Phytogeographic retrospective in ecotonal areas guided by soil attributes

Abstract

Re-establishing deforested ecosystems to pre-settlement vegetation is difficult, especially in ecotonal areas, due to lack of knowledge about the original physiognomy. Our objective was to use a soils database that included chemical and physical parameters to distinguish soil samples of forest from those of savannah sites in a municipality located in the southeastern Brazil region. Discriminant analysis (DA) was used to determine the original biome vegetation (forest or savannah) in ecotone regions that have been converted to pasture and are degraded. First, soils of pristine forest and savannah sites were tested, resulting in a reference database to compare to the degraded soils. Although the data presented, in general had a high level of similarity among the two biomes, some differences occurred that were sufficient for DA to distinguish the sites and classify the soil samples taken from grassy areas into forest or savannah. The soils from pastured areas presented quality worse than the soils of the pristine areas. Through DA analysis we observed that, from seven soil samples collected from grassy areas, five were most likely originally forest biome and two were savannah, ratified by a complementary cluster analysis carried out with the database of these samples. The model here proposed is pioneer. However, the users should keep in mind that using this technology, i.e., establishing a regional-level database of soil features, using soil samples collected both from pristine and degraded areas is critical for success of the project, especially because of the ecological and regional particularities of each biome.