For the majority of North American grasslands, climate change models forecast an increase in frequency and severity of growing season droughts by the end of the century. In grasslands, temporal stability in biomass production is maintained via greater spatial variability across local communities and lower spatial variability among local communities. In grassland landscapes where grazing is driven by fire (termed pyric herbivory), local communities occur as discrete patches that differ in structure and composition with time since focal disturbance. Temporal stability in biomass production therefore increases as a result of patch-contrast created via pyric herbivory. However, the robustness of the relationship between spatial variability and temporal stability has not been assessed under weather conditions consistent with climate change. In this study, we test whether temporal stability in plant biomass production continues to coincide with specific scales of spatial variability following one of the most severe drought events on modern record for the Great Plains. We conducted our study in three replicate experimental landscapes managed with fire and grazing for the past 15 years. We measured plant biomass and crude protein during 2011 and 2012 and compared it to data collected at the same site before the drought in 2009.
Drought reduced temporal stability in aboveground plant biomass production and switched the scales at which patterns of spatial variability manifested in this experimental landscape. Plant biomass was less predictable during drought and exhibited a highly oscillatory relationship with time since fire. These patterns contrast with the logistic growth of biomass that occurred with times since fire in the absence of drought. Consistent with predictions, reduced temporal stability during the drought corresponded with a significant decrease in variability in plant biomass among patches and a significant increase in variability within patches compared to pre-drought levels. NMDS ordination results revealed a shift in this experimental landscape from spatially distinct patches pre-drought to a convergence of patches during drought. Heterogeneity that manifested at the patch-scale before the drought became indistinguishable from patterns of variability within each patch, resulting in a more uniform landscape. The results of this study support an overarching hypothesis in global change research that climate change will cause ecosystem structure and function to converge at large scales. Further, complexity in the scales at which variability manifests during extreme events reinforces the importance of scale in the study of pattern-process relationships.