Scale-dependent controls on forest carbon uptake across hydroclimatic extremes

Laura Rez , Timo Vesala, Pasi Kolari, Eli Tziperman, Eyal Rotenberg, Rachamim Rubin, Dan Yakir

Conifer forests span some of the most climatically contrasting environments on Earth, from energy-limited Boreal systems to water-limited semi-arid ecosystems. Whether their carbon uptake is governed by universal drivers or by site-specific boundary conditions remains unresolved. Using more than two decades of eddy-covariance and multi-depth soil moisture measurements from two climatic end-members of evergreen needleleaf forests, Hyytiälä, Finland (Boreal) and Yatir, Israel (semi-arid Mediterranean), we isolate the ecophysiological controls on carbon uptake by restricting the analysis to photosynthetically active radiation (PAR)-saturated conditions and explicitly separating seasonal dynamics from daily residual variability. For each dataset, we apply Random Forest modelling (tested against baseline Generalized Linear & Additive Models) with SHAP analysis to identify dominant drivers and environmental thresholds.

At the seasonal scale, NEP was governed by distinct boundary conditions in each forest. In Hyytiälä, precipitation had dominant control, sustaining evapotranspiration and reflecting a radiation regime characterized by high diffuse fractions. In Yatir, deep soil water availability controlled both the timing and magnitude of productivity, with a critical threshold at ∼15.9 %vol in the deepest measured layer (∼45 cm). This threshold marked the transition from dry-season legacy constraints, where vertical soil potential gradients limit root uptake, to conditions permitting sustained transpiration and productivity.

At the daily residual scale, both forests showed strong sensitivity to shortwave radiation and vapour pressure deficit (VPD), despite contrasting climatologies. Elevated VPD reduced peak daily productivity by more than 50% at both sites, although sufficient deep soil moisture mitigated this effect in the semi-arid forest. Notably, both forests exhibited a similar optimal air temperature range (14–20 °C), indicating a conserved physiological optimum despite divergent hydroclimatic limitations.

Our results demonstrate that carbon uptake in evergreen needleleaf forests is structured by site-specific hydrological boundary conditions at seasonal scales, while atmospheric stress regulates short-term variability. This scale-dependent hierarchy of controls challenges the notion of universal productivity drivers and highlights the importance of subsurface hydrology and rainfall regime characteristics for predicting forest carbon responses to climate change.