Tree fine roots dynamics and carbon sequestration potential in a Mediterranean agroforestry system

According to the Food and Agriculture Organization, agriculture is responsible for 14 percent of global greenhouse gas emissions. Agriculture suffers the impacts of climate change, but “agriculture has the potential to be an important part of the solution, through mitigation” (FAO, 2012). The Intergovernmental Panel on Climate Change (IPCC) has estimated that soil carbon sequestration offers the greatest potential in agriculture for climate change mitigation. Agroforestry systems are known to sequester carbon into aerial biomass, but they could also play an important role by enhancing carbon inputs into deep soil horizons. Agroforestry trees have often a deep rooting system, due to the competition with the annual crop and to the plasticity of their roots (Mulia and Dupraz, 2006). Carbon inputs from roots (root turnover, root exudates) may constitute an important part of soil organic carbon as they are more easily stabilized through different processes like physical protection from microbial decomposers through aggregation, physico-chemical protection through the interaction with minerals and chemical resistance (Rasse et al., 2005). However, only a few authors have worked on this topic in temperate agroforestry systems (Oelbermann et al., 2004, Peichl et al., 2006), and none considered deep soil profiles. The ECOSFIX project aims at studying ecosystem services of shallow and deep roots such as hydraulic redistribution, carbon sequestration and soil fixation in different ecosystem structures and in a range of climates (Costa Rica, Laos, France). This work focusses on the experimental device that was implemented for this project in an intercropping system at Restinclières, France, established in 1995 (durum wheat and walnut at 13x8m spacing). A huge pit (5m long, 4m deep) was dug into the soil, perpendicularly to a tree row. Fine root impacts (< 2mm in diameter) were counted on each trench walls, cubes of soil (1 dm3 in volume) were sampled at different depth and a predictive model of root length density (RLD) from root intercept counts was established (Maurice et al., 2010). The model was quite good fitted, with a R2 of 0.59. This method allowed us to estimate root density on the whole profile. The rooting profile was surprisingly homogeneous, down to a depth of 4 m and across the cropped alley. Soil cylinders were collected in order to measure carbon content, bulk density, penetrometry and soil carbon fractionation. Roots were sampled at different depths to perform chemical analyses and study root traits. 16 minirhizotrons and temperature sensors were implemented in the pit at different depths and distances from the trees, in order to estimate fine root turnover of walnut trees. Images are collected every three weeks. 3 other 1.5 m deep pits were dug (in agroforestry, agriculture control and forestry control), and litterbags containing fine roots were installed at different depths before the pits were filled up. Chemical composition changes and mass loss will be analyzed after one year to determine the decomposition rate of roots. Root carbon input into the soil was also modeled with the Hi-sAFe model, an agroforestry model developed by INRA. (Texte intégral