Forest disturbance in China from 2000 to 2005: models of trend and variance in size-frequency distributions

Disturbances—including natural wind-throws, insect irruptions, and disease outbreaks, as well as anthropogenic logging—alter the physical and ecological properties of forest ecosystems. The size and frequency of disturbances affect the structural and functional characteristics of forests, and are important variables in ecosystem modeling. However, the relationships of disturbance size and frequency to their underlying mechanisms and processes are still not adequately understood, partly due to the challenges of quantitatively characterizing the patterns of events over large areas. Here, we present a hierarchical method to model the size-frequency distribution of forest disturbances using a power law to model the regional trend and gamma distributions to characterize the distribution of patch sizes around the trend. The method was tested for administrative regions in China that experienced substantial forest disturbance between 2000 and 2005. The hierarchical method represented the regional frequency distribution more precisely than simple ordinary least squares and maximum-likelihood estimators due to its accommodation of non-constant, heteroscedastic variance around the trend. The improved model fit across the full range of disturbance sizes in turn increased the accuracy of estimates of total disturbance area. The method achieved errors of <10% of total disturbance area when >30% of all disturbance patches were sampled. Forest disturbances from 2000 to 2005 over entire China were dominated (>50%) by disturbances smaller than 10 ha and large disturbances (>10 ha) were well controlled. Dominant disturbance events in the central and southern forests were smaller overall than those in the northern forests. We recommend including both the regional power law as well as its variation in ecosystem models in order to accurately describe the size-frequency distribution and to improve the accuracy of estimates of the total area of forest disturbances.