You are on page 1of 11

BMC Evolutionary Biology BioMed Central

Research article Open Access
Analysis of complete mitochondrial genomes from extinct and
extant rhinoceroses reveals lack of phylogenetic resolution
Eske Willerslev*1, M Thomas P Gilbert1, Jonas Binladen1, Simon YW Ho2,
Paula F Campos1, Aakrosh Ratan3, Lynn P Tomsho3, Rute R da Fonseca4,
Andrei Sher5, Tatanya V Kuznetsova6, Malgosia Nowak-Kemp7,
Terri L Roth8, Webb Miller3 and Stephan C Schuster*3
Address: 1Centre for Ancient Genetics, University of Copenhagen, Universitetsparken 15, DK-2100, Denmark, 2Centre for Macroevolution and
Macroecology, School of Biology, Australian National University, Canberra, ACT 0200, Australia, 3Pennsylvania State University, Center for
Comparative Genomics and Bioinformatics, 310 Wartik Building, University Park, PA 16802, USA, 4Department of Integrative Biology,
University of California, Berkeley, CA 9472-3140, USA, 5Severtsov Institute of Ecology and Evolution, Russian Academy of Sciences, 33 Leninsky
Prospect, 119071 Moscow, Russia, 6Department of Paleontology, Faculty of Geology, Lomonosov Moscow State University, Leninskiye Gory,
119992 Moscow, Russia, 7Oxford University Museum of Natural History, Parks Road, Oxford, OX1 3PW, UK and 8Center for Conservation and
Research of Endangered Wildlife (CREW), Cincinnati Zoo & Botanical Garden, 3400 Vine St, Cincinnati, OH 45220, USA
E-mail: Eske Willerslev* - ewillerslev@bio.ku.dk; M Thomas P Gilbert - mtpgilbert@gmail.com; Jonas Binladen - jbinladen@gmail.com;
Simon YW Ho - simon.ho@anu.edu.au; Paula F Campos - pcampos@bi.ku.dk; Aakrosh Ratan - ratan@cse.psu.edu;
Lynn P Tomsho - lap153@psu.edu; Rute R da Fonseca - rute.r.da.fonseca@gmail.com; Andrei Sher - mtpgilbert@gmail.com;
Tatanya V Kuznetsova - tatkuz@orc.ru; Malgosia Nowak-Kemp - malgosia.nowak-kemp@oum.ox.ac.uk;
Terri L Roth - terri.roth@cincinnatizoo.org; Webb Miller - webb@bx.psu.edu; Stephan C Schuster* - scs@bx.psu.edu
*Corresponding author

Published: 11 May 2009 Received: 3 October 2008
Accepted: 11 May 2009
BMC Evolutionary Biology 2009, 9:95 doi: 10.1186/1471-2148-9-95
This article is available from: http://www.biomedcentral.com/1471-2148/9/95
© 2009 Willerslev et al; licensee BioMed Central Ltd.
This is an Open Access article distributed under the terms of the Creative Commons Attribution License (http://creativecommons.org/licenses/by/2.0),
which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.

Abstract
Background: The scientific literature contains many examples where DNA sequence analyses
have been used to provide definitive answers to phylogenetic problems that traditional (non-DNA
based) approaches alone have failed to resolve. One notable example concerns the rhinoceroses, a
group for which several contradictory phylogenies were proposed on the basis of morphology,
then apparently resolved using mitochondrial DNA fragments.
Results: In this study we report the first complete mitochondrial genome sequences of the
extinct ice-age woolly rhinoceros (Coelodonta antiquitatis), and the threatened Javan (Rhinoceros
sondaicus), Sumatran (Dicerorhinus sumatrensis), and black (Diceros bicornis) rhinoceroses. In
combination with the previously published mitochondrial genomes of the white (Ceratotherium
simum) and Indian (Rhinoceros unicornis) rhinoceroses, this data set putatively enables reconstruc-
tion of the rhinoceros phylogeny. While the six species cluster into three strongly supported
sister-pairings: (i) The black/white, (ii) the woolly/Sumatran, and (iii) the Javan/Indian, resolution of
the higher-level relationships has no statistical support. The phylogenetic signal from individual
genes is highly diffuse, with mixed topological support from different genes. Furthermore, the
choice of outgroup (horse vs tapir) has considerable effect on reconstruction of the phylogeny. The
lack of resolution is suggestive of a hard polytomy at the base of crown-group Rhinocerotidae, and
this is supported by an investigation of the relative branch lengths.
Conclusion: Satisfactory resolution of the rhinoceros phylogeny may not be achievable without
additional analyses of substantial amounts of nuclear DNA. This study provides a compelling

Page 1 of 11
(page number not for citation purposes)

reject outright analyses of the constituent genes. woolly-Sumatran rhinoceros pairing using complete 12S resolution of the phylogeny of the five living rhinoceroses rRNA and partial cytochrome b gene sequences (max- using traditional (non-DNA) approaches has been con. and be an excellent illustration of the advantages of molecular the Javan (Rhinoceros sondaicus) and Indian (Rhinoceros sequence analysis over more traditional approaches. the geo. studies have been undertaken on the rhinoceroses in and extinct woolly. there are significant limitations with single-locus phylogenetics. Although it seems clear that the woolly and greatly increase the chance of finding the correct whole- Sumatran are closely related (for example both have two genome tree when sequence lengths are below 3. for example the woolly rhinoceros in analysed complete genomes and subsamples of them. large- scale sequencing of complete mitochondrial genomes becomes commonplace in evolutionary studies. their inferred phylogeny groups the woolly-Sumatran that has retained many ancestral morphological characters.146 bp of the combined 12S rRNA and cytochrome b genes. At the root of the problem is the placement of the Furthermore. troversial. We expect further examples of this to appear as next-generation. as well as individual could not. in agreement with the work of Tougard et al. Simpson (1945) Background analysed ancient DNA to confirm the monophyly of the Despite being a long-standing target of scientific research.1] [17] rhinoceroses.biomedcentral. this case. the clearly controversial information from the woolly rhinoceros has failed to phylogeny of the Rhinocerotidae. Tougard et al. the when resolving subtle phylogenetic questions. [8] found high support the living and one extinct member of the rhinocerotid (maximum likelihood bootstrap support of 97%) for family. a species [8]. (maximum parsimony bootstrap support of 57%) for four novel complete mitochondrial genomes (from the the extant rhinoceros phylogeny as outlined on the basis black. two horns of the Sumatran rhinoceros suggest that it should be placed with the similarly two-horned black and white Despite the successful results. The first such study generated. Orlando et al. 2. we have an attempt to resolve the phylogeny. With of horn morphology [7]. Cummings et al. [12] including fossil taxa. rather than with the single-horned Javan and lessons of the above is that even when using DNA data. the six genomes cover all sequences. questions the findings the horn topology). pair with the Javan-Indian pair. However.2]. however. using a Kishino-Hasegawa test. one of the key rhinoceroses. The variation in hard trichotomy has been proposed. but with <50% bootstrap among the broadly accepted sub-clades of the black (Diceros support. using our previously published approach of used restriction-digest mapping of mitochondrial DNA utilising keratinous tissues as a high-quality source of (mtDNA) ribosomal region to find weak support mtDNA for sequencing on the FLX platform [13-15].com/1471-2148/9/95 demonstration that. a yield falsely supported results [10]. on the one hand. woolly. however. Sumatran rhinoceros (Dicerorhinus sumatrensis). and the fact that produce a convincing resolution of the relationships previous studies have maximally been based on the among the three pairs. Indian rhinoceroses [1. Javan. and its amounts of sequence. this picture was white [GenBank:Y07726] [16] and Indian [Genbank: modified – using complete 12S rRNA and cytochrome b X97336. and Sumatran rhinoceroses). G. this has proven to be similarly concluding that small increases in sequence length will problematic. 9:95 http://www. On the other hand. "The human factor in classification is nowhere more evident than in dealing with this superfamily (Rhinocerotoidea). rhinoceroses. reflecting an effectively phylogenetic signal among mitochondrial genes in simultaneous divergence of the three lineages [4-6]. the addition of morphological base-pairs (bp). As larger amounts of data were the addition of the published mtDNA genomes of the incorporated into the analyses. unicornis) rhinoceroses. In light of this. Third. we have revisited the molecular analysis using six In response to these problems. elephantids provides a compelling illustration of this Attempts to resolve such questions can be made by problem [11]. It has previously been advocated close proximity with the two other living Asian species. [9] of the previous molecular reports. Thus the results of these later studies appeared to bicornis) and white (Ceratotherium simum) rhinoceroses. results can still be misleading without sufficiently large graphic distribution of the Sumatran rhinoceros. For example. We demonstrate that phylogenetic analysis of the the phylogeny as outlined by geography (although they complete mtDNA genomes. More recently.000 horns and a hairy pelt)." G. that phylogenies based on single genes can sometimes would indicate that they form a natural clade [3]. in spite of substantial sequence length. In addition. Page 2 of 11 (page number not for citation purposes) . imum likelihood bootstrap support between 93–100%).BMC Evolutionary Biology 2009. several DNA-based complete mtDNA genome sequences of the five extant. Specifically.

Hasegawa test p = 0. Similar for the three candidate topologies (Figures 2. Javan. Splits graph obtained using split decomposition coding genes are involved in oxidative phosphorylation.18]. Bayesian and maximum-likelihood phylogenetic ana- Despite some of the differences occurring in functionally lyses of the concatenated data set revealed mixed support relevant sites (boxed residues in the alignment. and co3 [23].154. We have mapped the amino acid differences between graphs were produced using split decomposition with the rhinoceroses on the available crystallographic structures other distance measures. SH-test p = 0. A very have been associated with the improvement of aerobic similar graph was obtained using LogDet-transformed capacity and adaptation to new thermal environments distances. topology test used. Regardless of the most morphological reports. all three candidate topologies were always included in the 95% confidence set. co2. SH-test p = Bayesian posterior probability and maximum-likelihood 0. and modifications in these genes maximum-likelihood estimates of model parameters. Indian. We find very strong support for each of the grouping of the white. Figures S1 and S2: Additional topological support varied according to outgroup choice. Sumatran. 22 Sumatran tRNA genes. In order to infer the position of the root. including LogDet-transformed for mitochondrial-encoded proteins. we were unable to observe any direct although support was generally weak. In only one case relationship between them and the markedly different (Bayesian analysis with horse outgroup) was there environments inhabited by the woolly. regardless of the choice of outgroup).716. black.biomedcentral. [19. 4). The GTR+I which is responsible for the production of up to 95% of the +G model of nucleotide substitution was assumed. FJ905814 (black rhinoceros). lihood-based tree topology tests. two ribosomal RNA genes. The relative spheres in the structure.20]. This is in agreement with previous molecular findings and KH-test p = 0. Tapirus terrestris. analysis of whole mitochondrial genomes. yielded no significant support for any particular tree topology. Furthermore.245). There was low support for the Equus caballus). bootstrap support.473). and woolly were analysed using Bayesian and likelihood-based phylo.BMC Evolutionary Biology 2009. rhinoceroses group together to the exclusion of the Indian genetic methods. Shimodaira- included in the analyses (tapir. Mitochondrial protein.337. Indian Rhinoceros unicornis 0.com/1471-2148/9/95 Results Tapir Tapirus terrestris Mitochondrial genome sequences The four newly sequenced mitochondrial genomes are similar to the two previously published rhinoceros mito- chondrial genomes. Like- extant rhinoceroses. There Phylogeny and speciation times of the rhinoceros was slight (but non-significant) support in favour of the The mitochondrial genomes of the six rhinoceros species topology in which the white. FJ905815 Rhinoceros sondaicus (Javan rhinoceros) and FJ905816 (Sumatran rhinoceros). consisting of 13 protein-coding genes. and horse. The sequences have been Equus caballus submitted to GenBank with accession numbers FJ905813 Javan (woolly rhinoceros). black. those for cytb [22] and distances (results not shown).1 sub/site Black Diceros bicornis White An obvious power of complete mtDNA genome sequencing Ceratotherium simum is that it enables functional assessment and comparison of Figure 1 the genes between the taxa [14.64). and woolly rhinoceroses (AU-test p = 0. the consecutive repeats are longer than the read length of the Roche FLX. so we Horse are unable characterise them. Sumatran. but is not shown here. mutations in mitochondrial genes have been implicated in exercise intolerance in humans [21]. Kishino-Hasegawa test p = 0. orange for three candidate topologies (Table 1). and Javan rhino- three sister-species pairings among the rhinoceroses (100% ceroses (AU-test p = 0. performed using the concatenated alignment with both outgroups. in contrast to the significant support for a particular tree topology. Page 3 of 11 (page number not for citation purposes) . Files 1 and 2). 9:95 http://www. and for the grouping of the Indian.802). Split decomposition produced a graph with good representation of the phylogenetic signal (fit value = Analyses using individual components of the mitochon- 89. and Javan rhinoceroses (approximately-unbiased test p = the sequences of two perissodactylan outgroup species were 0.151. and a control region. 3.663. using energy of eukaryotic cells. Woolly Dicerorhinus sumatrensis Coelodonta antiquitatis The exact length is difficult to determine due to the presence of variable tandem repeats in the control.398. KH-test p = 0. co1. but with an obvious lack of resolution among drial genome revealed wide variations in relative support the three pairs of rhinoceros species (Figure 1).

171 0.177 0.755 Both 0.8 atp8 Number of variable sites nd4l Posterior probability 500 cox2 0. B = both).657 Botha 0.342 0.165 0.474 0.084 0. T = tapir.681 a Control region excluded.074 0. Relative and the D-loop).BMC Evolutionary Biology 2009.032 0.961 0.938 0.361 0.003 0. Page 4 of 11 (page number not for citation purposes) . The line graph these components.013 0.362 0.com/1471-2148/9/95 Table 1: Support for three candidate topologies estimated using Bayesian and maximum-likelihood phylogenetic analysis Bayesian posterior probability Maximum-likelihood bootstrap support Outgroup Horse 0.050 0. regions of two rRNA genes. Each column probabilities of three candidate tree topologies.075 0.813 Bothc 0.2 cox1 nd4 H 100 cox3 cytb Black + White Indian + Javan Indian + Javan Black + White tb x3 5 2 x2 4l p8 6 4 A 1 A 6 3 D 1 op nd nd 12 atp nd 16 cox nd nd nd N N nd cy co co at -lo rR rR S S Sumatran + Woolly Sumatran + Woolly Figure 2 Figure 3 Column graph showing posterior probability support Ternary plot showing the relative Bayesian posterior for three candidate tree topologies.000 0.048 0. 9:95 http://www.401 Tapir 0.259 0.457 Bothb 0.480 0.biomedcentral.244 0.820 0. The represents the posterior probability of the favoured tree estimated support is shown for individual components of the for each component of the mitochondrial genome mitochondrial genome (13 protein-coding genes.4 nd6 B nd3 16S 200 0.0 700 600 0. c Control region and all third codon sites excluded. b All third codon sites excluded.244 0.167 0.035 0. with results from three analyses using indicates the number of variable sites in each component. loop (13 protein-coding genes. Sumatran + Woolly Sumatran + Woolly Sumatran + Woolly Indian + Javan Indian + Javan Indian + Javan Black + White Black + White Black + White Sumatran + Woolly Black + White Indian + Javan nd2 1. and the D-loop).469 0.257 0.724 0.6 T atp6 400 nd1 12S nd5 300 D-loop 0. The three shades of grey correspond to the support is also shown for a concatenated data set comprising three tree topologies shown above the graph.004 0.171 0. loop regions of two rRNA genes. different outgroup taxa (H = horse.

with a 95% highest atp8 posterior density (HPD) interval of about 28–33 Myr.9 (13. Indian + Javan and cox3 in the maximum-likelihood analyses.biomedcentral.4) 30.2) Black-White 15.7) Woolly-Sumatran 19. (13 protein-coding genes. Bayesian estimates of divergence times within Rhinocer- nd2 otidae.5) Node A 30.7 – 15.5) 15.3 (13.7 – 33.com/1471-2148/9/95 Sumatran + Woolly Several genes exhibited strong support for particular Black + White topologies: nd2. americanum) [11].4 (11.8 – 21. calculated using two fossil-based calibrations.9) 20.1 (13.5 – 18.7 – 14. Page 5 of 11 (page number not for citation purposes) .6) 13.5 – 14. Time is in millions of years before the present. These genomes were successfully used to resolve long-standing phylogenetic questions. and the mastodon (Mammut (H = horse. the mean time separating these two nodes represents only 3. and cytb in the Bayesian analyses.4 – 22.5 Myr afterwards. and Dinornis concatenated data set comprising these components.6 (29.8) 19.0 – 33. D-loop were similar across different tree topologies (Table 2). Black + White Indian + Javan Indian + Javan Black + White Sumatran + Woolly Sumatran + Woolly Discussion As ancient DNA techniques have advanced. T = tapir.0) *95% highest posterior density interval. loop regions of two rRNA genes.8 (17.2 – 34. The estimated support is shown for mtDNA genomes were generated using conventional individual components of the mitochondrial genome overlapping-PCR and Sanger sequencing techniques.2 (11.0–1.25].5 – 21. includ- ing the ratite and Elephantidae phylogenies.0 – 33.4) 15.6 (28. both questions were those for which prior analysis of smaller mitochondrial and nuclear fragments had yielded contradicting results [28-33].9 (11.27].8) 31.1 – 34. cox3.6 (17. 9:95 http://www. Initial complete ancient tree topologies. nd4l The estimated age of the common ancestor of the six cox2 nd1 atp6 rhinoceros taxa was around 30 Myr.5 (27.2) 12.0 (29.78% of the total tree height.BMC Evolutionary Biology 2009. the complete Figure 4 mitochondrial genomes of extinct taxa are now regularly Ternary plot showing the relative maximum- likelihood bootstrap support for three candidate being included in DNA analyses.5 (29.0) Node B 31. with giganticus) [24. yielding the mtDNA genomes of three moa species and the D-loop).3) 30. Relative support is also shown for a (Emeus crassus.4 (28.5 – 17. several woolly mammoths (Mam- results from three analyses using different outgroup taxa muthus primigenius) [26.0 – 34. Anomalopteryx didiformis.6 (18. B = both). with a very considerable B H nd3 overlap in the 95% HPD intervals of these two adjacent T 16S cox1 nodes (nodes A and B in Table 2). 12S nd5 The next divergence within the group occurred only nd6 around 1.6 – 16. In light of these Table 2: Rhinocerotid divergence times estimated using Bayesian phylogenetic analysis with a relaxed molecular clock. Notably.3) 32. On average across the nd4 cytb cox3 three topologies. assuming each of three candidate tree topologies Date estimate (mean and 95% HPD*) Divergence event Indian-Javan 13.

nail. Conclusion Extractions and sequencing We have demonstrated that. not far from the Arctic Ocean coast but can be identified as having occurred during the early (Yakutia. The considerable heterogeneity in pure mtDNA. 2. a question that land). 9:95 http://www. Finally. In these cases. The cause of the multiple defining ceros hair was excavated from the permafrost near the speciation events of the rhinoceros family is unknown. powdered nail was will need to be addressed using data from multiple obtained by drilling directly into the nails. we are unable to resolve the tree topology. The woolly rhino- of a hard polytomy.BMC Evolutionary Biology 2009. cleansed in 10% commercial bleach solution following drial genomes can be generated from old or ancient Gilbert et al. We have previously argued that the transmission. age of this specimen is not known. [13]. it is possible that the Museum. (Figures 1. mtDNA genomes.9 ± 0. is that used historic museum toenail samples. thus. [34] to explain the phyloge- Javan rhinoceroses (the latter not found in captivity) we netic conflicts among chloroplast genes in ferns. leaving hair shaft was used for the woolly rhinoceros. our inconclusive findings come as a Furthermore. [13]. In this study we have used hair shaft. dated to approxi- recombination might have occurred in the rhinocerotid mately 100 years old. restriction enzyme mapping studies.1 – 28. and the pellet partial mtDNA genomic sequences apparently resolved was thoroughly washed three times in ddH20 to remove the long-standing question as to the true phylogeny of all traces of the bleach. The discrepancy between the results of the present and previous studies led us to investigate the topologies Methods supported by each individual mitochondrial gene. ancient permafrost-preserved the three sister-species pairings of only ~1 Myr. Previous morphological studies.biomedcentral. DNA analyses have horn-based topology grouping the black and white not yet been able to provide a solution to the rhinoceroses with the Sumatran and woolly rhinoceroses evolutionary history of this challenging taxon. Olenyok River valley. Gilbert et al. Taimylyr village.7]. nail is Prior to DNA extraction. hair shaft was thoroughly an excellent substrate with which complete mitochon. the third possible candi. prior to pelleting through a centrifugation step.93°E) Oligocene (33. University of Copenhagen (Denmark). with 3–7 genes In our previous studies [13-15. A less likely in 2002 and is kept in the Lena Delta Reserve in Tiksi. Using these substrates we have generated the incubated in 10% commercial bleach solution for 5 sequences of four previously unpublished rhinoceros minutes. have supported the Thus. in addition to hair. respectively.2. 72. digestion of the hair shafts Page 6 of 11 (page number not for citation purposes) . details are given in Table 3. and ferns [34]. A possible explanation for the lack of have explored the properties of a similarly keratinous resolution could be the short divergence time among tissue. The powdered nail was likewise material. 3. data have favoured the geography-based topology. One would be the inclusion of DNA sequences recovered grouping all the Asian species together to the exclusion from a more closely related outgroup species. More recent studies of morphological and DNA possible steps might be taken to resolve this problem. Several [1. The Samples signal varied considerably among genes.61°N and 121. however. the age of known specimens likely precludes date grouping (the African species together with the the survival of DNA in them.com/1471-2148/9/95 successes. For the black and posited by Shepherd et al. for the rhinoceroses at least. Thus it is probable that Indian and Javan rhinoceroses) cannot be statistically satisfactory resolution of the phylogeny will instead rejected. but echoes similar results obtained for benefits derive principally from the keratin structure of elephantids [11] and for chloroplast genes in Asplenium the material.35.8]. Full sample nuclear loci. Both hair and nail material were the recent rhinoceros species. Specifically. Although several previous studies of Subsequently the bleach was poured off. we have reported supporting each of the three candidate topologies the benefit of using hair shaft as a source of enriched. the explanation for the lack of phylogenetic resolution. members of the Elasmotherium genus. and provided by the Zoological mitochondrial genome. from which complete mitochondrial topological support among mitochondrial genes is genomes can be rapidly sequenced using the Roche unusual due to the non-recombining mode of genomic FLX technology.1 Myr ago). the poor resolution might be indicative (USA) for the Sumatran rhinoceros.4 ± 0. even when using complete mitochondrial surprise. Unfortunately chondrial genomes. 4). and the mitochondrial tree does not provide an accurate reflec- Oxford University Museum of Natural History (Eng- tion of the underlying species phylogeny. while little time to accumulate mutations in the ancestral modern hair shaft was donated by Cincinnatti Zoo branches. require the incorporation of substantial amounts of nuclear DNA data. Specifically. however.36]. we have demonstrated that digested and DNA was subsequently purified following these estimates were probably misled by sampling error. therefore. such as of the African species [3. With the complete mito. along with sequences.

0).14 88.com/1471-2148/9/95 Table 3: Details of the samples and sequencing efforts Species Sample ID Material Weight (g) Age (years) Bases sequenced %mtDNA Length Coverage Genbank Black #36 Nail 0. The genomes could be quality and quantity of the sstDNA library was assessed assembled as single contigs for the woolly. 100 359.2 Fresh 199.644 1. Post rhinoceros sample is the highest observed from any digestion DNA was purified twice with phenol and once hair shaft studied to date. as (Equus caballus [GenBank:X79547] [38]) and lowland identified through alignment with the white rhinoceros tapir (Tapirus terrestris [GenBank:AJ428947]. Page 7 of 11 (page number not for citation purposes) .223 0. The DNA libraries from the four new samples yielded mitochondrial genome sequences from the horse between 0.2 12. 10 mM NaCl2.087 for black rhinoceros. Qiagen). was initially assembled into 15 contigs.6% mtDNA sequences (Table 3). with chloroform following standard protocols. 565. DNA-containing solution was concentrated to The mtDNA genomes of the taxa were assembled using 200 μl using Amicon Ultra-15 Centrifugal Filter Units the white rhinoceros mtDNA genome as a guide. Thus we are confident the data represents of the single-stranded template DNA (sstDNA). DNA sequencing. both to ensure the sequence accuracy of the beads without an amplified sstDNA fragment were regions that had low levels of FLX coverage (<3×). PCR- based) ancient mtDNA studies (in particular for the Library construction.BMC Evolutionary Biology 2009.41 100. Traditional clonally amplified within individual emulsion droplets. and their subsequent erroneous designation as constructed as previously described [37]..0). nail is comparable to mM Dithiothreitol (DTT.biomedcentral. and 96. and the beads with an amplified sstDNA fill in the missing data (Table S1: Additional File 3). The the true mtDNA sequence.1 5.5 Not dated 498. In total. that principally relate to the differences DNA polymerase. fragment were recovered for sequencing. Cleland's reagent) and 10% hair. however. T4 polynucleotide kinase. 2% w/v associated with a lack of closely related nuclear genome Sodium Dodecyl Sulfate (SDS). Furthermore.86 ca.748 6. and E.41 and 6. Sequencing reads from each project pairwise distances for the six mitochondrial genomes is were aligned using the complete mitochondrial genome provided in Table S2 (Additional File 4). 40 that. A matrix containing uncorrected Javan rhinoceros. and to removed. using mitochondrial genome. CA).8 FJ905813 Sumatran Suci* Hair 0. 2. black rhinoceroses.1 FJ905816 Javan 4139 Nail 0. The aqueous. with a 10 kD filter (Millipore). The variable number tandem repeats generated for woolly rhinoceros. due to difficulties 10 mM Tris-HCl (pH 8.56 ca.082 reads were aligned manually. The remaining sequence com- between 10 and 40 ml of the following digestion buffer: position was not analysed in detail. and assembly woolly rhinoceros [9]) is the PCR amplification of The DNA libraries from four rhinoceros species were numts. with details of from the white rhinoceros as a reference [GenBank: informative sites given in Table S3 (Additional File 5). and was set. To allow the position of the root to be inferred. and using the Agilent 2100 Bioanalyzer (Agilent Technolo. as a source of ancient mtDNA. leaving a data set rhinoceros.6 91.1 FJ905815 *Zoo studbook number 43. the mtDNA yield of the woolly proteinase K solution (>600 mAU/ml. pH 8. coli between the shot-gun nature of FLX emPCR and DNA polymerase.8 21. Sequence alignment and partitioning cing System (Roche Applied-Sciences. Each sstDNA library fragment was which yielded the lowest levels of mtDNA in this data captured onto a single DNA capture bead. A frequently reported problem in conventional (i. gies. The data indicate Ethylene-Diamine-Tetra-Acetic acid (EDTA). 100 85. was performed overnight at 55°C with rotation. The polished DNA fragments were sequencing and the targeted nature of PCR based then subjected to adapter ligation.5 30. Javan. 5 mM CaCl2.235 1. analogous to the method of Gilbert et al. IN) The mitochondrial genomes of the six rhinoceroses were as previously described.21 119. followed by isolation amplification. PCR and Sanger sequencing was thus applied to the The emulsions were disrupted using isopropanol. [13]. Indianapolis. This is not. 70. Santa Clara.900 for of 16. by shearing true mtDNA sequences.3 FJ905814 Woolly COEL LDR P75 Hair 1.e.5 mM against which to compare the results. 251. Y07726]. The recovered sstDNA beads were packed onto a 70 × 75 mm PicoTiterPlate™ and loaded onto the Roche FLX Sequen.323 aligned sites. the extract. a problem genomic DNA into fragments which were blunt-ended with FLX generated sequences for reasons argued and phosphorylated by enzymatic polishing using T4 previously [13]. while that of the Javan rhinoceros.065 for Sumatran (VNTRs) in the D-loop were removed. 9:95 http://www.

excluding the horse and In order to place a geological time-scale on the tapir sequences in turn. First. protein-coding genes. with substitution models selected assumed for each of the five data partitions. identified using a branch-and-bound search. rRNA gene (loops only).51]. includ. fossil record. mixing and convergence to the stationary distribution were checked using Tracer. using maximum- as for the analyses of the concatenated data set. mixing and convergence to the stationary distribution with the first 500 samples discarded as burn-in. Pairwise distances were estimated with the GTR+I tion criterion scores (Table S3). Posterior distributions of parameters. and (v) D-loop. different combinations of outgroup taxa were Analyses including additional representatives of Laur. relaxed-clock model in BEAST 1.000 MCMC steps over a total of 11. tion criterion scores (Table S3). two calibrations were taken from the to those of the original analysis.biomedcentral. The D-loop gawa test [49].7 [50. In each analysis. except analysis of the concatenated data described above.4 [44]. the maximum-likelihood tree was We constructed a concatenated data set (13.4. (iv) loop regions of 12S and 16S rRNA Topology tests were performed in a maximum-likelihood genes. Site-wise log likelihoods were omes. This program defined in accordance with the RNA secondary structural implements the approximately-unbiased test [47]. To investigate variations in phylogenetic signal among different genes. the models of Ceratotherium simum available from the Kishino-Hasegawa test [48]. bootstrap the 13 protein-coding genes. a lognormal prior (minimum 56 Myr. categories for the discrete gamma distribution. The We performed a Bayesian phylogenetic analysis of the alignment was analysed using an uncorrelated lognormal concatenated sequence alignment using MrBayes 3. software PAUP* 4b10 [45]. using both horse and tapir as the outgroup. using the representatives from all three families of Order Perisso. tapir.BMC Evolutionary Biology 2009.000 steps. tRNA genes. As with the Bayesian dactyla: Rhinocerotidae. Acceptable every 1. Samples from the posterior were drawn the first 1. analyses. including the RNA stems. were estimated using Markov chain with a separate analysis performed for each of the three Monte Carlo (MCMC) sampling with two chains (one candidate trees (see Table 2). were Maximum-likelihood phylogenetic analysis was also added to the alignment. The same settings were used +G model of nucleotide substitution. As with the A separate substitution model was assumed for each of MrBayes analysis. The GTR+I+G model of asiatheria were performed.000 steps. (ii) second codon sites of analysis was conducted with 1. ing the tree topology.500. The tree topology was fixed.1 [43]. Page 8 of 11 (page number not for citation purposes) . Rhinocerotid divergence times were estimated by Bayesian phylogenetic analysis of the concatenated sequence align- Phylogenetic analysis ment. relative confidence in three competing topological ium simum. with (MCMC) sampling. To examine the effect of outgroup choice. Both the alignment of complete mitochondrial genomes outgroup taxa (horse and tapir) were included. and the D-loop. with six rate different results. Each marised in Table S4 (Additional File 6). Posterior distributions of heated). The RNA loop regions were framework using the program Consel [46]. Bayesian and maximum-likelihood phylo- Split decomposition genetic analyses were performed for each protein-coding To examine the overall phylogenetic signal.000 pseudoreplicates. Using the split decomposition method [42]. with the same settings as for the intergenic sites were not used for further study. we analysed gene. are not presented here but are sum.000 samples discarded as burn-in. We used these to assess the was defined using the GenBank annotation of Ceratother.com/1471-2148/9/95 unpublished data deposition in Genbank [39]). in the split decomposition analysis. without yielding qualitatively nucleotide substitution was assumed. tested (horse.000 MCMC steps over a total of 5. To assess with the following five partitions: (i) first codon sites of levels of support for different nodes in the tree. Other sections of the mitochondrial gen. a separate substitution model was the five data partitions. All other settings were identical phylogenetic tree. (iii) third codon sites of protein- coding genes. likelihood estimates of model parameters. we ran two further phylogenetic analyses. with substitu- by comparison of Bayesian information criterion scores tion models selected by comparison of Bayesian informa- (Table S3). The resulting data set includes performed on the concatenated data sets.000. Acceptable were checked using Tracer 1. and calculated using PAUP. the data were canoni- cally decomposed into a sum of weakly compatible splits Divergence time estimation and represented in the form of a splits graph.714 bp). and both). Samples from the posterior were drawn every parameters were estimated using Markov chain Monte Carlo 1. among others. Substitution (excluding the VNTRs) using the software SplitsTree 4 models were chosen by comparison of Bayesian informa- [41]. 9:95 http://www. and Equidae. hypotheses (see Table 2). Tapiridae. data partition was allowed to have a unique substitution rate. and the Shimodaira-Hase- European Ribosomal RNA Database [40].

stig: stigmatellin). Nonsynonymous sites in mitochondrial co1/co2/co3 Curie Training Site (EW and PFC).E. Ursus arctos [GenBank:NC_003427]. MTPG. GENETIME – a Marie Figure S2. The Depart- represented as grey spheres. TVK. Asia. Click here for file or conclusions.doc] 4.com/content/supplementary/1471- 2148-9-95-S2. The prosthetic groups are represented in black. Europe with Table S1. 33:1–59. Second. and 2148-9-95-S6. NHGRI grant HG02238 (WM). when mutated in humans. [GenBank:AJ428947].com/content/supplementary/1471. Tougard C. MN-K and TLR coordinated sampling [GenBank:NC_001788]. 21:293–313. using nine laurasiatherian outgroup species. are responsible for exercise intolerance. Hanni C and Montgelard C: Phylogenetic relationships of the five extant rhinoceros species (Rhino- cerotidae. Acknowledgements Click here for file We thank the Lena Delta Reserve. Equus caballus [GenBank:X79547]. This project is funded in that located in a functionally relevant area represented as orange part (SCS) with the Pennsylvania Department of Health with the use of spheres. and Sorex unguiculatus [GenBank:NC_005435]. Additional file 6 Table S4. Mol Phylogenet Evol 1994. 15:599–605. Guérin C: Les Rhinocerotidae (Mammalia. AS. 2148-9-95-S3. with those located in [http://www. Matrix of uncorrected p-distances for the whole mitochondrial 5. and the University of Copenhagen [http://www. 2. Zeit [http://www. Support for three candidate topologies estimated using Authors' contributions Bayesian and maximum-likelihood phylogenetic analysis. a minimum age constraint of 16 Myr was placed on the Dicerorhinus lineage [55]. RRdF. Am Mus Nov 1995. Sites that. with Basic Research grant 07-04-01612 (AS and TK). interpretations. Click here for file 3. are depicted in red. All tetradactyla [GenBank:NC_004027]. Perissodactyla) based on mitochondrial cyto- Page 9 of 11 (page number not for citation purposes) . rhinoceros.000.biomedcentral. The Click here for file nonsynonymous sites are shown in white. Equus asinus analysis. Perissodactyla) du Miocene terminal au Pleistocene superieur d'Europe occi- dentale compares aux especes actuelles: Tendances evolu- Additional file 4 tives et relations phyletiques. Geobios 1982.pdf] References 1. read and approved the final manuscript.biomedcentral. Museum of Zoology for providing sampling access to the woolly and black 2148-9-95-S1.doc] surrounded by boxes in the alignment. Scripta Geol 1975.com/content/supplementary/1471- fossil belonging to either of the lineages descending 2148-9-95-S5. SYWH. and horse. Bull Am Mus Nat Hist 1945.pdf] rhinoceros samples. Prothero DR and Scoch RM: The Evolution of Perissodactyls. Manis interpreted the results in the biological context.com/content/supplementary/1471- functionally relevant areas represented as orange spheres.51 Myr) was specified for the age of the root [52. In addition to the six rhinoceros EW. and bound inhibitors in brown (ant: antimycin.com/content/supplementary/1471. the only exceptions being the use of the heuristic tree- bisection-reconnection instead of a branch-and-bound search in the Additional file 1 maximum-likelihood analysis. [http://www. MTPG and SCS conceived the study. Cerdeño E: Cladistic analysis of the family Rhinocerotidae given in bold. using an unpartitioned GTR+I+G substitution model. MN-K and TLR NC_002009]. authors wrote. Groves CP: Phylogeny of the living species of Rhinoceros. 3:128–134.BMC Evolutionary Biology 2009. tapir. mapped onto the bovine structure [PDF:1PPJ] [22]. mapped onto the bovine structure (SYWH). and surrounded by a box in the alignment. Table S2.com/content/supplementary/1471. JB. Details of the primers used for gapfilling of the Javan reference to the recent two-horned species of Africa and rhinoceros mitochondrial genome.biomedcentral. PFC. S.53]. The Additional file 2 Danish National Research Foundation (EW. and The Russian Foundation for [PDF:1V54] [23]. Artibeus jamaicensis [GenBank: of biological material.biomedcentral. 3134:1–24. AR. AS. Nonsynonymous sites in mitochondrial cytb sequences of support values from 200 pseudoreplicates rather than 1. ment specifically disclaims responsibility for any analyses.biomedcentral.com/1471-2148/9/95 mean 60 Myr. Zool Syst Evol 1983. Morales JC and Melnick DJ: Molecular systematics of the living [http://www. 2148-9-95-S4. Financial support is gratefully acknowledged from: The Danish Natural Science Research Council (EW and MTPG). MTPG. This is probably Additional file 5 the most appropriate summarisation of paleontological Table S3. Bos taurus [GenBank: NC_006853]. WM and SCS carried out data Hippopotamus amphibius [GenBank:NC_000889]. Click here for file age has a mode that is older than the age of the oldest [http://www. sequences. and the calculation of the bootstrap Figure S1. Delefosse T. genomes of six rhinoceros species.doc] from the node [54]. standard deviation 1. Click here for file 7. The nonsynonymous sites are shown in white. Details of the components of the aligned mitochondrial information because the probability density of the nodal genomes from six rhinoceroses.doc] 8. Loose H: Pleistocene Rhinocerotidae of W. with distances between sister species 6. New York: Oxford University Press. Simpson GG: The principles of classification and a classifica- Additional file 3 tion of mammals. (Perissodactyla). Phylogenetic analyses were performed on first and second codon positions only. JB).biomedcentral. the following outgroup taxa were included: Tapirus terrestris LPT and SCS carried out the molecular genetic studies. 85:1–350. Other Additional material settings in the Bayesian and likelihood-based analyses were as described in the main text. Prosthetic groups are Tobacco Settlement Funds appropriated by the legislature. TVK. 1989. The Australian Research Council sequences of rhinoceroses. rhinoceroses. 9:95 http://www.biomedcentral.

Axelsson E. 37. 11. Shimokata K. Nucleic Acids Res 2004. Ozawa T. likelihoods with applications to phylogenetic inference. 5:e207. Pusey B. Associates. et al: Sequencing the nuclear 17. Spriggs H. Nature 2008. mitogenomic tree of mammalian orders. Perrière G and Peer Van de V: The European ribosomal 20. 26:421–434. Mammalia): new 14. Andersen PK. Rhinoceros unicornis. Noro M. 34. 93:1190–1194. Malyarchuk BA. 38:1863–1888. Mol Biol Evol 2006. Miller W. Bandelt H-J and Dress AWM: Split decomposition: A new and and Bouchard C: The human gene map for performance and useful approach to phylogenetic analysis of distance data. BMC Genomics 2008. Perusse L. Cummings MP. Dear PH. Bemben LA. 1999. et al: The sequence of the white rhinoceros. Mol Phylogenet Sher A.com/1471-2148/9/95 chrome b and 12S rRNA genes. PLoS Biol 2006. H A. Perissodactyla. Gullberg A. Gilbert MTP. DNA sequence of the greater Indian rhinoceros. Altman WE. 13:1167–1173. et al: Intraspecific Evol 2003.Sunderland. 7:214. Janke A. Rohland N. Proc R Soc Lond B 2000. 9. Mol Phylogenet Evol 2008. Nature 2006. Janecka JE. Tung EY and Berry EA: Binding of the of phylogenetic trees. Rambaut A. J Mol Evol 1994. Rhinoceros 456:387–390. Tomsho LP. Ratan A. Bioinformatics 2001. 48. Katayama Y. the vicariance biogeography hypothesis. Grønnow B. Laudet V. Gene 1994. Otto SP and Wakeley J: Sampling properties of mitochondrial cytochrome b and 12S ribosomal RNA genes. 4: Derenko MV. Equus caballus: extensive hetero- 19. Molecular and morphological evidence on the phylogeny of 105:8327–8332. Phillips MJ and Rambaut A: Relaxed 27. Proc Natl Acad Sci USA 2008. 2000. Zhoa F and Sher A. Dalén L and Götherström A. Mammuthus primigenius. Lesk AM. Tomsho LP. Xu X. Cao Y. 4:e73. 17:754–755. relationships among eutherian orders estimated from 29. Kishino H and Hasegawa M: Evaluation of the maximum 409:704–707. Cooper A. Hofreiter M: Proboscidean mitogenomics: Chronology and 30. Laakkonen L. Holland BR and Perrie LR: Conflict amongst Reidla M. Qi J. Xu X and Arnason U: The for a carnivorous plant radiation. Slatkin M. Lister AM. 48:176–187. Huang LS. Wikstrom M and Albert VA: plasmy of the control region. Mol Biol Evol J Mol Evol 1998. 32:D101–D103. Med Sci Mol Phylogenet Evol 1992. Science 2008. PLoS Biol 2006. Dalén L. BMC Evol Biol 2007. Lesk AM. Proc Natl Acad Sci USA 17:1246–1247. phylogenetics and dating with confidence. and Artiodactyla (+ Cetacea). Yang Z and Lister AM: complete mitochondrial genomes. 21. genome shotgun sequencing of mitochondria from ancient 32. Bader JS. 317:1927–1930. 100:15304–15309. Page 10 of 11 (page number not for citation purposes) . Gilbert MTP. Ratan A. J Mol Evol 1989. 47. Chumakov I and Grigorenko AP: Complete e88. Possnert G and Pääbo S: Nuclear DNA 13. Drautz DI. and the branching order in genome sequences of extinct birds: ratite phylogenetics and Hominoidea. 16:1114–1116. (*and Other Methods). Moyes CD. Bioinformatics 2001. Mol Phylogenet Evol 1997. Gilbert MTP. respiratory chain inhibitor antimycin to the mitochondrial 44. 1:242–252. Nielsen R.. Shimodaira H: Another calculation of the p-value for the 24. Mol Biol Evol Tikhonov A. Zhang Y. Johnson WE. two extinct moas clarify ratite evolution. Ward R: Complete mitochondrial genome sequences of Stanford University. Orlando L. 101:18064–18068. Fredriksson E and C LS: Molecular RNA database. Rogaev EI. 45. 15. 12:814–822. Austin J and problem of regions using the scaled bootstrap resamplings. and the Phylogenetic relationship among Carni. Pääbo S and Hasegawa M: Phylogenetic Proc Natl Acad Sci USA 1996. Jobson RW. based on complete sequences of 12. Chen YJ and Chen Z. Yoshida MC and Kato M: mode of elephant evolution using mastodon as outgroup. 43. Muramoto K. 51. J Mol Evol 2006. DNA sequence data in phylogenetic analysis. Kuznetsova T and Kosintsev P. Rendulic S. Drautz DI. 1995. Matheus P and 44:406–413. Wolfarth B 42. sequences from late Pleistocene megafauna. unicornis. evolution of cytochrome c oxidase in high-performance fish 41. 437:376–380. Pääbo S and Hofreiter M: Multiplex 50.biomedcentral. Anderson S. Mol Phylogenet Evol 2001. Packard MD. Margulies M. Janke A. 36. Rankinen T. Tamm E. Krause J. Packard M. Kivisild T. 16. Proc R Soc Lond B 49. Metspalu E. Adaptive evolution of cytochrome c oxidase: Infrastructure 39. phylogenetic analysis of Siberian woolly mammoths using 33. O'Brien SJ. strated by mitochondrial DNA sequences. Sher A. 268:939–945. Mochizuki M. Moliaka YK. 7:189–194. Shimodaira H and Hasegawa M: Multiple comparisons of log- 2001. 31. Lesk AM. moth Mammuthus primigenius. mitochondrial genome and phylogeny of Pleistocene mam- 19:39–44. American mastodon (Mammut americanum) as an outgroup. 10. Proc Natl Acad Sci USA 2004. Drummond AJ and Rambaut A: BEAST: Bayesian evolutionary amplification of the mammoth mitochondrial genome and analysis by sampling trees. Hayashi S and Mikhelson VM: Phylogenetic position of inferred sequences of mitochondrial proteins: Instability of mammoth and Steller's sea cow within Tethytheria demon- a tree based on a single gene. Attiya S. Drautz DI. 35. Braverman MS. et al: 46. et al: chloroplast DNA sequences obscures the phylogeny of a Paleo-Eskimo MtDNA genome reveals matrilineal disconti. J Mol Biol 2005. 9:95 http://www. Shepherd LD. Yamashita E and Yao M. 39:519–527. Greenwood AD. Ceratotherium simum. Hagberg JM. 2004. The adaptive evolution of the mammalian mitochondrial 38. Bray MS. likelihood estimate of the evolutionary tree topologies 25. health-related fitness phenotypes: the 2005 update. Ho SYW. 9:119. Wuyts J. Debruyne R. 148:357–362. Götherström A and Campos PF. and mitochondrial genome sequence of the Tasmanian tiger comparison with the mtDNA sequence of the Indian (Thylacinus cynocephalus). 40. in evolutionary studies. Adachi J. Pollack JL. Drautz DI. the evolution of Elephantidae. 2003. Mol Biol Berka J. Molecular phylogenetic inference of the woolly mammoth PLoS Biol 2007. 46:314–326. et al: Genome Evol 1996. Shimada H. Xu X and Arnason U: The complete mitochondrial DNA genome. Masuda R. Leonard JA. Janke A and Arnason U: The complete mitochondrial genome of the extinct woolly mammoth. Slatkin M. Tomsho LP. 3:485–499. vora. 267:2493–2500. Qi J. 320:1787–1789. Xu X and Arnason U: The complete mitochondrial DNA Packard M. 62:319–331. Mol Phylogenet Evol 2003. Cobessi D. Egholm M. the Elephantidae. Sports Exer 2006. da Fonseca RR. 23:254–267. Ho SYW. Ratan A. Shimodaira H and Hasegawa M: CONSEL: for assessing the The low-spin heme of cytochrome c oxidase as the driving confidence of phylogenetic tree selection. Ramos MJ and Antunes A: tors. data and phylogenetic analyses of Elephantidae. Swofford DL: PAUP* Phylogenetic Analysis Using Parsimony 351:573–597. Massachusetts: Sinauer 23. 16:1466–1473. et al: Whole. Hsu C-H. group of Asplenium ferns. Mol 26. Lalueza-Fox C. Hagelberg E. M GE and Shoshani J: Phylogenetic resolution within Ancient DNA analysis reveals woolly rhino evolutionary the Elephantidae using fossil DNA sequence from the relationships. 2003. Biol Evol 1999. Guerin C and Hanni C: 28. Rambaut A and Drummond AJ: Tracer. Huson DH and Bryant D: Application of phylogenetic networks (teleostei: Scombroidei). Ebersberger I. element of the proton-pumping process. Unpublished. Nature 2005. Shinzawa-Itoh K. Yang H. Miller W. Nature 2001. Tsukihara T. rhinoceros. Dubrovo IA. Science 2007. J Mol Evol 1997. Barriel V and Tassy P: Mitochondrial cytochrome b hair shafts. McClellan LR and Qi J. Huelsenbeck JP and Ronquist F: MRBAYES: Bayesian inference 22. Capelli C. Malaspinas A-S. Drummond AJ. Harley EH. sequence of the horse. Genome Res 2009. Kondrashov FA. Dalziel AC. Roth SM. Pollack JL. Thomas MG. of the Lyakhov mammoth (Proboscidea. 29:170–179.BMC Evolutionary Biology 2009. intramolecular hydrogen-bonding pattern. 439:724–727. Haddrath O and Baker AJ: Complete mitochondrial DNA from DNA sequence data. Jones HB. Barnes I. nuity in Greenland. sequencing in microfabricated high-density picolitre reac- 18. Oxford: University of bc1 complex: a new crystal structure reveals an altered Oxford. Thenot A.

BMC Evolutionary Biology 2009. Publish with Bio Med Central and every scientist can read your work free of charge "BioMed Central will be the most significant development for disseminating the results of biomedical researc h in our lifetime.biomedcentral. Ho SYW: Calibrating molecular estimates of substitution rates and divergence times in birds. 24:26–53. Benton MJ and Donoghue PC: Paleontological evidence to date the tree of life. 55. Vertebrate Paleontology and Evolution New York: W. Edentates. Cancer Research UK Your research papers will be: available free of charge to the entire biomedical community peer reviewed and published immediately upon acceptance cited in PubMed and archived on PubMed Central yours — you keep the copyright Submit your manuscript here: BioMedcentral http://www. The Evolution of Perissodactyls New York: Oxford University Press: Prothero DM.com/info/publishing_adv. Scoch RM 1989. and Whales.biomedcentral. Mol Biol Evol 2007. 9:95 http://www. Carroll RL: Ungulates. Scoch RM: A Review of the Tapiroids." Sir Paul Nurse. 502–568. J Avian Biol 2007. 298–320. 54. Freeman and Co: Carroll RL 1988. H.asp Page 11 of 11 (page number not for citation purposes) . 38:409–414.com/1471-2148/9/95 52. 53.