Systematic Biology - Advance Access

Cryptic Failure of Partitioned Bayesian Phylogenetic Analyses: Lost in the Land of Long Trees

Marshall, D. C. — Tue, 17 Nov 2009 09:38:37 PST

Partitioned Bayesian phylogenetic analyses of routine genetic data sets, constructed using MrBayes (Ronquist and Huelsenbeck 2003), can become trapped in regions of parameter space characterized by unrealistically long trees and distorted partition rate multipliers. Such analyses commonly fail to reach stationarity during hundreds of millions of generations of sampling—many times longer than most published analyses. Some data sets are so prone to this problem that paired MrBayes runs begun from different starting trees repeatedly find the same incorrect long-tree solutions and consequently pass the most commonly employed tests of stationarity, including the average standard deviation of split frequencies (ASDSF) and the potential scale reduction factor (PSRF) statistics offered by MrBayes (Gelman and Rubin 1992). In these situations, failure to reach stationarity is recognizable only in light of prior knowledge of model parameters, such as the expectation that third-codon-position sites usually evolve fastest in protein-coding genes. The conditions that lead to the long-tree problem are frequently encountered in phylogenetic studies today, and I present 6 demonstration examples from the literature. Although the effects on tree length (TL) are often dramatic, effects on topology appear to be subtle. Susceptibility to the problem is sometimes predicted by the difference between the true TL and the starting TL. In some cases, the problems described here can be avoided or reduced by manipulation of the starting TL and/or by adjustments to the prior on branch lengths. In more difficult situations, accurate branch length estimation may not be possible with Bayesian methods because of dependence of the solution on the branch length prior.

Tinamous and Moa Flock Together: Mitochondrial Genome Sequence Analysis Reveals Independent Losses of Flight among Ratites

Phillips, M. J., Gibb, G. C., Crimp, E. A., Penny, D. — Fri, 13 Nov 2009 13:21:53 PST

Ratites are large, flightless birds and include the ostrich, rheas, kiwi, emu, and cassowaries, along with extinct members, such as moa and elephant birds. Previous phylogenetic analyses of complete mitochondrial genome sequences have reinforced the traditional belief that ratites are monophyletic and tinamous are their sister group. However, in these studies ratite monophyly was enforced in the analyses that modeled rate heterogeneity among variable sites. Relaxing this topological constraint results in strong support for the tinamous (which fly) nesting within ratites. Furthermore, upon reducing base compositional bias and partitioning models of sequence evolution among protein codon positions and RNA structures, the tinamou–moa clade grouped with kiwi, emu, and cassowaries to the exclusion of the successively more divergent rheas and ostrich. These relationships are consistent with recent results from a large nuclear data set, whereas our strongly supported finding of a tinamou–moa grouping further resolves palaeognath phylogeny. We infer flight to have been lost among ratites multiple times in temporally close association with the Cretaceous–Tertiary extinction event. This circumvents requirements for transient microcontinents and island chains to explain discordance between ratite phylogeny and patterns of continental breakup. Ostriches may have dispersed to Africa from Eurasia, putting in question the status of ratites as an iconic Gondwanan relict taxon.

Sparse Supermatrices for Phylogenetic Inference: Taxonomy, Alignment, Rogue Taxa, and the Phylogeny of Living Turtles

Thomson, R. C., Shaffer, H. B. — Wed, 11 Nov 2009 07:05:40 PST

As phylogenetic data sets grow in size and number, objective methods to summarize this information are becoming increasingly important. Supermatrices can combine existing data directly and in principle provide effective syntheses of phylogenetic information that may reveal new relationships. However, several serious difficulties exist in the construction of large supermatrices that must be overcome before these approaches will enjoy broad utility. We present analyses that examine the performance of sparse supermatrices constructed from large sequence databases for the reconstruction of species-level phylogenies. We develop a largely automated informatics pipeline that allows for the construction of sparse supermatrices from GenBank data. In doing so, we develop strategies for alleviating some of the outstanding impediments to accurate phylogenetic inference using these approaches. These include taxonomic standardization, automated alignment, and the identification of rogue taxa. We use turtles as an exemplar clade and present a well-supported species-level phylogeny for two-thirds of all turtle species based on a ~50 kb supermatrix consisting of 93% missing data. Finally, we discuss some of the remaining pitfalls and concerns associated with supermatrix analyses, provide comparisons to supertree approaches, and suggest areas for future research.

New Heuristic Methods for Joint Species Delimitation and Species Tree Inference

O'Meara, B. C. — Tue, 10 Nov 2009 12:14:41 PST

Species delimitation and species tree inference are difficult problems in cases of recent divergence, especially when different loci have different histories. This paper quantifies the difficulty of jointly finding the division of samples to species and estimating a species tree without constraining the possible assignments a priori. It introduces a parametric and a nonparametric method, including new heuristic search strategies, to do this delimitation and tree inference using individual gene trees as input. The new methods were evaluated using thousands of simulations and 4 empirical data sets. These analyses suggest that the new methods, especially the nonparametric one, may provide useful insights for systematists working at the species level with molecular data. However, they still often return incorrect results.

Unifying Vertical and Nonvertical Evolution: A Stochastic ARG-based Framework

Bloomquist, E. W., Suchard, M. A. — Mon, 09 Nov 2009 11:39:32 PST

Evolutionary biologists have introduced numerous statistical approaches to explore nonvertical evolution, such as horizontal gene transfer, recombination, and genomic reassortment, through collections of Markov-dependent gene trees. These tree collections allow for inference of nonvertical evolution, but only indirectly, making findings difficult to interpret and models difficult to generalize. An alternative approach to explore nonvertical evolution relies on phylogenetic networks. These networks provide a framework to model nonvertical evolution but leave unanswered questions such as the statistical significance of specific nonvertical events. In this paper, we begin to correct the shortcomings of both approaches by introducing the "stochastic model for reassortment and transfer events" (SMARTIE) drawing upon ancestral recombination graphs (ARGs). ARGs are directed graphs that allow for formal probabilistic inference on vertical speciation events and nonvertical evolutionary events. We apply SMARTIE to phylogenetic data. Because of this, we can typically infer a single most probable ARG, avoiding coarse population dynamic summary statistics. In addition, a focus on phylogenetic data suggests novel probability distributions on ARGs. To make inference with our model, we develop a reversible jump Markov chain Monte Carlo sampler to approximate the posterior distribution of SMARTIE. Using the BEAST phylogenetic software as a foundation, the sampler employs a parallel computing approach that allows for inference on large-scale data sets. To demonstrate SMARTIE, we explore 2 separate phylogenetic applications, one involving pathogenic Leptospirochete and the other Saccharomyces.

Phylogenetic Logistic Regression for Binary Dependent Variables

Ives, A. R., Garland, T. — Wed, 04 Nov 2009 10:58:54 PST

We develop statistical methods for phylogenetic logistic regression in which the dependent variable is binary (0 or 1) and values are nonindependent among species, with phylogenetically related species tending to have the same value of the dependent variable. The methods are based on an evolutionary model of binary traits in which trait values switch between 0 and 1 as species evolve up a phylogenetic tree. The more frequently the trait values switch (i.e., the higher the rate of evolution), the more rapidly correlations between trait values for phylogenetically related species break down. Therefore, the statistical methods also give a way to estimate the phylogenetic signal of binary traits. More generally, the methods can be applied with continuous- and/or discrete-valued independent variables. Using simulations, we assess the statistical properties of the methods, including bias in the estimates of the logistic regression coefficients and the parameter that estimates the strength of phylogenetic signal in the dependent variable. These analyses show that, as with the case for continuous-valued dependent variables, phylogenetic logistic regression should be used rather than standard logistic regression when there is the possibility of phylogenetic correlations among species. Standard logistic regression does not properly account for the loss of information caused by resemblance of relatives and as a result is likely to give inflated type I error rates, incorrectly identifying regression parameters as statistically significantly different from zero when they are not.

Relaxed Molecular Clocks, the Bias-Variance Trade-off, and the Quality of Phylogenetic Inference

Wertheim, J. O., Sanderson, M. J., Worobey, M., Bjork, A. — Tue, 13 Oct 2009 06:56:20 PDT

Because a constant rate of DNA sequence evolution cannot be assumed to be ubiquitous, relaxed molecular clock inference models have proven useful when estimating rates and divergence dates. Furthermore, it has been recently suggested that using relaxed molecular clocks may provide superior accuracy and precision in phylogenetic inference compared with traditional time-free methods that do not incorporate a molecular clock. We perform a simulation study to determine if assuming a relaxed molecular clock does indeed improve the quality of phylogenetic inference. We analyze sequence data simulated under various rate distributions using relaxed-clocks, strict-clocks, and time-free Bayesian phylogenetic inference models. Our results indicate that no difference exists in the quality of phylogenetic inference between assuming a relaxed molecular clock and making no assumption about the clock-likeness of sequence evolution. This pattern is likely due to the bias–variance trade-off inherent in this type of phylogenetic inference. We also compared the quality of inference between Bayesian and maximum likelihood time-free inference models and found them to be qualitatively similar.