Prediction of growth and feed efficiency performances in growing rabbits from their gut microbiota

Selecting for feed efficiency (FE) is challenging in rabbit breeding programs since individual feed intake (FI) of animals is not usually available. The current study raises the possibility that rabbits’ cecal microbiota could be considered to predict FE and its component traits, i.e., growth and FI. Our dataset comprised the individual average daily gain (ADG) and cage FI records of 425 kits raised in two farms and fed with the same diet supplemented or not with antibiotics but under different feeding regimes. A 16S rRNA gene amplicons MiSeq sequencing assessment was conducted on cecal samples collected from those kits at 66 days. Paired-end sequences were processed with QIIME2 software resulting in a final table of 2,638 sequence variants for 424 samples. We run cross-validations fitting a sparse partial least squares regression (sPLSR) model with a microbial effect to assess its predictive ability on phenotypic ADG of animals fed V (ADGv) and on animals fed R (ADGR), and on their residuals after correction by management factors. For traits ADG, FI and FE, we run cross-validations to compare two models differing by including or not the cecal microbial information. Our sPLSR model showed some predictive capacity for phenotypic ADGV (0.40) and ADGR (0.09), but this capacity becomes null for the prediction of the residuals of these traits. Although cecal microbiota explained more than 50% of the variation of ADG, fitting the microbial effect in the model did not improve the predictive accuracy of the recorded values. Cecal microbiota explained 51 and 59% of the variation of FI and FCR, respectively. Unlike growth, models that considered the microbial information improved the predictive accuracy for FI and FCR recorded performance values in 3 and 10%, respectively.