Table 1.
Statistical procedures used in the analysis. These are listed in the order described and reported in the Materials and methods and Results sections
| Statistical test or procedure | Objective | |
|---|---|---|
| 1. Comparisons of size and shape | ||
| a. anova for body mass | Comparison of mean body mass among strains. | |
| b. Principal components analysis | This analysis is used to produce new variables that summarize shape variation in the Procrustesof Procrustes coordinate data coordinate data. The variables that explain the most shape variation are then plotted for the three strains. Shape variation along principal components can also be visualized by obtaining the predicted Procrustes values along each component. | |
| c. EDMA FORM and SHAPE analysis | These tests compare the complete set of Euclidean distances (scaled and unscaled) among groups and provide boostrap based P-values for individual interlandmark distances. | |
| 2. Analysis of dominance | To test the hypothesis that the shape differences between the two parental strains exhibit significant dominance deviations. | |
| 3. Fluctuating asymmetry comparisons | ||
| a. Comparison of overall FA using Klingberg's object asymmetry method. | This method tests for the significance of the overall FA difference between between groups. | |
| b. EDMA Asymmetry Comparison | Tests for the significance of the FA difference for individual interlandmark distances. | |
| 4. Among individual variance | ||
| a. Levene's test on Procrustes mean-shape deviations. | Tests for differences in among-individual variation in overall shape among strains. | |
| b. Levene's test on interlandmark distances | Tests for differences in among individual variation for interlandmark distances among strains. | |
| 5. Integration of size and shape | ||
| a. Regressions of principal components against centroid size | To compare the amount of variation explained by size for particular shape components. | |
| b. Regression of shape mean-deviations against centroid size-deviations. | This method estimates the total percentage of shape variation that is explained by size. | |
| c. Matrix correlations with Mantel's test for significance for both the Procrustes and the interlandmark distance matrices. | This test is used to test the null hypothesis that the pairs of covariation matrices among the three groups (and by sex) are unrelated. | |
| d. anova for z-transformed correlations | To compare overall magnitudes of size integration among groups. | |
| e. Monte carlo test to compare the variances of eigenvalues. | To compare overall magnitudes of shape integration among groups. | |
| f. EDMA-based test for integration. | To compare magnitudes of size integration for particular interlandmark distances. | |
| g. Comparison of variances of eigenvalues for Procrustes centroid-size residuals. | To compare shape integration among groups after removing the allometric component of shape variation. | |