Applying survival analysis to a large-scale forest inventory for assessment of tree mortality in Minnesota

Tree mortality has traditionally been assessed in forest inventories through summaries of mortality by location, species, and causal agents. Although these methods have historically constituted the majority of tree mortality summarizations, they have had limited use in assessing mortality trends and dynamics. This study proposed a novel method of applying survival analysis for the purpose of analyzing tree mortality in forest inventories. Individual tree size and growth increments were used to estimate survival and hazard functions for a forest inventory for the state of Minnesota. These estimates provided regional mortality and variance estimates by diameter at breast height (DBH) and diameter growth classes (ΔDBH) between successive inventories. Comparisons of survival/hazard curves for various tree populations and tests of effects of covariates on individual survival curves were conducted allowing for mortality hypothesis testing across user-defined tree populations (i.e., species, location, stand conditions, and damage agents). Survival analysis techniques, facilitated by the variables of DBH and ΔDBH, may provide foresters with the ability to test tree mortality hypotheses and summarize regional tree mortality trends.