Evaluating spatial patterns of drought-induced tree mortality in a coastal California pine forest

In a coastal, fog-influenced forest on Santa Cruz Island in southern California, we observed mortality of Bishop pine (Pinus muricata D.Don) trees following a brief (2 year), yet intense, drought. While anecdotal evidence indicates that drought-induced Bishop pine mortality has occurred in the past in the stand we studied, this is the first attempt to capture the spatial distribution of mortality, and begin to understand the environmental drivers underlying these events. We used high spatial resolution remote sensing data to quantify the spatial extent of tree mortality using a 1 m true color aerial photograph and a I m LiDAR digital elevation model. We found the highest density of dead trees in the drier, more inland margins of the forest stand. We used the Random Forest decision tree algorithm to test which environmental variables (e.g., summertime cloud frequency, solar insolation, and geomorphic attributes) would best separate live and dead tree populations. We also included tree height as a variable in our analysis, which we used as a proxy for overall tree size and potential rooting distribution. Based on the Random Forest analysis, we generated a map of the probability of survival. We found tree survivorship after drought was best explained by the frequency of summertime clouds, elevation, and tree height. Specifically, survivorship was greatest for larger trees (similar to 8-10 m tall) in more foggy parts of the stand located at moderate elevation. We found that probability of survival was lowest at the inland extent of the stand where trees occur at the upper limit of their elevation range (similar to 400 m). The coexistence of these main factors with other landscape variables help identify areas of suitable habitat for Bishop pines across the stand, and extend our understanding of this species' distribution. (C) 2013 Elsevier B.V. All rights reserved.