4.3. Statistical Analysis

A generalized linear models procedure was used to evaluate the data. The effects of family, temperature, and family-temperature interaction (genotype-environment interaction, GEI) on TD and BW were examined using the following model:

where Y_ijk is the phenotypic observations of each individual, μ is the general mean, family_i is the family effect, temperature_j is the temperature effect, (family × temperature)_ij is the GEI, and e_ijk is the residual error.

In order to examine the family re-ranking effects between environments and detect the temperature effect on growth traits in each family, the family effect was re-examined in each temperature environment, and the temperature effect was re-examined in each family using one-way ANOVA. Duncan’s multiple comparisons were made between families. Variance components of family, temperature, and family-temperature interaction were evaluated using the restricted maximum likelihood method. Temperature differences between periods were examined by one-way ANOVA and Duncan’s multiple comparisons. The differences of BW between living and dead individuals were tested in the HE treatment using one-way ANOVA. Chi-square test was used to analyze the significance of SR between environments and between families. The temperature effect on the SGR of BW in each month was examined using one-way ANOVA with families as replicates. All analyses were conducted in SPSS version 16.0 (SPSS, Inc., Chicago, IL, USA). The significance level of all analyses was set as p < 0.05.