Table 1 Correlation comparisons between manual landmarking result and DAA for both the “Aligned-only” and “Poisson” meshes. The results show the correlations between matrices evaluated using the Mantel Test [42, 43], with results demonstrated using the Mantel r statistic, where a value of 0 shows no correlation, and 1 shows a perfect correlation, the Procrustes sum of squares, where a value of 0 indicates a perfect fit and 1 indicates high dissimilarity, and the Procrustes root mean squares estimates (high value means more error) using the PROcrustean randomisation TEST (PROTEST) [44, 45]. The correlation between the percentages of variation (eigenvalues) across each of the 321 axes is also given where an R2 value of 1 indicates a perfect fit. All results are significant (p < 0.01), with the best values for each statistic being highlighted in bold
From: Assessing the application of landmark-free morphometrics to macroevolutionary analyses
Data Set | Mantel r statistic | Procrustes sum of squares | Procrustes root mean squared error | Eigenvalue correlation (R2) |
|---|---|---|---|---|
Aligned-only 40 | 0.2799 | 0.762 | 0.049 | 0.942 |
Aligned-only 20 | 0.2647 | 0.666 | 0.045 | 0.962 |
Aligned-only 10 | 0.3098 | 0.608 | 0.043 | 0.975 |
Poisson 40 | 0.6574 | 0.503 | 0.040 | 0.993 |
Poisson 20 | 0.6107 | 0.418 | 0.036 | 0.978 |
Poisson 10 | 0.5879 | 0.430 | 0.037 | 0.962 |