Morphometry and size distribution of aggregates from a Typic Hapludult soil under natural secondary forest, implanted leguminous trees and degraded pasture in the Atlantic coast of Brazil

Revegetation of degraded land can in long-term improve the soil attributes compared to the initial condition. This study aimed to evaluate the efficiency of revegetation with leguminous trees to recover the structural quality of a Typic Hapludult soil. For this purpose, morphometric and size-classes distribution analyzes of soil aggregates were performed on soil samples collected from layers 0-0.10 m and 0.10-0.20 m depth. Three areas revegetated with leguminous trees 17 years earlier (Acacia auriculiformis, Mimosa caesalpiniifolia and Inga ssp.) and two reference areas (forest and pasture) were studied. The soil under leguminous trees showed larger aggregates than the soil under forest and pasture, with overall averages by grouping the former higher than those by grouping the latter around 31.6% for the weighted mean diameter and 47.6% for the geometric mean diameter. Aggregate morphometric variables obtained from image analysis were correlated to each other (such as circularity versus roundness, in all five size classes, with values ranging from R = 0.25* to R = 0.72**), and also to other soil physical and chemical attributes. As a general rule for all five aggregate size classes studied, both roundness and circularity increased with the increase in the aggregates diameter, the soil densification status, the soil richness in organic matter, the exchangeable cations content and the particle surface electronegativity. The aggregate morphometric variables of the two studied soil layers, analyzed by multivariate statistical tools, led to a consistent distinction between sites, grouping forests and pastures apart from the leguminous tree covers. Thus, it could be concluded that after 17 years of revegetation these leguminous trees changed the soil initially under pasture but in a divergent way in relation to the soil under forest. (C) 2020 Elsevier B.V. All rights reserved.