Ecological restoration is addressing concerns surrounding threats to ecosystem services and biodiversity. Support for this endeavor comes from substantial conceptual and practical advances, and policy initiatives and monetary incentives such as offsetting. Despite these advances, it remains unclear at fine scales what are the main factors influencing tropical forest restoration, particularly ecosystem functions and community assembly. This lack of knowledge can compromise the willingness to restore and waste limited resources. To improve the knowledge base and aid identification of appropriate restoration targets, we asked: How important are soil properties and forest adjacency for biomass uptake and community assembly during tropical forest restoration? We quantified woody aboveground biomass and tree species abundances across 32 sites, using four plots per site, six years after restoration initiation in an agricultural landscape of eastern Sao Paulo State, Brazil. The same species were used across sites, and differences in response variables could thus be related to soil properties and land coverage adjacent to plots given a site selection protocol removed other confounding variables. Using linear mixed models we evaluated predictors of biomass variability across the whole tree community. With multivariate abundance models, we evaluated predictors of community assembly based on species-trait groups.
Results/Conclusions
Biomass recovery was highly variable within and across sites, with empirical results ranging from total failure to 117 t/ha. Increasing percentage sand in a plot significantly reduced biomass sequestration, likely through its effects on species establishment and growth, but only in fertile soil conditions. In infertile conditions, biomass recovery was uniformly low. Increasing forest coverage, within 200m of a plot, tended to increase biomass recovery, and this was particularly noticeable for small trees (<14 m high) and those with softer wood (wood density < 0.58 g/cm3). Moist-forest specialist species were absent whenever sand percentage increased above 55%, possibly due to drought. Our results corroborate the need for fine scale evaluations to predict restoration outcomes, and suggest that degradation associated with agricultural practices constrains the likelihood of restoration to pre-disturbance conditions, particularly where soil fertility has been degraded and/or the percentage of sand has increased. Given projections for increasing drought in this area, restoration policies likely need to consider alternative restoration targets for these landscapes, coupled with a goal to improve soil conditions to allow moist tropical forest recovery in appropriate areas.