Dealing with uncertainty in qualitative models with a semi-quantitative approach based on simulations. Application to the Gironde estuarine food web (France)

In the context of ecosystem approach to fisheries, it is a critical issue to build management tools able to predict the possible trajectories of ecosystems under various human pressure or environmental variations, but also capable to point out influent and sensitive components. A major drawback of quantitative models is that they require many data to be implemented, which are often not available. Qualitative modelling with loop analysis needs only few data and offers the possibility to predict effect of sustained perturbations on an ecosystem. It considers direct and indirect interactions by inversing its qualitative community matrix. However, depending on the complexity of the studied ecosystem, predictions can be very inaccurate. Different approaches have been proposed to deal with uncertainty on the sign of the effects in loop analysis. However, as well as reducing uncertainty on their signs, we wanted to allow taking into account the strength of interactions. Thus, we developed a semi-quantitative method coupled with a simulation approach that also permits to use conjointly empirical data and expert opinion. The framework was applied to the Gironde estuarine ecosystem for which an Ecopath model has already been constructed. A sensitivity analysis permitted to show up the uncertainty on the results and revealed the gain of precision on the results while using the semi-quantitative method. Moreover, it pointed out that the precision was much better on strong effects, which allows the identification of influent and sensitive components of the ecosystem. Finally, the correlation between the results of the inverse community matrix and those of the mixed trophic impact routine of Ecopath allowed to place our semi-quantitative method as an intermediate step between loop analysis and Ecopath but also as a complementary approach for network analysis. (C) 2008 Elsevier B.V. All rights reserved.