Large-Scale Retrospective Analysis of Fresh Orange Production under Climate Variability: Implications for Industrial Biotechnology

Abstract

Climate variability is increasing uncertainty in citrus production and complicating industrial planning in Mediterranean systems. This study analyses a large industrial dataset of fresh orange production collected over nine growing seasons (2016/2017–2024/2025) in Palma del Río (southern Spain), integrating agronomic, spatial and climatic information from 169 orchards and 17 varieties. Agroclimatic indicators describing water availability, thermal accumulation and extreme temperatures were combined with yield data to assess interannual variability and develop a climate-based prediction model. A multiple linear regression explained a high proportion of seasonal yield variability (R² = 0.737) and showed good predictive performance, with low Root Mean Square Error (RMSE) (1.28 t ha?¹), low Mean Absolute Error (MAE) (1.10 t ha?¹), and negligible Mean Bias Error (MBE). Extreme heat, expressed as the number of days with Tmax > 35 °C, emerged as the main climatic driver of yield variability. These results highlight the potential of industrial datasets to support predictive tools for harvest planning and decision-making under increasing climate uncertainty.