Britta Tietjen, Andreas Huth

• The overuse of rainforests in the last century and its consequences necessitate a rethinking of logging policies. To this end models have been developed to simulate rainforest dynamics and to allow optional management strategies to be evaluated. Parameterisation of presently existing models for a certain site needs a lot of work, thus the parameterisation effort is too high to apply the models to a wide range of rainforests. Hence, in this paper we introduce the simplified model FORREG using the knowledge we have gained from a more complex model, FORMIX3-Q. The FORREG model uses differential equations to determine the volume growth of three successional species groups. Parameterisation is simplified by a genetic algorithm, which determines the required internal model parameters from characteristics of the forest dynamics. The new model is employed to assess the sustainability of various logging policies in terms of yield and damage. Results for three forests are discussed: (1) the tropical lowland rain forest in the Deramakot ForestThe overuse of rainforests in the last century and its consequences necessitate a rethinking of logging policies. To this end models have been developed to simulate rainforest dynamics and to allow optional management strategies to be evaluated. Parameterisation of presently existing models for a certain site needs a lot of work, thus the parameterisation effort is too high to apply the models to a wide range of rainforests. Hence, in this paper we introduce the simplified model FORREG using the knowledge we have gained from a more complex model, FORMIX3-Q. The FORREG model uses differential equations to determine the volume growth of three successional species groups. Parameterisation is simplified by a genetic algorithm, which determines the required internal model parameters from characteristics of the forest dynamics. The new model is employed to assess the sustainability of various logging policies in terms of yield and damage. Results for three forests are discussed: (1) the tropical lowland rain forest in the Deramakot Forest Reserve, (2) the Lambir National Park in Malaysia and (3) a subtropical forest in Paraguay. Our model reproduces both undisturbed forest dynamics and dynamics of logged forests simulated with FORMIX3-Q very well. However, the resultant volumes of yield and damage differ slightly from those gained by FORMIX3-Q if short logging cycles are simulated. Choosing longer logging cycles leads to a good correspondence of both models. For the Deramakot Forest Reserve different logging cycles are compared and discussed. (c) 2006 Elsevier B.V. All rights reserved.…