Sas version 9.2 2008

A randomized complete block design was used in this study with sex and initial BW as blocking criteria. Mixed procedure of SAS (version 9.2, 2008) [26 ] was used for statistical analysis. Dietary treatment was the fixed effect, while the block was the random effect. For performance, diarrhea rate and nutrient digestibility, individual pen was used as statistical unit, while for other data, the individual pig was used as statistical unit. The difference of diarrhea rate was analyzed by chi-square contingency test, while the statistical differences of other data except for the microbiota were analyzed via the Student–Neuman–Keul’s multiple range tests. Differences of microbiota abundance in feces were analyzed using a Kruskal–Wallis rank sum test. Significant difference was defined as p ≤ 0.05, and a trend of difference was determined as 0.05 < p ≤ 0.10.

The mixed model of SAS (version 9.2, 2008) [26 ] was used for variance analysis of all the data. The dietary treatments were fixed effects, while sex and body weight of pigs were the random effects. For the analysis of growth performance and diarrhea rate, the pen was treated as the statistical unit, whereas for the analysis of other data, individual piglets were taken as a statistical unit. The Student–Newman–Keul multiple range test was used for determining the statistical differences among all the treatments. Significant difference between the mean value was defined at p ≤ 0.05, while a trend for the significance between the mean value was designated at 0.05 < p ≤ 0.10.