Altering rainfall amount and pattern affects above- and belowground productivity differently among three US grassland types

Background/Question/Methods: By the end of the 21^st century, climate models forecast drastic shifts in precipitation amount and pattern (event size and number) globally. These responses will likely be highly variable among ecosystem types, highlighting the need for greater understanding of climate change-ecosystem responses among sites spanning regional biotic and abiotic gradients. Grasslands are experimentally tractable ecosystems that span much of the central US. We chose this biome to address the overarching question: How sensitive are different grassland types to changes in precipitation amount and pattern? During the 2011 and 2012 growing seasons, we experimentally augmented natural rainfall inputs into three different grassland types in two ways:  1) by adding small events weekly throughout the growing season (MS - many small treatment), and 2) by adding water on top of large ambient rainstorms a few times during the growing season (FL- few large treatment). Both treatments received the same total precipitation, and responses (above and belowground net productivity; ANPP and BNPP, respectively) were measured for comparison with ambient plots, in which no water was added.  The three grasslands selected differed in plant community composition, C₃ vs. C₄grass dominance, and annual precipitation amounts (321-840 mm). They were located at: the Central Plains Experimental Range near Nunn, CO (shortgrass steppe), the Fort Keogh Agricultural Research Station near Miles City, MT (northern mixed grass prairie), and the Konza Prairie Biological Station near Manhattan, KS (tallgrass prairie).

Results/Conclusions: ANPP was significantly increased by water addition in the shortgrass steppe and tallgrass prairie while only in the shortgrass steppe did precipitation pattern matter – the FL treatment increased productivity relative to ambient plots, but the MS treatment did not. Similarly, BNPP responses to rainfall amount also occurred in both shortgrass steppe and tallgrass prairie. Here, the MS treatment increased BNPP only in the shortgrass steppe whereas the FL treatment increased production only in tallgrass prairie relative to ambient. Total productivity (ANPP + BNPP) responded to precipitation amount in both shortgrass steppe and tallgrass prairie, with neither pattern of addition having an effect on total productivity.  In contrast to these two C₄ dominated grasslands, the C₃ dominated northern mixed grass prairie did not respond to precipitation amount or pattern in either year.  This suggests that the cooler northern mixed grass prairie may rely more on non-growing season precipitation inputs and thus that there is substantial variability in central US grassland sensitivity to forecast changes in precipitation regimes.