Microbial and Genomic Information Synergistically Contribute to Predicting Swine Performance Across Production Systems.

Abstract

Microbiota composition represents a promising tool in precision farming, simultaneously serving as a benchmark of environmental challenge, a predictor of animal physiological status, and a direct target for host selection. In this paper, we compared the ability of microbiota composition and genomic information to predict swine performance in two production settings, namely a purebred nucleus (NU) and a terminal cross commercial population (TE). Microbiota consistently predicted all traits in both scenarios (NU-TE: training on NU to predict TE; TE-NU: training on TE to predict NU) and at two time points: mid-test and off-test. The highest correlation (i.e., prediction accuracy) was achieved for back fat, with values of 0.08 and 0.04, and 0.30 and 0.23 for mid and off-tests, predicting from nucleus to terminal, and vice versa. Similarly, daily gains correlations were 0.05 and 0.04, and 0.18 and 0.15 for the same time points and scenario combinations. Including genomic information yielded correlations ranging from low for loin area to moderate for back fat (0.19 nucleus to terminal, 0.16 for the opposite). Microbiota had higher prediction accuracies than genomic for back fat both from nucleus to terminal and vice versa (+0.11, +0.07) and daily gain (+0.08, +0.02) at off-test. Lower accuracies were obtained for the IMF. Including genomic and microbial information produced higher accuracies than microbiota or genomic alone for back fat (0.37 and 0.29 for nucleus to terminal and opposite) and daily gain (0.19 and 0.21 for nucleus to terminal and opposite). Results for other traits differed for different scenarios. Results show that microbiota composition effectively predicted most growth and carcass traits, particularly growth and fat deposition, across production systems, prediction scenarios (NU-TE and TE-NU), and time points (mid-test and off-test). These findings highlight the potential of microbiota profiles to predict phenotypes across production systems and support their use as a tool for selecting animals in environments they have not been exposed to.