Behind the paper: eDNA has great potential as a wildlife survey tool, but should be used properly

As many people know, two of my favourite topics are bandicoots and environmental DNA (eDNA). So I’m very excited about the online debut of my latest paper “A framework for developing and validating taxon-specific primers for specimen identification from environmental DNA” at Molecular Ecology Resources, which includes both bandicoots and eDNA.

eDNA analysis is the analysis of DNA from environmental samples, which can include soil, water and animal faeces. eDNA studies can make valuable contributions to wildlife management, including management of invasive species, and conservation. Some animals are elusive, so relying on direct observations alone may not provide a true picture of where these species occur or which habitats they prefer to spend time in. eDNA detection, in conjunction with physical trapping, camera traps, and other survey data, has potential to improve our understanding of wildlife distributions. We may also want to learn about interactions between different species: predator-prey interactions are one example. This is where scat DNA is especially useful, as we can detect DNA from food remains in scats from a wide range of animals, including predators and herbivores.

However… if we are going to use eDNA to guide management actions, which often require considerable investments of time and resources, we need to make sure that our eDNA studies are well-designed and reliable. Many eDNA tests use PCR and DNA sequencing methods to detect a single species or a group of closely-related species. These rely on the development of species-specific or taxon-specific PCR primers. We need to be confident that these primers will reproducibly detect the target species AND that they won’t also detect DNA from other wildlife that will be mistaken for the target species.

In our paper, we outline a nine-step framework that encompasses the different stages of the primer design and validation process, including a series of bioinformatic and laboratory evaluations. We hope that this framework will be useful for others who design DNA tests in future, especially students or researchers moving into genetics from other fields. As a case study, we illustrated our framework with the development of a DNA test to detect bandicoot DNA from predator poo. Our aim was to design a tool that can be used to assess the impacts of introduced predators on these threatened marsupials.

sbb_simon_700_4510
A southern brown bandicoot in Tasmania – photo credit Simon Troman
The

validation process we describe may seem like a lot of work, but it provides information that will be useful when it comes to making sense of results. For example, early evaluation of primers may prevent valuable research dollars from being wasted on a DNA test that is doomed to failure, by indicating that a primer re-design is needed before anyone even picks up a pipette. And perhaps most importantly, by informing wildlife managers about the degree of confidence we have in each DNA test, they can then decide how best to incorporate DNA detection data into their management plans.

We set out to design a test to detect DNA from all six (now seven!) bandicoots within the sub-family Peramelinae, that could discriminate among bandicoot species based on the specific DNA sequence detected. We identified the mitochondrial ND2 gene as a good candidate for this task. In our ND2 phylogeny of the Peramelemorphia, we could clearly discriminate among most of our target species, the exception being one group that includes a known taxonomic uncertainty. We also used the fabulous R package SPIDER to determine which species were at risk of mis-identification. The greatest ambiguity arose in that same group, between sequences from Isoodon auratus and mainland Australian individuals of Isoodon obesulusSo we now know that our bandicoot DNA test can clearly distinguish the two Tasmanian bandicoot species (I. obesulus and Perameles gunnii), but in some parts of the Australian mainland it may only be possible to assign some DNA sequences to the genus level.

Our laboratory evaluations demonstrated that the new bandicoot primers are highly specific. We were able to use them to detect DNA from all of the target bandicoot species, but they did not detect any DNA from 42 other mammal species. They are also highly sensitive and able to detect bandicoot DNA reliably above a threshold equivalent to ~0.32 pg / μl (that’s a really low quantity of DNA). Finally, we demonstrated that we can successfully detect bandicoot DNA from scat samples. This is important because environmental samples might suffer from DNA degradation or contain substances that can inhibit DNA analysis, so they may have had a lower success rate than the good quality DNA we had used previously.

Unfortunately(!) we didn’t have access to scats from captive predators that had been fed bandicoots, but we did have 22 field-collected scats that we had already sequenced using other genes and a more costly method. Two of these scats (one from a cat and one from a devil) contained some bandicoot DNA sequences. Our new DNA test detected bandicoot DNA from those same two scats, but not from the 20 other samples tested. When we sequenced the bandicoot DNA, it was a perfect match for DNA from P. gunnii (the eastern barred bandicoot), providing further evidence that cats and devils find this threatened species rather tasty.

So, at the end of the day, we have developed a nice test that can be used to survey for bandicoot DNA, but we have also provided some guidelines on how to apply and interpret the results of this test with confidence. Of course the design of robust primers is only one component of a reliable eDNA survey. Other considerations include sampling design, use of replicates and controls, choice of sequencing method, and choice of data analysis parameters. It will become increasingly important for researchers to demonstrate their use of reliable eDNA methods and provide enough information for end users to make judgements about the strengths and limitations of each eDNA dataset.

Advertisements

One Comment Add yours

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s