Abstract
Restored plant communities provide a uniquely simplified context for studying difficult questions such as how ecological processes structure plant communities. Spatial pattern among individuals is an important component of community structure, and plays a role in determining ecosystem properties and restoration trajectories in plant communities. Nevertheless, the role of spatial pattern has previously received scant attention in a restoration context. We compared 87 plant species patterns from two samples of vegetation restored after sand-mining, with 233 patterns of species in four nearby natural reference sites in fire-prone, species-rich shrublands in south-west Australia. Spatial tests were performed at each site, both pooling species, and for each species with >20 individuals. Between 44 and 86% of species were aggregated in natural sites, with pattern in the remainder not distinguishable from random: unexpectedly, the restored sites lay at the upper bound of this variation. Comparison across species groups showed little difference in restored sites, but large differences in natural sites – with aggregation less frequent among large-seeded species, serotinous species and resprouting species. Observed differences among species groups may be attributed to contrasting dispersal mode patterns, spatial heterogeneity associated with fires, and multi- versus single-generational dispersal and these differences in observed patterns may influence future restoration trajectories. Other differences between natural and restored vegetation included a higher incidence of gaps, and lower stem density of restored vegetation. The increased abundance of gaps in restored vegetation suggests a major role of spatial processes in restoration failure.