Abstract: Semiarid grasslands in the USA Southwest are experiencing widespread transitions into shrublands with far-reaching consequences for the ecosystem services provided by the region. Prior research has uncovered many factors driving the transitions, including increased precipitation variability and overgrazing by cattle. To evaluate the relative strength of precipitation, defoliation, and their potential interactions on semiarid grass cover, we conducted a field manipulation experiment in the Jornada Basin Long-Term Ecological Research site in New Mexico, USA. We utilized an automatic rainfall manipulation system (ARMS) to create 3 levels of precipitation (ambient, ambient plus 80%, and ambient reduced by 80%) and 2 levels of grass defoliation (unclipped and clipped every 2 weeks). We monitored responses of Bouteloua eriopoda, a perennial native prairie grass, to our manipulations. B. eriopoda cover was assessed at the end of the growing seasons using the line intercept method. Precipitation and other climate data were recorded in nearby meteorological stations. B. eriopoda cover responses to precipitation and clipping were analyzed using pseudo F-ratios derived from PERMANOVA. To account for pre-existing levels of grass cover, relative changes in B. eriopoda cover in plots between years (percent changes, which divides the difference in cover between years by the initial value) were analyzed.
B. eriopoda mean cover peaked in the unclipped treatment at 44% under 220 mm of growing season precipitation and in the clipped treatment at 20% under 377 mm of growing season precipitation. Forty percent of the plots experienced declines in B. eriopoda cover, but the majority experienced increases. Cover was positively correlated with growing season precipitation (r = .35) and negatively correlated with defoliation (r = -.49). Both precipitation and defoliation significantly impacted relative change in B. eriopoda cover (P < .001 for precipitation, P = .031 for defoliation), and an interaction between the treatments was found (P = .028, F = 3.5), with the mean response of B. eriopoda cover to growing season precipitation being 55% greater in clipped plots than in unclipped. Spatial effects from plots were significant (P = .022), indicating that other environmental variables, such as animal activity or vegetation patch dynamics, also affected B. eriopoda cover. Our results suggest that precipitation variability and grazing will continue to impact B. eriopoda cover, but some negative effects of defoliation may be offset by increased water availability. Future research should include the analysis of drought duration as well as intensity.