Investigating relationships between the host genome, rumen microbiome, and dairy cow feed efficiency using mediation analysis with structural equation modeling

Abstract

ABSTRACT: The rumen microbiome is crucial for converting feed into absorbable nutrients used for milk synthesis, and the efficiency of this process directly affects the profitability and sustainability of the dairy industry. Recent studies have found that the rumen microbial composition explains part of the variation in feed efficiency traits, including dry matter intake, milk energy, and residual feed intake. The main goal of this study was to reveal relationships between the host genome, the rumen microbiome, and dairy cow feed efficiency using structural equation models. Our specific objectives were to (1) infer the mediation effects of the rumen microbiome on feed efficiency traits, (2) estimate the direct and total heritability of feed efficiency traits, and (3) calculate the direct and total breeding values of feed efficiency traits. Data consisted of dry matter intake, milk energy, and residual feed intake records, SNP genotype data, and 16S rRNA rumen microbial abundances from 448 mid-lactation Holstein cows from 2 research farms. We implemented structural equation models such that the host genome directly affects the phenotype (GP → P) and the rumen microbiome (GM → P), and the microbiome affects the phenotype (M → P), partially mediating the effect of the host genome on the phenotype (G → M → P). We found that 7% to 30% of microbes within the rumen microbial community had structural coefficients different from zero. We classified these microbes into 3 groups that could have different uses in dairy farming. Microbes with heritability <0.10 but significant causal effects on feed efficiency are attractive for external interventions. On the other hand, 2 groups of microbes with heritability ≥0.10, significant causal effects, and genetic covariances and causal effects with the same or opposite sign to feed efficiency are attractive for selective breeding, improving or decreasing the trait heritability and response to selection, respectively. In general, the inclusion of the different microbes in genomic models tends to decrease the trait heritability rather than increase it, ranging from −15% to +5% depending on the microbial group and phenotypic trait. Our findings provide more understanding to target rumen microbes that can be manipulated, either through selection or management interventions, in order to improve feed efficiency traits.

Keywords

• dry matter intake
• microbiome
• structural equation model
• total heritability