Airborne pollen and meteorological-related variables have proved to be a good indicator of flowering, olive fruit production and to predict harvest of upcoming crop, thus enabling efficient management and marketing strategies. This study describes the first forecasting models of the olive fruit production based on pre-peak airborne annual pollen integral (APIn) from Olea europaea L. and meteorological data prior and during the flowering and ripening olive trees in Tétouan (NW of Morocco) over a period of 11 years (2008–2018). Aerobiological sampling was conducted using Burkard volumetric Hirst trap. The data were analyzed by multiple regression analysis. Several forecasting models developed were validated using data of 2018 (not included in the models) and compared with real olive crop data obtained from the Provincial Directions of Agriculture of Tétouan. The main factors influencing the final olive crop were the rainfall registered prior to flowering (March) and during fruit growing and minimum temperatures in July and June. The most accurate forecast models for the 2018 harvest showed the highest coefficient of determination (R2 = 0.98; p < 000.1) and predicted the lowest RMSE between expected and observed data (452.80 and 398.75). The models developed provide efficient olive crop forecasting using independent variables which can be previously obtained. However, despite that the APIn is a reliable bio-indicator of regional crop yield forecasting in intensive farming areas it was not strongly representative in the regression equation probably due to the low airborne pollen concentrations recorded in Tétouan.

中文翻译：

---

使用得土安（摩洛哥西北部）的空气生物学和气象变量预测橄榄（Olea europaea L.）产量

空气中的花粉和气象相关变量已被证明是开花、橄榄果产量和预测即将收成的良好指标，从而实现有效的管理和营销策略。本研究描述了橄榄果实产量的第一个预测模型，该模型基于来自 Olea europaea L. 的峰值前空气传播年度花粉积分 (APIn) 以及得土安（摩洛哥西北部）橄榄树开花和成熟之前和期间的气象数据。 11 年（2008-2018 年）。使用 Burkard 容积式 Hirst 捕集器进行空气生物采样。数据采用多元回归分析。使用 2018 年的数据（未包含在模型中）验证了开发的几个预测模型，并与从得土安省农业方向获得的实际橄榄作物数据进行了比较。影响橄榄最终收成的主要因素是开花前（3 月）和果实生长期间的降雨量以及 7 月和 6 月的最低气温。2018 年收获的最准确预测模型显示出最高的决定系数（R2 = 0.98；p < 000.1），并预测了预期数据和观察数据之间的最低 RMSE（452.80 和 398.75）。开发的模型使用先前可以获得的自变量提供有效的橄榄作物预测。然而，