Predicting the Rate of Development of Black Sigatoka (Mycosphaerella fijiensisvar.difformis) Disease in Southern Taiwan

Publisher Website
Google Scholar
Abstract
A multiple regression equation relating disease incidence, accumulated rainfall, and days of RH .gtoreq. 90% to the rate of disease increase of banana leaf spot cuased by Mycosphaerella fijiensis var. difformis was developed from 5 yr of field data in three locations of southern Taiwan. The model is y = 0.124-0.00489 Z1 + 0.0000462 A1 + 0.0000515 I16, in which y is the relative rate of disease increase at any given time; Z1 is disease incidence represented at percentage of leaf infected at the time of the previous assessment; A1 is the square of Z1; and I16 is a composite variate, the product of P4 and R1, where P4 is the accumulated precipitation in a 4-wk period and R1 is the number of days with RH .gtoreq. 90% in a 1-wk period before the calculated relative rate. The equation accounted for 76% of the total variation in relative rate and performed best at two of the three locations and in years where there was some fluctuation in rate. Model selection and validation were based on the use of different regression analysis techniques including Mallow''s Cp statistics, the variance inflation factor, and methods of data splitting.