Remotely sensed assessment of increasing chronic and episodic drought effects on a Costa Rican tropical dry forest

Abstract Tropical dry forests (TDFs) have experienced pronounced droughts and increased temperatures for the last century. To assess whether these climatic shifts have influenced dry forest vegetation and ecosystem functioning, we integrated ground observations from a Costa Rican long‐term forest dynamics monitoring plot with remotely sensed measures of forest productivity and canopy functioning from a diverse set of satellite data. Previously reported long‐term climate data (1921–1997) show a reduction in annual rainfall, but since 1980 there has been no directional change in mean annual precipitation. However, the 2015 El Niño Southern Oscillation (ENSO)‐induced drought was unprecedented. Temperatures have increased by 1.1°C since 1931. However, the Landsat wet season (1987–2017) Enhanced Vegetation Index (EVI) (canopy greenness) and the dry season (1985–2017) fraction of non‐photosynthetic canopy cover all indicate that TDFs have become more deciduous but also more productive during the wet season. These changes are consistent with a shift in the functional composition observed in the long‐term plot as more drought‐deciduous tree species have increased in abundance. Nonetheless, more continuous 16‐d MODIS (the Moderate Resolution Imaging Spectroradiometer) measures of the EVI over the past 17 yr (2001–2017) showed no change in the total annual forest productivity. Further, while the 2015 ENSO event temporarily reduced forest EVI, it did not cause a longer‐term impact on forest productivity. Instead, high spatial resolution Worldview‐2 satellite imagery showed that forest phenology shifted in the subsequent years even though the region returned to normal precipitation. Our results indicate that while the species composition of TDFs may be sensitive to the long‐term trend of gradually increasing temperatures and aridity, the annual forest functioning has so far been resilient to long‐term drying and a large episodic extreme drought event. This study demonstrates the feasibility of synthesizing satellite images of different characteristics to study the vegetation dynamics of a long‐term forest dynamics plot. Our synoptically sensed results show that the longer‐term changing climate has been and is currently shifting the ecological functioning, and also provide a baseline to assess the impacts of an extreme drought year on TDFs.