CORRESPONDENCE Evolution of Tibetan wild boars a Sulawesi warty pig To the Editor: (S. celebensis) The analysis presented by Li et al.1 in ~2 their report of the genome sequence of Java warty pig the Tibetan wild boar provides interesting (S. verrucosus) insights into the genetic architecture of Tibetan wild boar high-altitude adaptation in this species. (S. scrofa) However, despite the large volume of novel data, we found shortcomings in several parts 4.1 (5.3–3.5) North Chinese wild boar 2.5 (4–1.3) (S. scrofa) of the study, suggesting that some specific 14 (28–8)* findings presented by Li et al. result from Sourth Chinese wild boar overinterpretation of the data. In addition, (S. scrofa) 1.2 (2–0.8) several of their conclusions contradict those ? 2–5 6.9 (21.9–3.1) reported in previous analyses . Duroc reference genome More specifically, the authors infer that ~15 9.8 (12–7.5) ~0.8 (S. scrofa) 10.8 (14–7) Tibetan wild boar and Duroc breeds (Sus 26 (50–13)* European wild boar scrofa Ssc10.2 reference genome) diverged (S. scrofa) during the Miocene, ~6.8 million years ago. This estimated date is nearly ten times Sumatran wild boar more ancient than the recently reported split (S. scrofa) between Asian and European wild boars ~5.5 2,4 Warthog (0.8–2 million years ago) . In addition, (P. africanus) 0.004 substitutions/site previous studies2,3 estimated the divergence b Nature America, Inc. All rights reserved. America, Inc. Nature 5 time between S. scrofa and other Sus species Demographic history of Eurasian wild boars ) 4.0 from Island Southeast Asia (outgroups in 4 10 3.5 © 201 Fig. 2b,e of ref. 1) to be 1.3–5.3 million years × ago. Li et al.1 do not describe the details of the 3.0 2.5 Tibet molecular clock analysis other than stating South China 2.0 North China 6 Europe npg that PAML (MCMCtree) was used with 1.5 three molecular clock–based calibrations 1.0 (as opposed to fossil-based calibrations), 0.5 and they neither specify which nodes population size ( Effective 0 were calibrated nor did they include the 104 105 106 107 uncertainties of these calibrations in their Years age priors. Moreover, we believe that the tree used for the molecular clock analysis Figure 1 Evolutionary history of Sus species. (a) Maximum-likelihood phylogenetic tree for Suinae was too sparsely sampled to be informative (Supplementary Note). All nodes besides the node for Tibetan or northern Chinese wild boars (support of 47%) are supported by 100% bootstrap replicates. Node labels represent time estimates in millions for the Duroc (S. scrofa reference genome) of years from refs. 2, 3 and 1, respectively; an asterisk indicates divergence times that were converted and Tibetan split time. Indeed, different using branch lengths estimated in this study and the times reported by Li et al.1 (Supplementary mutation rates are expected for the deep Note). Gray dots represent well-known fossils, with time in millions of years. (b) Demographic history of internal branches separating mammalian Eurasian wild boar (Supplementary Note). Generation time (g) = 5, mutation rate (+) = 2.5 × 10–8. orders and the short branches separating the two subspecies of S. scrofa2,4,7. Hence, resulting in a gross overestimation of the constructed a phylogenetic tree using data estimating a subspecies split time using rates subspecies divergence time. from a Tibetan wild boar1 together with estimated from the divergence of mammalian Furthermore, the authors do not take into seven other Sus samples (Fig. 1a) that were orders is nearly certain to bias the estimate, account the suid paleontological literature used in two previous studies2,4 and the leading to an incorrect conclusion. We that provides specific examples contradictory study of Li et al.1 (Supplementary Note). therefore believe that this analysis is likely to their conclusions (Fig. 1a) and would have We further annotated the tree with well- to have been influenced both by prior age provided useful calibrations (see Additional known fossils and our own time estimates2,3,8 misspecification and biased taxon sampling, File 6 in ref. 2). To illustrate our concerns, we (Fig. 1a). Our results demonstrate that 188 VOLUME 47 | NUMBER 3 | MARCH 2015 | NATURE GENETICS CORRESPONDENCE Tibetan wild boar clusters together with claims, we reanalyzed the data of Li et al. by AUTHOR CONTRIBUTIONS Chinese wild boar (as also suggested by the aligning short-read sequences from a single L.A.F.F. and M.A.M.G. designed the study with input ancestry and phylogenetic analyses of Li Tibetan wild boar that was used for de novo from H.-J.M., O.M. and G.L. L.A.F.F. analyzed the data with help from H.-J.M., Y.P., M.B., R.P.M.A.C. et al.). On the basis of our molecular clock assembly (over 10× coverage; Supplementary and J.G.S. L.A.F.F. wrote the manuscript with input analyses2,3, we conclude that Tibetan and Note) to the Ssc10.2 reference genome and from all authors. European wild boars are not different species conducted a similar pairwise sequentially but instead are closely related subspecies Markovian coalescence (PSMC) analysis to COMPETING FINANCIAL INTERESTS The authors declare no competing financial interests. that diverged during the Pleistocene. In the one carried out by Li et al.1. Our results addition, we show that the evolutionary demonstrate that Tibetan and southern Laurent A F Frantz1,5, Ole Madsen1, time scale proposed by Li et al.1 for the Sus Chinese wild boars have similar demographic Hendrik-Jan Megens1, Joshua G Schraiber2,3, genus implies a significantly older speciation histories (Fig. 1b). In particular, we show that Yogesh Paudel1, Mirte Bosse1, timeframe that contradicts the fossil record Tibetan wild boar underwent fluctuations Richard P M A Crooijmans1, 4,5 1 (Fig. 1a). Furthermore, according to our in population size during the Pleistocene, Greger Larson & Martien A M Groenen branch length estimates (Supplementary contrary to the conclusions of Li et al. This 1Animal Breeding and Genomics Group, Note), a divergence time for European and analysis shows that the claim that Tibetan Wageningen University, Wageningen, the Asian wild boars of 6.8 million years ago1 wild boar experienced no demographic Netherlands. 2Department of Integrative Biology, (confidence interval of 2.4–12.9 million years fluctuations owing to the presence of refugia University of California, Berkeley, Berkeley, 3 ago) would imply that African and Eurasian in Tibet is incorrect. California, USA. Department of Genome Suinae diverged roughly 26 million years In light of the Assemblathon 2.0 (ref. 10), Sciences, University of Washington, Seattle, Washington, USA. 4Durham Evolution and ago (confidence interval of 13–50 million we believe that it is important not to Ancient DNA, Department of Archaeology, years ago), an estimate that bears no overinterpret data from de novo assemblies Durham University, Durham, UK. 5Present correspondence with the fossil timeframe of next-generation short-read sequencing address: Palaeogenomics and Bio-Archaeology 8 for Suinae . The erroneous time inference data, especially when comparing these to an Research Network, Research Laboratory for presented by Li et al. contributes to their assembly built from Sanger sequencing reads Archaeology and the History of Art, University of generally implausible results. combined with a high-resolution linkage Oxford, Oxford, UK. Another point of concern is the results map. Here we show that the overconfidence e-mail: [email protected] or of their demographic analysis9, shown in of the authors in their de novo assembly [email protected] 1 Figure 2e of Li et al. , which highlight the low and divergence time estimates misled their 1. Li, M. et al. Nat. Genet. 45, 1431–1438 (2013). coverage of the data set and the shortcomings interpretation of the evolutionary history of 2. Frantz, L.A.F. et al. Genome Biol. 14, R107 (2013). 3. Gongora, J. et al. Zool. Scr. 40, 327–335 (2011). of a next-generation sequence–based Tibetan wild boars. Furthermore, we believe 4. Groenen, M.A.M. et al. Nature 491, 393–398 (2012). genome assembly. The authors present the that some results presented in the study 5. Paudel, Y. et al. BMC Genomics 14, 449 (2013). demographic history as inferred both from by Li et al.1, in particular, large-scale gene 6. Yang, Z. Mol. Biol. Evol. 24, 1586–1591 (2007). 7. Ho, S.Y.W. & Larson, G. Trends Genet. 22, 79–83 boar genomes resequenced and aligned to family expansion and/or contraction since (2006). their assembly (green and pink lines in Fig. the European-Tibetan wild boar split, are 8. Orliac, M.J., Pierre-Olivier, A. & Ducrocq, S. Zool. Scr. Nature America, Inc. All rights reserved. America, Inc. Nature 39, 315–330 (2010). 5 2e in ref. 1) and previously sequenced boar dubious. The interpretation of these results 9. Li, H. & Durbin, R. Nature 475, 493–496 (2011). genomes aligned to the high-quality draft as being due to a genuine biological signal 10. Bradnam, K.R. et al. Gigascience 2, 10 (2013). © 201 reference genome4 (Ssc10.2; blue and black rather than assembly artifact might have been lines in Fig. 2e in ref. 1). Surprisingly, the misled by the gross overestimation of the Li et al. reply: results show that the alignments of Chinese evolutionary timeframe of S. scrofa. Frantz et al. reanalyzed our data1 by aligning npg wild boar to Ssc10.2 and the de novo assembly We suggest that the experimental design a small part of our short-read sequences significantly disagree (compare South China of studies describing the resequencing of from a Tibetan wild boar that were used for with Southwest China in Fig.