Correlating the oral swab microbial community with milk production metrics in Holstein dairy cows

Abstract

Oral swabs of dairy cows have been suggested as a proxy for direct ruminal sampling, and this approach can identify the presence of up to 70% of the rumen microbial community. Here, we further extend the utility of this approach by correlating the bacterial community of swabs collected from 226 dairy cows on a research farm in Wisconsin, USA, with average milk yield and days in milk, two phenotypes previously associated with differences in the ruminal microbiome. We then obtained milk production efficiency data for a subset of these animals (gross feed efficiency [GFE] and residual feed intake [RFI]) and correlated these metrics against their associated microbial data. We found that when using the oral swabs, we could identify correlations between bacterial genera and days in milk (P < 0.05). We further show that the ruminal microbiota was associated with average milk yield and days in milk for animals in their first lactation. Differential abundance testing identified amplicon sequence variants (ASVs) associated with these metrics (P < 0.05). Our comparison of bacterial communities between high and low efficiency groups, as determined by GFE and RFI, identified a significant difference in Shannon's diversity in second lactation cows (P < 0.05). We also found that RFI was significantly correlated with the bacterial community in second lactation animals (P < 0.05). Differential abundance analysis identified multiple oral- and rumen-associated ASVs correlated with GFE and RFI (P < 0.05). This study further establishes the utility of oral swabs as a ruminal proxy.IMPORTANCEImproving milk production efficiency is a key goal in the dairy industry and is traditionally pursued through genetic selection, diet optimization, and herd management practices. The ruminal microbiome, essential for digesting feed, has been linked to milk production efficiency, suggesting that microbiome modulation could improve efficiency. However, the integration of rumen microbiology into current management practices is hampered by the difficulty of large-scale rumen sampling, as proxies like fecal samples do not accurately reflect the ruminal microbiota. Traditional methods, like cannulation and stomach tubing, are labor-intensive and impractical for extensive sampling. Our research demonstrates the potential use of oral swabs as a scalable, effective method for characterizing the microbiome and its associations with milk production metrics, recapitulating established associations obtained through traditional ruminal sampling methods.

Keywords: milk production efficiency; milk yield; next-generation sequencing; oral community; oral swab; rumen community.

Fig 1

Spearman’s correlation heatmap of the top 75 genera against days in milk. The relative abundances of the top 75 genera with >0.1% relative sequence abundance in at least one animal and over 50% prevalence in all animals were correlated against days in milk. Correlations were determined for all animals combined and for the individual parities. The darkness of the color indicates the strength of the negative (purple) or positive (red) correlation score, with an asterisk indicating a significant relationship between days in milk and the relative abundance with false discovery rate correction for multiple tests (P < 0.05).

Fig 2

Linear modeling of days in milk (A and B) and milk yield (C and D) by PCA axes for first lactation animals. The location of the first lactation animals on the PCA axes was extracted and plotted against the animals’ respective days in milk or average milk yields. The lm() function in R was used to test for a significant relationship (P < 0.05) between the values.

Fig 3

Linear modeling of residual feed intake by the PCA X-axis for second lactation animals. The location of the second lactation animals for which we have residual feed intake values was plotted against the X-axis of the PCA plot. The lm() function in R was used to test for a significant relationship between the values (P < 0.05).