Temporal stability of the rumen microbiome and its longitudinal associations with performance traits in beef cattle

Abstract

The rumen microbiome is the focus of a growing body of research, mostly based on investigation of rumen fluid samples collected once from each animal. Exploring the temporal stability of rumen microbiome profiles is imperative, as it enables evaluating the reliability of findings obtained through single-timepoint sampling. We explored the temporal stability of rumen microbiomes considering taxonomic and functional aspects across the 7-month growing-finishing phase spanning 6 timepoints. We identified a temporally stable core microbiome, encompassing 515 microbial genera (e.g., Methanobacterium) and 417 microbial KEGG genes (e.g., K00856-adenosine kinase). The temporally stable core microbiome profiles collected from all timepoints were strongly associated with production traits with substantial economic and environmental impact (e.g., average daily gain, daily feed intake, and methane emissions); 515 microbial genera explained 45-83%, and 417 microbial genes explained 44-83% of their phenotypic variation. Microbiome profiles influenced by the bovine genome explained 54-87% of the genetic variation of bovine traits. Overall, our results provide evidence that the temporally stable core microbiome identified can accurately predict host performance traits at phenotypic and genetic level based on a single timepoint sample taken as early as 7 months prior to slaughter.

Fig. 1

Schematic representation of the experimental timeline. Data was recorded during the 56-day performance testing period and methane emissions during the 3-day-long respiration chamber period. T1 to T6 represent each of the timepoints in which the rumen digesta was sampled from live animals. T6 corresponds to shortly after slaughter.

Fig. 2

Rumen microbial genera. Stacked bar chart of the ten microbial genera with the highest average relative abundances over all timepoints and animals. Rpt refers to repeatability, analysed across both diets. Microbial genera without an Rpt value were not significantly repeatable in this analysis.

Fig. 3

Clustering of samples based on beta diversity. Non-metric multi-dimensional scaling (NMDS) plot of Bray–Curtis distances between samples at a microbial genera and b microbial genes level. The stress values of < 0.2 above the panels indicate that the two-dimensional plot is a fair representation of the original high-dimensional distances with some acceptable distortion.

Fig. 4

Correlations between microbiome profiles of different timepoints. Pearson correlations between vectorized matrices of (a) 515 microbial genera (MT) and 417 microbial genes (MG), (b) 412 MT and 376 MG from animals fed concentrate (CON), and (c) 223 MT and 165 MG from animals fed forage (FOR), respectively. MT and MG abundances were centred and additive log-ratio transformed and then centred based on their mean abundance throughout all timepoints. Correlations of MT profiles from different timepoints are represented above the diagonal and MG are represented below the diagonal.

Fig. 5

Proportion of variance of each trait explained by repeatable microbiome features in six timepoints. Phenotypes (feed conversion ratio, average daily gain, daily feed intake, residual feed intake, methane yield and daily methane production) were predicted using partial least squares models including as explanatory variables different sets of microbial genera (all (i.e., 1050), 515 repeatable across both diets, 412 repeatable within concentrate-fed animals, 223 repeatable within forage-fed animals, and stableMT) or microbial genes (all (i.e., 1901), 417 repeatable across both diets, 376 repeatable within concentrate-fed animals, 165 repeatable within forage-fed animals, and stableMG). Each pie chart is read counterclockwise, in a cumulative manner; for example, using 1050 microbial genera from timepoints T3, T6, T5, T4, T1, and T2 explains 64.4%, 67.9%, 71.4%, 73.1%, 76.2% and 77.2%, respectively, of the variance in feed conversion ratio.

Fig. 6

Microbial genera and genes with temporally stable and strong relationship with host traits. Heatmaps show the direction of longitudinal associations (positive or negative) between selected a microbial genera and microbial genes and host performance traits. Bar charts present the average relative abundance and standard deviation of each b microbial genera and c microbial genes over six sampling timepoints.