Understanding that absence of proof isn’t proof of absence
Tracy Rout and Michael McCarthy (University of Melbourne)
After a couple of years with no physical trace of foxes, the Tasmanian Government has decided to wind back its fox eradication program.
Focus will shift from poison baiting and monitoring across core fox habitat to follow-up monitoring of the fox sightings that continue to be reported by the public. Opinions differ on whether this decision is conservative, sensible or unnecessarily risky. Clearly the stakes are high and this story is a potent reminder of the risks involved in declaring success in any program of eradication. How should the risks surrounding the declaration of success be weighed up?
Invasive species, particularly predators such as foxes, are a major threat to Australia’s biodiversity. Complete eradication of these species is desirable but eradication is an extremely challenging thing to accomplish. And demonstrating it’s been achieved can be just as difficult because individual plants and animals cannot always be detected in the wild.
When individuals are no longer detected, management efforts are reduced. Confidence in the absence of a species will increase with every search that comes up empty, but complete certainty is impossible when a species is detected imperfectly. The costs of mistakenly assuming success can be high. The species can bounce back and render the eradication attempt pointless; wasting months, years, or even decades of management effort (and the precious and limited resources that went into it). The species can even escape the area to which it’s previously been confined, threatening new ecosystems and industries.
Many eradication programs fail. Some never come close to success, and were possibly misguided from the outset. In some cases, however, the failure was brought about because of winding up the program prematurely. Take for example the attempt to eradicate Asian musk shrews from the Mauritian island of Ile aux Aigrettes (see Decision Point #25). Beginning in July 1999, the animals were trapped continually for 49 days. Then, after only eight days with no captures, the program was scaled back. This was later recognised as premature. Following the decision, the number of animals captured increased again until the program was abandoned in February 2000. A later study estimated that the probability of successful eradication after those 8 days without detection was only 27%! Of course, in this case the horse (shrew) had already bolted so the value in such a retrospective analysis only lies in our capacity to learn from it.
On the other end of the spectrum is the Australian campaign to eradicate bitterweed. If ingested by stock, bitterweed is toxic and leads to production of bitter undrinkable milk. Bitterweed was discovered at an isolated site in Queensland in 1953 and an eradication program was launched. In 1992, after four years and nine surveys with no detections, the weed was declared eradicated and regular surveys stopped. Many years later, March 2007 to be precise, a small infestation of bitterweed was discovered at the site of original occupancy. Control activities
were re-instigated, and the site is still being monitored.
“We found it will take six years of continued management without seeing any foxes to be 90% certain they have been eradicated [on Phillip Island], and nine years of continued management to be 95% certain.”
Understanding that absence of proof isn’t proof of absence By Tracy Rout and Michael McCarthy (University of Melbourne) These examples illustrate how differences in the characteristics of a species – whether it’s a plant or an animal, how quickly and often it reproduces, and how difficult it is to find – will affect the amount of time and effort that’s needed before we can be sufficiently confident it has been successfully eradicated. Of course, ‘sufficiently confident’ is a relative term. What does it actually mean? Given the costs of continued monitoring and management, and the economic and environmental costs of stopping an eradication program prematurely, how confident do we need to be before winding back an eradication program? One simple approach to defining this was framed by Regan et al., 2006, who suggested we stop looking when the expected costs outweigh the expected benefits.
Balancing risk on Phillip Island. If there are huge negative consequences to declaring eradication prematurely, it would be sensible to set the threshold level of confidence very high and keep managing for longer. Conversely, if management is very expensive and the consequences of mistakenly declaring eradication are insubstantial, it would be best to set a much lower threshold and declare success sooner. Rather than using guesswork and implicit value judgements, these risks and consequences can be considered explicitly using decision theory.
For example, we recently completed a study analysing the fox eradication program on Phillip Island in Victoria (see ‘The tale of the Phillip Island fox’). Lying just off the coast of Victoria, Phillip Island is a significant chunk of land having an area of approximately 100 km2. Foxes were first seen on the island in the early 1900s, and threaten much of the island’s wildlife. Indeed, they are the number one threat to little penguins which, due to fox predation, have been reduced from ten colonies to one. This remaining colony is a popular tourist attraction, attracting around 500,000 visitors per year who pay to watch the nightly Penguin Parade, when penguins return to their nests after sunset.
Fox management, involving searching, hunting, baiting, and trapping, has been conducted across most of the island since 1986. Effort has been ramped up since 2006, with the aim of eradication. This campaign seems to be working, with the number of foxes detected having decreased substantially in recent years. Managers expect that in the near future they will stop detecting foxes. They wanted to estimate the current fox population, and plan ahead by working out how long management should continue after foxes are no longer detected.
Using data on the number of foxes and staff-hours spent on management, we built a model of the fox population on Phillip Island (Rout et al., 2014). We estimated that there may have been as many as 200 foxes on the island in 1996, but in June 2012 there were approximately 11 remaining. We then projected our model into a hypothetical future where no further foxes are detected. We found it will take six years of continued management without seeing any foxes to be 90% certain they have been eradicated, and nine years of continued management to be 95% certain.
“This has significant implications for the fox eradication campaign on Phillip Island and resourcing in general. Three years is a long time to maintain fox control across the island without seeing any foxes or their spoor.”
Risk and cost
Declaring eradication when foxes are still present means they could bounce back to pre-2006 numbers, jeopardising six years of eradication efforts costing around $160,000 a year. To minimise these expected management costs, it’s best to continue managing for three years with no detections, when we can be around 70% certain of successful eradication. If we also consider the other costs of foxes bouncing back, such as decreases in numbers of penguins, shorebirds, and paying eco-tourists, it’s best to continue management for longer and be much more certain before declaring foxes eradicated.
So what does all this mean for foxes in Tasmania? Well, without public access to the report the State Government based its decision on, it’s impossible to say how they’ve assessed the risks and consequences of this decision. We know that foxes are extremely hard to detect, and there appears to have been periods in the past when foxes were present in Tasmania but no physical evidence was being found. We would therefore expect the certainty of successful eradication to increase very slowly with the time since foxes were last detected. It’s not clear whether the certainty of success has actually been estimated, as we did for Phillip Island.
The consequences of mistakenly declaring foxes eradicated in Tasmania are potentially huge, depending on how well the newly ‘wound back’ monitoring scheme can detect foxes at low numbers. If the scheme fails and numbers increase to levels seen before the eradication program, then the multi-million dollar investment (cited as around $50 million in recent reports) in the program would be wasted. If numbers increase to a level where future eradication is impossible, it is likely there will be catastrophic impacts for Tasmania’s biodiversity. Foxes are among the most destructive invasive species in Australia, and a main cause of our extraordinarily high mammal extinction rate.
Let’s hope, for Devil’s sake too, that the Tasmanian Government has got this decision right.
Regan TJ, MA McCarthy, PWJ Baxter, F Dane Panetta & HP Possingham (2006). Optimal eradication: when to stop looking for an invasive plant. Ecology Letters 9:759–766.
Rout TM, R Kirkwood, DR Sutherland, S Murphy & MA McCarthy (2014). When to declare successful eradication of an invasive predator? Animal Conservation 17: 125–132.