Phenology and yield of strawberry (Fragaria X ananassa Duch.) under varied weather conditions

Abstract

Strawberry (Fragaria × ananassa Duch.), a widely consumed fruit of the Rosaceae family, is an octoploid hybrid (2n=8x=56) derived from Fragaria chiloensis and Fragaria virginiana. Renowned for its adaptability to diverse climatic conditions, the crop thrives across temperate to tropical regions. This study aimed to investigate the phenology of the strawberry variety 'Winter Dawn' under varying weather conditions and analyze the crop weather relationship to predict yield using statistical and artificial intelligence techniques. Experiments were conducted at the Regional Agricultural Research Station, Ambalavayal, employing a randomized block design (RBD) in open-field conditions and a completely randomized design (CRD) in polyhouse conditions. Five planting dates (1st September, 15th September, 30th September, 15th October and 30th October) were evaluated in the open field, while a single planting date (30th September) was tested under two polyhouse conditions. The study revealed that planting time significantly influenced the yield and quality of strawberries, with optimal weather conditions enhancing vegetative growth and fruit development. Phenological observations indicated six distinct stages: inflorescence emergence, first flower opening, fruit set, full flowering, fruit development, and fruit ripening. September 30th planting recorded superior growth parameters, including plant height, spread, number of leaves and crown number, and achieved the highest yield across both open field and polyhouse conditions. Among polyhouse treatments, the Indian polyhouse demonstrated superior performance in terms of growth and yield compared to the Dutch polyhouse, attributed to its higher light intensity and lower humidity levels, which promoted photosynthesis and nutrient assimilation. Weather parameters critically influenced phenophases and yield. Correlation analysis indicated negative impacts of higher temperatures and suboptimal humidity during later growth stages, while increased bright sunshine hours and reduced minimum temperatures during early phenophases positively influenced vegetative and reproductive development. In polyhouses, the Dutch polyhouse's higher humidity and lower light intensity curtailed growth and yield compared to the Indian polyhouse. For yield prediction, Random Forest emerged as the most effective model among machine learning techniques evaluated, including AdaBoost, Gradient Boosting, and Stacking. Feature importance analysis underscored the pivotal role of weather variables in determining yield across phenophases. This study highlights the importance of optimizing planting time and microclimatic conditions to enhance productivity and highlights the potential of machine learning models for accurate yield forecasting, thereby aiding informed agricultural decision-making .