Carbon and nutrient balance are essential components of yield determinism in oil palm agrosystems and as such is crucial to optimize cropping practices and monitor fertilisation for sustainable production. Carbon balance at the plantation scale has received much attention, in particular carbon sequestration in global change context and crop modelling. My work provided new estimates of roots biomass, respiration loss (autotrophic and heterotrophic at tree scale) net and gross primary production (NPP and GPP) from planting to adult stage. My work further considered spatial variability inherited from cropping practices and temperature increase (global warming) in Indonesia (North Sumatra) and North Benin. In both countries, I have demonstrated that efficient recycling of fronds (dead leaves) is a key for agrosystem sustainability. I also have undertaken crop modelling (carried out as in studies from Wageningen) to look for new estimates of NP (net photosynthesis) and maximal stomatal conductance in potential ecological conditions (i.e. North Sumatra with no water deficit), coupled with phenology to predict (with two determinist models, SIMPALM and PHENOPALM) carbon allocation to fruits. In order to understand further mechanisms controlling C allocation I took advantage of natural isotope 13C abundance use to follow photosynthetic products (soluble sugars, starch) from leaves to fruits and identify C reserve pools. A metabolic model at tree scale was built from this study to estimate allocation fluxes. Because it is well-known that potassium (K) is essential for phloem transport and thus sugar circulation, I then explored links between carbon metabolism and K availability with metabolomics, and analyses of yield and leaf K content and the relationship between them. This work has shown that K can circulate quickly within the tree, supporting sucrose transport from leaves to fruits, while surprisingly, glucose is a major sugar that declines along with starch when yield in