Deep Learning and Microbiome Analysis Reveal the Preservation Mechanism of Cinnamomum cassia for Strawberry.

Strawberry preservation remains a critical challenge due to rapid postharvest microbial spoilage. This study investigated the preservative efficacy of Cinnamomum cassia and Punica granatum peel aqueous extracts, prepared via spray drying, on strawberries over 5 days of storage, with a specific focus on their regulatory impact on the fruit surface microbiome. Preservation tests demonstrated that the C. cassia extract was more effective in reducing visible mold development. High-throughput sequencing revealed that the C. cassia treatment reshaped microbial community structures, decreasing the relative abundance of spoilage-associated bacteria and the primary pathogenic fungus Botrytis (94.37%), while enriching potentially beneficial or antagonistic genera such as Sphingobium (28.72%), Sphingomonas (9.52%), and Cladosporium (0.62%). Using a probability threshold of 0.7, 121 compounds were identified as potential active candidates from a library of 675 C. cassia constituents. These compounds predominantly have a molecular weight between 100 and 250 and are characterized by prevalent functional groups including alkene (49.60%), hydroxyl (38.80%), and benzene rings (36.40%). In vitro antibacterial assays confirmed the inhibitory activity of vanillin and its isomers, validating the reliability of the computational predictions. These findings suggest that the preservative mechanism of C. cassia is likely mediated by the collective action of a multi-component matrix that modulates the microecological balance on the fruit surface, rather than the isolated effect of a single compound. This integrated approach provides an effective framework for developing plant-derived preservation strategies by combining microbiome dynamics with machine learning.