Ecology, economics and ethics: the three Es required for sustained and effective vertebrate pest management
In New Zealand some introduced vertebrates are managed to limit their impacts on indigenous biodiversity, agricultural production, and infrastructure, and others for commercial and recreational harvest. Most of this management requires the use of lethal tools and techniques, and as a result wildlife managers are often challenged to justify their policies and actions on the basis of pest ecology, population dynamics and ethical acceptability. Unfortunately, most managers are poorly equipped to enter into informed discussion on what are often complex ethical and philosophical issues. For example, choosing and defending lethal control poses significant ethical challenges, and defending why populations need to be managed at all can raise significant philosophical challenges.
Most decisions that managers make related to ecology, economics and ethics have varying degrees of uncertainty, and here Bruce Warburton and Dean Anderson explore whether a probabilistic modelling approach can help managers frame and formalise adaptive management that integrates ecology, economics and ethics. Such formalisation will encourage managers to maximise the probability of achieving sustainable and effective wildlife management outcomes.
Ecological uncertainty
Managing wildlife can be inherently complex because individual species are part of broader multi-species communities, which in turn contribute to higher-level ecosystem processes. There are many examples of successful management actions delivering desired outcomes. For example, long-term suppression of ship rats, brushtail possums and stoats has resulted in significant population recoveries of threatened bird species, while reducing possum numbers alone has resulted in significant reductions in the number of cattle herds infected with bovine tuberculosis.
However, there are many failed programmes. Failures occur because managers often have insufficient knowledge about (1) species interactions, resulting in the release of meso-predators whose negative impact is as great or greater than that of the species controlled; (2) the relationship between pest density and impacts, resulting in inadequate reductions in pest density, and (3) the effectiveness of the control tools and strategies used.
Economic uncertainty
Managing wildlife has costs that managers need to consider when planning control programmes, either to decide how best to allocate limited funds across priority areas or how to allocate sufficient funds to achieve the desired outcome. There is a range of fixed and variable costs associated with managing wildlife, including planning, implementation, non-target mitigation, addressing public needs, monitoring, compliance, and, where cost–benefit analyses are required, the monetising of conservation assets and benefits.
Ethical uncertainty
Wildlife management programmes often generate strongly polarised dialogue because of the diverse value sets held by stakeholders. Managers need to be aware of such differing values and recognise them as another level of uncertainty to account for. Managers faced with manipulating vertebrate species, especially if lethal methods are proposed, have both animal welfare and ethical issues to address. All lethal control methods have welfare costs (i.e. they cause pain or distress), ranging from fast-acting vertebrate toxins and kill traps that have the least welfare cost, to anticoagulant toxins and leg-hold traps that have significant welfare costs.
As a result, managers can be the target of vociferous opposition from stakeholders citing ethical issues, and although they are often passionate about protecting conservation or production values, they are often not well enough informed to respond to animal rights advocates and moral philosophers who claim the moral high ground by advocating non-lethal control methods. Utilitarians attempt to compare the costs to the benefits of managing pests, but in most ecological cases, although the welfare costs can be accounted for, it is often extremely difficult to measure the benefits, especially when they relate to non-sentient organisms such as plants, providing managers with another level of uncertainty.
Integrating ecological, economic and ethical uncertainty
Effective management strategies for wildlife issues hinge on the success of the 3Es for the duration of the project; that is, the continued success of ecological outcomes, financial and institutional support, and broad acceptance of the ethics of the actions undertaken. Given the inherent complexity and associated uncertainty, managers should modify their strategies as experience and new information are obtained.
Adaptive management guided by Bayesian uncertainty modelling is an evidence-based mechanism for arriving at and revising strategies to maximise the probability that management actions will achieve their objectives. Adaptive management provides a framework for formally and quantitatively incorporating the likely acceptance of the ethics of management actions into strategy development, especially when applied to the management of wild animals. Dean and Bruce have developed a conceptual approach for incorporating the 3Es into an adaptive management programme informed by Bayesian uncertainty modelling. Quantitative predictions of the probability of successful outcomes of each of the 3Es are made, and the probability of the overall success of a programme is the product of each of the 3E-component probabilities:
P(overall success) = P(ecology)P(economics)P(ethics)
The individual 3E-component probabilities are derived from data-driven or expert-parameterised models. Data-driven models are preferred and should replace expert-parameterised models as management programmes progress. Bayesian statistics are desirable because they take advantage of existing independent parameter estimates and expert insight of the system, and they update parameter distributions as more data are collected. Parameters are incorporated using probability distributions, not point estimates, to capture uncertainty in the understanding of the processes and predictions.
Three hypothetical examples are presented below to illustrate how this approach will lead to an overall probability of project success.
- In the first example, the mean ecological, economic and ethical probabilities of success are 0.9, 0.6, and 0.5 respectively (see graph, scenario A, upper panel). To obtain an overall probability of success that incorporates uncertainty, random variates from each component probability distribution are used to calculate the product (see equation). This is repeated for 2,000 iterations to obtain the resulting distribution of the probability of success (see graph, scenario A, lower panel). With two of the 3E-component probabilities equal to or slightly better than a coin toss, the overall probability of success has a likely value of 0.24.
- In a second example, the mean probabilities of continued economic and ethical support are 0.9 and 0.85 respectively, but the mean probability of ecological success is only 0.2 (see graph, scenario B, upper panel). Clearly the weak link is that the predicted ecological outcome for this strategy is unlikely to be sufficient, and the equation gives a most likely combined probability of success of 0.12 (see graph, scenario B, lower panel).
- A third example illustrates that even with relatively high probabilities of success for the 3Es (means = 0.95, 0.85, and 0.80; see graph, scenario C, upper panel), the most likely combined probability of success for the program is 0.70 (scenario C, lower panel).
Do these relatively low joint probabilities mean that such programmes are unjustified? Not necessarily, as many will succeed, and adaptive management and research will discover innovative ways to increase each of the 3E-component probabilities to increase the combined probability of project success.
Conclusions
Managers will benefit from adopting this simple framework for managing vertebrate pests, and even if their first probabilities are based on expert opinion or degrees of belief, adaptive management using a Bayesian model will allow these probabilities to be updated. If nothing else, recognising that all three components have to be addressed when developing a sustainable pest management plan will be an improvement on the status quo.
This work was funded by Strategic Science Investment Funding.
Bruce Warburton
warburtonb@landcareresearch.co.nz
Dean Anderson