Abstract: Comparative analyses in ecology and evolution often face the challenge of controlling for the effects of shared ancestry (phylogeny) from those of ecological or trait-based predictors on species traits. Phylogenetic Generalized Linear Models (PGLMs) address this issue by integrating phylogenetic relationships into statistical models. However, accurately partitioning explained variance among correlated predictors remains challenging. The phylolm.hp R package tackles this problem by extending the concept of “average shared variance” to PGLMs, enabling nuanced quantification of the relative importance of phylogeny and other predictors. The package calculates individual likelihood-based R² contributions of phylogeny and each predictor, accounting for both unique and shared explained variance. This approach overcomes limitations of traditional partial R² methods, which often fail to sum the total R² due to multicollinearity. We demonstrate the functionality of phylolm.hp through two case studies: one involving continuous trait data (maximum tree height in Californian species) and another focusing on binary trait data (species invasiveness in North American forests). The phylolm.hp package offers researchers a powerful tool to disentangle the contributions of phylogenetic and ecological predictors in comparative analyses.

Key words: Average shared variance, Comparative analysis, Hierarchical portioning, Phylogenetic signal, Variation partitioning

摘要： Comparative analyses in ecology and evolution often face the challenge of controlling for the effects of shared ancestry (phylogeny) from those of ecological or trait-based predictors on species traits. Phylogenetic Generalized Linear Models (PGLMs) address this issue by integrating phylogenetic relationships into statistical models. However, accurately partitioning explained variance among correlated predictors remains challenging. The phylolm.hp R package tackles this problem by extending the concept of “average shared variance” to PGLMs, enabling nuanced quantification of the relative importance of phylogeny and other predictors. The package calculates individual likelihood-based R² contributions of phylogeny and each predictor, accounting for both unique and shared explained variance. This approach overcomes limitations of traditional partial R² methods, which often fail to sum the total R² due to multicollinearity. We demonstrate the functionality of phylolm.hp through two case studies: one involving continuous trait data (maximum tree height in Californian species) and another focusing on binary trait data (species invasiveness in North American forests). The phylolm.hp package offers researchers a powerful tool to disentangle the contributions of phylogenetic and ecological predictors in comparative analyses.

关键词: Average shared variance, Comparative analysis, Hierarchical portioning, Phylogenetic signal, Variation partitioning