Biotic and abiotic  heterogeneity show contrasting GPP and growth trends in mediterranean forests

The Mediterranean basin is a well-known drought-prone region, making forest ecosystems potentially vulnerable to drought episodes, heat waves and dry spells. In the last two decades, extreme weather events affected different regions of Europe including mediterranean áreas. This led to significant impacts on forest ecosystems, with extensive mortality events, episodes of crown dieback, and identified reduced tree growth at local level. The predictive abilities to depict early warning signals of negative extreme-induced impacts, well before that the mortality event might occur, are pivotal for monitoring Mediterranean forests. The 3D-CMCC-FEM model, a detailed ecophysiological process-based model, is here applied at gridded level over the Basilicata region, in southern Italy. The model is run on a regular 1x1 km grid for the period 2005-2019 to simulate, among others, gross primary productivity, carbon allocation and tree growth, processes which are controlled by abiotic, e.g. meteorological conditions, and biotic factors, e.g. trees reserve pools, in a mechanistic manner. This modeling approach allows discriminating the degree of decoupling of  carbon assimilation, and tree growth , e.gcarbon woody accumulation. As a result of such interaction between processes and factors, the model  highlighted different emerging patterns of the system under investigation. In particular, results show pronounced differences between European beech dominated areas and oaks dominate areas of the region. Generally, despite a significant drop of summer GPP in beech forests and an overall negative GPP trend, in accordance with remote-sensing based data, the tree growth rate is still positive. Oppositely, the oaks dominated forests show contrasting patterns, with areas where positive trends in GPP can be accompanied by positive but even negative trends in tree growth. The 3D-CMCC-FEM is shown to identify areas which might be prone to statistically significant negative trends in tree growth and, thus, likely to be prone to higher mortality risk in the near future. Indeed, these negative growth trends can not be explained only in terms of forest aging, but also in terms of abiotic factors. Our results show how the diverse degree of coupling between assimilation and growth between different species might be predicted by the model and leading to increase our capability to detect early signals of declining growth, which might already occur in spite of an apparent full recovery after a drought event at canopy level.