Distribución espacial de las propiedades físicas y de la curva de retención de humedad en suelos con cultivo de banano

Abstract

The present work was carried out in the Province of El Oro, located in the south of Ecuador, in the canton of Pasaje, parish "La Peaña" in the farm "La Flor", parish dedicated to banana production. The bad agricultural practices have provoked alterations in certain physical properties, of the soils, thus it demonstrates it studies carried out by Vásquez, Baena and Menjivar in 2010 where a great variability of the Physical properties in the first 25 cm of the soil is evidenced. In the study area, a planimeter survey of the farm was carried out with the use of a GPS. This geographic information was processed in the AutoCAD 2017 software and a grid of 75 * 75 m was carried out, where 15 points were identified, which were made staircase type calicatas to obtain samples that were later processed in the soil laboratory of the Machala Technical University of the Faculty of Agricultural Sciences from April 5 to May 2 of the current year, where the texture, apparent density and real density were determined. The hydrophysical variables field capacity (CC), and permanent wilt point (PMP) were performed in the soil laboratory of the State University of Cuenca, with which moisture retention curves were constructed. In addition, maps of spatial variability were elaborated, synthesizing them into a single document where the state of the soil can be quickly interpreted and the geostatistical model or models defined (exponential, circular, gauss and spherical of the kriging method) that best adapts to the area under study through the use of the pearson correlation coefficient and the square average error. The descriptive analysis of data was used the IBM SPSS software Estatistics 22 and for the geostatistical analysis the ArcMap10.5 software. In Microsoft Excel 2013 the tables to calculate the Nassh-Sutcliffe coefficient were elaborated, and the tables where the geographical coordinates of the 15 points plus the observations of each variable in the different sampling depths were entered, these tables were loaded in ArMap 10.5 and analyzed by means of the Geostatistical Wizard tool where the distribution maps of the physical properties of the soil were designed by Kriging's method. The semivariogram models used were spherical, Gaussian, Circular and exponential, of which the exponential and spherical semivariogram models were more precise and efficient in the different physical variables at different depths, for apparent density (Da), Percentage of porosity the model that best fit was the spherical, as for real Density, percentage of silt and sand was the exponential as for the percentage of clays there were 2 models that better adjusted the spherical and Gaussian because in general terms they presented correlation coefficients of Pearson and the coefficient of Nash equal to 1, likewise the average quadratic error was the lowest, all the variables studied presented strong spatial dependence (less than 25%).