Insensitivity of ecosystem productivity to predicted changes in fine‐scale rainfall variability

Changes in rainfall associated with climate change are expected to affect the tightly coupled water‐carbon ecosystem dynamics. Here, we study the effects of altered rainfall at 33 sites in North America, as projected by the high‐resolution/high‐fidelity (∼4 km, 1 hr) continental‐wide Weather Research Forecasting (WRF) convection‐permitting model under a high‐emission scenario (RCP 8.5). We make use of a stochastic weather generator to extend WRF outputs, accounting for natural variability and simultaneously separate the changes in total rainfall, its seasonality, and its intraseasonal pattern. We used these rainfall scenarios to study ecosystem responses with the state‐of‐the‐art Tethys‐Chloris terrestrial biosphere model. Model simulations suggest that increases in mean annual rainfall dominate ecosystem responses at dry sites, while wet sites are less sensitive to rainfall changes. Sites of intermediate wetness face reductions in productivity, due to reduced growing season rainfall and increased water losses under altered seasonality, which outpace any possible benefits induced by increases in mean annual totals. Changes in the fine‐scale temporal structure of rainfall have an insignificant impact on ecosystem productivity and only alter hydrological dynamics, contradicting expectations based on some field experiments, which, however, are not tailored to directly quantify climate change impacts, but rather to understand the mechanisms leading to ecosystem responses. We further demonstrate how approaches following the “fewer but larger rainfall events” concept might exacerbate ecosystem responses.