Abstract
Penicillium expansum is a major postharvest pathogen causing fruit decay and mycotoxin contamination. This study investigated the mycotoxin production and metabolic profiles of 91 P. expansum strains using metabolomic analysis. Six mycotoxins were identified, with patulin (PAT) and chaetoglobosin A being particularly prevalent at 77.56 and 45.58 mg·kg(-1) respectively. Untargeted metabolomics profiled 506 metabolites, revealing a decrease in major metabolites during cultivation due to fungal assimilation. Comparative analysis between high- and low-PAT samples showed distinct metabolic signatures in organic acids, benzenoids, organoheterocyclic metabolites, which are linked to mycotoxin production pathways. These differential metabolites were used to build discriminant models, with random forest model achieving 98 % accuracy in distinguishing high- and low-PAT samples. Metabolites, mainly 3, 4-dihydroxybenzoic acid and 4-ketopimelic acid, were explored as potential biomarkers for PAT contamination diagnosis. This research enhances the understanding of P. expansum's metabolic diversity and supports improved fruit quality and safety control.