Correlação linear entre componentes da produção e produtividade do arroz de terras altas em sistema plantio direto

The rice is one of the main sources of the humanity's feeding. During the agricultural year 2009/2010, in Selviria County, Mato Grosso do Sul State, in the Brazilian Savannah, an experiment was installed with rice upland in a Dystropherric Red Latosol (Typic Acrustox) under no-tillage, irrigated by central pivot, with the purpose of selecting the best components production to explain the variability the irrigated rice yield upland. The geostatistical grid was installed, to collect the data, with 120 sampling points, in an area of 3.0 ha and and homogeneous slope of 0.055 m m(-1). The medium rice yield was of the 5980 kg ha(-1). For the simple lineal regressions, the number of spikelets grenades for panicle presented the best direct potential correlation with the yield rice, given for: PGO = 115,5.NEG(0,770). However, for the multiple lineal regressions, the equation equacao PGO = 2754,30-411,55.NEG-461,07. NEC+436,59. NET it was the one that better she came to esteem the yield rice. However, spatial, it was not possible to establish correlation between the yield rice and the components production, once none of those it presented spatial dependence in their data.