Incidence-severity relationships in non-treated and fungicide-treated wheat head blast epidemics in Brazil

Abstract

The relationship between wheat head blast incidence (I; proportion of diseased heads in a sample) and severity (S; the average diseased area in a sample of heads) was studied with the aim of determining whether severity could be reliably predicted from incidence. Data were collected from two main sources: a) the EPAMIG (non-treated) field experiments spanning seven years (2013-2019 with three cultivars) designed to evaluate the effect of planting dates on wheat blast (no fungicide use); and b) the UFTs (fungicide-treated) field experiments conducted by researchers of the network of uniform fungicide trials conducted during nine years (2012-2020) across 10 locations (totaling 14 cultivars). An ordinary linear regression model fitted to the complementary log-log transformation of the data provided a good fit to the data based on the square correlation of the predicted Ŝ, and the observed S (R2): The R2 was 0.95 for the pooled (all years) non-treated dataset and 0.88 for the pooled fungicide-treated dataset. A covariance analysis indicated that year, but not cultivar, significantly affected the parameters of the I-S relationship in the non-treated dataset. We conclude that wheat head blast S can be predicted from I but more reliably in non-treated epidemics given the larger influence of chemical treatments on the relationship than cultivar or year, a surrogate for varying weather.