Publications des agents du Cirad


Modelling the forest dynamics at Paracou : the contributions of four models

Picard N., Gourlet-Fleury S., Favrichon V.. 2004. In : Gourlet-Fleury Sylvie (ed.), Guehl Jean-Marc (ed.), Laroussinie Olivier (ed.). Ecology and management of a neotropical rainforest : lessons drawn from Paracou, a long-term experimental research site in French Guiana. Paris : Elsevier, p. 281-296.

Four models of forest dynamics were built using the Paracou data base: (i) an individual-based space-dependent model, SELVA, which describes in detail the behaviour of the trees on the basis of diameter increment, standing death, treefall and recruitment; (ii) a simplified individual-based space-dependent model derived from gap models to explore theoretical questions on aggregation; (iii) a distribution-based model based on a Usher matrix; (iv) a hybrid model mixing a matrix model for trees < 40 cm DBH and an individual-based space-independent model for trees => 40 cm DBH. The four models were compared in terms of their ability to predict the steady-state characteristics of the Paracou stands, as well as their dynamics following two levels of disturbance, and formal links between models showing various degrees of detail (according to whether they take into account individual trees and spatial interactions) were established using the simplified gap-like model. Among other results, we show that SELVA performs best at predicting the total number of trees and basal area at steady state and following a moderate level of disturbance such as conventional logging. The distribution-based and hybrid models show very similar behaviours. They perform better at describing the diameter distribution at steady state and the reaction of the stands after a high level of disturbance. The return to steady state after clear-cutting is predicted to be 250 years by these models, and 300 years by SELVA. The study of formal links between models led to the conclusion that individual-based and distribution-based models are equivalent provided they both take, or do not take , space-dependence into account. However, models working at the same level of stand description, either individual-based or distribution-based, do not yield the same predictions if they are, or are not, space-dependent. The respective interest and limitations, for managers, of the various models developed are discussed in the light of these results.

Mots-clés : forêt tropicale humide; dynamique des populations; modèle mathématique; aménagement forestier; Écologie

Documents associés

Chapitre d'ouvrage

Agents Cirad, auteurs de cette publication :