Bayesian estimation of fossil phylogenies and the evolution of early to middle Paleozoic crinoids (Echinodermata) David F. Wright* School of Earth Sciences, The Ohio State University, Columbus, OH 43215, USA 〈wright.1433@osu.edu〉 Abstract.—Knowledge of phylogenetic relationships among species is fundamental to understanding basic patterns in evo- lution and underpins nearly all research programs in biology and paleontology. However, most methods of phylogenetic inference typically used by paleontologists do not accommodate the idiosyncrasies of fossil data and therefore do not take full advantage of the information provided by the fossil record. The advent of Bayesian ‘tip-dating’ approaches to phylo- geny estimation is especially promising for paleosystematists because time-stamped comparative data can be combined with probabilistic models tailored to accommodate the study of fossil taxa. Under a Bayesian framework, the recently developed fossilized birth–death (FBD) process provides a more realistic tree prior model for paleontological data that accounts for macroevolutionary dynamics, preservation, and sampling when inferring phylogenetic trees containing fossils. In addition, the FBD tree prior allows for the possibility of sampling ancestral morphotaxa. Although paleontologists are increasingly embracing probabilistic phylogenetic methods, these recent developments have not previously been applied to the deep-time invertebrate fossil record. Here, I examine phylogenetic relationships among Ordovician through Devonian crinoids using a Bayesian tip-dating approach. Results support several clades recognized in previous analyses sampling only Ordovician taxa, but also reveal instances where phylogenetic affinities are more complex and extensive revisions are necessary, parti- cularly among the Cladida. The name Porocrinoidea is proposed for a well-supported clade of Ordovician ‘cyathocrine’ cla- dids and hybocrinids. The Eucladida is proposed as a clade name for the sister group of the Flexibilia herein comprised of cladids variously considered ‘cyathocrines,’‘dendrocrines,’ and/or ‘poteriocrines’ by other authors. Introduction Modern macroevolutionary research resides at the nexus of paleontology and phylogenetic comparative biology. The fossil record provides a spectacular temporal window into the vicissi- tudes of life’s history, and paleontologists have long used its pat- terns to investigate large-scale trends in diversification dynamics and morphologic evolution over timescales inaccessible to experimental manipulation or field-based investigation (Simpson, 1944; Sepkoski, 1981; Hunt et al., 2008; Alroy, 2010). Similarly, biologists armed with molecular phylogenies of extant species and tree-based statistical techniques have increasingly become inter- ested in addressing macroevolutionary questions traditionally stu- died by paleontologists (e.g., O’Meara et al., 2006; Bokma, 2008; Rabosky, 2009; Rabosky and McCune, 2009; Harmon et al., 2010; Pennell et al., 2014). Although differences between paleontologic and biologic perspectives remain, attempts to bridge disciplinary gaps between fields have wide-reaching implications for assem- bling a more synthetic macroevolutionary theory (Jablonski, 2008; Slater and Harmon, 2013; Hunt and Slater, 2016). Instances of integration between fields, such as paleontology and molecular phylogenetics, often provide opportunities for reci- procal illumination. For example, fossils play a major role in dating divergences among extant species. Without external information to constrain absolute ages, branch length estimation is confounded by the fact that both rates of molecular sequence evolution and elapsed time contribute to observed distances among species. Thus, the construction of a time-calibrated molecular phylogeny requires information on fossil morphologies and their temporal distributions to provide a numerical timescale for testing alternative models of macroevolutionary dynamics (Donoghue and Benton, 2007; dos Reis et al., 2016; Ksepka et al., 2015). Equally illuminating for paleontologists, many probabilistic methods originally developed by molecular phylogeneticists can be modified and applied to paleontologic data (Wagner, 2000a; Wagner and Marcot, 2010; Lee and Palci, 2015; but see Spencer and Wilberg, 2013). For example, Lewis (2001) developed a k-state Markov model for calculating likelihoods of discrete, morphologic characters based on a generalization of the Jukes-Cantor model of molecular sequence evolution. Although simplistic, Lewis’s (2001) ‘Mk’ model has recently been demonstrated in a Bayesian context to outperform other phylogenetic methods under a range of condi- tions present in real data sets, including missing character data, high rates of character evolution (and therefore homoplasy), and rate heterogeneity among characters (Wright and Hillis, 2014; O’Reilly et al., 2016). The recent resurgence of ‘total-evidence’ (Pyron, 2011; Ronquist et al., 2012) approaches in phylogenetics coincides with a renewed interest among biologists in phenotypic evolution and the utility of morphologic phylogenetics in an age of ‘post-molecular systematics’ (Lee and Palci, 2015; Pyron, 2015). * Present address: Department of Paleobiology, National Museum of Natural History, The Smithsonian Institution, P.O. Box 37012, MRC 121, Washington, DC 20013-7012, USA 〈wrightda@si.edu〉 Journal of Paleontology, page 1 of 16 Copyright © 2017, The Paleontological Society 0022-3360/15/0088-0906 doi: 10.1017/jpa.2016.141 1 https://doi.org/10.1017/jpa.2016.141 Downloaded from https:/www.cambridge.org/core. IP address: 122.59.106.201, on 09 Feb 2017 at 19:40:25, subject to the Cambridge Core terms of use, available at https:/www.cambridge.org/core/terms.