In this post Anna MacDonald discusses her recent paper with Stephen Sarre ‘Species assignment from trace DNA sequences: an in silico assessment of the test used to survey for foxes in Tasmania’

Which species occur in an area and how do they interact with one another? These are crucial questions for ecologists and wildlife managers to address, yet answering them is rarely straightforward. Many animals display cryptic behaviours, making them difficult to detect by human observation alone, especially where they occur at low densities. In other cases, closely related species can be difficult to distinguish morphologically. Such challenges explain why many researchers are turning to DNA tests, to assign unknown specimens to a species of origin and to detect wildlife from non-invasive samples, including scats, hairs, soil and water.

DNA detection can provide valuable insights into wildlife ecology, but DNA is no “magic bullet”. As with any approach, DNA tests can be prone to biases and errors, and it is essential that these are evaluated to understand their implications for management decisions.

One potential source of error in DNA tests concerns mis-identification of samples because of unrecognised sequence similarities among different species. This may be especially true for trace DNA samples, which are prone to degradation.

As DNA degrades it becomes fragmented, so PCR amplification will be more successful for shorter than longer sequences. Likewise, amplification is often more successful for mitochondrial genes than nuclear genes, because there are many more copies of the mitochondrial genome per cell, and the circular structure of the mitochondrial genome may help to protect it from degradation. The tradeoff is that the shorter the DNA sequence analysed, the fewer opportunities there are to observe diagnostic sequence differences between closely related species. For some taxonomic groups, it may not be possible to reliably distinguish among species. The challenge, then, is to determine when DNA detection of a target species is likely to be reliable and which species are more likely to be mis-identified.

In our paper, we investigate potential sources of error in the DNA test used to detect the red fox, Vulpes vulpes, in Tasmania. In this test, fox-specific PCR primers are used to amplify 134 bp of the cytochrome b gene from scat DNA. This amplicon is then sequenced to confirm that the DNA sequences obtained from scats match those from known fox specimens. Here, we investigate the reliability of species detection based on this short cytochrome b sequence. We ask what is the possibility that another species might be mistakenly identified as fox because of high sequence similarity, and what is the possibility that a genuine fox DNA sequence might be mistakenly assigned to another species?

We evaluated the fox DNA test using a reference DNA database we constructed from publicly available cytochrome b sequences from 74 vertebrate species. These included the majority of Tasmania’s terrestrial mammals, representatives of all major Australian mammal groups, and multiple sequences from the red fox and its closest relative in Australia, the dog. We used the R package SPIDER to conduct an analysis of DNA barcoding efficacy, for example by identifying instances where sequences from more than one species were separated by less than 3% divergence, which makes it difficult to clearly delineate these species based on this DNA sequence alone.

The good news is that we found no fox sequences that were mis-identified as another species and no sequences from other species that were mis-identified as fox. This gives us high confidence that this DNA test can reliably discriminate fox DNA from that of other Australian species.

The approach we have used here requires little specialised genetics training and can be implemented using publicly available DNA sequences and bioinformatics tools. When the results of DNA testing contribute to the development of wildlife management plans, it is important for managers to understand the likely sources of error in species assignments based on DNA sequences. It is even better if the risks of species mis-identification can be determined before laboratory work commences, so that resources are not wasted on DNA tests that will not provide the required resolution.