Quantifying tree mortality risk and spatial pattern in a temperate conifer forest.

摘要

Forests today face a suite of novel stressors, varying from the introduction of exotic pests to a changing climate. If we are to understand how forests will change, we must understand how the trees within them die and how those deaths affect future forest structure. Tree death, however, is a complex process that can involve a multitude of interacting factors. Furthermore, tree mortality is often an infrequent event, making robust datasets quite rare.; In this work, robust datasets from the Sierra Nevada of California were used: to generate improvements in the assessment of growth-related mortality risk; to assess the importance of spatial elements of mortality risk; to quantify the effect of mortality on forest structure; and to test the generality of mortality models across site and species. First, an information theoretic approach was used to select growth-mortality models that included indices derived from tree-ring data (average growth, growth trend, counts of abrupt growth declines) over varying time intervals. These more cumulative indices were found to improve the assessment of mortality risk over the frequently used average recent growth alone. Secondly, the importance of spatial elements of mortality risk was assessed by quantifying improvement of models with the addition of spatial indices, such as competition index or proximity to disease vectors. These spatial indices were found to be important for assessing mortality risk, even after growth rate and size had been taken into account. Thirdly, second order spatial analyses were performed to quantify the effect of mortality and recruitment on future forest structure. These analyses demonstrated that density-dependent mortality did not always dominate changes in structure and that both recruitment and other forms of mortality must be considered. Finally, the generality of growth models developed in the first analysis was tested by transferring these models to a new site and to a different species. This transfer demonstrated that models developed with a more comprehensive use of the growth record often showed increased generality. In addition, knowledge about how measures of mortality risk can vary with site was critical for adequately assessing the generality of a given model.