ANPP-precipitation relationships in multi-year drought experiments in natural ecosystems
Predicting the effects of a reduction in precipitations on ecosystem productivity confronts an uncertainty: the relationship between aboveground net primary productivity (ANPP) and precipitation differs if the focus is spatial, driven by the climatic mean annual precipitation (MAP) variability among sites, or temporal, driven by the interannual variability in annual precipitation (AP) within sites. In observational studies, the slope of the spatial fit (the ANPP-MAP relationship) is much stronger than the slope of the temporal fit (the ANPP-AP relationship) in most of ecosystems, indicating higher sensitivity to climate versus weather. The spatial fit reflects the variability in ecosystem states and incorporates the long-lasting effects of structural and functional traits that constrain the ANPP of ecosystems. The temporal fits reflect the reversible adjustments in plant physiology and the transient changes in ecosystems structure and functioning that can recover within one or two years. We analyzed results from eleven multi-year (4 or more years) rainfall reduction experiments to determine whether ANPP responded to drying treatments either differently or as expected from the temporal relationship in the non-manipulated treatment. According to the spatial vs temporal framework, we expect that, should the drying treatment induce a response to precipitation different from the responses under the non-manipulated treatment as described by the temporal fit under control conditions, it would be reflected, according to the spatial fit, by a reduction in the intercept of the temporal fit that would indicate a change in the state of the ecosystem.
Results show that none of the elevenexperiments presented a lower intercept under the drying treatment, providing no evidence for changes in ecosystem state and indicating that, at annual timescale, effects of reducing experimentally the precipitation are not different from the effects of natural variability in precipitation within the AP ranges covered by the experiments. Past manipulation experiments do not unravel the thresholds beyond which temporal fits under non-manipulated conditions are inadequate. Knowing such thresholds will improve the benchmarks for the validation of productivity in carbon cycle models. Therefore, future experiments should include stronger treatments aiming to cross the precipitation thresholds and disrupt the current ANPP-AP relationships. These treatments may be unrealistic in terms of climate change but will display the mechanisms that contribute to regime shifts in ecosystems.