ENSO, SEASONAL RAINFALL PATTERNS AND SIMULATED MAIZE YIELD VARIABILITY IN ZIMBABWE

A correlation between ENSO (El Nino/Southern Oscillation) and rainfall in southern Africa has been recognized for at least a decade. This re cognition has led to the expectation that ENSO-based climate predictio ns will have significant applications in agricultural management. This study is an analysis of the potential for using ENSO predictions to r educe risk in agricultural production associated with rainfall variabi lity at the site level. Records of sea-surface temperatures in the equ atorial Pacific during November, December and January were used to def ine El Nino, La Nina and neutral years. Climate data from four sites i n four of the five agroecological zones (AEZ) in Zimbabwe were analyze d with respect to ENSO phases and used to drive a maize growth simulat ion model parameterized for soil conditions typical of each area, usin g two nitrogen fertilizer treatments and three planting dates. The fou r sites (Karoi, AEZ II; Gweru, AEZ III; Masvingo, AEZ IV; and Beitbrid ge, AEZ V) all showed a decrease in seasonal precipitation associated with the El Nino phase, compared to both neutral and La Nina years. At sites in zones II and III, within-season rainfall variability increas ed for both El Nino and La Nina years relative to neutral years. While average simulated maize yields were generally lowest in El Nino years , variability in rainfall pattern and standard deviation of yields at the site level was high within each ENSO phase, indicating that more p recise seasonal climate predictions would be necessary for forecasts t o be valuable in crop management decisions in Zimbabwe. However, simul ation results point towards the relative importance of predicting favo rable cropping seasons as opposed to poor ones with respect to better nitrogen management and yield improvement for the more marginal sites. (C) 1998 Elsevier Science B.V.