Spatial Patterns of Resilience and Mortality in Semi-arid Riparian Woodlands During Extreme Drought

Abstract:

Riparian woodlands in southern California are adapted to local mesic conditions, and the dominant tree species rely on groundwater to maintain basic physiologic function throughout the summer dry season. Extreme droughts reduce groundwater recharge and can cause the alluvial water table to decline. Reduced groundwater availability can lead to water stress and mortality in riparian tree species. In this analysis, we used remote sensing to monitor the health of riparian woodlands during the 2012-2019 California drought. We examined floodplain woodlands along the Santa Clara River, which supplies water for agriculture, urban development, and the natural riparian ecosystem. Imagery from AVIRIS was used to calibrate a spectral mixing model, which was then used to unmix Landsat imagery from 2011 to 2019. The model estimated the cover of green vegetation, dead/woody vegetation, and soil in each Landsat pixel. The resulting data set indicated changes in land cover across the entire floodplain from 2011 to 2019. The remote sensing data was combined with data from groundwater monitoring wells and field observations to identify the drivers of resilience and mortality in riparian woodlands during the drought. Our analysis revealed complex spatial patterns of resilience versus stress and mortality across individual riparian sites. The largest changes in vegetation cover occurred in the areas that experienced large water table declines. In some areas, the water table declined more than 10 meters, far deeper than the rooting depths of riparian tree species. Other sites were relatively unaffected by the drought because they received water subsidies from agricultural runoff, water treatment plant effluent, or geomorphic features such as bedrock intrusions. The remote sensing analysis also revealed a distinct spatial and temporal pattern of riparian woodland mortality at a landscape scale. Mortality first occurred at arid inland sites that experienced rapid water table declines. Over a period of three years, a wave of mortality moved toward the coast where groundwater declines occurred more slowly and plant water stress was buffered by coastal climatic conditions. Our study demonstrates the value of landscape-scale analyses facilitated by remote sensing for modeling plant-water relations during drought conditions.