Spatial variability of soil physical properties and weeds variables on two soilmanagement systems

Abstract

Variables for soil physics and weeds have been quantified and described as spatially dependent. This has been carried out by means of geoestatisticals tools, allowing detection and description of spatial dependence and to estimate by kriging and/or co-kriging for ansamplet places. The objective of this research is to establish and to describe the spatial variability of the following parameters: soil bulk density (Ds); macro (Mac); micro (Mic) and total soil porosity (Pt); aeration capacity of soil matrix (CAMS); soil water content at matric potential of 100 cm water column - defined as field capacity (CC), biomass of the weeds (biomass), densities of weeds (Dp), density of Commelina benghalensis (commelina) and density of Bidens pilosa (bidens) under no-tillage (PD) and conventional tillage (PC). Sampling were made in 128 points for both soil management systems, in a 3 x 5 m and 3 x 2,5 m grid, formed by the crossings of X and Y coordinates. Undisturbed soil samples were collected in the 0 -15 cm depth layer. Weed densities were collected in an area of 1 m2. Weed biomass was obtained by drying in an oven until constant weight was obtained (dry mass). The exploratory data analysis showed the existence of trend for soil physics variables in PC and for weed variables in both sistems, with exception for density of commelina in PD. Data were transformed into natural log and outliers were excluded before spatial variability was studied using geostatistics, by analising semivariograms and cross semivariograms, interpolation by kriging and constructions of maps. Variables presented spatial dependence, with exception of density of commelina in PC. Range values of semivariograms were higher in PC than PD, except for biomass and CAMS. Cross semivariograms among soil physics variables showed that the standard of spatial dependence was mantained, with higher ranges for PC, except for range of Ds x CC. The ranges of the cross semivariograms were higher for PD. It was possible to conclude that the parameters of soil physic sand weeds investigated in this study are spatial dependent, widh exception for commelina density in PC, and also that the spatial dependent parameters can be estimated by co-kriging.