The increasing incidence and intensity of extreme climate events are a major threat to global crop production. Sugarcane represents the main source of sugar and ethanol in tropical and sub-tropical regions. Here, we assessed the impacts of climatic variability and extreme events on sugarcane yield gaps in Reunion Island which is characterised by extreme spatial contrasts in climate, and is regularly subjected to cyclones or drought. We combined a process-based crop model with spatial databases of soil, climate and management data, to predict the annual sugarcane yield in each field over the whole island from 1998 to 2018. Simulated yields were compared to actual sugarcane yields from sugar producers in five agro-climatic zones. Extreme climate indices were calculated for extreme daily rainfall, temperature and drought, and their influence on yield gap was assessed using a Principal Component Analysis (PCA). The average total yield gap (YG_T, i.e. the difference between potential sugarcane yield and actual yield, Mg ha^-1) over Reunion Island was 96 Mg ha^-1, explained by 34% throw crop management and hazards. YG due to crop management reached 65 Mg ha^-1 with increasing rainfall and nitrogen mineralization rate and up to 88 Mg ha^-1 with increasing temperature and radiation. Yield gap was correlated with the first axis of the PCA, highlighting how extreme rainfall induced increase in YG_T. The actual yield loss was defined as the difference between the actual yield from sugarcane producers in a given year and the highest yield obtained during the whole studied period in the same zone. The annual actual yield loss increased by 0.6 Mg ha^-1 per rainfall days higher than 29 mm d^-1 along the annual crop cycle. Using this approximation, average actual yield loss due to heavy rain was estimated as 5.5 Mg ha^-1 but it was higher than 10 Mg ha^-1 for 25% of years in the East coast region. Our results highlight the need for improving sugarcane crop yield response to extreme rain in process-based crop models, in order to better assess climate change impacts in future.

极端气候事件的发生率和强度不断增加，是对全球作物生产的主要威胁。甘蔗是热带和亚热带地区糖和乙醇的主要来源。在这里，我们评估了气候变化和极端事件对留尼汪岛甘蔗产量差距的影响，留尼汪岛的特点是气候空间差异极大，经常遭受飓风或干旱。我们将基于过程的作物模型与土壤、气候和管理数据的空间数据库相结合，预测了 1998 年至 2018 年全岛每个田的甘蔗年产量。将模拟产量与五个糖生产商的实际甘蔗产量进行比较农业气候区。极端气候指数是针对极端的日降雨量、温度和干旱计算的，并使用主成分分析 (PCA) 评估它们对产量差距的影响。平均总产量差距（YG_T，即潜在甘蔗产量与实际产量之间的差异，Mg ha ^-1 ) 在留尼汪岛为 96 Mg ha ^-1，解释为 34% 的作物管理和危害。YG由于作物管理达到65镁公顷^-1随降雨量和氮的矿化速率和高达88镁公顷^-1随温度和辐射增加。产量差距与 PCA 的第一轴相关，突出了极端降雨如何导致 YG _T增加. 实际产量损失定义为给定年份甘蔗生产者的实际产量与同一地区整个研究期间获得的最高产量之间的差异。每年的实际产量损失在每年的作物周期中每高于 29 mm d ^{-1 的}降雨天增加 0.6 Mg ha ^-1。使用这个近似值，大雨造成的平均实际产量损失估计为 5.5 Mg ha ^-1但在东海岸地区有 25% 的年份高于 10 Mg ha ^-1。我们的结果强调需要在基于过程的作物模型中改善甘蔗作物产量对极端降雨的响应，以便更好地评估未来的气候变化影响。