Statistical analysis
Linear mixed models (LMM) were fitted with ASReml-R (Butler et al., 2018) in the analysis of the frost and growth stage phenotypic traits, where the variance parameters in the mixed model are estimated using the residual maximum likelihood (REML) procedure (Patterson & Thompson, 1971). Residual diagnostics were performed to examine the validity of the model assumption (normality and homogeneity of variance). Both best linear unbiased predictions (BLUPs) and best linear unbiased estimations (BLUEs) were obtained for the phenotypic traits.
Phenotypic data were analyzed by multivariate analysis of variance (MANOVA) using the general linear model implemented in IBM SPSS statistics 24 (https://www.ibm.com/au-en/products/spss-statistics). Wilks Lambda was used as the multivariate test statistic. Post-hoc Tukey’s Multiple Range tests were used to identify significant groupings. Pearson was used for correlation analysis.