2.5. Statistical Analysis

Comparisons among groups were done by analysis of variance, using the general linear model procedure of SAS (GLM; SAS Institute Inc., Cary, NC, USA), after normality testing of the data. Maternal body weight and body condition score changes were analyzed through repeated measures analysis, considering pregnancy rank (single or twin) and nutrition (natural pasture or natural pasture plus supplement) as independent effects. Fetal data were analyzed using a linear model including the fixed effects of pregnancy rank, the nutritional plane, the sex of the fetus, as well as their interactions and the random effects of ewes, to adjust for twin pregnancy. The effects of fetal sex and of the interactions of treatment and sex were not significant (p > 0.05) and, therefore, were removed from the model. In addition, Pearson correlations between cord blood PO₂, O₂ content, and oxidative stress biomarkers with fetal and placental traits were also calculated. The differences were considered significant when p < 0.05. The results were expressed as means ±SEM.