Greater sensitivity to hotter droughts underlies juniper dieback and mortality in Mediterranean shrublands.

Drought-induced dieback episodes have been globally reported. However, few studies have jointly examined the role played by drought on growth of co-occurring shrub and tree species showing different dieback and mortality. Here, we focused on dieback events affecting Mediterranean shrublands dominated by the Phoenician juniper (Juniperus phoenicea) since the middle 2000s in three sites across a wide geographical and climatic gradient in Spain. We compared their growth responses to climate and drought with coexisting tree species (Pinus pinea, Pinus pinaster and Quercus faginea), which did not show dieback in response to drought. We characterized the major climatic constraints of radial growth for trees, surviving and dead junipers by quantifying climate-growth relationships. Then, we simulated growth responses to temperature and soil moisture using the process-based VS-Lite growth model. Growth of shrubs and trees was strongly reduced during extreme droughts but the highest negative growth responsiveness to climate and drought was observed in trees followed by dead junipers from the most xeric and cold sites. Growth of dead junipers responded more negatively to droughts prior to the dieback than co-occurring, living junipers. Growth was particularly depressed in the dead junipers from the warmest site after the warm and dry 1990s. The growth model showed how a steep precipitation reduction in the 1980s triggered soil moisture limitation at the driest sites, affecting growth, particularly in the case of dead junipers and mainly in warm and dry sites. The asynchrony in the simulated seasonal timing of drought events caused contrasting effects on growth of co-occurring shrubs and tree species, compromising their future coexistence. Junipers were particularly vulnerable to hotter droughts during the early growing season. The presented projections indicate that de-shrubification events in response to hotter droughts will be common but conditioned by site conditions. Our modelling approach provides tools to evaluate vulnerability thresholds of growth under similar drought-induced dieback and mortality processes.