Can decision rules simulate carbon allocation for years with contrasting and extreme weather conditions? A case study for three temperate beech forests

The standard DR allocation scheme captured the average annual dynamics of C allocation and wood growth at beech stands with contrasting climate and standing stock. However, the allocation model required high quality GPP input and errors (even modest) in GPP resulted in large errors in the growth of the tree organs lowest in the modelled sink hierarchy (woody organs). The ability of the standard DR allocation model to simulate year-to-year variability was limited. The amended DR allocation scheme improved the annual simulations and allowed capturing the stand growth dynamics at Hesse during the extreme 2003 summer and its important lag effect on next year's wood production. Modelling of drought-induced changes in fine root dynamics and of short-term thermal acclimation of maintenance respiration should not be overlooked when simulating the C cycle of forests, particularly for sites likely to experience extreme drought and heat waves. The most relevant model bias was the inaccurate estimation of leaf biomass production (up to 15%) and a poor description of its interannual variability. Future studies should focus primarily on this limitation.