Análisis geoestadístico y multivariado de atributos físicos de un suelo en cultivo de banano finca Márquez la Peaña cantón Pasaje

Abstract

The spatial variability of the physical properties of agricultural soil is caused by soil formation processes or induced by crop management practices. Given the need to understand the behavior of the variation of some physical properties associated with the banana crop, multivariate statistical techniques were carried out in order to optimize production processes. The present research was carried out in the canton of Pasaje, parish of La Peaña, "Márquez" farm. The spatial variability of the following properties were studied: texture, true density (Dr), bulk density (Da), total porosity (Pt), field capacity (CC) and permanent wilt point (PMP). A systematic 70 x 70 grid sampling was performed giving a total of 15 points of interest, the samples were taken in a 50 cm deep profile in strata every 10 cm, establishing 5 levels for sampling of 0-10, 10-20, 20-30, 30-40 and 40-50 cm processing a total of 75 samples in 6.24 ha. The analysis of the physical properties that were carried out in the laboratory of the Technical University of Machala are: texture, Dr, Da and Pt, the CC and PMP were carried out in the laboratory of soil physics of the State University of Cuenca. The data analysis was performed in several approaches, univariate (descriptive statistics) using IBM SPSS Estatistics 23 and Microsoft Excel 2013 software, exploratory charts were used as mustache box diagrams that allowed to visually appreciate the behavior of each variable. In the geostatistical analysis Surfer 16 software was used for the determination of experimental semivariograms, where theoretical, spherical, Gaussian and exponential models were adjusted for the different depths of each variable, being the most precise and efficient the spherical and Gaussian model, which presented a Nash coefficient and a determination coefficient (R^2) close to or equal to 1, likewise the lowest values of the root mean square error (RMSE) were obtained. In the ArcGis 10.3 software, parameters of the semivariogram were used for interpolation by the kriging method that allowed the design of maps of spatial variability of the physical properties of the soil. The results showed the existence of moderate spatial dependence for Da and CC in the first 10 cm of soil and with the same trend in the 20 -30 cm stratum for PMP and the last stratum of 40-50 cm Pt, on the other hand a strong spatial dependence was presented in the first 10 cm of clay, Silt, Sand, Dr, Pt and PMP, for the 10-20 cm stratum, Clay, Silt, Sand, Dr, Da and Pt, of 20-30 cm maintains the same strong dependence of the previous stratum including the 30-40 cm CC presents the same strong spatial dependence of the 10-20 cm stratum and in the last 40-50 cm stratum, Clay, Silt, Sand, Dr and Da. It did not present at any depth a weak dependence. The ranges that explain the existence of a spatial dependence are between 80 and 180 m. For the statistical analysis by main components, the software IBM SPSS Estatistics 23 was used, 8 variables were associated, clay, silt, sand, Dr, Da, Pt, CC, PMP. The results indicate that all variables were grouped into three main components and these represent 82.65% of the variance. The first component obtained a higher correlation coefficient Dr and Are, the second, CC and PMP and the last component Clay. This method provides a visual idea of the interactions between the variables, allowing most of the information contained in each of them to be explained.