Remotely Sensed Tree Mortality Rates in Mesic Forests of the US Southwest During an Extended Drought

Abstract Persistent drought since the early 2000s has increased forest vulnerability to disturbances such as insect outbreaks, wildfire, and climate-induced tree die-off events across the western United States (US). Mesic forests of the US Southwest, characterized by geographic and climatic isolation and rich biodiversity, face heightened risks due to limited connectivity with similar forest ecosystems, restricted population migration potential, and upslope range contraction of high-elevation tree species. We manually interpreted high-resolution aerial imagery from 1076 forest stands to quantify recent tree mortality due to insect outbreaks and drought in isolated, mesic forests of the Southwest, then combined these data with Landsat time series and Random Forest (RF) modeling to create a 30-m map of cumulative mortality (i.e., percentage of initial canopy cover that died) from 1997 to 2023. Consistent with prior research, changes in shortwave infrared reflectance (e.g., 1.55–1.75 μm) were the best predictors of tree mortality. Overall mortality rates (0.4 % canopy cover year -1) were comparable to field-derived rates of background mortality for large trees throughout the western US (i.e., 0.5 – 2 % year -1). Low mortality (< 30 %; < 1 % year -1) occurred in 94 % of the total area, while higher mortality (> 80 %; > 2.5 % year -1) occurred in fragmented, high-elevation patches. Notably, the RF model had low-to-moderate accuracy (R² = 0.28, RMSE = 18 % mortality), and some forest management and fire occurrences were difficult to omit from disturbance maps using existing data, highlighting challenges in capturing low-severity disturbance effects using Landsat imagery. Since ca. 2000, wildfire may have been a larger driver of change in mesic forest systems of the US Southwest than drought-caused tree die-off and background mortality. This study emphasizes the need for continued monitoring of background tree mortality rates and provides insights into monitoring change in other vulnerable, isolated forest ecosystems.