Multivariate technique for determination of soil pedoenvironmental indicators in Southern Amazonas

ABSTRACT. The use of statistical techniques to evaluate soil attribute behaviour is an important tool for choosing the most adequate form of soil management. Thus, the aim of this study was to jointly assess the physical and chemical attributes and the magnetic susceptibility characteristics of three environments and to use multivariate statistics to define which attributes have the greatest potential as environmental change indicators. The study was conducted in three areas: one with archaeological black earth (and planted maize); one with pasture; and one of the forest. Sixty-four (64) deformed and non-deformed soil samples each were collected within these areas, with a 70 x 70 m grid spaced every 10 meters at depths of 0.0-0.2 and 0.4-0.6 m, to determine the chemical attributes (fertility), as well as the physical attributes (texture) and magnetic susceptibility. The data were analyzed through multivariate statistics. The archaeological dark earth areas have different behaviours between pastures and forests: chemical attributes have greater interaction in archaeological dark earth due to the high fertility of the anthropic horizons. The attributes that may indicate environmental changes are calcium, potential acidity and clay for archaeological dark earth, forest and pasture areas, respectively.