Search

Highlights: [•] Rapid bootstrapping can provide inflated support values for polytomies. [•] The SH-like aLRT outperforms the bootstrap when applied to polytomies. [•] The 70% bootstrap cutoff does not equate to a 5% error rate. [•] Rapid bootstrapping with GTRCAT can generate values with very low precision. [Copyright &y& Elsevier]

Abstract: A supermatrix of 272 terminals from Rubiaceae tribe Spermacoceae that were scored for up to 10 gene regions (two nrDNA, eight plastid) was used as an empirical example to quantify sources of error in heuristic parametric (Bayesian MCMC and maximum likelihood) phylogenetic analyses. The supermatrix includes dramatic disparities in which terminals were sampled for which gene regions. The sources of error examined include poor quality tree searches, requiring a single fully resolved optimal tree, undersampling-within-replicates and frequency-within-replicates bootstrap artifacts, and extrapolation from one character partition to another such that synapomorphies that would only be ambiguously optimized by parsimony are optimized with high probability by parametric methods. Four of our conclusions are as follows. (1) The resolution and support provided by parametric methods for clades that lack unambiguously optimized (by parsimony) synapomorphies are less robust to the addition of terminals and characters than those clades that have unambiguously optimized synapomorphies. (2) Those tree-search methods which can create phylogenetic artifacts (frequency-within-replicates resampling, undersampling-within-replicates resampling, requiring a single fully resolved optimal tree, non-independent resampling among replicates) provide the greatest resolution and support irrespective of whether that resolution or support is corroborated by more conservative and better justified methods. (3) Partitioning data matrices cannot be relied upon to consistently obviate potentially dubious resolution and support caused by missing-data artifacts in likelihood analyses when the models require linked branch lengths among partitions. (4) Undersampling-within-replicates and frequency-within-replicates resampling artifacts are not unique to parsimony and should be accounted for in likelihood analyses by allowing multiple equally likely trees to be saved within each resampling pseudoreplicate and applying the strict-consensus bootstrap rather than the frequency-within-replicates bootstrap. [Copyright &y& Elsevier]

Abstract: We used random sequences to determine which alignment methods are most susceptible to aligning sequences so as to create artifactual resolution and branch support in phylogenetic trees derived from those alignments. We compared four alignment methods (progressive pairwise alignment, simultaneous multiple alignment of sequence fragments, local pairwise alignment, and direct optimization) to determine which methods are most susceptible to creating false positives in phylogenetic trees. Implied alignments created using direct optimization provided more artifactual support than progressive pairwise alignment methods, which in turn generally provided more artifactual support than simultaneous and local alignment methods. Artifactual support derived from base pairs was generally reinforced by the incorporation of gap characters for progressive pairwise alignment, local pairwise alignment, and implied alignments. The amount of artifactual resolution and support was generally greater for simulated nucleotide sequences than for simulated amino acid sequences. In the context of direct optimization, the differences between static and dynamic approaches to calculating support were extreme, ranging from maximal to nearly minimal support. When applied to highly divergent sequences, it is important that dynamic, rather than static, characters be used whenever calculating branch support using direct optimization. In contrast to the tree-based approaches to alignment, simultaneous alignment of sequences using the similarity criterion generally does not create alignments that are biased in favor of any particular tree topology. [ABSTRACT FROM AUTHOR]

Abstract: We used simulations to compare the performance of 10 approaches that have been used for treating unambiguously aligned gaps in phylogenetic analyses. We examined how these approaches perform under the ideal conditions of correct alignments, as well as how robust they are to errors caused by use of inferred alignments. Our results indicate that 5th-state coding dramatically outperformed all other coding methods, which in turn all outperformed treating gaps as missing data or excluding gapped positions. Simple indel coding (SIC) and modified complex indel coding (MCIC) performed about the same, and generally outperformed the other indel-coding methods. The high performance of 5th-state coding was largely found to be a weighting artifact. We suggest that MCIC-coded gap characters be scored for all unambiguously aligned gaps in parsimony-based molecular phylogenetic analyses. When the number of terminals sampled precludes the use of MCIC, SIC may be used as an effective substitute. [Copyright &y& Elsevier]