Choice and development of decision support tools for the sustainable management of deer–forest systems
Situations where a natural resource is both an asset, as well as a threat, to the integrity of ecosystem function and biodiversity are difficult to manage sustainably. One such situation happens when native deer populations, which are managed for sport are overexploiting forests to a point where they severely compromise natural forest regeneration. Managers facing those situations need support from the scientific community to analyse and synthesise information on deer-forest relationships and thus help to predict the potential outcomes of different management options for both the deer and the forests. Research scientists are increasingly expected to provide expertise and support into the decision-making process. One way to achieve this is to develop decision support tools (DSTs) based upon sound, scientific understanding of the deer-forest systems. Our objective is to explore a range of approaches that have been used for the development of DSTs for deer-forest management and to propose criteria for selecting a specific approach or combination of approaches for specific situations. DST and research-oriented models were catalogued according to two modelling paradigms: bottom-up models, which simulate systems through inductive inference, by scaling up from fundamental processes to the inherent behaviour of the system-the best known applications are forest gap and individual-based models; and top-down models which proceed by deductive, rule-based inference-they include expert systems, qualitative simulation models, frame-based models, Markovian process models and Bayesian networks. Uncertainty assessment in both modelling paradigms is discussed. The analysis is put in the context of two very different examples of deer-forest systems currently requiring DST development to guide their management: (1) the upland red/roe deer-fragmented temperate/boreal forest system of Scotland; and (2) the white-tailed deer-eastern boreal forest system of Anticosti Island, Quebec, Canada. We conclude that a top-down approach with explicit uncertainty assessment should be aimed for, as a deliverable product to the end-users, keeping in mind that simulation models from the bottom-up family may be required to gain insights about the underlying mechanisms. (C) 2003 Elsevier B.V. All rights reserved.