2.8. Statistical analysis
Data were analyzed as a randomized complete block design by using
the GLM procedure (SAS Inst. Inc., Cary, NC, USA). Paddock was the experimental unit, and the model effects included treatment. Orthogonal
contrasts were used to determine the linear, cubic and quadratic effects
of glycerin and 0% glycerin vs. glycerin treatment, with significance considered at P b 0.05.