In this study, the CERES-Maize model was calibrated and evaluated using data from 60 farmers’ fields across Sudan (SS) and Northern Guinea (NGS) Savannas of Nigeria in 2016 and 2017 rainy seasons. The trials consisted of 10 maize varieties sown at three different sowing densities (2.6, 5.3, and 6.6 plants m−2) across farmers’ field with contrasting agronomic and nutrient management histories. Model predictions in both years and locations were close to observed data for both calibration and evaluation exercises as evidenced by low normalized root mean square error (RMSE) (≤15%), high modified d-index (> 0.6), and high model efficiency (>0.45) values for the phenology, growth, and yield data across all varieties and agro-ecologies. In both years and locations and for both calibration and evaluation exercises, very good agreements were found between observed and model-simulated grain yields, number of days to physiological maturity, above-ground biomass, and harvest index. Two separate scenario analyses were conducted using the long-term (26 years) weather records for Bunkure (representing the SS) and Zaria (representing the NGS). The early and extra-early varieties were used in the SS while the intermediate and late varieties were used in the NGS. The result of the scenario analyses showed that early and extra-early varieties grown in the SS responds to increased sowing density up to 8.8 plants m−2 when the recommended rate of N fertilizers (90 kg N ha−1) was applied. In the NGS, yield responses were observed up to a density of 6.6 plants m−2 with the application of 120 kg N ha−1 for the intermediate and late varieties. The highest mean monetary returns to land (US$1336.1 ha−1) were simulated for scenarios with 8.8 plants m−2 and 90 kg N ha−1, while the highest return to labor (US$957.7 ha−1) was simulated for scenarios with 6.6 plants m−2 and 90 Kg N ha−1 in the SS. In the NGS, monetary return per hectare was highest with a planting density of 6.6 plants m−2 with the application of 120 kg N, while the return to labor was highest for sowing density of 5.3 plants m−2 at the same N fertilizer application rates. The results of the long-term simulations predicted increases in yield and economic returns to land and labor by increasing sowing densities in the maize belts of Nigeria without applying N fertilizers above the recommended rates.

中文翻译：

---

优化尼日利亚北部小农玉米农场不同施氮量的播种密度管理决策

在本研究中，使用 2016 年和 2017 年雨季尼日利亚苏丹 (SS) 和北几内亚 (NGS) 稀树草原的 60 个农田的数据对 CERES-Maize 模型进行了校准和评估。试验包括以三种不同播种密度（2.6、5.3 和 6.6 株米）播种的 10 个玉米品种。-2) 跨越农田，农艺和养分管理历史形成鲜明对比。年份和地点的模型预测与校准和评估练习的观测数据接近，低归一化均方根误差 (RMSE) (≤15%)、高修正 d 指数 (> 0.6) 和高模型效率证明了这一点(>0.45) 所有品种和农业生态的物候、生长和产量数据的值。在年份和地点以及校准和评估练习中，观察到的和模型模拟的谷物产量、生理成熟天数、地上生物量和收获指数之间的一致性非常好。使用 Bunkure（代表 SS）和 Zaria（代表 NGS）的长期（26 年）天气记录进行了两个单独的情景分析。SS 使用早期和超早品种，NGS 使用中间和晚期品种。情景分析的结果表明，SS 种植的早熟和超早熟品种对播种密度增加至 8.8 株 m-2当氮肥的推荐比例（90 kg N ha-1) 被应用。在 NGS 中，观察到产量反应高达 6.6 株 m-2施用 120 kg N ha-1为中晚期品种。土地的最高平均货币回报（1336.1 美元）-1) 模拟了 8.8 株植物 m 的情景-2和 90 kg N ha-1，而劳动回报率最高（957.7 美元-1) 模拟了 6.6 株植物的情景-2和 90 公斤 N ha-1在SS。在 NGS 中，每公顷的货币回报最高，种植密度为 6.6 株 m-2施氮量 120 kg 时，播种密度 5.3 株 m 的劳动回报率最高-2在相同的氮肥施用量下。长期模拟的结果预测，通过增加尼日利亚玉米带的播种密度而不使用高于推荐比例的氮肥，产量和土地和劳动力的经济回报会增加。