Résumé
Several decomposition experiments were used to explore the potential of near infrared reflectance spectroscopy (NIRS) for predicting litter decomposability. A first experiment was conducted on a calibration set, and predictive equations were established between the NIR spectral data of the initial litters and the decay descriptors. These equations were then applied to two validation sets. The 34 litters of the calibration set were incubated for eight weeks in microcosms in the laboratory. The values of litter mass remaining (LMR) in relation to incubation time were fitted to the single-exponential decay model (LMR = e-kt). The litter mass loss (LML) values at different sampling dates and the rate constant k were strongly correlated with the initial litter characteristics. The calibrations carried out between measurements of the NIR spectra on the initial litters and the different descriptors of the decomposition patterns of the 34 litters showed that the LML values at the different sampling dates and the rate constant k were correlated with the spectra of the initial litters more closely than with their initial characteristics. The descriptors of litter decomposition patterns could therefore be predicted by NIRS. Among these descriptors, which were all correlated with one another, the easiest to measure (LML after one week incubation in the microcosm) was tested as a litter decomposability index (LDI). LDI was predicted by NIRS on the initial litters in two validation sets, the first consisting of 12 litters incubated under the same conditions in microcosms over a longer period (14 mo), the second consisting of four litters incubated in the field for one year at three stations situated along a climatic gradient. The results showed that the values of LDI predicted by NIRS were significantly correlated with the decay rate constant k in the two validation sets. The values of LDI predicted by NIRS therefore expressed a decomposability scale that was valid for litters decomposing over a longer period under the same incubation conditions and for litters subjected to different incubation conditions. The possibility of predicting litter decomposability by NIRS provides many opportunities, firstly for studying spatial and temporal variability in the rate of recycling of organic matter, and secondly for characterizing the gradual changes in litter quality during decomposition.