The applicant's objective is to learn and develop new methods, especially hyperspectral remote sensing, for integrating plant functional diversity changes into landscape-scale carbon cycle models. Climate and land-use changes affect patterns of plant functional diversity, which in turns are strong drivers of the carbon cycle. Plant functional diversity allows to consider carbon cycle-related processes operating on large time and spatial scales - from local and transient ecosystem response to long-term impact of vegetation changes on carbon budgets. Here, we propose to explicitly incorporate temporal and spatial patterns of plant functional diversity into landscape-scale terrestrial-carbon-cycle models using hyperspectral remotely sensed data. As a first step, remote sensing will be used for the mapping and landscape-modelling of key functional traits ¿ canopy nitrogen and water content, light-use efficiency etc ¿ known to determine major carbon-cycle related processes. As a second step, we will integrate these distribution models of functional diversity into terrestrial biogeochemical models with the aim to extend local measurements of CO2 fluxes both spatially and temporally. The project will be conducted in two contrasted Australian landscapes - tropical savannas and evergreen Eucalyptus forests - for which ecosystem properties and continuous on-ground measurements of CO2 fluxes are available. This project aims at coupling the applicant's expertise in plant functional ecology with biogeochemistry, climatology and remote sensing. The competencies acquired in a world leading place for carbon cycle science - CSIRO, Australia - will open new research avenues for the applicant when returning to his host institution ¿ University of Grenoble I, France. This project meets the concern of the European Community environmental policies relating to the understanding and forecasting of the carbon cycle, including its feedbacks with human components.