Sistema de alerta e relação de variáveis ambientais com o progresso da mancha de Phoma do cafeeiro

Abstract

The study of the progress curve of the Phoma leaf spot of coffee using statistical models can help to define control strategies to reduce fungicide applications and, consequently, minimize costs and environmental impacts. In addition, the understanding of the spatial distribution of this disease and of the environmental and host variables too can be integrated into its management. So, the objective of this study was to evaluate the temporal and spatial distribution of Phoma leaf spot of coffee (Phoma spp.), its relationship with different elevations, meteorological variables, soil texture and mineral nutrients in soil and leaves, plant foliage and production. The experiment was conducted for two years, from September 2013 to August 2015, with monthly evaluations in a plantation of Coffea arabica L. A mesh of 7.65 ha with 85 georeferenced points was used. The disease progress curves were modeled using time series techniques, nonlinear regression models (NLRM) and multiple linear regression models (MLRM), considering the overall average of 85 points in each month. In addition, progress curves were constructed for the mean elevations of 1130.54, 1140.65, 1143.18 and 1143.40 m. The spatial distribution of the disease and environmental and host variables correlated with this was modeled with geostatistics techniques. The progress curves of the disease presented a variable behavior within each year, as well as between the evaluated years. Higher elevations provided higher values of disease intensity. Only the incidence and severity progress curves at elevation of 1143.18 m showed autocorrelation over time, with adjustment of autoregressive models. The Gompertz NLRM was adjusted for progress curves of disease incidence and severity from February to June 2014. Were adjusted 126 MLRM to the incidence progress curve, considering the overall average of the 85 points, using the meteorological variables. Four of these presented high precision and accuracy, being possible to estimate the disease with two of them, two weeks in advance. For geostatistical analyzes, monthly data of disease incidence and severity were converted into area under disease progress curve (AUDPC). There was higher AUDPC to incidence and severity in the last 12 months of the study. The elevation, P-rem and Ca present in the soil, and the P and N present in the leaves had a significant positive correlation with incidence in the form of AUDPC. However, K, Cu and Mn present in the leaves, plant foliage and production, correlated negatively with incidence in the form of AUDPC. The exponential model of semivariogram was the most appropriate to model the spatial autocorrelation of the analyzed variables, except for the elevation.