Multivariate analysis of latosol attributes in the amazon-caatinga transition zone under an agropastoral system

This study aims to analyze the dynamic relationship between the physicochemical attributes of a Yellow Dystrophic Latosol, which a particular focus on their direct and indirect influences on soil electrical conductivity. Soil samples were collected from 50 georeferenced points within the municipality of Brejo – MA, Brazil, to evaluate these interactions using multivariate stastistical approaches. The path analysis process was conducted using the GENES software, a statistical tool essential for estimate linear correlation values and decomposing the data. The evaluated soil attributes included electrical conductivity, altitude, humidity, soil density, porosity, organic matter, pH, clay, silt, total sand, very coarse sand, coarse sand, medium sand, fine sand, very fine sand, aggregates larger than 2.00 mm³, aggregates larger than 1.00 mm³, aggregates larger than 1.00 mm³, aggregates larger than 0.10 mm³, weighted mean diameter, and mean internal diameter. The approach was adopted due to the high occurrence of multicollinearity, which can lead to interpretative misunderstanding. For this purpose, data decomposition was performed, using electrical conductivity as primary variable - to mitigate the mutual influence among soil attributes. The analysis revealed that pH is the most influential attribute in determining the electrical conductivity of the studied Latosol.