On the comparison of the strength of morphological integration across morphometric datasets

• DOI: 10.1111/evo.13045
• Publication Date: 2016-09-05T05:53:09Z
• Authors: Dean C. Adams; Michael L. Collyer
• Keywords: 0106 biological sciences; 0301 basic medicine; 570; Models, Genetic; Evolution; Ecology and Evolutionary Biology; 610; Datasets as Topic; Genetic Variation; Lizards; 01 natural sciences; 6. Clean water; Statistical Methodology; Evolution, Molecular; 03 medical and health sciences; Animals; Body Size; Least-Squares Analysis; 10. No inequality; Algorithms

Evolutionary morphologists frequently wish to understand the extent which organisms are integrated, and whether strength of morphological integration among subsets phenotypic variables differ taxa or other groups. However, comparisons across datasets difficult, in part because summary measures that characterize these patterns (RV coefficient rPLS) dependent both on sample size number variables. As a solution this issue, we propose standardized test statistic (a z-score) for measuring degree between sets The approach is based partial least squares analysis trait covariation, its permutation-based sampling distribution. Under null hypothesis random association variables, method displays constant expected value confidence intervals differing sizes variable number, thereby providing consistent measure suitable datasets. A two-sample also proposed statistically determine levels datasets, an empirical example examining cranial shape Mediterranean wall lizards illustrates use. Some extensions procedure discussed.