Rapport de mission en Martinique du 11 au 13 juin 2006

Understanding the behaviour of individual-based models of forest dynamics becomes difficult as their complexity increases. A useful strategy consists in simplifying parts of the model, then scrutinizing for possible differences between the original model's predictions and those of the simplified version. This strategy is adopted to understand the role of space in a complex individual-based space-dependent model of tropical rain-forest dynamics in French Guiana. The model is made of three components: growth, recruitment and mortality. Two aggregation techniques are used to simplify space-dependent interactions into space-independent interactions: the mean-field approximation and an ad hoc approximation that considers pair interactions. Each technique is applied to one of the components of the model, the other two being left untouched, or to all components. Nine versions of the model (including the original one) are thus compared on the basis of the predicted diameter distribution at stationary state. The fully distance-independent models are consistent with the original one, whereas growth and recruitment significantly modify the model's predictions when simplified by the aggregation techniques. The interplay between the spatial distribution of trees at the population level and the dynamic processes at the individual level is thus understood, and the excessive impact of the mortality component that favours the establishment of clusters of big trees is diagnosed. © 2006 Elsevier B.V. All rights reserved.

Bibliographic Details

Main Author:	Pouzet, Denis
Format:	monograph biblioteca
Language:	fre
Published:	CIRAD
Subjects:	U10 - Informatique, mathématiques et statistiques, F62 - Physiologie végétale - Croissance et développement,
Online Access:	http://agritrop.cirad.fr/534967/ http://agritrop.cirad.fr/534967/1/document_534967.pdf
Tags:	Add Tag No Tags, Be the first to tag this record!