Why does restoration fail? Though scientists’ and managers’ ability to design and implement strategies to promote recovery of biodiversity and ecosystem function has improved in recent years, outcomes frequently fall short of restoration goals. Sometimes called “novel ecosystems” or “alternative vegetation states”, these habitats may be missing key components or fail to provide key ecosystem services. If we are to design strategies that bring degraded habitats closer to their full potential, we need to better understand the pathways to successful restoration and the barriers that stand in their way. We investigated these topics in an inland pitch pine barren in New York that had experienced invasion by the N-fixing black locust (Robinia pseudoacacia) tree. Ongoing management has included removal of the trees, replanting of native species, and the use of prescribed fire. We surveyed 47 sites that have experienced a range of land use histories (remnant never-invaded, currently-invaded, and invaded-but-restored) and fire regimes. Age since restoration ranged from 0-17 years. We tested whether there were consistent community types associated with particular site histories, and whether differences in soil chemistry could explain observed community differences.
We found significant differences in species composition between the three types of site (overall and pairwise comparisons using multivariate ANOVA p < 0.001). Within the restored stands, multivariate ANOVA and NMDS analysis found a significant effect of stand age on community composition (p < 0.01). Older restored sites were more similar to remnant stands than were younger restored sites, though even the oldest (17 year-old) stands still occupied distinctive places in ordination space from never-invaded stands. There was only a weak effect of prescribed fire on community composition (p = 0.06). We also found that, though net N cycling rates were similar in restored sites compared to never-invaded ones, there were significantly lower soil C and N pools in restored sites (p < 0.01). In conclusion, significant differences in community composition persist between remnant and restored sites even more than a decade after restoration. We suggest that differences in soil C and N pools may act as barriers to successful restoration and contribute to the creation of these putative alternative vegetation states.