EVOLUTION & DEVELOPMENT 16:4, 189–196 (2014) DOI: 10.1111/ede.12081 Toward consilience in reptile phylogeny: miRNAs support an archosaur, not lepidosaur, affinity for turtles Daniel J. Field,a,b,* Jacques A. Gauthier,a Benjamin L. King,c Davide Pisani,d,e Tyler R. Lyson,a,b and Kevin J. Petersonf,* a Department of Geology and Geophysics, Yale University, 210 Whitney Avenue, New Haven, CT 06511, USA b Department of Vertebrate Zoology, National Museum of Natural History, Smithsonian Institution, Washington, DC 20560, USA c Mount Desert Island Biological Laboratory, Salisbury Cove, ME 04672, USA d School of Earth Sciences, University of Bristol, Queen's Road, Bristol BS8 1RJ, United Kingdom e School of Biological Sciences, University of Bristol, Woodland Road, Bristol BS8 1UG, United Kingdom f Department of Biological Sciences, Dartmouth College, Hanover, NH 03755, USA *Authors for correspondence (e‐mail: [email protected], daniel.fi[email protected]) SUMMARY Understanding the phylogenetic position of for a turtle lepidosaur sister‐relationship; instead, we recover þ crown turtles (Testudines) among amniotes has been a source strong support for turtles sharing a more recent common of particular contention. Recent morphological analyses ancestor with archosaurs. We further test this result by suggest that turtles are sister to all other reptiles, whereas analyzing a super‐alignment of precursor miRNA sequences the vast majority of gene sequence analyses support turtles as for every miRNA inferred to have been present in the most being inside Diapsida, and usually as sister to crown recent common ancestor of tetrapods. This analysis yields a Archosauria (birds and crocodilians). Previously, a study topology that is fully congruent with our presence/absence using microRNAs (miRNAs) placed turtles inside diapsids, but analysis; our results are therefore in accordance with most as sister to lepidosaurs (lizards and Sphenodon) rather than gene sequence studies, providing strong, consilient molecular archosaurs. Here, we test this hypothesis with an expanded evidence from diverse independent datasets regarding the miRNA presence/absence dataset, and employ more rigorous phylogenetic position of turtles. criteria for miRNA annotation. Significantly, we find no support INTRODUCTION et al. 2005; Shen et al. 2011; Tzika et al. 2011; Chiari et al. 2012; Crawford et al. 2012; Fong et al. 2012; Lu et al. 2013; Shaffer The phylogenetic position of turtles represents one of the most et al. 2013; Wang et al. 2013). In the absence of a well‐ recalcitrant problems in vertebrate biology, with contrasting resolved phylogenetic hypothesis for Amniota, outstanding hypotheses arising from different datasets. In recent years, three macroevolutionary questions, including those regarding the mutually‐exclusive hypotheses have been put forth for the acquisition of the unique turtle body plan, cannot be adequately phylogenetic placement of turtles: (i) turtles represent the sister addressed. group to all diapsid reptiles (mainly supported by morphological Although in contradiction to most molecular studies, the datasets and developmental data, e.g., Gauthier et al. 1988; miRNA data supporting a turtle lepidosaur clade (Lyson et al. þ Lee 1997; Werneburg and Sánchez‐Villagra 2009; Lyson 2012) was not entirely unexpected (Rieppel and deBraga 1996; et al. 2010, 2013a); (ii) turtles are the sister group to Lepidosauria Becker et al. 2010). miRNAs are approximately 22‐nucleotide (Sphenodon and lizards, including snakes; supported mainly by noncoding RNA molecules that have been heralded as especially expressed miRNAs (Lyson et al. 2012), as well as some useful phylogenetic characters due to their continuous addition to morphological analyses, e.g., Rieppel and deBraga 1996; animal genomes through time, comparatively low rates of deBraga and Rieppel 1997; Rieppel and Reisz 1999; Li secondary loss, and the largely conservative nature of the mature et al. 2008); and (iii) turtles are the sister taxon to, or are nested gene product’s primary sequence. These characters have been used within, Archosauria (birds and crocodilians; supported mainly to reconstruct the phylogenetic interrelationships of numerous by gene‐sequence datasets, e.g., Zardoya and Meyer 1998; animal clades at all levels in metazoan phylogeny (Sperling and Hedges and Poling 1999; Kumazawa and Nishida 1999; Iwabe Peterson 2009; Tarver et al. 2013). Lyson et al. (2012) showed that © 2014 Wiley Periodicals, Inc. 189 190 EVOLUTION & DEVELOPMENT Vol. 16, No. 4, July–August 2014 the lizard Anolis carolinensis and the turtle Chrysemys picta shared METHODS four putative miRNAs, and that these nucleotide sequences were not recovered in a small RNA library derived from a total RNA Total RNA (Wheeler et al. 2009) was extracted from homo- preparation of an alligator, nor present in any sequenced bird genized individuals; preparations were made from single late‐ genome. On the basis of these apparent synapomorphic miRNAs, stage embryos of the pigeon Columba livia, an adult gecko these authors concluded that turtles were likely the extant sister Coleonyx variegatus, an adult xantusiid Xantusia wigginsi, and a group of the lepidosaurs. juvenile snake Chionactis occipitalis, following standard animal As with any dataset, however, the quality of the characters care protocols (IACUC number 2009‐11302). Small RNA used directly dictates the robustness of the analysis. With libraries were prepared at the Yale W. M. Keck Facility miRNAs, care must be taken in distinguishing them from other according to manufacturer’s instructions, and sequenced on an types of RNA molecules including other small RNAs (e.g., Illumina Genome Analyzer II platform. The number of reads piRNAs, tRNAs), and fragments of larger RNA molecules (in sequenced per library is detailed in Table 1. particular, fragments of rRNAs and mRNAs). Recent clarifica- An updated version of miRMiner (Wheeler et al. 2009) was tions of the criteria for miRNA annotation have challenged the used to identify both orthologues of previously identified diagnosis of many sequences previously identified as miRNAs miRNAs (miRBase v. 19; Kozomara and Griffiths‐Jones 2011) (Tarver et al. 2012). According to these more rigorous and novel miRNA families from these four taxa, in addition to specifications, four miRNA sequences identified as turtle reanalyzing the raw data from Lyson et al. (2012) for the turtle þ lepidosaur synapomorphies by Lyson et al. (2012) do not meet Chrysemys picta, the alligator Alligator mississippiensis, and the the minimal criteria established for miRNA annotation lizard Anolis carolinensis. Because published genomes are now (Kozomara and Griffiths‐Jones 2011; Tarver et al. 2012), available for Chrysemys picta (Shaffer et al. 2013) and Alligator especially as none of the four putative miRNAs exhibited mississippiensis (St John et al. 2012), the near‐complete expression of both arms of the hairpin (see below). Thus, the complements of miRNAs from these two taxa were assembled discordance between the miRNA dataset of Lyson et al. (2012) (see Electronic Supplementary Material: ESM Files 1–2, and most sequence‐based datasets to this point, including the respectively). In addition, the reads from Columba livia were recent phylogenomic analysis of Chiari et al. (2012), could be used to query the recently released pigeon genome (Shapiro due to mistaken miRNA homologies in Lyson et al. (2012). To et al. 2013), and its near‐complete miRNA repertoire was address this issue, we characterized the near‐complete miRNA assembled (ESM File 3). Finally, the near‐complete ancestral repertoire of the turtle Chrysemys picta using both small RNA miRNA complement of macrostomate snakes (Lee et al. 2007) library reads and genomic sequences, and compared this was assembled (ESM File 4) using the reads from the snake repertoire to the near‐complete repertoires of the snake Python Chionactis occipitalis and the genome of the python Python bivittatus, the crocodilian Alligator mississippiensis, and the bivittatus (Castoe et al. 2011). avian Columba livia, in addition to previously published lizard Next, the miRNAs constituting each of these complements (Lyson et al. 2012) and bird data (miRBase v.19; Kozomara and (ESM Files 1–4) were used as queries to search the genomes of Griffiths‐Jones 2011). We also sequenced small RNA libraries three other turtles (the cheloniid Chelonia mydas, and the two from three additional species representing major lineages from trionychids Pelodiscus sinensis and Apalone spinifera), the across the lizard tree—the gecko Coleonyx variegatus, the Chinese alligator Alligator sinensis, and the coelacanth xantusiid Xantusia wigginsi, and the snake Chionactis occipi- talis—and queried the genomes of one additional crocodilian (Alligator sinensis), and three other turtles (the cheloniid Table 1. Read count and genome assembly information Chelonia mydas, and the trionychids Pelodiscus sinensis and Apalone spinifera). Our analyses fully support an archosaur Collapsed Genome assembly affinity for turtles: the original ‘miRNAs’ identified by Lyson Species Total reads reads1 accession number et al. (2012) appear to be spurious, whereas we demonstrate that turtles share several bona fide miRNAs with archosaurs not Alligator mississippiensis 21,731,314 97,477 AKHW00000000.1 found or expressed in lepidosaurs, mammals, or any other Columba livia 45,635,579 86,204 AKCR00000000.1 Chrysemys picta 23,765,521 104,168 AHGY00000000.1 metazoans. Additionally, this conclusion