A method for assessing phylogenetic least squares models for shape and other high-dimensional multivariate data.

Area of Science:

• Evolutionary biology
• Comparative biology
• Bioinformatics

Background:

• Phylogenetic regression methods like independent contrasts and phylogenetic generalized least squares are standard for assessing trait covariation.
• These methods are limited with high-dimensional data due to the ratio of variables to sample size, hindering statistical analysis.
• This limitation forces biologists to simplify trait quantification or alter research hypotheses.

Purpose of the Study:

• To propose a novel statistical procedure for phylogenetic ANOVA and regression models capable of handling high-dimensional datasets.
• To provide a method that overcomes the limitations of current phylogenetic regression techniques for complex trait data.

Main Methods:

• The new approach leverages the statistical equivalence between covariance matrix-based parametric methods and distance matrix-based methods.
• Simulations were conducted under a Brownian motion model to evaluate the procedure's performance.
• The method was tested for Type I error rates and statistical power in high-dimensional scenarios.

Main Results:

• The proposed method demonstrated appropriate Type I error rates and statistical power in simulations.
• In contrast, standard parametric procedures showed decreasing statistical power as data dimensionality increased.
• The new procedure effectively assesses trait covariation in high-dimensional datasets.

Conclusions:

• This new statistical procedure offers a robust solution for analyzing trait covariation in high-dimensional phylogenetic datasets.
• It enables macroevolutionary biologists to test hypotheses related to adaptation and phenotypic change with complex data.
• The method expands the analytical capabilities for comparative biologists working with multivariate trait data.