Current understanding of patterns and controls of belowground net primary productivity is weaker than that for aboveground net primary productivity. However, belowground productivity accounts for a larger flow of carbon than aboveground productivity in most water-limited ecosystems (deserts to grasslands). Our work explores hypotheses regarding how changes from year to year in water availability at one location affect the fraction of belowground net productivity relative to total production and how this temporal effect of water availability changes across spatial gradients of long-term mean precipitation from desert to humid grasslands. In order to test these hypotheses, we carried out a multi-site field manipulative experiment from a Desert grassland, to the Shortgrass steppe, to the Tallgrass prairie. At each site, we manipulated incoming precipitation with a combination of rainout shelters and automatic irrigation systems achieving five levels of precipitation ranging from 10th percentile to 90th percentile of historic rainfall records. We estimated belowground productivity using ingrowth cores, standing root biomass and minirhizotron images and aboveground productivity through biomass harvests.
Our results show that the fraction of primary productivity allocated to belowground components changed along a precipitation gradient. At the Desert grassland site, both aboveground and belowground net primary production increased with precipitation (F1,38 = 18.31, P < 0.001; F1,38 = 9.58, P = 0.004). The fraction of net primary production allocated belowground did not change with local rainfall manipulations (F1,38 = 0.4, P = 0.53). At the Tallgrass prairie site, aboveground net primary production increased with precipitation (F1,38 = 9.13, P = 0.004) while belowground productivity showed non-significant responses (F1,38 = 1.25, P = 0.27). The fraction of biomass allocated belowground decreased with precipitation (F1,38 = 8.59, P = 0.005) suggesting that competition for other factors different from water may be driving plant allocation patterns at the wettest site of our gradient. At the shortgrass steppe, we found intermediate responses indicating that regional productivity allocation patterns change along precipitation gradients. Our results highlight the importance of within biome differences in biomass allocation. This study contributes to a better understanding of the functioning of grasslands and their ability to sequester carbon.