Development of linear equations for predicting wheat rust epidemics in New Halfa, Sudan

Multiple regression analysis was used to develop equations predictive of the severity of leaf rust (Puccinia recondita f.sp. tritici) and stem rust (P. gram in is f.sp. tritici) of bread wheat ,and to estimate weekly cumulative urediniospore numbers of both rust species in New Halfa region. Equations were generated for three overlapping periods to identify and quantify biological and meteorological variables which might provide clues of high predictive value for development of !disease epidemics. Analysis of the combined data showed that the derived multiple regression models varied with prediction period, prediction duration and rust species. On the whole, progress in time (X1) invariably significantly contributed to variation observed in rust severity. Other significant variables were found to be components of atmospheric humidity: minimum RH (X4), maximum RH (Xs), and hours RH 80% (X6), followed by maximum temperature (X3) towards the end of the growing season but for leaf rust only. Minimum temperature (X2) and wind speed (X7), generally, did not significantly contribute to variation in rust severity .Simple linear regression analysis of overall disease severity revealed that leaf rust and combined rust exhibited highly significant and positive relationships with progress in time from 15 Ian. to 31 March, whereas stem rust had a non-significant negative relationship. The combined influence of three biological variables, namely progress in time (T), wheat growth stage (GS), and weekly trapped rust spore numbers (WSN), accounted for only 53 and 21.4% of the total variation in leaf rust and stem rust severity (DS), respectively. Studies with 'mechanical rust spore trapping (MRST) showed that 40 and 53.7% of the total variation in P. recondita, and P. graminis weekly trapped urediospore numbers (WSN) can be explained by a function involving: progress in time (T), wind direction (WD), and presence or absence of wheat crop (C). Wind direction exhibited a significant (5%-1%) negative effect on P. gram in is, but non-significant influence on P. recondita, WSN. The inclusion of wheat growth stage (OS) in the model did not significantly improve the explanatory power of the function. Analysis of MRST data collected over ten weeks, from 15 Ian. to 31 Mar., indicated that 68.4 and 74.7% of the total variation in P. recondita and P. gram in is weekly trapped spore number (WSN), respectively, can be accounted for by a function involving: progress in time (T), wheat growth stage (OS), and disease severity (DS). The inclusion of wind speed (WS) as a fourth variable in the prediction model significantly ( 5% ) increased the amount of variation with leaf rust, and had no effect on stem rust spore counts.