Identification of gut microbiota features of diarrheic calves using the full-length 16S rRNA gene amplicon sequencing and machine learning.

Abstract

BACKGROUND: Gut microbial dysbiosis is a common cause of calf diarrhea, a major contributor to calf mortality. Additionally, the early gut microbiota in calves is considered a hotspot for antimicrobial resistance, potentially leading to the failure of antibiotic treatments. Most prior research investigating the associations between calf microbiota and diarrhea has relied on partial 16S rRNA gene amplicon sequencing, which offers limited taxonomic resolution. In this study, we collected fecal samples from 25 Holstein-Angus crossbred neonatal calves on days 4, 7, 11, 14, and 30 and analyzed the gut microbiota using full-length 16S rRNA gene amplicon sequencing. A machine-learning approach was employed to identify potential bacterial markers associated with diarrhea. RESULTS: A significant increase in gut microbial diversity (P < 0.05) was observed with the calves’ age; however, days 11 and 14 showed similar diversity levels, coinciding with the peak prevalence of diarrhea. Differences in gut microbiota features between healthy (n = 5) and diarrheic calves (n = 8), determined by fecal scores on days 11 and 14, were also detected at earlier (day 4) and later (day 30) stages of life. Seven diarrhea-associated species (e.g., Fusobacterium spp. and Bacteroides fragilis) and four health-associated species (e.g., Megasphaera elsdenii and Enterococcus faecium) were consistently identified by both Linear Discriminant Analysis Effect Size (LEfSe) analysis and Boruta feature selection. Machine learning models trained on Boruta-selected bacterial features using Random Forest and Support Vector Machine algorithms achieved high classification accuracy (87–100%) in distinguishing diarrheic from healthy fecal samples. Moreover, the concentration of gut bacteria resistant to the cephalosporin antibiotic cefotaxime was significantly higher in diarrheic calves that harbored a greater abundance of pathogens (P < 0.05), particularly on days 11 and 14, when the prevalence of diarrhea was high, which may be attributed to the co-localization of virulence genes and antimicrobial resistance genes. CONCLUSIONS: This study provided a comprehensive characterization of the neonatal calf gut microbiota, identified potential bacterial species as biomarkers to predict diarrhea and probiotic candidates, and indicated early, immediate, and long-term microbiota changes associated with diarrhea. Additionally, it highlighted the increased risk of antimicrobial resistance to antibiotic treatment in diarrheic calves.