Phylogenetic ANOVA: Group-clade aggregation, biological challenges, and a refined permutation procedure

Area of Science:

• Macroevolutionary biology
• Phylogenetic comparative methods
• Quantitative genetics

Background:

• Phylogenetic regression is well-studied, but phylogenetic ANOVA requires further investigation for accurate macroevolutionary inferences.
• Current methods for comparing multivariate phenotypes across groups using phylogenies can lead to incorrect statistical and biological conclusions.
• Understanding how phylogenetic structure influences group comparisons is crucial for robust macroevolutionary analyses.

Purpose of the Study:

• To develop and validate a refined permutation procedure for phylogenetic ANOVA and regression.
• To assess the statistical properties and power of the proposed method compared to existing techniques.
• To investigate the impact of phylogenetic group aggregation on statistical inferences in comparative studies.

Main Methods:

• A novel permutation procedure, Randomizing Residuals in a Permutation Procedure (RRPP), was developed.
• RRPP was applied to both phylogenetic ANOVA and regression models for univariate and multivariate datasets.
• Statistical properties and power were evaluated, including comparisons with phylogenetic simulation methods.
• The influence of phylogenetic group dispersion on statistical power and biological interpretation was analyzed.

Main Results:

• RRPP demonstrates appropriate statistical properties for phylogenetic ANOVA and regression across various data types.
• RRPP exhibits higher statistical power for phylogenetic ANOVA than methods using phylogenetic simulation.
• Highly aggregated groups along the phylogeny reduce statistical power and can bias biological interpretations.
• The method effectively conditions data on the phylogeny for robust group comparisons.

Conclusions:

• RRPP is a statistically sound and powerful method for phylogenetic ANOVA and regression.
• Researchers should be cautious about phylogenetic group aggregation, as it can compromise inferences.
• RRPP is recommended for phylogenetic comparative studies of continuous traits to assess trait variation sources.