[Differences of driving factors for carbon sink changes in different carbon pools of Larix gmelinii natural forests].

Carbon balance of the tree layer in natural forests is affected by three carbon pools: tree growth, morta-lity, and recruitment. However, the dynamics of the sink of each carbon pool and the driving factors are still unclear. To this end, we used stepwise regression method and structural equation model to assess the effects of biotic (stand and diversity) and abiotic (soil, topography and climate) factors on three dynamic processes of carbon sinks, namely, stand growth, recruitment and mortality, in the natural forests of Larix gleminii, based on the data from the seventh and eighth national continuous forest inventory of the Greater Khingan Mountains. The results showed that the average growth, recruitment and mortality of L. gleminii stand were 0.73, 0.17 and 0.41 t·hm^-2·a^-1. The growth of carbon sink in forest stands increased with the slope aspect index (the total path coefficient β=0.135), stand basal area (β=0.132) and slope position (β=0.085), but decreased with the increases of average stand age (β=-0.225) and altitude (β=-0.196). Recruitment was negatively correlated with stand basal area (β=-0.368), average stand age (β=-0.240), soil pH (β=-0.184), soil total phosphorus (β=-0.115) and tree species richness (β=-0.012), and was positively correlated with soil total nitrogen (β=0.258). The larger important factors affecting mortality were the average tree height (β=0.276) > Simpson index (β=0.232) > soil total phosphorus (β=0.195) > stand basal area (β=0.154) > soil pH (β=0.084). Stand age, basal area and average tree height were the most important factors affecting stand carbon sink growth, recruitment and mortality. Topography and soil conditions were also the key to affect stand dynamics. Therefore, the structure and soil conditions of the forest can be adjusted through reasonable thinning in the management process, which in turn promotes the sustained fulfilment of forest carbon sink capacity.