Koellner, Thomas and Grét-Regamey, Adrienne and Marchamalo Sacristán, Miguel and Vignola, Raffaele
Bayesian modeling of ecosystem services in human enviroment systems..
In: "ACES 2008. A Conference on Ecosystem Services.", 08/12/2008-11/12/2008, Florida, USA.
The adaptive management of ecosystem services requires knowledge about the interdependence of land use decision-making and the ecosystem features in a given landscape; and how this coupled humanenvironment system is influenced by drivers of global change. The problem in this context is, that both decisionmaking processes and the ecosystem changes are subject to large uncertainties and incomplete information. Furthermore, trade-offs between different ecosystem services and biodiversity exist and actors tend to maximize only one feature. The adaptive management of an entire system thus needs to find a solution, which optimizes all ecosystem services given uncertain information. For this purpose, we develop a Bayesian Network BN of the humanenvironment system allowing evaluating simultaneously the effect of different decision-making processes on ecosystem responses and updating the results when better information becomes available. We test the approach in a case study in the Swiss Alps, where we focus on integrating the value of different ecosystem services as a support for landscape planning. Results show that if uncertainties are not explicitly integrated into the modeling framework, the information provided to the decision-makers might be misleading. For a case study in a Costa Rican watershed, we expand the BN with exogenous drivers from market (e.g., change in price for crops), policy (e.g., change in national park border) and climate (e.g., change in frequency of heavy rainfall). Policy instruments like command and control, park zoning and payments for cosystem services can help reaching a more balanced management of a watershed. For the planning of those instruments, however, it is helpful to have a model which shows how the manager of individual land units, takes policy measures, together with expected market changes and climate change into account in his land use decision-making. For each management unit, the prior probability of a specific land use and cover is updated with a posterior probability, when additional information about the management unit (e.g., slope, soil type, governance) is available. This type of model can be used to plan and simulate new policy measures like payments for ecosystem services, because it simultaneously takes the ecosystem, socio-economic system and the policy system into account. The model allows identifying management units with high and low values for each ecosystem services and thus the targeting of available financial funds can be optimized. First working steps show that such a BN provides a robust modeling environment, useful for better informed and participatory decision-making.