Gray sufu, a traditional fermented food derived from soybeans, undergoes a complex fermentation process. This study aimed to investigate the dynamics of the microbial community during sufu fermentation and its relationship with key quality characteristics. Through systematic sampling of sufu at different phases of fermentation, 143 bacterial genera and 84 fungal genera involved in the process were identified. Among these, Chishuiella, Enterococcus, Lactococcus, and Weissella emerged as the predominant bacterial communities. After seven days of ripening fermentation, Trichosporon supplanted Diutina as the predominant fungus, accounting for more than 84% of all fungi. Using redundancy analysis, significant correlations between microbiota and physicochemical properties were uncovered. Chishuiella and Empedobacter displayed positive relationships with pH, soluble protein, and amino nitrogen content. In addition, five biogenic amines were detected, and it was determined that tyramine accounted for more than 75% of the total biogenic amines in the final gray sufu products. Spearman correlation analysis revealed significant positive relationships between Lactococcus, Enterococcus, Tetragenococcus, Halanaerobium, and Trichosporon and the five biogenic amines examined. These findings shed light on the complex interactions between microorganisms and biogenic amines during the fermentation of gray sufu, thereby facilitating the development of microbial regulation strategies for better quality control.

Lactobacillus species are widely used in various food products, including conventional food products, dairy products, and health food products. To achieve the desired functional properties, manufacturers commonly incorporate two or more distinct Lactobacillus species during production. In this study, a multiplex PCR detection method was developed for four Lactobacillus species commonly used in food based on TaqMan real-time fluorescent PCR technology, enabling the efficient and rapid identification of multiple Lactobacillus strains in food matrices. The research team selected and validated four representative species—Lactobacillus rhamnosus, Lactobacillus plantarum, Lactobacillus acidophilus, and Lactobacillus paracasei—as targets for the multiplex PCR assay, designing specific primer–probe combinations for each. The accuracy and reliability of the detection method were rigorously evaluated through a series of validation experiments, including the assessment of primer–probe specificity, optimization of fluorescent signal labeling chemistries, determination of the limits of detection for individual strains, evaluation of the method’s repeatability, and analysis of commercial food samples. The results demonstrated that the selected primer–probe sets exhibited no cross-reactivity in the multiplex system and specifically amplified their target Lactobacillus species, with no amplification observed for non-target strains. The established method achieved a minimum LOD for L. acidophilus of 10² CFU/g and showed high repeatability across replicates. Furthermore, the successful detection of labeled Lactobacillus strains in commercial products confirmed the method’s practical applicability. Therefore, the developed multiplex real-time PCR assay provides a reliable, sensitive, and high-throughput tool for the simultaneous detection of multiple Lactobacillus species in complex food products and holds potential for application in quality control, product authentication, and regulatory compliance monitoring.