Taming the wild uncertainty of a multi-species reintroduction project

Reintroduction projects are becoming more complex, often involving the translocation of multiple species. In their latest research, Peterson and colleagues use ensembles of ecosystem models to compare 23 alternative reintroduction strategies on Dirk Hartog Island in Western Australia 

The idea of “rewilding” has gained popularity worldwide, and there is an interesting dialogue at play around the meaning of the term – some may allude to “playing God”. However, they are often our best-bet to restore environments after damage from human activities.

Predicting the outcome of single species reintroductions is difficult enough, but when multiple species are being considered, this adds a whole other level of complexity. So, when faced with the challenge of reintroducing 12 mammals and one bird to a 63,000ha island off the coast of Western Australia, we, as applied ecologists, turned to those more mathematically tuned to help us understand how the order, timing and location of releases on the island may influence the success of the program.

Dirk Hartog Island, a former sheep station, became a national park in 2009 and has since been the subject of the largest ecological restoration project of its kind in Australasia. Following the removal of sheep, goats and feral cats (an enormous achievement in itself), the next step was the return of locally extinct fauna, along with the ecosystem services they provide. The overarching goal is to restore the island to a healthy ecological state, much like it was when the first Europeans arrived in 1616.

A project of this size requires extensive planning, and we needed to be sure that any changes made would not be counter-productive to the goals of the project. With our analytical collaborators, we undertook an exercise in ensemble modelling to simulate the outcome (or abundance trajectory) of a range of reintroduction scenarios (or strategies), that also considers the highly unpredictable nature of dynamic ecosystems.

As “experts” our role was to decide what strategies would be most plausible given ecological and logistical constraints, and how might species interact with each other, as this may influence the outcome.

The western quoll or chuditch (Dasyurus geoffroii). Photo: Jeff Pinder DBCA.

For example, boodies (or burrowing bettongs) were the only truly fossorial mammal that occurred on Dirk Hartog Island and are thought to have an important role in restoring ecosystem function. However, they are also thought to compete with other species of higher conservation concern. And what about the apex predator, the western quoll or chuditch? How will the timing of this species’ return to the island affect outcomes for all the potential prey species? Finally, there are predators and prey species already living on the island, which all needed to be factored into the overall melting pot. How would a few tweaks to the ‘status quo’ recipe change the flavour of the final dish?

Our findings were reassuring, but certainly provided food for thought. The model predicted that almost all reintroductions will be successful (12.5 out of 13 species would successfully establish, on average), regardless of choice of strategy. Interestingly, we found that the worst performing strategy was choosing not to reintroduce boodies, so perhaps the benefit provided by this ecosystem engineer outweighs the negative effect of competition.

Figure 3
Example time series outputs for the ‘status quo’ reintroduction strategy and the consensus interaction matrix model ensemble. Solid line indicates the median result. Shaded areas enclose 95% and 80% of the ensemble predictions at each time
Figure 5
Strengths of positive and negative interspecific interactions in models where dibbler reintroduction failed, relative to models where it succeeded (for the consensus matrix). Bar colour indicates whether the interaction was positive (e.g. consumption) or negative (e.g. competition) for dibbler. Interactions with Shark Bay bandicoots, varanids and rodents were unusually strong in the models where dibbler translocation failed

The approach we used also indicated which strategies were good, bad or indifferent for each species. Some species were consistently prone to success or failure, while others did well or badly according to the strategy used. We were also able identify potential causes of failure.

For example, when we considered the dibbler (a small carnivorous marsupial) the strongest association with reintroduction failure was competition with other small mammals. In contrast, predation by chuditch was relatively weak. This then focuses our attention on monitoring those components of the ecosystem that pose a potential risk.  

Dibbler (Parantechinus apicalis). Photo: Jason McDonnell DBCA

Our work shows that should we need to change our status quo to one of the alternatives, this is likely to be okay. In effect, the ensemble ecosystem modelling approach we used is an exercise in risk analysis that can be used to identify options that are most likely to achieve the desired outcome. It also helped to simplify a very complex problem.

A further advantage is that as the reintroductions proceed, any new information can be adaptively fed back into the models to help improve the accuracy of predictions providing further confidence our decisions.  

Read the full paper Reconstructing lost ecosystems: A risk analysis framework for planning multispecies reintroductions under severe uncertainty in Journal of Applied Ecology.

One thought on “Taming the wild uncertainty of a multi-species reintroduction project

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 )

Facebook photo

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

Connecting to %s