Résumé : Cette these etudie les effets de la conversion de la foret tropicale en parcelles agricoles, sur les emissions de n2o par les sols. Ce travail est realise a l’echelle du cycle cultural sur le dispositif experimental de combi (littoral guyanais), a l’aide de mesures in situ et de simulations (modele noe) des flux de n2o. La foret tropicale a ete comparee a des parcelles de foret converties en terres agricoles par la methode ‘chop-and-mulch’. La methode de deforestation ‘chop-and-mulch’ associe coupe mecanique de la vegetation, broyage puis enfouissement des residus forestiers dans le sol. Les terres agricoles etaient soient des parcelles de prairie non paturee soient des parcelles cultivees (maïs fertilise/soja) conduites en semis conventionnel (semis apres travail du sol, sans plante de couverture) ou en semis direct (sans travail du sol, avec plantes de couverture).Les principaux resultats de cette etude sont : le sol de la foret tropicale de combi est un faible emetteur de n2o ; la conversion par la methode ‘chop-and-mulch’ de cette foret en prairie ne conduit pas a une augmentation significative des emissions de n2o entre le 19eme et le 31eme mois suivant la conversion ; la conversion de la foret en parcelles cultivees induit une augmentation significative des emissions de n2o due a la fertilisation et a la modification des parametres edaphiques (densite apparente, temperature, humidite volumique) ; la technique sans travail du sol n’engendre pas de flux de n2o significativement plus eleves que la technique avec travail du sol ; l’introduction du phenomene d’hysterese hydrique dans le modele noe presente un reel potentiel pour l’estimation des emissions de n2o in situ.
This study investigates the effects of the conversion of tropical forest to cultivation on soil n2o emissions. The study was carried out over a complete crop cycle at the experimental site combi (french guianese coast). Nitrous oxide fluxes were obtained in the field and by conducting simulations with the noe model. Undisturbed tropical rainforest was compared to rainforest that had been converted to agricultural land using the ‘chop-and-mulch’ method. The ‘chop-and-mulch’ method is a fire-free method used for vegetation clearing combining the mechanical felling of trees with the mulching of small vegetation. Agricultural land included either mowed grassland or soybean/fertilised maize crop rotation. For croplands the two cultivation practices employed were: conventional seeding (using an offset disc harrow, without cover plants) or direct seeding (no till, with cover plants).The main results of this study are: rainforest soil at combi produced low n2o emissions; rainforest converted to mowed grassland using the 'chop-and-mulch’ method did not lead to a significant increase in n2o emissions between the 19th and 31st months after conversion; the conversion of rainforest to croplands induced a significant increase in soil n2o emissions due to the application of fertiliser and the modification of soil parameters (bulk density, temperature, volumetric moisture); n2o emissions from agricultural practices with no-till were no higher than those produced by conventional agricultural practices using an offset disc harrow; and, the introduction of an hydric hysteresis into the noe model constitutes a promising improvement to estimate in situ n2o emissions.

http://www.theses.fr/2013AGUY0610/document