Dissecting Phylogenetic Support: Unified Decay Indices, AU Tests, and Branch-Site Specific Visualizations.

Phylogenetic inference relies on robust measures of branch support to assess the reliability of evolutionary relationships. While bootstrap resampling and Bayesian probabilities have become the predominant support metrics in molecular phylogenetics, decay indices (also known as Bremer support) provide an alternative perspective by quantifying the amount of evidence required to collapse specific clades. We present panDecay, a unified software framework that extends the classical parsimony-based decay index concept to maximum likelihood (ML) and Bayesian inference approaches. panDecay automates the calculation of ML-based decay indices using log-likelihood differences, performs Approximately Unbiased (AU) statistical tests, and computes Bayesian decay indices through marginal likelihood comparisons using stepping-stone sampling. The software integrates with established phylogenetic tools (PAUP* and MrBayes) and provides comprehensive output including site-specific support analysis, multiple annotated tree formats, and detailed statistical reports. panDecay supports diverse data types (DNA, protein, and discrete morphological characters), offers flexible model specification, and provides parallel processing capabilities for computationally intensive Bayesian analyses. By unifying decay index calculations across parsimony, likelihood, and Bayesian frameworks within a single command-line interface, panDecay enables researchers to obtain complementary support metrics that provide deeper insights into phylogenetic signal strength and distribution across sequence alignments. The software is implemented in Python 3.8+, distributed via PyPI, and released under the MIT license.