Written Paper

Evaluation of multiple regression models based on epidemiological factors to forecast Bemisia tabaci and mungbean yellow mosaic virus  [2006]

Khan, M.A. (University of Agriculture, Faisalabad (Pakistan). Dept. of Plant Pathology) Rasheed, S. (University of Agriculture, Faisalabad (Pakistan). Dept. of Plant Pathology) Ali, S. (University of Agriculture, Faisalabad (Pakistan). Dept. of Plant Pathology)

Access the full text

Stepwise regression was used to develop pest and disease predictive models based on two years (2003 and 2004) epidemiological data. Weekly air temperatures (max/min), rainfall and relative humidity were employed as independent variables while, whitefly population and mungbean yellow mosaic virus (MBYMV) percent plant infection served as dependent variables. Interactions of years, dates of sowing, varieties and weeks of disease and pest ratings were significant. There was a significant correlation of whitefly population with MBYMV in both crop seasons. The degree of correlation of environmental variables differed according to years, dates of sowings and varieties. Stepwise regression indicated the significant influence of air temperatures, rainfall and relative humidity on whitefly population and MBYMV severity. According to dates of sowing and varieties different models predicted pest population and disease severity, however, some models were not of full rank and the degree of prediction decreased greatly when data were split by sowing dates and varieties. During 2003, a full model consisting of all the environmental variables explained 92% of the variability in whitefly population and 81% of the variability in MBYMV disease development. During 2004, the same four environmental variable model explained 91% of the variability in pest and disease development. The observed and model predicted values of whitefly and MBYMV were in close conformation.