Host-plant phenology and weather based forecasting models for population prediction of the oriental fruit fly, Bactrocera dorsalis Hendel

Predicting oriental fruit fly, Bactrocera dorsalis Hendel, populations well in advance with maximum accuracy will contribute to the success of IPM programs in India. The present study explored the scope of using host-plant phenology variables in addition to abiotic variables for fine tuning the current system of fruit fly population prediction. Variables representing host-plant (Guava, Psidium guajava L.) phenology and weather were used and compared as components in step-wise regression to develop a comprehensive forecasting model for the pest. Significant associations (P < 0.05) were observed between host-plant phenology and synoptic weather variables. Among all the combinations of step-wise linear regression models, the model that used the availability of small immature P. guajava fruits as a single independent variable gave the best-fit as it explained the highest variability (R2 = 0.78) in the trap catch with R2 being 78% and 80% for linear and 6th order polynomial regression, respectively. Thus, the simple linear regression model derived for small immature P. guajava fruits had the strongest relationship with fruit fly trap catch and can be derived easily from visual scoring data. The best single predictor, small immature P. guajava fruits, is proposed as an accurate indicator suitable for forecasting the changes in B. dorsalis population well in advance. The predictive performance of linear regression models involving both host-plant phenology and weather variables and their prescriptive use is discussed.

► A forecasting model to predict Bactrocera dorsalis population accurately was attempted.
► Host-plant phenology and weather variables were compared in step-wise regression.
► The small immature guava fruits exhibited significant positive correlation with fruit fly trap catch.
► This variable was found to be the best phenological indicator for predicting fruit fly trap catch.