Abstract

SILVA, Geovani José et al. Soil strength in degraded pasture assessed by multivariate analysis techniques. Cienc. suelo [online]. 2020, vol.38, n.2, pp.236-248. ISSN 1850-2067.

The objective of this work was to select similarity measures and agglomeration methods most appropriate to cluster analysis to characterize the vertical and horizontal variability of resistance to penetration in Latossolo Vermelho Distrófico under degraded pasture. The PR was sampled at intervals of 10 mm until 300 mm depth, in 4 points spaced 0.50 m from the 30 vertices of a 6 x 10 m mesh and at 3 more sites with 18 radially distanced points by 0.125 m. AF and HCA were applied to evaluate the soil PR variability in depth and by HCA its horizon- tal dispersion too. The highest PR values was between 20 and 60 mm deep in 94.22% of the evaluated profiles and in the remaining 5.8%, thicker and deeper, this layer it was established between 60 and 190 mm. The HCA technique by algorithm of arithmetic mean linkage between group and, Euclidean distance coefficient proved efficient, as well as, AF in the evaluation of the vertical PR variability, resulting a stratification of the resistance profile more coherent with the dynamics of the PR in depth. The PR values in the soil under degraded pasture are higher than the critical limit, with the layer of 20 to 60 mm deep having the highest mechanical impedance.

Keywords : Penetrometry; principal components; hierarchical cluster.

        · abstract in Portuguese     · text in Portuguese     · Portuguese ( pdf )