Brendan A. Wintle

Predicting species distributions for conservation decisions.

Species distribution models (SDMs) are increasingly proposed to support conservation decision making. However, evidence of SDMs supporting solutions for on-ground conservation problems is still scarce in the scientific literature. Here, we show that successful examples exist but are still largely hidden in the grey literature, and thus less accessible for analysis and learning. Furthermore, the decision framework within which SDMs are used is rarely made explicit. Using case studies from biological invasions, identification of critical habitats, reserve selection and translocation of endangered species, we propose that SDMs may be tailored to suit a range of decision-making contexts when used within a structured and transparent decision-making process. To construct appropriate SDMs to more effectively guide conservation actions, modellers need to better understand the decision process, and decision makers need to provide feedback to modellers regarding the actual use of SDMs to support conservation decisions. This could be facilitated by individuals or institutions playing the role of ‘translators’ between modellers and decision makers. We encourage species distribution modellers to get involved in real decision-making processes that will benefit from their technical input; this strategy has the potential to better bridge theory and practice, and contribute to improve both scientific knowledge and conservation outcomes.

Prioritizing multiple-use landscapes for conservation: methods for large multi-species planning problems.

Across large parts of the world, wildlife has to coexist with human activity in highly modified and fragmented landscapes. Combining concepts from population viability analysis and spatial reserve design, this study develops efficient quantitative methods for identifying conservation core areas at large, even national or continental scales. The proposed methods emphasize long-term population persistence, are applicable to both fragmented and natural landscape structures, and produce a hierarchical zonation of regional conservation priority. The methods are applied to both observational data for threatened butterflies at the scale of Britain and modelled probability of occurrence surfaces for indicator species in part of Australia. In both cases, priority landscapes important for conservation management are identified.

When to stop managing or surveying cryptic threatened species.

Threatened species become increasingly difficult to detect as their populations decline. Managers of such cryptic threatened species face several dilemmas: if they are not sure the species is present, should they continue to manage for that species or invest the limited resources in surveying? We find optimal solutions to this problem using a Partially Observable Markov Decision Process and rules of thumb derived from an analytical approximation. We discover that managing a protected area for a cryptic threatened species can be optimal even if we are not sure the species is present. The more threatened and valuable the species is, relative to the costs of management, the more likely we are to manage this species without determining its continued persistence by using surveys. If a species remains unseen, our belief in the persistence of the species declines to a point where the optimal strategy is to shift resources from saving the species to surveying for it. Finally, when surveys lead to a sufficiently low belief that the species is extant, we surrender resources to other conservation actions. We illustrate our findings with a case study using parameters based on the critically endangered Sumatran tiger (Panthera tigris sumatrae), and we generate rules of thumb on how to allocate conservation effort for any cryptic species. Using Partially Observable Markov Decision Processes in conservation science, we determine the conditions under which it is better to abandon management for that species because our belief that it continues to exist is too low.

PRECISION AND BIAS OF METHODS FOR ESTIMATING POINT SURVEY DETECTION PROBABILITIES

Wildlife surveys often seek to determine the presence or absence of species at sites. Such data may be used in population monitoring, impact assessment, and species– habitat analyses. An implicit assumption of presence/absence surveys is that if a species is not detected in one or more visits to a site, it is absent from that site. However, it is rarely if ever possible to be completely sure that a species is absent, and false negative observation errors may arise when detection probabilities are less than 1. The detectability of species in wildlife surveys is one of the most important sources of uncertainty in determining the proportion of a landscape that is occupied by a species. Recent studies emphasize the need to acknowledge and incorporate false negative observation error rates in the analysis of site occupancy data, although a comparative study of the range of available methods for estimating detectability and occupancy is notably absent. The motivation for this study stems from the lack of guidan...

ESTIMATING AND DEALING WITH DETECTABILITY IN OCCUPANCY SURVEYS FOR FOREST OWLS AND ARBOREAL MARSUPIALS

Abstract Surveys that record the presence or absence of fauna are used widely in wildlife management and research. A false absence occurs when an observer fails to record a resident species. There is a growing appreciation of the importance of false absences in wildlife surveys and its influence on impact assessment, monitoring, habitat analyses, and population modeling. Very few studies explicitly quantify the rate of these errors. Quantifying the rate of false absences provides a basis for estimating the survey effort necessary to assert that a species is absent with a pre-specified degree of confidence and allows uncertainty arising from false absences to be incorporated in inference. We estimated the rate of false absences for 2 species of forest owl and 4 species of arboreal marsupial based on 8 repeat visits to 50 survey locations in south-eastern Australia. We obtained estimates using a generalized zero-inflated binomial model. We presented detectability curves for each species to convey the number of visits required to achieve a specified level of confidence that resident species will be detected. The observation error rates we calculated were substantial but varied between species. For the least detectable species, the powerful owl (Ninox strenua), our standard surveys returned false absences on 87% of visits. However, our surveys of the more detectable sugar glider (Petaurus breviceps) returned a 45% false absence rate. We predict that approximately 18 visits would be required to be 90% sure of detecting resident owls and approximately 5 visits would provide 90% confidence of detecting resident sugar gliders. We fitted hierarchical logistic regression models to the data to describe the variation in detection rates explained by environmental variables. We found that temperature, rainfall, and habitat quality influenced the detectability of most species. Consideration of observation error rates could result in important changes to resource management and conservation planning.

Modeling species-habitat relationships with spatially autocorrelated observation data.

Spatial autocorrelation in wildlife observation data arises when extrinsic environmental processes and patterns that influence the spatial distribution of wildlife are themselves spatially structured, or when species are subject to intrinsic population processes, causing contagion or dispersion effects. Territoriality, Allee effects, dispersal limitations, and social clustering are examples of intrinsic processes. Both forms of autocorrelation can violate the assumptions of generalized linear regression models, resulting in biased estimation of model coefficients and diminished predictive performance. Such consequences may be avoided for extrinsic autocorrelation when autocorrelated environmental variables are available for use as model covariates, whereas intrinsic spatial autocorrelation requires an alternative modeling approach. The autologistic model provides an approach suited to the binary observations often obtained in wildlife surveys, but its performance has not been tested across widely varying sampling intensities or strengths of intrinsic spatial structure. Here we use simulated data to test the autologistic model under a range of sampling conditions. The autologistic model obtains better fits and substantially better predictive performance than the standard logistic regression model over the full range of sampling designs and intensities tested. We provide a simple Bayesian implementation of the autologistic model, which until now has not been achieved with standard statistical software alone. A step-by-step procedure is given for characterizing and modeling spatial autocorrelation in binary observation data, along with computer code for fitting autologistic models in WinBUGS, a freeware Bayesian analysis package. This approach avoids normal approximations to the pseudo-likelihood, in contrast to previous Bayesian applications of the autologistic model. We provide a sample application of the autologistic model, fitted to survey data for a gliding marsupial in southeastern Australia.

Counting the books while the library burns: why conservation monitoring programs need a plan for action

Conservation monitoring programs are critical for identifying many elements of species ecology and for detecting changes in populations. However, without articulating how monitoring information will trigger relevant conservation actions, programs that monitor species until they become extinct are at odds with the primary goal of conservation: avoiding biodiversity loss. Here, we outline cases in which species were monitored until they suffered local, regional, or global extinction in the absence of a preplanned intervention program, and contend that conservation monitoring programs should be embedded within a management plan and characterized by vital attributes to ensure their effectiveness. These attributes include: (1) explicit articulation of how monitoring information will inform conservation actions, (2) transparent specification of trigger points within monitoring programs at which strategic interventions will be implemented, and (3) rigorous quantification of the ability to achieve early detection of change.

Active adaptive conservation of threatened species in the face of uncertainty

Adaptive management has a long history in the natural resource management literature, but despite this, few practitioners have developed adaptive strategies to conserve threatened species. Active adaptive management provides a framework for valuing learning by measuring the degree to which it improves long-run management outcomes. The challenge of an active adaptive approach is to find the correct balance between gaining knowledge to improve management in the future and achieving the best short-term outcome based on current knowledge. We develop and analyze a framework for active adaptive management of a threatened species. Our case study concerns a novel facial tumor disease affecting the Australian threatened species Sarcophilus harrisii: the Tasmanian devil. We use stochastic dynamic programming with Bayesian updating to identify the management strategy that maximizes the Tasmanian devil population growth rate, taking into account improvements to management through learning to better understand disease latency and the relative effectiveness of three competing management options. Exactly which management action we choose each year is driven by the credibility of competing hypotheses about disease latency and by the population growth rate predicted by each hypothesis under the competing management actions. We discover that the optimal combination of management actions depends on the number of sites available and the time remaining to implement management. Our approach to active adaptive management provides a framework to identify the optimal amount of effort to invest in learning to achieve long-run conservation objectives.

Ignoring imperfect detection in biological surveys is dangerous: a response to 'fitting and interpreting occupancy models'.

In a recent paper, Welsh, Lindenmayer and Donnelly (WLD) question the usefulness of models that estimate species occupancy while accounting for detectability. WLD claim that these models are difficult to fit and argue that disregarding detectability can be better than trying to adjust for it. We think that this conclusion and subsequent recommendations are not well founded and may negatively impact the quality of statistical inference in ecology and related management decisions. Here we respond to WLDs claims, evaluating in detail their arguments, using simulations and/or theory to support our points. In particular, WLD argue that both disregarding and accounting for imperfect detection lead to the same estimator performance regardless of sample size when detectability is a function of abundance. We show that this, the key result of their paper, only holds for cases of extreme heterogeneity like the single scenario they considered. Our results illustrate the dangers of disregarding imperfect detection. When ignored, occupancy and detection are confounded: the same naïve occupancy estimates can be obtained for very different true levels of occupancy so the size of the bias is unknowable. Hierarchical occupancy models separate occupancy and detection, and imprecise estimates simply indicate that more data are required for robust inference about the system in question. As for any statistical method, when underlying assumptions of simple hierarchical models are violated, their reliability is reduced. Resorting in those instances where hierarchical occupancy models do no perform well to the naïve occupancy estimator does not provide a satisfactory solution. The aim should instead be to achieve better estimation, by minimizing the effect of these issues during design, data collection and analysis, ensuring that the right amount of data is collected and model assumptions are met, considering model extensions where appropriate.

Strategic foresight: how planning for the unpredictable can improve environmental decision-making

Advanced warning of potential new opportunities and threats related to biodiversity allows decision-makers to act strategically to maximize benefits or minimize costs. Strategic foresight explores possible futures, their consequences for decisions, and the actions that promote more desirable futures. Foresight tools, such as horizon scanning and scenario planning, are increasingly used by governments and business for long-term strategic planning and capacity building. These tools are now being applied in ecology, although generally not as part of a comprehensive foresight strategy. We highlight several ways foresight could play a more significant role in environmental decisions by: monitoring existing problems, highlighting emerging threats, identifying promising new opportunities, testing the resilience of policies, and defining a research agenda.