Semi-arid rangelands have been historically susceptible to persistent ecosystem state changes as a result of rapid environmental change (e.g., conversion by annual Bromus spp., shrub encroachment due to grazing). Climate change is likely to exacerbate these challenges resulting in new, persistent species assemblages. This is especially true in semi-arid rangelands of the American Colorado Plateau where grazing, invasive species, and other ecological factors interact with climate to alter species composition. Understanding stability in species composition and the causes of compositional shifts are important aspects of managing ecological systems and predicting potential future changes in communities. Moreover, understanding and interpreting compositional shifts requires knowledge of patterns and processes across spatial scales. We used a multivariate control chart analysis developed by Anderson and Thompson and extend its application to cross-scale analysis by calculating control limits of compositional shifts at different scales (analogous to alpha, beta, and gamma diversity) to study a decade of annual shifts in species composition of semi-arid rangeland communities of southern Utah in both grazed and ungrazed plots. Control limits were calculated based on bootstrapped samples of previous compositional shifts in species abundance to identify shifts in species composition outside of normal variability. We discuss our results in the context of extending use of this analysis for other rangeland monitoring programs throughout the southwestern U.S.
Results/Conclusions
In southern Utah, control charts detected compositional shifts that were influenced by both grazing and precipitation. Further, grazed plots appeared less compositionally stable than ungrazed plots. These shifts were driven by changes in the abundance of perennial grasses, exotic annual plants, and shrubs. Control charts allowed identification of both pulse events (e.g., regional drought) and more gradual changes (e.g., recovery following livestock removal). Extension of these methods; however, indicated that sampling frequency and timespan of the dataset can substantially influence the ability to detect these changes. Further testing and refinement of this approach could enhance the ability to detect, predict, and identify the trigger of, major perturbations in ecosystem structure and function, providing more information in the practice of range management.