Background/Question/Methods Ecological restoration efforts often involve only the manipulation of abiotic factors at the local scale. However, processes external to the boundaries of a restoration site determine the range of local conditions within the site, constraining the level of control that can be achieved by on-site manipulations. We applied a method of constrained ordination combined with variation partitioning among sets of predictor variables to study plant species composition in 28 restored wetlands in Illinois, USA. We partitioned the variation in species composition among three spatially hierarchical sets of predictor variables: (1) macro-scale landscape features reflecting site setting within regional landscapes and hydroscapes, (2) meso-scale landscape features reflecting nearby propagule sources and buffers from landscape disturbances, and (3) local environmental factors including site age, fertility, and hydrology. We determined the unique and shared contributions of each of these sets of predictor variables to three aspects of plant communities: (1) overall species composition, (3) functional group composition, and (3) the composition of species groups categorized on the basis of their conservation value (floristic quality groups).

Results/Conclusions All three predictor sets together explained 65.8% of the variation in species composition, and the sets of predictors at the three scales had independent and nearly equally strong relationships to species composition. In contrast, when species were aggregated into functional groups the unique contribution of local predictors (24.9%) was greater than the unique contributions of macro-scale or meso-scale landscape predictors (14.9% and 18.6%, respectively), suggesting convergence of functional group composition in sites with similar environmental conditions. Macro-scale and local predictors both explained a significant amount of variation in floristic quality group composition, but a large amount of the variation (23.4%) attributed to local predictors could be explained by large-scale landscape factors, indicating that landscape constraints on local environmental conditions limited the level of floristic integrity achievable in restored wetlands. The appropriate scale at which to focus restoration efforts will vary depending on restoration objectives. Restoration of species composition may require simultaneous consideration of multiple scales of influence, but for other restoration goals, practitioners might more effectively target processes operating at certain spatial scales.