Publications des agents du Cirad

Cirad

Kinetics of added organic matter decomposition in a Mediterranean sandy soil

Thuriès L., Pansu M.A., Feller C., Herrmann P., Remy J.C.. 2001. Soil Biology and Biochemistry, 33 (7-8) : p. 997-1010.

Carbon mineralization kinetics of 17 organic materials were studied in a Mediterranean sandy soil. These added organic matters (AOM) used in the organic fertilizer industry differed in their origin and composition: plant residues from the agri-food industry, animal wastes, manures (plant and animal origin), composts at different composting times and organic fertilizers. The mixtures AOM-soils were incubated under aerobic conditions at 28°C during 6 months. Soil moisture was maintained at 75% water holding capacity and respired-CO2 was regularly trapped into alkali media in closed chambers, then checked by HCl titration. Analyses of CO2 were performed in triplicate at 17 sampling occasions. The mineralized AOM fraction (MAOMF) varied according to the AOM origin: from 12¿33% of added C for composts, to 65¿90% for animal-originated AOM, with many intermediate patterns for plant-originated AOM. Seven decomposition models from the literature were fitted to actual MAOMF: (a) three consecutive models with two 1st-order-kinetic compartments and three parameters (m1, humification; m2, exchange; m3, decomposition), (b) three parallel models (m4, with two compartments and three parameters; m8, a 1st-order plus 0-order model with three parameters; m5, a three-compartment model with four parameters), and (c) m7, a model with one 2nd-order-kinetic compartment and two parameters. Additionally, m6, a simplified version of m5 was proposed. Models m2 and m7 did not match with actual data or gave a poor fit. By the correlation parameters, the most simple model m4 was chosen instead of the consecutive models m1 and m3. Residual sums of squares were always greater¿but not significantly¿in m8 than in m4, which confirmed the superiority of the models with two 1st-order compartments against 1st-order plus 0-order models for incubation times higher than 100 days. Model m5 (most of its parameters being not correlated) gave the best predictions of our data. The proposed m6 version gave predictions with similar precision as m4 and appeared powerful with only two parameters (very labile and stable fractions of the AOM). A compromise between the precision of the predictions and the simplicity of the formulae allowed the recommendation of the well-known m4 model, and above all the simpler m6 model. (Résumé d'auteur)

Thématique : Chimie et physique du sol; Fertilisation; Méthodes mathématiques et statistiques

Documents associés

Article de revue

Agents Cirad, auteurs de cette publication :