Across large geographic gradients, variability in aboveground net primary production (ANPP) can be explained largely by mean annual precipitation (MAP). Within a site, however, annual rainfall amount explains less temporal variability in ANPP in part because of unpredictable precipitation patterns (i.e. rainfall event size and timing) that differ from year to year. In semi-arid grasslands, for example, experiments that contrast fewer, larger rainfall events vs. many small events indicate that large event size can increase ANPP. However, rainfall patterns are typically imposed in an even pattern, which fails to capture intra-annual precipitation stochasticity. The aim of this study was to assess the importance of even vs. variable patterns of rainfall within a growing season. Rainfall exclusion shelters were constructed in a shortgrass prairie in Colorado and irrigation treatments were imposed to replicate the naturally stochastic rainfall pattern of a past year (2005) with average precipitation amount and ANPP. Adjacent plots received the same amount of precipitation yet the pattern of precipitation (either event size, event timing, or both) was altered to create perfectly even patterns.
Results/Conclusions:
Our experimental manipulations resulted in four treatments: historically stochastic (HIST), even event size (EES), even event timing (EET), and even event size and timing (EEST). Annual coefficient of variation (CV) in soil moisture reflected the evenness of precipitation inputs for each treatment with the highest CV of soil moisture in the HIST treatment. ANPP was significantly lower in the HIST treatment compared to the EEST treatment, with event timing having a greater effect than event size on ANPP. Thus, large rainfall events may only increase ANPP when dispersed evenly throughout the growing season. Physiological (e.g. photosynthesis and leaf water potential) and phenological (e.g. growth rate) measurements of the dominant plant species (Bouteoula gracilis) indicate higher plant stress with increased rainfall variability. Below-ground processes, such as soil respiration, were more responsive to rainfall event size than timing. An understanding of how ecosystems respond to rainfall event size and timing is important given that years with fewer, but larger rainfall events are expected to become more common with climate change. When implementing rainfall manipulation experiments, however, it is important to consider and incorporate the natural stochasticity of precipitation within the ecosystem of interest and avoid including a hidden evenness treatment.