Revealing beliefs: using ensemble ecosystem modelling to extrapolate expert beliefs to novel ecological scenarios. Issue 8 (19th December 2016)

• Record Type: Journal Article
• Authors: Bode, Michael; Baker, Christopher M.; Benshemesh, Joe; Burnard, Tim; Rumpff, Libby; Hauser, Cindy E.; Lahoz‐Monfort, José J.; Wintle, Brendan A.
• Editors: Chisholm, Ryan
• Is Part Of: Methods in ecology and evolution. Volume 8:Issue 8(2017)
• Page Start: 1012
• Page End: 1021
• Publication Date: 2016-12-19
• Subjects: ensemble forecasting -- expert judgement -- interaction networks -- qualitative modelling -- uncertainty analysis; Ecology -- Periodicals; Evolution -- Periodicals; 577
• DOI: 10.1111/2041-210X.12703
• Languages: English
• ISSNs: 2041-210X

Abstract

Summary: Ecosystem‐based management requires predictive models of ecosystem dynamics. There are typically insufficient empirical data available to parameterise these complex models, and so decision‐makers commonly rely on beliefs elicited from experts. However, such expert beliefs are necessarily limited because (i) only a small proportion of ecosystem components and dynamics have been observed; (ii) uncertainty about ecosystem dynamics can result in contradictory expert judgements and (iii) elicitation time and resources are limited. We use an ensemble of dynamic ecosystem models to extrapolate a limited set of stated expert beliefs into a wider range of revealed beliefs about how the ecosystem will respond to perturbations and management. Importantly, the method captures the expert uncertainty and propagates it through to predictions. We demonstrate this process and its potential value by applying it to the conservation of the threatened malleefowl ( Leipoa ocellata ) in the Murray mallee ecosystems of southern Australia. In two workshops, we asked experts to construct a qualitative ecosystem interaction network and to describe their beliefs about how the ecosystem will respond to particular perturbations. We used this information to constrain an ensemble of 10 9 community models, leaving a subset that could reproduce stated expert beliefs. We then interrogated this ensemble of models to reveal experts' implicit beliefs about management scenarios that were not a part of the initial elicitation exercises. Our method uses straightforward questions to efficiently elicit expert beliefs, and then applies a flexible modelling approach to reveal those experts' beliefs about the dynamics of the entire ecosystem. It allows rapid planning of ecosystem‐based management informed by expert judgement, and provides a basis for value‐of‐information analyses and adaptive management.