Phylogeny of Paleozoic limbed vertebrates reassessed through revision and expansion of the largest published relevant data matrix David Marjanovic1 and Michel Laurin2 1 Science Programme “Evolution and Geoprocesses”, Museum für Naturkunde—Leibniz Institute for Evolutionary and Biodiversity Research, Berlin, Germany 2 Centre de Recherches sur la Paléobiologie et les Paléoenvironnements (CR2P), Centre national de la Recherche scientifique (CNRS)/Muséum national d’Histoire naturelle (MNHN)/Sorbonne Université, Paris, France ABSTRACT The largest published phylogenetic analysis of early limbed vertebrates (Ruta M, Coates MI. 2007. Journal of Systematic Palaeontology 5:69–122) recovered, for example, Seymouriamorpha, Diadectomorpha and (in some trees) Caudata as paraphyletic and found the “temnospondyl hypothesis” on the origin of Lissamphibia (TH) to be more parsimonious than the “lepospondyl hypothesis” (LH)—though only, as we show, by one step. We report 4,200 misscored cells, over half of them due to typographic and similar accidental errors. Further, some characters were duplicated; some had only one described state; for one, most taxa were scored after presumed relatives. Even potentially continuous characters were unordered, the effects of ontogeny were not sufficiently taken into account, and data published after 2001 were mostly excluded. After these issues are improved— we document and justify all changes to the matrix—but no characters are added, we find (Analysis R1) much longer trees with, for example, monophyletic Caudata, Diadectomorpha and (in some trees) Seymouriamorpha; Ichthyostega either crownward or rootward of Acanthostega; and Anthracosauria either crownward or Submitted 16 January 2016 Accepted 12 August 2018 rootward of Temnospondyli. The LH is nine steps shorter than the TH (R2; Published 4 January 2019 constrained) and 12 steps shorter than the “polyphyly hypothesis” (PH—R3; Corresponding author constrained). Brachydectes (Lysorophia) is not found next to Lissamphibia; David Marjanovic, instead, a large clade that includes the adelogyrinids, urocordylid “nectrideans” [email protected] and aïstopods occupies that position. As expected from the taxon/character ratio, Academic editor most bootstrap values are low. Adding 56 terminal taxa to the original 102 Erik Seiffert increases the resolution (and decreases most bootstrap values). The added Additional Information and taxa range in completeness from complete articulated skeletons to an incomplete Declarations can be found on lower jaw. Even though the lissamphibian-like temnospondyls Gerobatrachus, page 164 Micropholis and Tungussogyrinus and the extremely peramorphic salamander DOI 10.7717/peerj.5565 Chelotriton are added, the difference between LH (R4; unconstrained) and TH Copyright 2019 Marjanovic and Laurin (R5) rises to 10 steps, that between LH andPH(R6)to15;theTHalsorequires several more regains of lost bones than the LH. Casineria,inwhichwetentatively Distributed under identify a postbranchial lamina, emerges rather far from amniote origins in a Creative Commons CC-BY 4.0 gephyrostegid-chroniosuchian grade. Bayesian inference (Analysis EB, settings as in R4) mostly agrees with R4. High posterior probabilities are found for How to cite this article Marjanovic D, Laurin M. 2019. Phylogeny of Paleozoic limbed vertebrates reassessed through revision and expansion of the largest published relevant data matrix. PeerJ 6:e5565 DOI 10.7717/peerj.5565 Lissamphibia (1.00) and the LH (0.92); however, many branches remain weakly supported, and most are short, as expected from the small character sample. We discuss phylogeny, approaches to coding, methods of phylogenetics (Bayesian inference vs. equally weighted vs. reweighted parsimony), some character complexes (e.g. preaxial/postaxial polarity in limb development), and prospects for further improvement of this matrix. Even in its revised state, the matrix cannot provide a robust assessment of the phylogeny of early limbed vertebrates. Sufficient improvement will be laborious—but not difficult. Subjects Biodiversity, Ecology, Evolutionary Studies, Paleontology, Zoology Keywords Phylogenetics, Data matrix, Morphology, Tetrapoda, Amphibia, Lissamphibia, Evolution, Temnospondyli, Lepospondyli, Phylogeny INTRODUCTION This ancient inhabitant of the coal swamps of Nova Scotia, was, in short, as we often find to be the case with the earliest forms of life, the possessor of powers and structures not usually, in the modern world, combined in a single species. It was certainly not a fish, yet its bony scales, and the form of its vertebræ, and of its teeth, might, in the absence of other evidence, cause it to be mistaken for one. We call it a batrachian, yet its dentition, the sculpturing of the bones of its skull, which were certainly no more external plates than the similar bones of a crocodile, its ribs, and the structure of its limbs, remind us of the higher reptiles; and we do not know that it ever possessed gills, or passed through a larval or fish-like condition. Still, in a great many important characters, its structures are undoubtedly batrachian. It stands, in short, in the same position with the Lepidodendra and Sigillariæ under whose shade it crept, which though placed by palæo-botanists in alliance with certain modern groups of plants, manifestly differed from these in many of their characters, and occupied a different position in nature. In the coal period, the distinctions of physical and vital conditions were not well defined—dry land and water, terrestrial and aquatic plants and animals, and lower and higher forms of animal and vegetable life, are consequently not easily separated from each other. – Dawson (1863:23–24) about Dendrerpeton acadianum Homoplasy Is even More Common than I, or perhaps Anyone, Has ever Imagined – section headline in Wake (2009: 343) What is required is a more complete discussion of the character coding of previous data matrices, and a thorough reanalysis based on those matrices. This would enable a well-founded discussion of lissamphibian origins in light of a supermatrix based on all the current and pertinent data. – Sigurdsen & Green (2011: 459) Giant morphological data matrices are increasingly common in cladistic analyses of vertebrate phylogeny, reporting numbers of characters never seen or expected before. Marjanovic and Laurin (2019), PeerJ, DOI 10.7717/peerj.5565 2/191 However, the concern for size is usually not followed by an equivalent, if any, concern for character construction/selection criteria. Therefore, the question of whether quantity parallels quality for such influential works remains open. – Simões et al. (2017: abstract) Not too surprisingly, as it is yet a youthful paradigm shift, modern phylogenetic systematics is still evolving to improve on the lack of precision, rigour and objectivity it inherited from the pre-cladistic period. Furthermore, transforming a descriptive science (morphological description) bounded by language as a means of outlining empirical observations into hopefully objectively delimited characters and character states is a difficult task; every effort to do so is to be commended, while at the same time rigorously scrutinized and improved upon. – Simões et al. (2017: 215) Phylogenetic trees are hypotheses that attempt to explain a data matrix. Much work has gone, to great success, into the methods for generating and testing phylogenetic hypotheses from a given data matrix; and a matrix of molecular data can, apart from the problem of alignment, be largely taken for granted as a set of observed facts. But molecular data are not always available. In many cases phylogeneticists have to rely on morphological data—and a matrix of morphological data is a matrix of hypotheses. The characters, their states, and the relationships between the states are hypotheses that rely on hypotheses about homology, about independent evolution from other characters, about ontogeny and even preservation (especially in the case of fossils); the terminal taxa (operational taxonomic units—OTUs) are hypotheses that rely on hypotheses of monophyly, ontogeny and again preservation; and even given all these, each cell in a data matrix is still a hypothesis that relies on hypotheses of homology, ontogeny and preservation—some are close enough to observed facts, others less so. In addition, morphological data matrices can only be compiled by hand—there is no equivalent to sequencer machines or alignment programs. This makes human error inevitable. Consequently, morphological data matrices must not be taken for granted as sets of objective data; the hypotheses of which they consist must be identified and carefully tested. The analysis by Ruta & Coates (2007—hereinafter RC07) has played a large role in shaping current ideas on the phylogeny and early evolution (Late Devonian to Cisuralian, with a few younger taxa) of the limbed vertebrates, including the origins of amniotes and lissamphibians. Being based on the largest matrix so far applied to this problem, its results have been widely treated as a consensus and even used as the basis for further work in evolutionary biology (Bernardi et al., 2016; MacIver et al., 2017). However, several conflicting phylogenetic results have persisted in other analyses based on different matrices (Vallin & Laurin, 2004; Marjanović & Laurin, 2008, 2009; Sigurdsen & Green, 2011; Pardo et al., 2017; Pardo, Small & Huttenlocker, 2017: fig. 2, S6, S7). Although the large differences in character sampling between any two of these analyses