Publications

The litterbag method (LB) widely used in decomposition studies was compared to a direct observation method (DO) of the litter collected in the forest floor in three European Scots pine forests. In this last method, needles were sorted out of blocks of the forest floor and separated into 5 categories using morphological criteria representing different layers and decomposition levels. The residence time of the needles in each layer was calculated using the litterfall input, the size of the layer in the forest floor and its level of decomposition. The mass loss of each layer was determined using the linear mass density method. To estimate decomposition rates, the DO method might be more advantageous than the LB method because it is done without any manipulation or microclimatic artifact and because it integrates spatial and temporal variability. It is also convenient because the sampling does not have to be done simultaneously on different sites to allow site comparisons. Indeed, the collected litter blocks include the whole site history until the needle collapses which has been shown to take place after a period of up to 10 years in the present study. After 3 years, the mass loss was lower measured by the DO method than by the LB method. The decrease of litter mass measured by the DO method was best described by a simple asymptotic function of time. This model has a mechanistic signification because it underlines the possible long-term accumulation of recalcitrant products which is not the case with the exponential function. This accumulation of recalcitrant products is a key process in the storage of carbon in the organic matter reservoir of forest soils. Nevertheless this method is time consuming and cannot replace the LB method for large data set collection. This paper recommends when using the LB method to be aware of its inherent bias that can lead to misestimating decomposition in the long run.