Issue 55:2 of Journal of Applied Ecology features a Spotlight on Decision making under uncertainty. Senior Editor, Michael Bode discusses these Spotlight papers, chosen by the Editors to showcase the latest research in this area.
Professor John Shepherd famously compared fisheries management to forestry, except that the trees were invisible and were constantly moving around. Most applied ecologists would sympathise, while simultaneously casting an envious eye towards the long-term catch and effort datasets available to fisheries science, uncertain though they are.
Uncertainty permeates decision making in ecological systems. This doesn’t make applied ecology unique – ask any meteorologist – but ecology faces a particularly twisted set of challenges. Ecological systems are complex, comprised of a large number of nonlinearly-interacting species and processes that lack a solid mechanistic understanding. As Shepherd pointed out, they’re also particularly difficult to observe in situ, and even harder to isolate, replicate, or to experiment on. Finally, ecological management is an (increasingly) underfunded discipline. Managers don’t have a lot of free resources to direct towards monitoring, either before, during, or after they undertake interventions.
Calls for additional funding, for long-term monitoring programs and distributed observation networks are enjoyable, but none of these facts are about to change in the near future. In the face of uncertainty – sometimes in the face of flat-out ignorance – applied ecologists and conservation scientists need to make irreversible decisions that will decide the fate of species and ecosystems. This month, Journal of Applied Ecology focuses on questions, methods and solutions in this space. Our Spotlight this issue is on decision making under uncertainty, and it includes a series of articles that delve into the effects of uncertainty on management, going beyond the standard phrase in the discussion that begins: “Future research should focus on …”.
Risk analysis offers a primary solution to the problem of uncertainty, taking into account its potential influence on management outcomes, without necessarily attempting to reduce that uncertainty. It’s also a precise solution to uncertain problems (rather than an approximation) – if the future states of the system can be completely and accurately characterised by a probability distribution, then risk analyses will identify the best decision. A primary strength of risk analytic tools is their common use and broad comprehension and acceptance, across applied science, economics, government and the private sector. The Spotlight contains two papers that focus on the application of existing risk analytic tools to environmental regulations, with Warwick-Evans et al. applying a model-based risk analysis tool to the placement of wind farms, and Roy et al. using the same approaches to provide recommendations for international policy on invasive alien species management. Meffin et al. also engage with risk analyses. Focusing on invasive alien species, they assess whether the standard unit of invasion risk analysis – the individual invasive species – deserves as much attention as it currently attracts. Although their analysis focused on a set of relatively similar species (varieties of three Brassica species), their results show that varietal differences explain very little of the variation in performance – an order of magnitude less than is explained by environmental and microsite differences.
Ecological decision-makers often face incredibly complex problems, particularly when their objectives explicitly include the large number of species, locations, and processes that make up ecosystems. Rather than attempt to based our decisions on every one of these uncertain factors (expensive & difficult), many papers in applied ecology instead suggest that we can instead base them on a subset, or an amalgamation of these features, expressed as an index or surrogate. However, while they’re easier to measure, these indices generally have an uncertain or untested functional relationship with the full set of management objectives. Giljohann et al. evaluate such an index – the geometric mean of species’ abundances – in the context of fire management in the pyrophilic ecosystems of southern Australia. Meanwhile, Yamane et al. introduce and analyse a new index, the diversity deficit, which they argue will help make decisions that improve stock stability. Both analyses reveal both the strengths and weaknesses of indices. The distillation of complex system states into simple, unidimensional values will never be perfect; rather than measure the accuracy of the indices, these papers both correctly attempt to understand the types of decisions for which their respective metrics perform well, and when they perform poorly.
The final two articles in the Spotlight section approach the coupled issues of decision making and uncertainty from very different angles. Eriksen et al. are not focused on our uncertain understanding of ecological systems, nor on our limited ability to accurately observe those systems. Instead, they consider the uncertain implementation of management policies by individual actors in the system – fishers, harvesters, managers, etc. Using a combination of empirical regression and simulation-based risk analyses, they show that the presence of substantial implementation uncertainty should drive a markedly different approach to management. Meanwhile, de Bie et al. interrogate the idea of “triggers” in ecological management – ecosystem events (e.g., the decline of a threatened species’ population below some threshold abundance) that instigate predefined management interventions. Conservation scientists are attracted to trigger points because they might help avoid the all-too-common phenomenon of species being monitored to extinction. However, in data-poor situations, actions are often delayed because they are uncertain about whether the system has indeed fallen below a trigger point (as it often retrospectively proves to be). de Bie et al. identify the steps required to define these trigger points, with a focus on the role played by data (and therefore by uncertainty), in choosing the particular trigger state and response.
Uncertainty is the nemesis of good decision making. However, this fact cannot excuse applied ecologists from incorporating this uncertainty in their decision making processes and tools. Each of these articles, in different ways, has explicitly and quantitatively placed uncertainty close to the heart of their analyses and thinking. Despite the discomfort this provokes, this is where uncertainty belongs.
Find out more about Spotlights here.