Análisis geoestadístico y multivariado de propiedades químicas de un suelo en cultivo de banano, finca marquez, La Peaña, cantón Pasaje

Abstract

Chemical properties are constantly changing as a consequence of soil formation processes and induced by poor management of field activities, caused by traditional agriculture. The knowledge of spatial variability not only fulfills one more stage of precision agriculture, but rather leads to the correct use of agricultural inputs such as fertilizers, amendments and even insecticides, providing positive changes for the benefit of the environment. With the objective of studying the variability of the soils, the research was carried out in a banana farm located in the parish "La Peaña" canton of Pasaje, province of El Oro, whose coordinates are: 3°19'02.3 "S and 79°51'15.7 "W, at an altitude of 20 meters above sea level. A geostatistical and multivariate analysis of chemical properties of the soil; pH, E.C., W.M., N, P, K, Ca, and Mg, was made, first making a descriptive statistic calculating mean, median, asymmetry, variance and coefficient of variation, and for geostatistics the classic semivariance accompanied by semivariograms that allow to appreciate the most representative results, maps were obtained using the interpolation method called Kriging that show the behavior of each variable on the ground. A systematic grid sampling was carried out with a distance between points of 70 meters, in an area of 6.27 ha, from which 15 points were distributed equidistantly. Each sample was collected at a depth of 30 cm; subdivided every 10cm, which were processed in the soil laboratory of the Technical University of Machala and in the INIAP soil laboratory. With the data, the traditional statistics were carried out in the IBM SPSS Estatistics 22 software, for the geostatistics the Surfer program was used to elaborate semivariograms and their theoretical adjustment with Gaussian, spherical, linear and exponential models, ArcMap 10.2 was used for interpolation once the geostatistical parameters such as sill, nugget and range were obtained. According to the analysis of the data of descriptive statistics, the coefficient of variation of the variables N and P, resulted with values higher than 60%, which indicates a greater variation among all the variables, unlike the pH and Ca that obtained the lowest percentages. With respect to pH, high values were determined, with an average of 7.90 in the three depths, resulting in slightly alkaline soils, high MO content, and an EC in slightly saline ranges. As for the geostatistical approach, Nugget presented values in lower proportion to the structural variance in most attributes, which determines a moderate and strong spatial dependence in the study variables. The semivariogram model that best adjusted to the study area was the exponential one in most of the variables, the range presented by the geostatistical analysis was within the maximum distance (310m) in the sampling area established for the calculation of semivariograms, variables such as MO and K presented the range value close to what was established. The variability in the farm was high as can be seen in the interpolation maps, and this type of study is very useful to determine the changes produced in the soil by the effect of management.