Abstract
This paper aimed to classify the forest species related to the demands on soil fertility and to identify groups with similar characteristics using multivariate statistical techniques. To do so, we analyzed soil samples for the texture and for chemical attributes and we carried out a phytosociological survey of the region. In the principal component analysis, the relations of the complex soil were summarized into four components, which together explained 80% of data variance, with most significant relationships between the sum of bases, base saturation, total cation exchange capacity, aluminum saturation, Ca + Mg, Al and clay, with correlation coefficient greater than 90%. In hierarchical cluster analysis, 70% is similar to soil attributes; three distinct groups of forest species were formed. Multivariate analysis was effective in reducing the original variables, separation and classification of groups and identification of soil chemical properties most required by forest species. Groups 1 and 2 contain species more demanding in fertility, with preferences for medium textured soils and rich in calcium and magnesium bases. While Group 3 seems to be less demanding, with a predominance in sandy and acidic soils, with large cation exchange capacity and high aluminum saturation. Therefore, the species of the latter group can be indicated for the recovery of the degraded areas. The species Copaifera sp. and Goupia glabra deserve more studies about the mechanisms of adaptation to low soil fertility conditions.
Keywords:
Principal component analysis; Cluster analysis; Discriminant analysis; Reforestation
Thumbnail
