences. In particular, DNA analyses can be used to characterize several attributes of biological evidence including sex, individual and species identification. Additionally, genetic assignment testing has enabled forensic biologists to identify the local population from which an individual may have originated. This technique has proved useful in situations where animals have been illegally harvested from areas/populations where hunting is prohibited. For this report, we used individual-based clustering (IBC), in the program Structure 2.2, under both "supervised" and "unsupervised" approaches to assess whether three suspected, illegally harvested moose originated from an endangered population. Atypical circumstances, with Nova Scotia having two moose sub-species in its jurisdiction, enabled strong IBC assignment testing results to determine the source population of the suspected samples. We found differences between the "unsupervised" and "supervised" modeling approaches to define genetic structure among the a priori characterized populations in our data set. Our findings illustrate the fact that individual clustering assignment tests can assist wildlife forensic cases to identify the source population of illegally harvested animals. However, the accuracy of results are highly dependant on the model choice used to define genetic clusters, as well as on the availability of a thorough database of samples throughout the managed area to accurately identify all genetic populations. Further, it is clear from our analyses that political jurisdictions do not accurately reflect isolated populations and we recommend using unsupervised IBC modeling for biological accuracy.