Recovering the riparian vegetation depends, at first, on rescuing the attributes of the historical ecosystem in each particular site. The restoration target, at least in terms of vegetation structure, shall be settled on the basis of that information, since water-related ecosystem services and carbon sequestration are closely related to that structure. Setting the target ecosystem, however, is not an easy task when forest and non-forest ecosystems are alternative stable states, such as in the forest-savanna boundaries in tropical regions. We implemented an experiment aiming at riparian tropical forest restoration by planting nursery-raised seedlings of 20 species arranged in pure or mixed stands, in the Cerrado ecoregion (south-eastern Brazil), where either wet grasslands or gallery forests can occur along the streams. The outcomes of eight treatments were assessed after 26 years and compared to both reference ecosystems (gallery forests and wet grasslands). We sampled all trees (planted or not) as well as native plants spontaneously regenerating in the experimental plots. From these data we obtained community structure, plant richness, carbon storage and rain interception in different treatments. The question is: how to decide what is good or bad if there are alternative stable states as reference ecosystems?
Results/Conclusions
In those treatments where tree-seedlings survived, a forest vegetation was obtained and, favoured by shade, tree species colonized the understory. Some treatments resulted in tree biomass similar or higher than that in the native forests, but rain interception was proportionally high, far above that in the wet grasslands. In those treatments where tree-seedlings died before providing shade, a slow process of colonization by native grasses, forbs and shrubs occurred, eventually re-establishing the wet grassland. Despite the successful forest restoration, with high carbon sequestration and storage, we discuss if restoring a grassland could be preferable, since water is a regionally scarce resource and the experimental area is around the springs of the water-supply system of a city. We argue that the already known trade-off between carbon storage and water supply should be the driver of riparian restoration planning in regions where forests, grasslands or savannas are alternative states.