Sugarcane production evaluated by the state-space approach

The effects of soil properties on crop growth and yield have traditionally been analyzed using classical statistics methodologies. These methodologies do, however, not consider sampling position coordinates and assume independence between samples. This study had the objective of using the state-space approach, which considers sampling position, to evaluate and to discuss a spatial process using variables related to the soil-plant system. For this, six data sets were collected in a sugarcane experiment carried out on a Dark Red Latosol (Rhodic Kandiudalf), at Piracicaba, State of Sao Paulo, Brazil. The sugarcane was planted on a 0.21 ha field, comprising 15 rows, 100 m long, spaced 1.4 m apart, with three treatments (mulching, bare soil and straw burning before harvest) and four replicates, forming a transect of 84 points. In this way, the relationships between the number of canes per meter of row and available soil P, Ca and Mg, clay content and aggregate stability were studied using a first order state-space model. Results show that all of the used state-space equations described the spatial distribution of number of canes better than the equivalent multiple regression equations. It was also identified that the soil clay content spatial series has an effective contribution to describe the number of canes in this study, because it is related to the best performance in each different scenario. (C) 2002 Elsevier Science B.V. All rights reserved.