PADRÕES E MÉTODOS PARA PREDIÇÃO DA DISTRIBUIÇÃO ESPACIAL DE NEMATOIDES EM SISTEMA DE SUCESSÃO SOJA-ALGODÃO

Abstract

The present study was developed with the purpose of comparing methods of spatial interpolation of population density of phytonematodes as well as to evaluate the effect of crop succession, soybean and cotton, on the population density and the spatial distribution of phytonematodes in a commercial field. The study was carried out in the city of Pedra Preta - MT. The sampling was done in a grid with regular distances of 25 m, collecting one sample per m², in 20 geo-referenced sample points. Soil and root samples were collected at three times: before planting (initial population), soybean at stages R1 and R6 and cotton at stages R1 and R6. The following methods were used to construct infestation maps: the circular variable-radius moving window method (CVRMW), which is proposed here, the inverse squared-distance weighting and ordinary kriging. The mean absolute percentage error of each map was calculated. In none of the scenarios we found a reasonable fit of semivariance models, so that kriging was not done. The CVRMW method presented, in general, smaller interpolation errors. There was no effect of the soybean-cotton succession on the phytonematodes of the species Pratylenchus brachyurus and Rotylenchulus reniformis.