Assessing Role of Drought Indices in Anticipating Pine Decline in the Sierra Nevada, CA

Tree mortality in Sierra Nevada's 2012-2015 drought was unexpectedly excessive: similar to 152 million trees died. The relative performance of five drought indices (DIs: SPEI, AI, PDSI, scPDSI, and PHDI) was evaluated in the complex, upland terrain which supports the forest and supplies 60% of Californian water use. We tested the relative performance of DIs parameterized with on-site and modeled (PRISM) meteorology using streamflow (linear correlation), and modeled forest stand NDVI and tree basal area increment (BAI) with current and lagged year DI. For BAI, additional co-variates that could modify tree response to the environment were included (crown vigor, point-in-time rate of bole growth, and tree to tree competition). On-site and modeled parameterizations of DIs were strongly correlated (0.9), but modeled parameterizations overestimated water availability. Current year DIs were well correlated (0.7-0.9) with streamflow, with physics-based DIs performing better than pedologically-based DIs. DIs were poorly correlated (0.2-0.3) to forest stand NDVI in these variable-density, pine-dominated forests. Current and prior year DIs were significant covariates in the model for BAI but accounted for little of the variation in the model. In this ecosystem where trees shift seasonally between near-surface to regolithic water, DIs were poorly suited for anticipating the observed tree decline.