Nonlinear responses of desert grassland NPP to climate mean and variance

Background/Question/Methods A strong relationship between aboveground net primary production (NPP) and mean annual precipitation (MAP) has been documented along extensive spatial gradients in grassland biomes worldwide. This implies that MAP strongly controls NPP spatially, yet the relationship between annual precipitation (PPT) and NPP within a site over time is often quite weak. Thus, factors in addition to PPT must regulate the local temporal dynamics of NPP. Seasonal PPT or legacy effects improve this relationship, but considerable temporal variation in NPP still remains to be explained. In many grassland regions, climate models predict increased mean annual temperature (MAT) and precipitation variability in the future. Although some have reported that MAT has limited impact on NPP, higher MAT may interact non-linearly with changes in PPT amount or variability to alter the temporal dynamics of NPP.  Moreover, indices such as the Standardized Precipitation Evaporation Index (SPEI) explicitly integrate temperature and precipitation. We used non-linear mixed effects models on 17 years of data from native, ungrazed desert grassland and shrubland communities in central New Mexico (Sevilleta) to model the non-linear responses of NPP to temperature and precipitation variability, and ask how these responses might drive grassland dynamics and shrub encroachment.

Results/Conclusions

Our system spans ecotones from blue grama- to black grama-dominated grassland, and black grama grassland to creosotebush shrubland. Since 1989, SPEI has declined (r=-0.38) at Sevilleta while interannual variance in SPEI has increased (r=0.50). Over time, black grama increased in abundance relative to blue grama, and creosotebush has encroached into black grama grassland, a pattern consistent with increasingly negative (more arid) SPEI. We found significant non-linear responses of grasslands to annual precipitation and temperature, indicative of sensitivity to interannual variance. Essentially, NPP of black grama grassland benefits from increased variance under low SPEI (p=0.019) whereas blue grama NPP declines under increased variance at low SPEI (p=0.0001). Specifically, NPP of black grama grassland benefited from increased precipitation variability during warmer years and did poorly under high variance in cooler years. The opposite pattern occurred for blue grama. Creosotebush shrubland was essentially unresponsive to higher variability or changes in summer temperature. These patterns suggest that warmer temperatures when combined with increased precipitation variability will favor black grama replacement of blue grama grassland, likely paving the way for further shrub encroachment. Our results have important implications for how climate change will drive temporal dynamics and impact the carbon budget of aridland ecosystems.