Uso da geoestatística na avaliação da implantação de pastagem com a brachiária híbrida mavuno

Abstract

Geostatistics is a valuable tool for analyzing, understanding and helping to make decisions regarding the use and management of agricultural production areas. It contributes to the efficient management of natural resources, territorial planning, risk analysis and improvement of strategies in various areas, promoting a more precise approach to dealing with the complexity of spatial and temporal phenomena. It is possible to apply the geostatistics tool in the agricultural segment, especially in pasture areas, allowing more precise and sustainable planning of the activity to improve production performance. In this context, this study aimed to evaluate the productive characteristics during the initial phase of implementation and establishment of Brachiaria Híbrida Mavuno, using geostatistics. The results were obtained through the analysis of experimental data that were collected in the municipality of Coração de Maria – BA, on a livestock property. The preparation of the area, correction, sowing and fertilization followed agronomic recommendations. For data collection and sampling, a rectangular sampling grid containing 75 georeferenced points was installed in the area and the evaluations were carried out in three periods at 30, 60 and 90 days after sowing. The variables analyzed were: plant height, number of tillers, soil resistance to penetration, occurrence of clods, green phytomass and dry mass of phytomass. Data from the three periods were processed by applying classical descriptive statistics – position and dispersion measurements – to analyze the occurring variability and Geostatistics to verify and quantify spatial dependence, calculating the semivariance and generating response surface maps with the aid of the GS+ software. There is a strong positive correlation between plant height at 90 days and green mass at 90 days. The variables number of tillers at 30 days and green mass at 90 days showed high variability. Geostatistical mapping revealed distinct patterns for the analyzed variables. The models that best fit to estimate spatial variability include the spherical, linear and linear model with plateau.