Testing the efficacy of tree-ring methods for detecting past disturbances

Radial-growth averaging and absolute-increase methods had lower levels of type I and II error in detecting disturbance events with our datasets. Parameter settings play a key role in the accuracy of reconstructing disturbance history regardless of the method. Time-series and radial-growth averaging methods require the least amount of a priori information, but only the time-series method quantified the subsequent growth increment related to a reduction in competition. Finally, we recommend yearly binning of releases using a kernel density estimation function to identify local maxima indicating disturbance. Kernel density estimation improves reconstructions of forest history and, thus, will further our understanding of past forest dynamics.