Bradford, L. M.; Yao, L.; Anastasiadis, C.; Cooper, A. L.; Blais, B.; Deckert, A. E.; Reid-Smith, R. J.; Lau, C. H.-F.; Diarra, M. S.; Carrillo, C.; Wong, A.

In order to prevent the spread of foodborne illnesses, the presence of pathogens in the food chain is monitored by government agencies and food producers. The culture-based methods currently employed are sensitive but time- and labour- intensive, leading to increasing interest in exploring culture-independent diagnostic tests (CIDTs) for pathogen detection. However, sensitivity and reliability of these CIDTs relative to current approaches has not been well established. To address this issue, we conducted a comparison of the limit of detection (LOD50) for Salmonella between a culture-based method and three CIDT methods: qPCR (targeting invA and stn), metabarcode (16S) sequencing, and shotgun metagenomic sequencing. Samples of chicken feed and chicken caecal contents were spiked with Salmonella serovar Enteritidis and subjected to culture- and DNA-based detection methods. To explore the impact of non-selective enrichment on LOD50, all samples underwent both immediate DNA extraction and an overnight enrichment prior to gDNA extraction. In addition to this spike-in experiment, feed and caecal samples acquired from the field were tested with culturing, qPCR, and metabarcoding. In general, LOD50 was comparable between qPCR and shotgun sequencing methods. Overnight microbiological enrichment resulted in an improvement in LOD50 with up to a three log decrease, comparable to culture-based detection. However, Salmonella reads were detected in some unspiked feed samples, suggesting false-positive detection of Salmonella. Additionally, the LOD50 in feeds was three logs lower than in caecal contents, underscoring the impact of background microbiota on Salmonella detection using all methods.