Harnessing gastrointestinal microbial co-association networks to predict feed efficiency and methane emissions across...

Enteric methane emissions from cattle pose a significant environmental concern and represent a substantial energy loss for the animal, necessitating the development of effective mitigation strategies. The gastrointestinal microbiota plays a crucial role in determining both feed efficiency and methane production. Still, the specific microbial signatures that predict these traits across different production systems remain poorly understood. This study aimed to identify common predictive microbial biomarkers for feed efficiency and methane emissions using co-association network analysis across contrasting cattle production systems. Rumen liquid and faecal microbiota from 55 Charolais heifers (beef) and 56 Holstein cows (dairy) were analysed using 16S rRNA gene amplicon sequencing. Phenotypic data included feed efficiency, methane yield and acetate/propionate ratio. Co-association networks were constructed using Partial Correlation and Information Theory to identify amplicon sequence variants (ASVs) directly connected to phenotypes. Multiple regression analysis determined the minimal ASV sets required to achieve optimal predictive accuracy. Rumen microbiomes consistently showed superior predictive performance compared to faecal communities across all traits. Network-selected ASVs explained substantial phenotypic variance across traits and production systems (R²=0.45-0.84), consistently outperforming randomly selected ASVs by 0.13-0.44 R² units. The ACET:PROP ratio showed the highest predictive accuracy (R²=0.84 in Charolais rumen, 0.77 in Holstein rumen), while minimal ASV sets achieved 93-97% of the full model's performance using 33-65% fewer ASVs (6-17 ASVs). Bacteroidaceae was consistently enriched across phenotypes in rumen networks, regardless of production system. Contrary to expectations, most associations were production system-specific, with notable exceptions including negative correlations between the ACET:PROP ratio and Prevotella/Ruminococcus genera and negative associations with members of the Succinivibrionaceae family for methane-related traits. The anatomical site-specific and production system-specific nature of most associations underscores the importance of context-specific approaches.