Forecasting bean yield losses under weed and Fusarium impacts from field plot statistical modeling

Abstract

A better understanding of bean dynamic threats and pathogens is needed to improve accuracy of field measurement for breeding and mitigation programs. Thus, a two-year field plot study examined associations of bean production, Fusarium root rot, and weed in two cultivars sown under different dates and weed control treatments. We examined six standard curves to model bean leaf area, disease incidence, and weed biomass development across 256 plots. Gaussian, linear-by-linear and exponential models were fitted to bean-disease-weed development data, respectively, and then model parameters besides maximum values were used in the remainder of analyses. There were significant correlations between bean-disease-weed development descriptors. Principal component analysis showed bean-disease-weed development descriptors accounted for 84% of total data variance. Using six principal components as variables, a multivariate regression model showed the descriptors leaf area, Fusarium root rot, and weed biomass development explained 31% of variations in bean yield.