Modeling hybrid and sowing date effects on potential grain yield of maize in a humid temperate region

Maize is cultivated extensively in the humid temperate region of Argentina but little is known about potential grain yields under nonstress conditions. In this study, grain yields were modeled for hybrids of different cycle (FAG 430 to FAO 630), sown on four dates, at Pergamino (33 degrees 56'S, 60 degrees 33'W), Argentina, under no limitations for water or nutrients. Twenty-four years of climatic records were included in an analysis with two models: (i) a correlative model built with experimental data obtained during two growing seasons; and (ii) the CERES-Maize model. The correlative model was based upon relationships between: (i) photosynthetically active radiation intercepted by the crop (IPAR) and accumulated temperature above 8 degrees C (thermal units = TU); (ii) shoot biomass and accumulated IPAR; and (iii) harvest index. The correlative model gave good agreement between observed independent data sets and simulated values of grain yield (root mean square error RMSE = 15% of the mean value). Predictions using CERES-Maize were less accurate (RMSE = 22% of the mean value). The correlative model estimated grain yields greater than 16000 kg ha(-1) (15.5% moisture) in 50% of the years with a FAO 630 hybrid sown in September. Frost could eliminate the crop in 8 out of 24 years with August sowings. With November sowings, median yields lower than 14160 kg ha(-1) could be expected. The long-cycle cultivars outyielded the short-cycle ones by an average of 2460 kg ha(-1). The consistency of these trends is discussed.