Mark A. Burgman

A TAXONOMY AND TREATMENT OF UNCERTAINTY FOR ECOLOGY AND CONSERVATION BIOLOGY

Uncertainty is pervasive in ecology where the difficulties of dealing with sources of uncertainty are exacerbated by variation in the system itself. Attempts at classifying uncertainty in ecology have, for the most part, focused exclusively on epistemic uncertainty. In this paper we classify uncertainty into two main categories: epistemic uncertainty (uncertainty in determinate facts) and linguistic uncertainty (uncertainty in language). We provide a classification of sources of uncertainty under the two main categories and demonstrate how each impacts on applications in ecology and conservation biology. In particular, we demonstrate the importance of recognizing the effect of linguistic uncertainty, in addition to epistemic uncertainty, in ecological applications. The significance to ecology and conservation biology of developing a clear understanding of the various types of uncertainty, how they arise and how they might best be dealt with is highlighted. Finally, we discuss the various general strategies for dealing with each type of uncertainty and offer suggestions for treating compounding uncertainty from a range of sources.

PREDICTIVE ACCURACY OF POPULATION VIABILITY ANALYSIS IN CONSERVATION BIOLOGY

Population viability analysis (PVA) is widely applied in conservation biology to predict extinction risks for threatened species and to compare alternative options for their mangement. It can also be used as a basis for listing species as endangered under World Conservation Union criteria. However, there is considerable scepticism regarding the predictive accuracy of PVA, mainly because of a lack of validation in real systems. Here we conducted a retrospective test of PVA based on 21 long-term ecological studies—the first comprehensive and replicated evaluation of the predictive powers of PVA. Parameters were estimated from the first half of each data set and the second half was used to evaluate the performance of the model. Contrary to recent criticisms, we found that PVA predictions were surprisingly accurate. The risk of population decline closely matched observed outcomes, there was no significant bias, and population size projections did not differ significantly from reality. Furthermore, the predictions of the five PVA software packages were highly concordant. We conclude that PVA is a valid and sufficiently accurate tool for categorizing and managing endangered species.

Limits to the use of threatened species lists

Threatened species lists are designed primarily to provide an easily understood qualitative estimate of risk of extinction. Although these estimates of risk can be accurate, the lists have inevitably become linked to several decision-making processes. There are four ways in which such lists are commonly used: to set priorities for resource allocation for species recovery; to inform reserve system design; to constrain development and exploitation; and to report on the state of the environment. The lists were not designed for any one of these purposes, and consequently perform some of them poorly. We discuss why, if and how they should be used to achieve these purposes.

Eliciting Expert Knowledge in Conservation Science

Expert knowledge is used widely in the science and practice of conservation because of the complexity of problems, relative lack of data, and the imminent nature of many conservation decisions. Expert knowledge is substantive information on a particular topic that is not widely known by others. An expert is someone who holds this knowledge and who is often deferred to in its interpretation. We refer to predictions by experts of what may happen in a particular context as expert judgments. In general, an expert-elicitation approach consists of five steps: deciding how information will be used, determining what to elicit, designing the elicitation process, performing the elicitation, and translating the elicited information into quantitative statements that can be used in a model or directly to make decisions. This last step is known as encoding. Some of the considerations in eliciting expert knowledge include determining how to work with multiple experts and how to combine multiple judgments, minimizing bias in the elicited information, and verifying the accuracy of expert information. We highlight structured elicitation techniques that, if adopted, will improve the accuracy and information content of expert judgment and ensure uncertainty is captured accurately. We suggest four aspects of an expert elicitation exercise be examined to determine its comprehensiveness and effectiveness: study design and context, elicitation design, elicitation method, and elicitation output. Just as the reliability of empirical data depends on the rigor with which it was acquired so too does that of expert knowledge.

ROBUST DECISION-MAKING UNDER SEVERE UNCERTAINTY FOR CONSERVATION MANAGEMENT

In conservation biology it is necessary to make management decisions for endangered and threatened species under severe uncertainty. Failure to acknowledge and treat uncertainty can lead to poor decisions. To illustrate the importance of considering uncertainty, we reanalyze a decision problem for the Sumatran rhino, Dicerorhinus sumatrensis, using information-gap theory to propagate uncertainties and to rank management options. Rather than requiring information about the extent of parameter uncertainty at the outset, information-gap theory addresses the question of how much uncertainty can be tolerated before our decision would change. It assesses the robustness of decisions in the face of severe uncertainty. We show that different management decisions may result when uncertainty in utilities and probabilities are considered in decision-making problems. We highlight the importance of a full assessment of uncertainty in conservation management decisions to avoid, as much as possible, undesirable outcomes.

Bias in species range estimates from minimum convex polygons: implications for conservation and options for improved planning

Minimum convex polygons (convex hulls) are an internationally accepted, standard method for estimating species’ ranges, particularly in circumstances in which presence-only data are the only kind of spatially explicit data available. One of their main strengths is their simplicity. They are used to make area statements and to assess trends in occupied habitat, and are an important part of the assessment of the conservation status of species. We show by simulation that these estimates are biased. The bias increases with sample size, and is affected by the underlying shape of the species habitat, the magnitude of errors in locations, and the spatial and temporal distribution of sampling effort. The errors affect both area statements and estimates of trends. Some of these errors may be reduced through the application of αhulls, which are generalizations of convex hulls, but they cannot be eliminated entirely. α-hulls provide an explicit means for excluding discontinuities within a species range. Strengths and weaknesses of alternatives including kernel estimators were examined. Convex hulls exhibit larger bias than α-hulls when used to quantify habitat extent and to detect changes in range, and when subject to differences in the spatial and temporal distribution of sampling effort and spatial accuracy. α-hulls should be preferred for estimating the extent of and trends in species’ ranges.

Scientific Foundations for an IUCN Red List of Ecosystems

An understanding of risks to biodiversity is needed for planning action to slow current rates of decline and secure ecosystem services for future human use. Although the IUCN Red List criteria provide an effective assessment protocol for species, a standard global assessment of risks to higher levels of biodiversity is currently limited. In 2008, IUCN initiated development of risk assessment criteria to support a global Red List of ecosystems. We present a new conceptual model for ecosystem risk assessment founded on a synthesis of relevant ecological theories. To support the model, we review key elements of ecosystem definition and introduce the concept of ecosystem collapse, an analogue of species extinction. The model identifies four distributional and functional symptoms of ecosystem risk as a basis for assessment criteria: A) rates of decline in ecosystem distribution; B) restricted distributions with continuing declines or threats; C) rates of environmental (abiotic) degradation; and D) rates of disruption to biotic processes. A fifth criterion, E) quantitative estimates of the risk of ecosystem collapse, enables integrated assessment of multiple processes and provides a conceptual anchor for the other criteria. We present the theoretical rationale for the construction and interpretation of each criterion. The assessment protocol and threat categories mirror those of the IUCN Red List of species. A trial of the protocol on terrestrial, subterranean, freshwater and marine ecosystems from around the world shows that its concepts are workable and its outcomes are robust, that required data are available, and that results are consistent with assessments carried out by local experts and authorities. The new protocol provides a consistent, practical and theoretically grounded framework for establishing a systematic Red List of the world’s ecosystems. This will complement the Red List of species and strengthen global capacity to report on and monitor the status of biodiversity

Mapping epistemic uncertainties and vague concepts in predictions of species distribution

Most habitat maps are presented as if they were a certain fact, with no indication of uncertainties. In many cases, researchers faced with the task of constructing such maps are aware of problems with the modelling data and of decisions that they make within the modelling process that are likely to affect the output, but they find it difficult to quantify this information. In some cases they attempt to evaluate the modelled predictions against independent data, but the summary statistics have no spatial component and do not address errors in the predictions. It is proposed that maps of uncertainty would help in the interpretation of these summaries, and to emphasize patterns in uncertainty such as spatial clustering or links with particular covariates. This paper reviews the aspects of uncertainty that are relevant to habitat maps developed with logistic regression, and suggests methods for investigating and communicating these uncertainties. It addresses the problems of subjective judgement, model uncertainty and vague concepts along with the more commonly considered uncertainties of random and systematic error. Methods for developing realistic confidence intervals are presented along with suggestions on how to visualize the information for use by decision-makers.

Sensitivity analysis for models of population viability

A method of sensitivity analysis for population viability models is presented that uses logistic regression to evaluate the importance of model parameters that influence the risks of extinction. This approach is used to evaluate the importance of fecundity parameters and the initial number of non-breeding birds in a stochastic stage-structured model of helmeted honeyeater Lichenostomus melanops cassidix population dynamics. The regression analysis indicates which model parameters have the greatest impact on the risk of population decline. The results demonstrate that a simple expression containing the parameters of the model can encapsulate predictions of risk. This technique is proposed as an efficient alternative method of sensitivity analysis for population viability models. Of four fecundity parameters, the mean fecundity of intact pairs had the greatest influence on the risks faced by the helmeted honeyeater population. Mean fecundity of split pairs and the sex ratio of offspring were also important parameters. Over the range of parameters considered in this paper, environmental variation in fecundity and the initial number of non-breeding birds had little influence on the risks of decline. The importance of interactions between parameters was analysed.

Reducing overconfidence in the interval judgments of experts.

Elicitation of expert opinion is important for risk analysis when only limited data are available. Expert opinion is often elicited in the form of subjective confidence intervals; however, these are prone to substantial overconfidence. We investigated the influence of elicitation question format, in particular the number of steps in the elicitation procedure. In a 3-point elicitation procedure, an expert is asked for a lower limit, upper limit, and best guess, the two limits creating an interval of some assigned confidence level (e.g., 80%). In our 4-step interval elicitation procedure, experts were also asked for a realistic lower limit, upper limit, and best guess, but no confidence level was assigned; the fourth step was to rate their anticipated confidence in the interval produced. In our three studies, experts made interval predictions of rates of infectious diseases (Study 1, n = 21 and Study 2, n = 24: epidemiologists and public health experts), or marine invertebrate populations (Study 3, n = 34: ecologists and biologists). We combined the results from our studies using meta-analysis, which found average overconfidence of 11.9%, 95% CI [3.5, 20.3] (a hit rate of 68.1% for 80% intervals)-a substantial decrease in overconfidence compared with previous studies. Studies 2 and 3 suggest that the 4-step procedure is more likely to reduce overconfidence than the 3-point procedure (Cohens d = 0.61, [0.04, 1.18]).