Modelling comparison to evaluate the importance of phenology and spring frost damage for the effects of climate change on growth of mixed temperate-zone deciduous forests

The importance of 3 phenological types of deciduous tree, and the effects of the occurrence of frost damage on growth of mixed-species forests, were evaluated using the models FORGRO and HYBRID, The climate change scenarios used were a doubling of the CO2 concentration (700 mu mol mol(-1)) and an increase in temperature ranging from 0 to 7 degrees C. Both FORGRO and HYBRID are mechanistic models treating eco-physiological processes in detail. FORGRO highlights potential growth in managed forests where all individuals of one species are of the same age and size, whereas HYBRID highlights growth in natural forests, including regeneration and mortality of individual trees that differ in age and size. Furthermore, the importance of inaccurate prediction of phenological events and frost hardiness for growth in mixed-species stands was evaluated by comparing dynamic models to regression models. The dynamic models predict the timing of phenological events annually and the progression of frost hardiness during dormancy, whereas the regression models represent empirical relationships between the change in the average date of phenological events with a rise in mean winter temperature and the level of frost hardiness at the moment of leaf unfolding. The results of the climate change scenarios indicate for both FORGRO and HYBRID that: (1) the differences in net primary production (NPP) of the 3 phenological types considered are enhanced when grown in a mixed-species stand compared to a monospecies stand; and (2) the effects of frost damage on growth are more prominent in mixed-species stands than in monospecies stands. Regarding the accuracy of the dynamic approach compared to the regression approach for predicting the timing of leaf unfolding and spring frost damage, the dynamic approach for leaf unfolding results in a similar response of NPP to the regression approach, both for the monospecies and the mixed-species situation. The dynamic approach, however, yields larger differences in the NPP between the phenological types because the model predicts a greater advancement of leaf unfolding than does the regression approach. Comparing the regression approach to the dynamic approach with regard to frost hardiness, the regression approach shows a greater frequency of frost damage; because, according to the dynamic approach the minimum level of frost hardiness is attained after the date of leaf unfolding, thus reducing this frequency.