The potential of microbiota information to better predict efficiency traits in growing pigs fed a conventional and a high-fiber diet

Table 1 Influence of five design effects on prediction accuracy: degrees of freedom, F-value, and associated P-value in combined analyses of variance

Effects	Degrees of freedom	F-value	P-value
Scenario^a	4	10.07	4.67 × 10^–5
SBE^b	1	109.49	< 2.2 × 10^–16
Type of trait (growth/feed vs. digestive)	1	634.61	< 2.2 × 10^–16
Model^c	2	109.49	< 2.2 × 10^–16
FEM^d	1	0.81	0.37
Residuals	1670

^aFive scenarios were analysed—COref/COval: pigs were fed a conventional diet in the reference and the validation populations; HFref/HFval: pigs were fed a high-fiber diet in the reference and the validation populations; COref/HFval: pigs were fed a conventional diet in the reference population and a high-fiber diet in the validation population; HFref/COval: pigs were fed a high-fiber diet in the reference population and a conventional diet in the validation population; COHFref/COHFval: pigs were fed a conventional + a high-fiber diet in the reference population and a conventional + a high-fiber diet in the validation population
^bSBE: sire breeding environment; the impact of a common sire breeding environment between the reference and validation populations was tested
^cThree models were used: Micro (with only microbiota information), Gen (with only genomic information), and Micro+Gen (with both microbiota and genomic information)
^dFEM: fixed effects in the model, for each trait, two phenotypes were studied: phenotypes corrected or not for environmental effects (diet and batch)

ISSN: 1297-9686