Exploring Phylogenetic Signal in Multivariate Phenotypes by Maximizing Blomberg's K

Area of Science:

• Evolutionary Biology
• Phylogenetics
• Quantitative Genetics

Background:

• Phylogenetic signal describes the tendency of related species to share similar traits.
• Measuring phylogenetic signal for single traits is established, but challenging for multivariate data.
• Modern biological studies frequently involve complex, multivariate phenotypic data.

Purpose of the Study:

• To develop a novel method for exploring phylogenetic signal in multivariate phenotypes.
• To introduce interpretable components and summary statistics for multivariate phylogenetic signal.
• To assess the performance of new statistics against existing methods.

Main Methods:

• Decomposition of multivariate data into linear combinations (K-components) maximizing/minimizing phylogenetic signal (Blomberg's K).
• Development and algebraic/statistical characterization of new summary statistics KA and KG.
• Simulation studies comparing KA and KG with the statistic Kmult.
• Empirical application to vertebrate cranial shape data (crocodyliforms and papionins).

Main Results:

• The new method allows for biological interpretation of components reflecting phylogenetic signal.
• KA and KG demonstrated higher statistical power than Kmult, particularly for low or concentrated signal.
• Significant phylogenetic signal was detected in specific dimensions of vertebrate cranial shape.

Conclusions:

• The proposed method effectively quantifies multivariate phylogenetic signal.
• New statistics KA and KG provide a powerful tool for evolutionary analyses.
• Phylogenetic signal can be unevenly distributed across trait dimensions, impacting phylogenetic inference.