Abstract
Experimental plots of Lu-26, Pak-81 and Fsd-85 were established during 1995-96 and 1996-97 wheat growifl9 seasons. The crop was artificially inoculated with leaf rust urediniospores and natural inoculum was also relied upon for infection. Step wise regression was used to develop multiple regression models by employing weekly maximum and minimum air temperatures, rainfali, relative humidity, wind speed and 24 hr wind movement as independent variables while leaf rust severity served as dependent variable. R2, Mallows C, and mean square error were used to select the best model. Environmental conditions and leaf rust severity recorded on wheat varieties differed in two seasons. During 1995-96, two multiple regression models containing weekly maximum and minimum temperatures and maximum temperature and relative humidity explained more than 93% of the variability in leaf rust development on Fsd-85 and Pak-81 respectively. During 1996-97, weekly minimum temperature and relative humidity explained more than 90% of the variability in disease development on three varieties. Observed leaf rust severity values and those predicted by these models conformed to each other for most of the varieties.
