Contrasting plant-functional type responses to increased interannual precipitation variability
Climate change will result not only in changes in mean precipitation but also in its variability. IPCC scenarios project dry regions to get dryer and wet regions to get wetter with high uncertainty. However, there is general agreement among models projecting increases in the frequency of extreme events. Consequences of increased precipitation variability for the functioning of ecosystems have received considerable less attention than assessments of the effects of changes in mean precipitation and temperature. Here, we assess the effect of increased precipitation variability on primary productivity and its stability. In order to address this objective we combined approaches at the local and global scale. Locally, we use a six year-long manipulative experiment where we increased interannual precipitation variability by applying sequences of wet and dry years. Globally, we analyzed long-term data including 92 datasets from 37 sites worldwide. These two approaches complement each other; the manipulative experiment allows for causation testing but it is limited in its spatial scale. The analysis of long term data allows for comparison among sites located worldwide and sequences of years with low and high precipitation variability but does not test for causation.
At the local scale, plant-functional types showed contrasting responses and different growth strategies. Perennial-grass productivity significantly decreased (F(1,48) = 33.85, P < 0.001) while shrub productivity increased (F(1,48) = 5.50, P = 0.02) as a result of opposite non-linear responses to precipitation amount. Rare species showed no response to interannual precipitation variability (F(1,48) = 2.15, P = 0.15) supported by a strong linear relationship between annual species productivity and growing season precipitation. Overall ecosystem response resulted from the aggregated response of plant-functional types with contrasting strategies that contributed to increase ecosystem stability. At the global scale, our results indicate that there is significant effects of precipitation amount (t(1254) = 16.42, P < 0.0001) and variability (t(1254) = 4.22, P < 0.001) on ANPP. However, we found a significant interaction effect between precipitation coefficient of variation and site mean annual precipitation (t(1254) = -4.86, P < 0.0001) with an inflection point around 340 mm mean annual precipitation. Therefore, increased interannual precipitation variability had a positive effect on ANPP in sites where mean annual precipitation is lower than 340 mm and a negative effect on sites where the mean annual precipitation is higher than 340 mm. Ecosystem stability was also different in sites above and below the interaction threshold.