RRphylo

RRphylo detects morphological convergence across phylogenetically distant taxa by computing phenotypic evolutionary rates and testing whether morphological similarities exceed expectations given phylogenetic distances, supporting analysis of ultrametric and non-ultrametric phylogenies and multivariate datasets.

Key Features:

• Phenotypic evolutionary rates and ancestral states: Computes phenotypic evolutionary rates, estimates ancestral states, and identifies shifts in rates across phylogenetic trees.
• Vector comparison: Compares phenotype vectors and their associated evolutionary-rate vectors among species or clades.
• Model-free approach: Operates without assuming a specific evolutionary model, enabling analysis of paleontological (non-ultrametric) phylogenies.
• Performance of search.conv: Implements the core convergence test in the search.conv function, which is reported to be approximately three orders of magnitude faster than competing methods for testing morphological convergence.
• Incorporation of fossil data: Can incorporate known phenotypes as reconstructed most recent common ancestors within clades to maximize use of fossil information.
• Phylogeny and data support: Works with both ultrametric and non-ultrametric phylogenies and with multivariate datasets.
• Implementation environment: Implemented in the R programming environment.

Scientific Applications:

• Ungulate grazing adaptations: Applied to investigate the evolution of grazing adaptations in ungulates characterized by high-crowned molars.
• Sabertooth mandibular evolution: Used to study changes in mandibular shape among sabertooth cats.
• Anole ecomorphs: Used to analyze the development of discrete ecomorphs among Caribbean anoles.

Methodology:

Implemented in R, the search.conv method assesses whether morphological similarities between unrelated clades exceed expectations given their phylogenetic distances and evaluates significance via simulations that reported ~95% success in identifying simulated convergent scenarios and a type I error rate of 4–6%.