Molecular Phylogenetics and Evolution 78 (2014) 25–35

Contents lists available at ScienceDirect

Molecular Phylogenetics and Evolution

journal homepage: www.elsevier .com/ locate /ympev

A review of molecular-clock calibrations and substitution ratesin liverworts, mosses, and hornworts, and a timeframefor a taxonomically cleaned-up genus Nothoceros

http://dx.doi.org/10.1016/j.ympev.2014.04.0141055-7903/� 2014 Elsevier Inc. All rights reserved.

⇑ Corresponding author.E-mail address: jcarlos.villarreal@gmail.com (J.C. Villarreal).

Juan Carlos Villarreal ⇑, Susanne S. RennerSystematic Botany and Mycology, University of Munich (LMU), Germany

a r t i c l e i n f o

Article history:Received 31 January 2014Revised 30 March 2014Accepted 15 April 2014Available online 30 April 2014

Keywords:Bryophyte fossilsCalibration approachesCross validationNuclear ITSPlastid DNA substitution ratesSubstitution rates

a b s t r a c t

Absolute times from calibrated DNA phylogenies can be used to infer lineage diversification, the origin ofnew ecological niches, or the role of long distance dispersal in shaping current distribution patterns.Molecular-clock dating of non-vascular plants, however, has lagged behind flowering plant and animaldating. Here, we review dating studies that have focused on bryophytes with several goals in mind, (i)to facilitate cross-validation by comparing rates and times obtained so far; (ii) to summarize rates thathave yielded plausible results and that could be used in future studies; and (iii) to calibrate a species-level phylogeny for Nothoceros, a model for plastid genome evolution in hornworts. Including the presentwork, there have been 18 molecular clock studies of liverworts, mosses, or hornworts, the majority withfossil calibrations, a few with geological calibrations or dated with previously published plastid substitu-tion rates. Over half the studies cross-validated inferred divergence times by using alternative calibrationapproaches. Plastid substitution rates inferred for ‘‘bryophytes’’ are in line with those found in angio-sperm studies, implying that bryophyte clock models can be calibrated either with published substitutionrates or with fossils, with the two approaches testing and cross-validating each other. Our phylogeny ofNothoceros is based on 44 accessions representing all suspected species and a matrix of six markers ofnuclear, plastid, and mitochondrial DNA. The results show that Nothoceros comprises 10 species, ninein the Americas and one in New Zealand (N. giganteus), with the divergence between the New Zealandspecies and its Chilean sister species dated to the Miocene and therefore due to long-distance dispersal.Based on the new tree, we formally transfer two species of Megaceros into Nothoceros, resulting in thenew combinations N. minarum (Nees) J.C. Villarreal and N. schizophyllus (Gottsche ex Steph.) J.C. Villarreal,and we also newly synonymize eight names described in Megaceros.

� 2014 Elsevier Inc. All rights reserved.

1. Introduction

Molecular clock dating has become an indispensable tool inevolutionary biology and has utterly transformed the field of bio-geography. Absolute times from calibrated phylogenies can beused to infer the climate or geological context of lineage diversifi-cation (Schneider et al., 2004), the role of long distance dispersal inshaping current distribution patterns (Renner, 2005; Heinrichset al., 2009a; Schaefer et al., 2009; Linder et al., 2013), or the originof new ecological niches (Schneider et al., 2004; Schuettpelz andPryer, 2009; Bytebier et al., 2011; Crisp et al., 2011). The use ofdated phylogenies in evolutionary studies of non-vascular plants,however, has lagged behind flowering plant and animal dating

studies. Part of the explanation may lie in the scarcity of liverwort,moss, or hornwort fossils suitable for calibration, although recentwork has brought to light characteristic hornwort spores(Chitaley and Yawale, 1980; Graham, 1987; Archangelsky andVillar de Seone, 1996) and many amber-preserved mosses(Heinrichs et al., 2013). The problem of a scarce fossil record, ofcourse, is not restricted to non-vascular plants. In other clades withpoor fossil records, researchers have overcome the difficulty byapplying substitution rates taken from related plant groups, prefer-ably with similar generation times. Substitution rates of plastidand nuclear markers, especially of the internal transcribed spacerregions of ribosomal DNA (ITS1 and ITS2), are available from manyfossil-calibrated phylogenies (e.g., Richardson et al., 2001; Kayet al., 2006). Alternatively, workers have used geological eventsas calibration points, usually the age of oceanic islands with ende-mic radiations, which obviously cannot be older than the island

26 J.C. Villarreal, S.S. Renner / Molecular Phylogenetics and Evolution 78 (2014) 25–35

(e.g., Schaefer et al., 2009). Another way around a lack of calibra-tion fossils is to use a two-tiered approach, in which the clade ofinterest’s root node is assigned an age obtained in another datamatrix, in which fossil calibration was possible because it con-tained a different taxon sampling (as done in Heinrichs et al. 2007).

A combination of these approaches to calibrate branch lengths(genetic distances) in phylogenies has been used to infer diver-gence times also in bryophytes. Here, we review such dating stud-ies with several goals in mind. One is that a review will facilitatecross-validation by comparing rates and divergence times obtainedso far; another is that plastid or nuclear ITS rates that have yieldedplausible results could be used in future studies; a third is that wewanted to calibrate a species-level phylogeny for Nothoceros, theonly hornwort genus with transcriptome data and a completeplastid genome (Villarreal et al., 2013; Li et al., 2014).

The genus Nothoceros contains approximately ten species, andtogether with Dendroceros, Megaceros, and Phaeomegaceros it makesup the family Dendrocerotoceae (Duff et al., 2007). Except for theNew Zealand species N. giganteus, Nothoceros is restricted to theAmericas, where it occurs in southernmost South America (N. fuegi-ensis and N. endiviifolius), the foothills of the Andes, Central Ameri-can and Caribbean mountains above 400 m (N. vincentianus, N.canaliculatus, N. renzagliensis and N. superbus), high Andean páramosand punas, and the Southern Appalachian Mts. in Eastern NorthAmerica (N. aenigmaticus; Villarreal et al., 2010a, 2012). Its sisterclade consists of Dendroceros, which has�43 species and a pantrop-ical and south temperate distribution, and Megaceros, which occursin Africa, Australia, New Zealand, and South America, at least as tra-ditionally circumscribed. The sister to all three genera is Phaeome-gaceros, with 9–10 species in tropical Asia, New Zealand, australand tropical America (Villarreal et al., 2010b). The placement ofthe New Zealand species of Nothoceros close to one of its two Chileanspecies in a previous molecular phylogeny (Villarreal et al., 2012)that included eight species of Nothoceros raised the question of theage of this disjunction. To answer this question, we generatedsequence data from up to seven loci from the three plant genomiccompartments and sampled all species of Nothoceros, mostly withmultiple representatives per species.

2. Materials and methods

2.1. Review and partial reanalysis of published bryophyte clock datingstudies

For all published molecular clock studies on bryophytes that weare aware of, we summarized key aspects, such as size of the datamatrix, clock model and dating program used, calibrationapproach, and whether cross validation by different calibrationmethods was attempted. We also obtained the dataset ofMcDaniel and Shaw (2003) and re-analyzed it under a relaxedmolecular clock with either a plastid substitution rate of5.0 � 10�4 subst./site/my (cf. Section 2.4) and a strict clock or arelaxed clock constrained with the age of the Atacama desertifica-tion (14 ± 1 Ma, using a normally distributed prior), which was thecalibration used in the original study for the node separating Pata-gonian populations from more northern Neotropical populations.

2.2. Plant material, DNA extractions, PCR amplifications, sequencingand alignment

We used DNA sequences from 44 Nothoceros collections (21 ofthem first sequenced here) to represent all species. Outgroups wereMegaceros and M. flagellaris from Australia, and Dendroceros difficilisfrom Fiji; these outgroups were selected based on Duff et al. (2007).Tables S1 and S2 list all sequenced collections with species names

and taxonomic authors, voucher details, and GenBank accessionnumbers. We sequenced four plastid markers, viz. the rbcL gene,the trnL-F intron and spacer, the rps4-trnS spacer, and a portion ofmatK; a fragment of the mitochondrial nad5–nad4 spacer; and thenuclear 5.8S gene and ITS2 ribosomal RNA. Methods for DNAextraction, PCR protocols and sequencing are detailed in Villarrealet al. (2012) and Villarreal and Renner (2013); primer sequencesare shown in Table S3. The first author identified all material basedon his study of the types of most names published in Nothoceros andMegaceros. The total alignment included 47 accessions and 5993base pairs (bp). After excluding regions of ambiguous alignment(exclusions sets available from TreeBase accession number15666) the matrix consisted of 4660 aligned nucleotides.

2.3. Phylogenetic analyses

Phylogenetic analyses were performed under likelihood (ML)optimization using RAxML 7.2.8 (Stamatakis et al., 2008) and theGTR + C substitution model, with six unlinked partitions for the dif-ferent markers. Additional analyses were made with three unlinkedpartitions (one partition per genomic compartment). Statistical sup-port was assessed via 500 ML bootstrap replicates under the samesubstitution model. Bayesian analyses were conducted in MrBayes3.2 (Ronquist et al., 2012), using the default two runs and four chains(one cold and three heated), with default priors on most parametersexcept that we modified the prior on branch lengths using theapproach of Brown et al. (2010) for branch length exponential priors(k ¼ � lnð0:5Þ

brl) where brl is the average branch length obtained from

the maximum likelihood tree. We again used six or three data parti-tions, applying best-fit models found with MrModel test v.2(Nylander, 2004; all are shown in Table S4). Model parameters forstate frequencies, the rate matrix, and gamma shape were unlinked,and posterior probabilities of tree topologies were estimated fromthe partitions. Tracer 1.5 (Rambaut et al., 2013; part of the BEASTpackage) was used to assess convergence and to judge the percent-age of burn-in for tree constructions. Convergence was usuallyachieved in MrBayes after 4 � 106 generations, with trees sampledevery 1000th generation for a total length of 40 � 106 generations;we discarded 10% of each run and then pooled runs.

2.4. Molecular clock dating

To convert genetic branch lengths to absolute times, we appliedthree approaches: (A) A strict clock calibrated with a plastidgenome substitution rate of 5.0 � 10�4 subst./site/my for landplants calculated for the entire single copy region and invertedrepeats by Palmer (1991). This dating approach was used for amatrix of 16 hornwort accessions (and the three outgroups listedunder Section 2.2.) and 3515 aligned nucleotides of the plastidregions rbcL, the trnL-F intron and spacer, the rps4-trnS spacer,and a portion of matK. (B) A relaxed clock calibrated with a horn-wort-specific plastid substitution rate of 5.0 � 10�4 with a SD ofa 2.0–8.0 � 10�4 subst./site/my, using a normal distribution forthe rate prior. We had calculated this rate (and its 95% HPDinterval) for the hornwort clade in a fossil-calibrated 72-taxon rbcLdataset that included representatives of liverworts, mosses,lycophytes and ferns (Villarreal and Renner, 2012). This datingapproach was applied to the same matrix as approach A. (C) Alog-normally distributed relaxed clock model (UCLN) was appliedto the 72-taxon-2442 nucleotide hornwort dataset of Villarrealand Renner (2012) with the same four constraints as used in ourearlier study, except that we constrained the Chilean and NewZealand species of Nothoceros to be monophyletic as they are in aML tree from now available sequences.

Dating relied on Bayesian divergence time estimation as imple-mented in BEAST 1.7.4 (Drummond et al., 2012), using a Yule tree

J.C. Villarreal, S.S. Renner / Molecular Phylogenetics and Evolution 78 (2014) 25–35 27

prior and the GTR + C substitution model with 3 unlinked data par-titions. We ran strict clock and UCLN relaxed clock models. TheMCMC chains were run for 70 million generations, with parame-ters sampled every 7000th generation. Tracer 1.5 (part of theBEAST package) was used to assess effective sample sizes (ESS)for all estimated parameters and to judge the percentage ofburn-in for tree constructions. Trees were combined in TreeAnno-tator 1.6.1 (part of the BEAST package), and maximum clade cred-ibility trees with mean node heights were visualized using FigTree1.40 (Rambaut, 2007). We report highest posterior densities (HPD)intervals (the interval containing 95% of the sampled values).

3. Results

3.1. Bryophyte clock calibrations and published rates

We found 17 molecular clock studies for liverworts, mosses, orhornworts, not including the present analysis of Nothoceros(Table 1). Except Wall (2005) and Hartmann et al. (2006), all stud-ies used relaxed molecular clock models. Most inferred recentdiversification (Eocene-Pleistocene) of bryophyte lineages and aprominent role for distant dispersal in shaping current biogeo-graphic patterns (Wall, 2005; Devos and Vanderpoorten, 2009;Shaw et al., 2010). The only study inferring geographic vicariancedue to the break-up of East Gondwana is that of the moss Pyrrhobr-yum mnoides by McDaniel and Shaw (2003). Our re-analysis of theMcDaniel and Shaw dataset, using either Palmer’s (1991) plastidsubstitution rate or their own Atacama Desert calibration, how-ever, yielded much younger ages for the relevant split, namely20–26 vs. 80 Ma in the original study (Table 1).

Six of the 17 studies applied previously published plastid sub-stitution rate for calibration purposes. The rates used ranged from5.0 � 10�4 to 5.6 � 10�4 subst./site/my (Table 1). The lowest ratescome from our own study of hornworts (Villarreal and Renner,2012). Another three studies used published ITS substitution rates,in one case 1.35 � 10�3 subst./site/my (0.27%/my) and in two cases1.4 � 10�2 subst./site/my (Table 1 provides sources for these rates).Geological calibrations were used in four studies (Table 1), namelythe origin of the volcanic islands of Sta. Maria Island (Azores) at�5.1 Ma (Devos and Vanderpoorten, 2009), Mo’orea at 1.8 Ma,Samoa at 2.9 Ma (Wall, 2005), and the Mascarene Islands at8–35 Ma (Table 1, Pokorny et al., 2011). The formation of theAtacama Desert in the Miocene at �14 Ma was used in one study(McDaniel and Shaw, 2003).

Ten of the 17 studies cross-validated results by trying alterna-tive geological calibrations (Wall, 2005), alternative clocks and treetopologies using Bayes Factors (Pokorny et al., 2011), or compari-son with published substitution rates and fossil calibrations(Aigoin et al., 2009; Shaw et al., 2010; Villarreal and Renner,2012). In the latter three studies, ages obtained with either fossilcalibrations or plastid substitution rates more or less agreed,except that in two of these studies (Shaw et al., 2010; Villarrealand Renner, 2012) root ages obtained with Palmer’s (1991) plastidrate were considerable younger than those obtained with fossil cal-ibrations (our Table 2). In Shaw et al. (2010), the root was the splitbetween liverworts and mosses; in Villarreal and Renner (2012),the root was the split between liverworts and the remaining landplants, in neither study was the root age the focus of the analysis.

3.2. Phylogeny and divergence times of Nothoceros

The length and number of informative sites for each of the sixmarkers used for the Nothoceros phylogeny are reported inTable S5. Three- and six-partition analyses produced similarresults; we report here on the six-partition analyses. Bayesian

and ML analyses of the Nothoceros data matrix resulted in a well-supported phylogeny, with major ingroup nodes receiving poster-ior probability values >0.94 and bootstrap values >79% (Fig. 1). Thetwo Chilean species N. fuegiensis and N. endiviifolius and the NewZealand species N. giganteus (referred to as the Austral clade inFig. 1) together form the sister clade to the more northern NewWorld species. Nothoceros renzagliensis (from the ColombianAndes) is sister to the remaining Neotropical species (Fig. 1).Megaceros minarum from Brazil and Uruguay and Megaceros schizo-phyllus from Costa Rica, the Dominican Republic and Panama areboth embedded in Nothoceros (in Fig. 1 they are marked with‘‘comb. nov.’’; see Section 4.3. Taxonomic implications). Nothocerosaenigmaticus plants from the USA and morphologically similarplants from high elevation tropical areas (>3000 m alt.) in Mexico,Costa Rica, Colombia, Venezuela, and Bolivia form a clade in whichtwo samples of N. aenigmaticus from Costa Rican páramos (Cerro dela Muerte and the top of Volcan Turrialba) come out as sister to theother collections, including some from high elevations in CostaRica (Fig. 1). Unexpectedly, the species Megaceros alatifrons,M. cristisporus, M. guatemalensis, M. martinicensis, M. mexicanusand M. solidus are all nested in Nothoceros vincentianus (Fig. 1, theyare marked with ‘‘syn. nov.’’, see Section 4.3).

Divergence times and HPD intervals obtained with the threecalibration approaches for the three nodes of interest (Fig. 2) aresummarized in Table 2. The stem age of Nothoceros (Node 1 inFig. 2) was dated to 81.2 Ma using the strict clock and Palmer’s(1991) plastid rate on 3515 aligned nucleotides; 83.5 Ma usingthe hornwort-specific rbcL rate (with a SD) on 3515 alignednucleotides; or 68.4 Ma with a relaxed clock applied to a72-taxon-2442-nucleotide fossil-calibrated phylogeny (Fig. S1).The Nothoceros crown group (node 2) was dated to 35.0 Ma withPalmer’s (1991) plastid rate, 43.8 Ma using the hornwort-specificrbcL rate, and 54.0 Ma in the fossil-calibrated phylogeny. Thesplit between N. endiviifolius from Chile and its sister speciesN. giganteus from New Zealand (Node 3) was dated to 5.3 Ma withthe strict clock and Palmer’s (1991) plastid rate, 6.3 Ma with thehornwort rbcL rate, and 20.7 Ma with the fossil-calibrated tree.

4. Discussion

4.1. Substitution rates of bryophytes compared to other land plants

Examples of substitution rates obtained in seed plants are listedin Table 3. Among bryophytes, our earlier study of hornworts(Villarreal and Renner, 2012) seems to be the only one reportinga substitution rate for rbcL, namely 5.0 � 10�4 (HPD 2–8 � 10�4)subst./site/my, which is identical to a plastid genome substitutionrate calculated by Palmer (1991) for whole plastid genomes. Thissubstitution rate falls entirely within the range of plastid ratessince published for seed plants (Table 3), contradicting the notionof generally slower substitution rates in bryophytes. This notioncomes from a relative rate test of three markers (rbcL, nad5 and18S) obtained for 10 mosses, which as expected for relative ratetesting was not calibrated (Stenøien, 2008) so that no substitutionsrates could be calculated. Obviously, rate heterogeneity among lin-eages is real (Lewis et al., 1997; Yoder and Yang, 2000; Smith andDonoghue, 2008; Rothfels and Schuettpelz, 2014), and where suchheterogeneity is punctuated, it can sometimes be accommodatedby local clocks, i.e., separate strict clocks in different parts of thetree (Yoder and Yang, 2000; Rothfels and Schuettpelz, 2014;Bellot and Renner, in preparation). Rate heterogeneity is expectedto increase with the size of datasets, and the desire to includenodes suitable for fossil calibration often requires the inclusionof distant outgroups, potentially introducing further rate variation.In such cases, it may be better to rely on published plastid substi-tution rates (if the focal dataset is also based on plastid markers).

Table 1Dating methods, calibration approaches, and inferred divergence times for liverworts, hornworts, and mosses. The programs mentioned in the table are PAUP* (Swofford, 2002), r8s (Sanderson, 2003), and BEAST (Drummond et al., 2012).For ease of comparison, all rates are expressed as subst./site/my = substitutions/nucleotide site/million years. Cross validation refers to obtaining ages with at least two calibration approaches (see text).

Taxon Number of tips Aligned nucleotides Program used Clock model Calibration approach Cross validation Reference

LiverwortsBryopteris 51 1078 nucleotides;

ITS region (nuclear)Phylogeny usingmaximum likelihoodand calculatingdistances using aTamura-Nei model(PAUP⁄)

Strict clock Method I:– Break-up of Gondwana as calibration for the separa-

tion of B. filicina and B. gaudichaudii. Not a calibra-tion, use for comparison

Yes. Conflicting results. Authors preferredthe rate calibration

Hartmann et al.(2006)

Method II:– 1A prior rate for ITS of 1.35 � 10�3/subst./site/my

Plagiochila,Proskauera

118 1256 nucleotides;rps4 (plastid); ITSregion (nuclear)

Non-parametric ratesmoothing (r8s)

Relaxed clock – A fossil as a minimum constraint No. The same fossil was used on differentnodes (in different analyses) to discussalternative diversification scenarios. Theauthors discussed the uncertainty of theplacement of the fossil

Heinrichs et al.(2006)

Dataset I:LiverwortsDataset II:Leafy liverworts

Dataset I: 56Dataset II: 86

Dataset I: 1854nucleotides; rbcL/rps4 (plastid)Dataset II: 3042nucleotides rbcL/rps4 (plastid)

Penalized likelihood(r8s)

Relaxed clock Dataset I:– Land plants (maximum age constraint 475 Ma)– Split of vascular plants (fixed age constraint 430 Ma)– Seven fossil calibrations (as minimum estimates)

No Heinrichs et al.(2007)

Dataset II:– Split of Metzgeriidae and Jungermanniidae based on

dataset I (300–316 Ma)– Ten fossils (as minimum constraints)– To obtain confidence intervals for fossil calibrations,

their minima and maxima were usedLejeunaceae 135 3773 nucleotides;

rbcL/psbA/trnL-Fregion (plastid); ITS(nuclear)

Penalized likelihood(r8s)

Relaxed clock – Root based on Heinrichs et al. (2007) (Node 19,130.1 Ma)

– Thirteen fossils (as minimum constraints)– To obtain confidence intervals for fossil calibrations,

the minimum age of 120.3 Ma and maximum of139.8 Ma was used for each analysis

No Wilson et al.(2007)

Leptoscyphus 36 1672 nucleotides;atpB/rbcL/ trnL-Fregion/rps4 (plastid)

Uncorrelated log-normal clock (BEAST)

Relaxed clock – Geological calibration on the stem node of L. azoricusand L. porphyrius (neo-endemism in L. azoricus) usingthe oldest age of Sta. Maria Island (Azores) with anormal distribution (5.1, SD 1.5 Ma, truncated witha lower and upper bound of 5.1 Ma and 0respectively)

– Same calibration on the crown node of L. azoricus

No Devos andVanderpoorten(2009)

Marchesinia 31 938 nucleotides; ITSregion (nuclear)

Uncorrelated log-normal clock (BEAST)

Relaxed clockand strictclock

– Deepest split between M. brachiata and M. robustabased on Wilson et al. (2007), uniform priors(33.9–38.1 Ma).

– One fossil (as minimum constraint)

No Heinrichs et al.(2009b)

Dataset I: LiverwortsDataset II:Lepidoziaceae

Dataset I: 157Dataset II: 212

Dataset II: 3602nucleotides; rbcL/rps4 /psbA) (plastid)

Uncorrelated log-normal clock (BEAST)

Relaxed clock Dataset I:– Root (0–475 Ma uniform, normal or exponential dis-

tribution prior)– Nine fossils (as minimum constraints and an uni-

form prior to the fossil age to 475 Ma)– The frequency distribution of age estimates where

compared against the prior to identify the activeconstraints on the analyses

No Cooper et al.(2012)

Dataset II:– Root with a normal distribution of 475 (SD12.5) Ma– Nine fossils (as minimum constraints and an uni-

form prior to the fossil age to 475 Ma)

28J.C.V

illarreal,S.S.Renner/M

olecularPhylogenetics

andEvolution

78(2014)

25–35

Table 1 (continued)

Taxon Number of tips Aligned nucleotides Program used Clock model Calibration approach Cross validation Reference

Cephaloziineae Dataset: 94 2084 nucleotides;rbcL/psbA (plastid)

Uncorrelated log-normal clock (BEAST)

Relaxed clock – Root based on Heinrichs et al. (2007) using uniform(158–500 Ma) and normal (171.1, SD 8.3 Ma) priors

– Three fossils (uniform priors)

Yes. Different root calibrations with andwithout fossils

Feldberg et al.(2013)

MossesPyrrhobryum 39 1902 nucleotides.

atpB-rbcL spacer/rps4/trnL –F region(plastid)

Penalized likelihood(r8s)

Relaxed clock – Atacama desert aridification (14 Ma) No McDaniel andShaw (2003)

Mitthyridium 80/26 1007 nucleotides;glyceraldehyde 3-phosphatedehydrogenase(gapd) (nuclear)

Non-parametric ratesmoothing (r8s)

Strict clock – Maximum island age: Mo’orea (1.8 Ma), and Samoa(2.9 Ma)

Yes, using individual islands Wall (2005)

Bryophyta and Landplants

151 2006 nucleotides;rbcL/rps4 (plastid)

Penalized likelihood(r8s)

Relaxed clock – Land plants (maximum age fixed at 450 Ma)– Split between Magnolia and two monocot taxa (min-

imum age at 123 Ma)– Seed plant crown (minimum age at 310 Ma)– Monilophyte crown (minimum age at 354 Ma)– Euphyllophyte crown group (minimum age at

380 Ma)– Split between euphyllophytes and lycopsids (mini-

mum age at 408 Ma)– Crown group lycopsids (minimum age at 390 Ma)– Split between Sphaerocarpales and the complex

thalloid liverworts (minimum age at 203 Ma)

No Newton et al.(2007)

Homalothecium 2203 nucleotides;rpl16/atpB-rbcL(plastid), ITS(nuclear)

Uncorrelated log-normal clock (BEAST)

Relaxed clock Method I:Opening of the North Atlantic Ocean as a calibrationpoint with a normal distribution (60 SD 3 Ma), yieldedan ITS rate of 1.15 � 10�4/subst./site/ my. The rates forrpl16 and atpB-rbcL were 5.61 � 10–5 and 5.51 � 10–5

subst./site/myMethod II:

– Plastid rate: 5.0 � 10�4 (SD 1.0 � 10�4) /subst./site/my, with a normal distribution2

– Nuclear rate: 1.4 � 10�2 (SD 0.5 � 10�2 /subst./site/my) with a normal distribution3

YesIncongruent. Comparing substitution rates

Huttunen et al.(2008)

Helicodontioideae 46 1940 nucleotides;trnL-trnF/atpB-rbcL/psbT-psbH/psbA-trnH (plastid)

Uncorrelated log-normal clock (BEAST)

Relaxed clock Method I:– Two fossils on internal nodes

Yes. Congruent results using either fossils orsubstitution rates

Aigoin et al.(2009)

Method II:– Plastid rate: 5.0 � 10�4/subst./site/my (SD

1.0 � 10�4/subst./site/my) with a normaldistribution3

Sphagnum 60 8503 nucleotides;psbA, rbcL, rps4,trnG, trnL-F(plastid), nad7(mitochondrion),18S, 26S (nuclear)

Uncorrelated log-normal clock (BEAST)

Relaxed clock Method I:– Bryophyte ancestor with normal distribution (405

SD 43 Ma)

Yes. Highly congruent, using either a prioron root node (bryophyte ancestor) or thepublished substitution rate

Shaw et al.(2010)

Method II:– Plastid rate: 5.6 � 10�4, 5.0 � 10�4 subst./site/my

with a normal distribution– Mitochondrial rate: 1.09 � 10�4 subst./site/my with

a normal distribution– Nuclear rate: 1.09 � 10�4 subst./site/my with a nor-

mal distribution– Overall rate: 3.0 � 10�4 (SD 1.0 � 10�4)4, subst./site/

my with a normal distribution

(continued on next page)

J.C.Villarreal,S.S.R

enner/Molecular

Phylogeneticsand

Evolution78

(2014)25–

3529

Table 1 (continued)

Taxon Number of tips Aligned nucleotides Program used Clock model Calibration approach Cross validation Reference

Calyptrochaeta Dataset I (C.apiculata & C.brownii): 16Dataset II (C.asplenioides &C. cristata): 38Dataset III(Hookeriales):123

Dataset I (C.apiculata & C.brownii): 1751nucleotides;trnL-F & trnG(plastid), ITS(nuclear)Dataset II (C.asplenioides & C.cristata): 1872nucleotides;trnL-F & trnG(plastid), ITS(nuclear)Dataset III(Hookeriales): 4176nucleotides;trnL-F (plastid), ITS(nuclear)

BEASTTrial runs:

– Clock models:relaxed vs. strictclock

– Tree models: Allmodels availablein v1.6.2 werecompared usingBayes factors

Final analysis:– Clock model:

Uncorrelated log-normal clock.

– Tree models: coa-lescent Bayesianskyline for data-sets I & II andspeciation Yuleprocess for data-set III

Relaxed clockvs. Strictclock

Trial runs:– Method I, geographical time constraint: Maximum

island age of Mascarene Islands, Mauritius and LaRéunion. (lognormal distribution 8 SD 1 Ma)

– Method II, substitution rates from the literature:– Plastid rate: 5.0 � 10�4 (SD 1.0 � 10�43subst./site/

my) with a lognormal distribution.– Nuclear rate: 1.4 � 10�2 (SD 0.5 � 10�2 /subst./site/

my) with a lognormal distribution4

Final analysis:– Method III: Datasets I & II are nested within dataset

III (by linking substitution and clock models) and thelatter is calibrated using time constraints fromNewton et al. (2007). The constrained nodes werethe origin of Hypnodendrales (69.79 ,[ 58.98–100.59] Ma;) origin of the Ptychomniales (82.03[72.21–100.44] Ma), and origin of the Hyp-nales + Hookeriales clade (183.61 [ 130.75–1 65.1]Ma) using a gamma distribution

Yes. Comparing calibration methods andclock and tree models using Bayes Factors.The model using the mean substitutionrates from the literature, an uncorrelatedlognormal relaxed clock, and a coalescentBayesian skyline tree model, performedbetter for dataset II

Pokorny et al.(2011)

Orthotrichum 95 2281 nucleotides;atpB/rbcL/ trnL-Fregion/rps4, trnG(plastid)

Uncorrelated log-normal clock (BEAST)

Relaxed clock Plastid rate:5.0 � 10�4 (SD 1.0 � 10�4 /subst./site/my) with a normaldistribution3.

No Patiño et al.(2013)

Hornworts72 2442 nucleotides;

rbcL (plastid), nad5(mt)

Uncorrelated log-normal clock (BEAST)

Strict clock,Relaxed clock

Method I:– Three fossil calibrations. Alternative analyses with

gamma or exponential prior

Yes. Congruent results using either fossils orsubstitution rates

Villarreal andRenner (2012)

Method II:– Average rate between plastid and mitochondrial

rate: 3.0 � 10�4 (SD 1.0 � 10�4) subst./site/ my, witha normal distribution5

1 The rate used by Hartmann et al. (2006) and attributed to Les et al. (2003 who in turn attribute it to three earlier studies) is 1.35 � 10�3/subst./site/my. This falls within the average angiosperm ITS rates reported by Kay et al.(2006) of 0.38 � 10�9 subst./site/year to 8.34 � 10�9 subst./site/year (=0.38 � 10�3 subst./site/my to 8.34 � 10�3 subst./site/my). The rate of 1.4 � 10�2 subst./site/my used by Huttunen et al. (2008) came from the green algaCladophora and seems unusually high.

2 The standard deviation of 1.0 � 10�4 subst./site/my was not calculated in a reproducible way.3 Huttunen et al. (2008) used a mean prior of 1.4 � 10�2 corresponding to the average between 0.8% and 2.0% Myr�1 from Bakker et al. (1995). A normal distribution was centered using the mean value of 1.4 � 10�2 and the

standard deviation includes minimum and maximum values estimated by Bakker et al. (1995).4 The standard deviation of 1.0 � 10�4 subst./site/my is reported in Table 2 (Shaw et al., 2010) but was not calculated from the four substitution rates listed in the study.5 The same prior substitution rate was used by Shaw et al. (2010).

30J.C.V

illarreal,S.S.Renner/M

olecularPhylogenetics

andEvolution

78(2014)

25–35

Table 2Nothoceros divergence dates (in Ma) from the three calibration schemes (seeSection 2); node numbers refer to Fig. 2. (A) Strict clock with a plastid genomesubstitution rate of 5.0 � 10�4 subst./site/my (Palmer, 1991) applied to 16 taxa and3515 aligned nucleotides; (B) Relaxed clock with a hornwort-specific rbcL substitu-tion rate of 5.0 � 10�4 and a SD of a 2–8 � 10�4 subst./site/my applied to 16 taxa and3515 aligned nucleotides; (C) UCLN relaxed clock with fossil calibrations applied to 72taxa and a 2442 aligned nucleotides (Fig. S1). In bold are mean values with 95%highest posterior density intervals (HPD) shown in brackets.

Calibrationscheme

Node 1 Stem ageof Nothoceros

Node 2 Crown ageof Nothoceros

Node 3 Split of New Zealand/Chile species pair

A 81.2 [73.3–88.8] 35.0 [30.0–39.8] 5.3 [2.7–7.9]B 83.5 [43.7–133.9] 43.8 [22.1–70.8] 6.3 [1.2–13.4]C 68.4 [42.4–101.7] 54.0 [30.1–78.9] 20.7 [4.4–39.9]

J.C. Villarreal, S.S. Renner / Molecular Phylogenetics and Evolution 78 (2014) 25–35 31

For the nuclear ITS region, several reviews have tabulated ratesfrom numerous studies of 21 different angiosperm families (.e.g.,Kay et al., 2006). The rates range from 4.13 � 10�3

(1.72–8.34 � 10�3 subst./site/my) in herbaceous angiosperms toapproximately half that in woody angiosperms from 2.15 � 10�3

(0.38 � 10�3–7.83 � 10�3 subst./site/my; our Table 3). These pub-lished nuclear rates, perhaps in a relaxed clock approach with con-fidence intervals (as done here: Table 2), can provide an additionaltest for ages inferred with plastid rates or geological or fossil cali-brations, each with their own caveats. Ideally, one would alwaysuse two or more calibration approaches and then compare theobtained ages to achieve cross validation (as done by Shaw et al.,2010; our Table 2).

In our case, Nothoceros ages obtained with a strict clock andPalmer’s (1991) plastome substitution rate and those from a relaxedclock with hornwort-specific rbcL rate (from a fossil-calibratedrelaxed clock, Fig. S1) agreed because the two rates happen to be

Fig. 1. Maximum likelihood phylogeny of Nothoceros based on seven loci of nuclear, plashown above branches. Species names are colored-coded. Accessions of Nothoceros aetransferred or synonymized here are marked accordingly. (For interpretation of the referarticle.)

identical. Yet Nothoceros ages obtained from a relaxed clock andfour fossil calibrations were much older (Table 2; Fig. S1, whichshows the four calibrations). One of the fossil calibrations is a fossilPhaeoceros from the Uscari Formation, Costa Rica (Lower Miocene15–23 Mya; Graham, 1987), which may force the Nothoceros agesto become older than they are in the rate-calibrated trees. Possiblythis fossil does not represent the node to which it was assigned.

The available plastid rates for bryophytes so far do not point to aconsistent difference between plastid substitution rates in horn-worts, liverworts, mosses, and vascular plants (Tables 1 and 3).Bryophyte molecular clock studies might therefore always carryout at least one relaxed or strict clock run with previously pub-lished plastid or ITS rates (obviously depending on the focalmatrix) and then use the results to cross-validate other calibrationapproaches. There is no theoretical reason not to use substitutionrates for related organisms when the goal is to translate geneticdistances into absolute times, and there are well documentedexamples of clock-like substitution accumulation (e.g., Korberet al., 2000). For small data matrices in terms of taxa or nucleo-tides, the strict clock model calibrated with a published substitu-tion rate is the statistically more appropriate and stronger modelbecause they do not contain sufficient information to calculatethe numerous parameters of a complex model. Alternatively, onemight decide not to use such matrices for dating.

4.2. Diversification and geographic expansion of Nothoceros

Nothoceros diverged from its sister clade, the genera Megaceros(in its new monophyletic circumscription; below) and Dendroceros,sometime in the Upper Cretaceous, and diversification within

stid, and mitochondrial DNA; ML bootstrap and Bayesian posterior probabilities arenigmaticus from high elevation (>3000 m) are labeled (P). Names that are newlyences to color in this figure legend, the reader is referred to the web version of this

Fig. 2. Time tree for the genus Nothoceros based on several plastid DNA loci and an average plastid substitution rate (see methods). Red color-coded species are found in theAmericas, yellow in Austral America and blue ones in Australasia and Africa. Bars at nodes indicate highest posterior density (HPD) intervals around node ages. The maprepresents the geographical distribution of the Nothoceros species sampled in the study. Red dots are Neotropical and US collections (N. aenigmaticus, N. canaliculatus, N.minarum, N. renzagliensis, N. schizophyllus, N. superbus, N. vincentianus) and yellow ones those from Austral America (N. endiviifolius and N. fuegiensis). A blue dot represents theNew Zealand N. giganteus. Nodes of biogeographic interest listed in Table 2 and discussed in the text are numbered. (For interpretation of the references to color in this figurelegend, the reader is referred to the web version of this article.)

32 J.C. Villarreal, S.S. Renner / Molecular Phylogenetics and Evolution 78 (2014) 25–35

Nothoceros began during the Eocene (36–54 Ma). The divergence ofthe New Zealand species N. giganteus from its Chilean sister speciesN. endiviifolius is here dated to 5.3 Ma, 6.3 Ma, or 20.7 Ma, depend-ing on the calibration and clock model used (Table 2, Fig. S1). NewZealand and New Caledonia broke away from West Antarctica�82 Ma and moved northwest, opening the Tasman Sea. SouthernSouth America and Antarctica remained in contact through theAntarctic Peninsula until the Oligocene (Wilson et al., 2012), whenthe Drake Passage opened between these continents. The inferreddivergence times (Table 2) thus imply that N. giganteus is the resultof a transoceanic dispersal event to New Zealand. In terms of sex-ual condition, N. endiviifolius is dioicous, N. giganteus monoicous,which would in principle have allowed intra-gametophytic selfingfollowing establishment in New Zealand. Sister species or clades inwhich one member of a pair occurs in New Zealand, the other insouthernmost South America have been reported also from the liv-erwort genus, Monoclea, with M. gottschei occurring in SouthernSouth America and its sister species M. forsterii in New Zealand(Meißner et al., 1998; without molecular clock dating). In flower-ing plants, there are at least 28 families with closely related speciesin Australasia and Southern South America (Chacón et al., 2012).

We did not conduct a statistical biogeographic analysis forNothoceros, but based on the distribution of its closest relatives,Dendroceros and Megaceros, Nothoceros is likely to have first estab-lished in Patagonia and Tierra del Fuego and then to have expandedits range northwards all the way to eastern North America andnortheast to Rio de Janeiro and Uruguay. No Nothoceros speciesgrow in lowland forests below 500 m altitude or dry habitats,and these mesophytic plants are not able to withstand desiccation.

Such northward expansion from Patagonia to Central America hasnot been reported among other hornworts, mosses, or liverwortsbut in angiosperms, a similar South to North range evolution hasoccurred in five families (Chacón et al., 2012, Alstroemeriaceae,Calceolariaceae, Cunoniaceae, Escalloniaceae and Proteaceae).

A concern in this study is that both Dendroceros and Megaceros(excluding the Neotropical species that are embedded in Nothocer-os; Fig. 1 and Section 4.3.) are insufficiently known. A monophy-letic Megaceros has seven species (based on ongoing taxonomicwork) that have highly disjunct ranges, at least as they are cur-rently circumscribed (Cargill et al., 2013). One occurs in WestAfrica (São Tomé and Tanzania, M. flagellaris); this same speciesand one other one (M. tjibodensis) occur in India, China, Japanand Pacific Islands; Australasia has five endemic species (M. austro-nesophilus, M. denticulatus, M. gracilis, M. leptohymenius and M. pel-lucidus) plus the widespread M. flagellaris (Hasegawa, 1983;Villarreal et al., 2010b; Cargill et al., 2013). Dendroceros has 43 spe-cies, with three species in Africa, three in China, 13 species in theNeotropics, and 16 on the Pacific Islands; two Neotropical speciesof Dendroceros seem to be widespread in Australasia and poten-tially on Atlantic islands. Clearly, more molecular phylogeneticwork is needed before the precise relationships between theseentities are understood. Based on the current sampling, the com-mon ancestor of Nothoceros, Dendroceros and Australian Megaceroslived in the Upper Cretaceous, at a time when East Gondwana hadlong broken up, while the two fragments of West Gondwana, Africaand South America, were still relatively close to each other.

The Central American species N. canaliculatus and the CentralAmerican and Cuban Megaceros schizophyllus, as well as N. superbus,

Table 3Typical substitution rates from land plants used in phylogenetic studies. For ease of comparison, we have expressed all rates as subst./site/my, subst./site/ my = substitutions/nucleotide site/million years.

Studies in chronological sequence Genomic region Plant clade Substitutionrate

Plastid ratesWolfe et al. (1987) atpA, atpB, atpE, atpH, psbH, orf62 Maize vs. wheat 0.2–0.3 � 10�3

subst./site/myWolfe et al. (1987) atpA, atpB, atpE, atpF, rbcL, psbA,

psbB, psbC, psbD, psbG, petA, petB,rps4, rpL16, eight genes1

Monocot (barley, maize, rice, Spirodelaoligorhiza, wheat) vs. dicot (alfalfa, pea,Oenothera sp., petunia, spinach, tobacco, Viciafaba)

2.1–2.9 � 10�3

subst./site/my

Wolfe et al. (1987) Whole plastid genomes Marchantia and tobacco 1.4–1.6 � 10�3

subst./site/myPalmer (1991) Whole plastid genomes Plastid genomes available in 1991 5.0 � 10�4

subst./site/myFrascaria et al. (1993) rbcL Fagaceae (red oak and chestnut) 1.34–

7.99 � 10�4

subst./site/myAlbert et al. (1994) rbcL 19 species of woody plants 1.03 � 10�4

subst./site/mySchnabel and Wendel (1998) rpl16 Gleditsia, Fabaceae 6.0 � 10�4

subst./site/myRichardson et al. (2001) trnL-F region Inga,Fabaceae 1.30 � 10�3

subst./site/myRichardson et al. (2001) trnL-F region Phylica, Rhamnaceae 4.87 � 10�4

subst./site/mySanderson (2002) psaA, psaB, rbcL psaA, psaB (19 land plants)

rbcL (37 land plants including Marchantia andthe algae Chara)

�5.6 � 10�4

subst./site/my

Lavin et al. (2005) rbcL Fabaceae 1.6–8.6 � 10�4

subst./site/myVillarreal and Renner (2012) rbcL Hornworts 5.0 � 10�4

(HPD, 2–8 � 10�4)subst./site/my

Lockwood et al. (2013) rbcL, matK, trnL-trnF, trnH-psbA,trnS-trnG

Pinaceae 1.0–1.9 � 10�4

subst./site/my

Nuclear ratesWolfe et al. (1987) Plastocyanin Spinach, Silene 15.8 to

31.5 � 10�3/subst./site/my

Wolfe et al. (1987) gapC, adH1 Monocot (barley, maize) vs dicot (Arabidopsisthaliana mustard, pea, tobacco)

5.8–8.1 � 10�3/subst./site/my

Les et al. (2003: 919) ‘‘We used the rate of 0.27% per millionyears because it falls within the narrower range that isestimated frequently for the ITS region. . .’’

Internal transcribed spacer (nrITS) Flowering plants 1.35 � 10�3/subst./site/my

Sanderson (2002) 26 nrRNA 37 land plants (including Marchantia and thealgae Chara)

1.09 � 10�4

subst./site/my

1 The genes or gene tracts are listed in Wolfe et al. (1987).

J.C. Villarreal, S.S. Renner / Molecular Phylogenetics and Evolution 78 (2014) 25–35 33

and N. renzagliensis from the Andean foothills (Fig. 1 ‘‘Neotropical/US clade’’) are restricted to montane forests below 3500 m alt(below the tree line). Only N. aenigmaticus occurs at high elevationsin alpine Bolivian punas and páramos, and this species may haveoriginated during the most recent uplift phase of the Andes(�5 Ma). The genetically structured grade of high-altitude acces-sions of N. aenigmaticus seen in Fig. 1 (labeled with a ‘‘P’’) suggestslocal differentiation and little gene flow between populations.From one such high altitude population, N. aenigmaticus may havereached alpine Mexico and the Appalachian Mts. in eastern NorthAmerica. Daily temperatures in páramos can range from �5 �C to25–30 �C (Herrera, 2005), suggesting that N. aenigmaticus waspre-adapted to seasonally freezing conditions in North Americanwinters.

4.3. Taxonomic implications

The 6-locus plastid phylogeny presented here (Fig. 1; Table S5)indicates that Nothoceros contains 10 genetically and morphologi-cally distinct entities that can be ranked as species, Megaceros min-arum, M. schizophyllus, Nothoceros aenigmaticus, N. canaliculatus, N.

endiviifolius, N. fuegiensis, N. giganteus, N. renzagliensis, N. superbus,and N. vincentianus. Table S6 lists their morphological differences.Additionally, there are six names that based on our phylogeny(Fig. 1) and study of the relevant types should be synonymizedunder N. vincentianus, namely Megaceros alatifrons, M. guatemalen-sis, M. cristisporus, M. martinicensis, M. mexicanus, and M. solidus.Their type specimens are identical to N. vincentianus in the onlytaxonomically useful characters in dried state: sexual condition,gametophyte shape, and spore ornamentation. With these newsynonymizations, most of the 17 names described in Megacerosfor tropical America (Stephani 1912–1917, 1923) are now allo-cated: Three were transferred to Nothoceros in a previous study,viz. Nothoceros aenigmaticus, N. fuegiensis, and N. vincentianus(Villarreal et al. 2010a); two are transferred into Nothoceros below;two are synonymized under N. minarum (also below); and sixremain unaccounted for (M. aneuraeformis, M. boliviensis, M. colum-bianus, M. flavens, M. jamaicensis, M. jamesonii).

In its new circumscription, Megaceros, the type species of whichis from Java (M. tjibodensis Campbell), has six species in Africa, Aus-tralasia and Asia (Hasegawa, 1983; Villarreal et al., 2010b; personalinformation from D.C. Cargill, Australian National Herbarium,

34 J.C. Villarreal, S.S. Renner / Molecular Phylogenetics and Evolution 78 (2014) 25–35

Canberra, 2013). However, only Megaceros flagellaris, M. spec.Cargill and Fuhrer 535 (CANB) and M. tjibodensis have beensequenced (Villarreal and Renner, 2013; this study).

5. New combinations

5.1. Nothoceros minarum (Nees) J.C. Villarreal, comb. nov

�Anthoceros minarum Nees, Naturgesch. Europ. Lebermoose 4,340 footnote, 1838, diagnosis in Nees in Martius, Flora BrasiliensisI, 304–305. 1833 = Megaceros minarum (Nees) Steph., Spec. Hep. 5,949, 1916 = Dendroceros minarum (Steph.) Hell, Univ. Sao PauloFac. Cienc. y Let. 25, 43, 1967. Type, Brazil, Minas Gerais, São Joãoda Baptista, C.F.P. Martius ex herbar J.F.C. Montagne (isotype M!).

=Megaceros wiemannii Steph., Spec. Hep. 5, 948, 1916. Type,Brazil, São Paulo, Alto da Serra, 900 m, V. Schiffner 991, 28 May1901 (holotype G!) syn. nov.

=Megaceros amoenus Steph., Spec. Hep. 5, 949, 1916. Type,Brazil, São Paulo, V. Schiffner 1652, without date (holotype G!),syn. nov.

Note, Megaceros minarum is based on a collection made by C.F.P.Martius in Minas Gerais, Brazil (Nees, 1833), and the sample fromRio de Janeiro that we sequenced (Appendix 1) represents Nees’sspecies.

5.2. Nothoceros schizophyllus (Gottsche ex Steph.) J.C. Villarreal, comb.nov

�Megaceros schizophyllus Gottsche ex Steph., Spec. Hep. 5, 949,1916. Type, Cuba, Mt. Verde, C. Wright 276, without date (HolotypeG!).

Of Megaceros schizophyllus (Fig. 1) we sequenced three acces-sions, one from Costa Rica, one from Panama and one from theDominican Republic. They agree in all respects with the typecollection from Mt. Verde, Cuba, and we therefore think that theyrepresent Stephani’s species.

6. New synonyms of Nothoceros vincentianus

Fresh samples of Nothoceros vincentianus may or may not con-tain a conspicuous pyrenoid. This trait is reflected in the clusteringof accessions with or without a pyrenoid in the phylogenetic tree(Fig. 1). This is remarkable because the phylogeny is based not juston plastid DNA, but instead on combined nuclear, plastid, andmitochondrial DNA loci. In dried condition, however, the pyrenoidcharacter cannot be used to distinguish species, and it also does notcorrelate with geographic provenience of samples (Fig. 1) or othermorphological characters. The six sequenced Caribbean and Cen-tral-South American accessions of Megaceros alatifrons, M. guate-malensis, M. cristisporus, M. martinicensis, M. mexicanus and M.solidus (Fig. 1) all cluster in the ‘‘pyrenoid present’’ clade of N. vin-centianus and since they all morphologically resemble N. vincenti-anus, we assume they have a pyrenoid and synonymize thesenames under N. vincentianus.

6.1. Nothoceros vincentianus (Lehm. & Lindenb.) J.C.Villarreal

=Anthoceros cristisporus Steph., Bull. Soc. R. Bot. Belgique 31,175. [1892] 1893.�Megaceros cristisporus (Steph.) Steph., Spec. Hep. 5, 948, 1916.

Type, Costa Rica, Forets de la Palma, 1550 m, H. Pittier in Pittier andDurand, Plantae Costaricense Exsic. 6, 18 Dec. 1888 (Holotype G!),syn. nov.

=Anthoceros alatifrons Steph. in Urban, Symb. Antill. 2, 469,1901. � Megaceros alatifrons (Steph.) Steph., Spec. Hep. 5, 947,

1916. Type, Guadalupe, A. Duss 308, without date (holotype G!, iso-type NY!), syn. nov.

=Megaceros mexicanus Steph., Spec. Hep. 5, 947, 1916. Type,Mexico, Colipa, F. Liebman 142, without date (Holotype G!, syn.nov.).

=Megaceros guatemalensis Steph., Spec. Hep. 5, 948, 1916. Type,Guatemala. Coban, Alta Verapaz, 1650 m, H. von Türckheim, Jan.1908, herb. Levier 6078 (Holotype G!, isotypes NY!, syn. nov.).

=Megaceros martinicensis Steph., Spec. Hep. 5, 949, 1916. Type,Martinique, A. Duss 574, 0.9.1901, (Holotype G!), syn. nov.

=Megaceros solidus Steph., Spec. Hep. 5, 950, 1916. Type, Guada-loupe, A. Duss 520, without date (Holotype G!), syn. nov.

Acknowledgments

We thank B. Goffinet, N. dos Santos, D.C. Cargill, L. Lavocat, Y.Queralta, A. Morales and A. Cruz for providing hornwort material;Michelle Price (G) for assistance in locating a type; S. McDaniel forthe data matrix used in his 2003 study; L. Montero, J. Patiño, A.Vanderpoorten, R. Medina, B. Shaw, J. Heinrichs, K. Feldberg andL.L. Forrest for information presented in Table 1; the late G. Hässelde Menéndez and M. Rubies for images and notes on type collec-tions; and the German Research Foundation for funding (DFG RE-603/14-1).

Appendix A. Supplementary material

Supplementary data associated with this article can be found, inthe online version, at http://dx.doi.org/10.1016/j.ympev.2014.04.014.

References

Aigoin, D.A., Devos, N., Huttunen, S., Ignatov, M.S., Gonzáles-Mancebo, J.M.,Vanderpoorten, A., 2009. What if Engler was not completely wrong? Evidencefor multiple evolutionary origins in the moss flora of Macaronesia. Evolution 63,3248–3257.

Albert, V.A., Backlund, A., Bremer, K., Chase, M.W., Manhart, J.R., Mishler, B.D.,Nixon, K.C., 1994. Functional constraints and rbcL evidence for land plantphylogeny. Ann. Miss. Bot. Gard. 81, 534–567.

Archangelsky, S., Villar de Seone, L., 1996. Estudios palinógicos de la formaciónBaqueró (Cretácico), provincia de Santa Cruz, Argentina. Ameghiniana 35, 7–19.

Bakker, F.T., Olsen, J.L., Stam, W.T., 1995. Evolution of nuclear rDNA ITS sequences inthe Cladophora albida/sericea clade (Chlorophyta). J. Mol. Evol. 40, 640–651.

Bellot, S., Renner, S.S., in preparation. Exploring new calibration approaches forparasites: The worldwide Apodanthaceae (Cucurbitales) as an example. Mol.Phylogen. Evol.

Brown, J.M., Hedtke, S.M., Lemmon, A.R., Lemmon, M.E., 2010. When trees grow toolong, investigating the causes of highly inaccurate Bayesian branch lengthestimates. Syst. Biol. 59, 145–161.

Bytebier, B., Antonelli, A., Bellstedt, D.U., Linder, H.P., 2011. Estimating the age offire in the Cape flora of South Africa from an orchid phylogeny. Proc. R. Soc. B278, 188–195.

Cargill, D.C., Vella, N.G.F., Sharma, I., Miller, J.T., 2013. Cryptic speciation and speciesdiversity among Australian and New Zealand hornwort taxa of Megaceros(Dendrocerotaceae). Aust. Syst. Bot. 26, 356–377.

Chacón, J., Camargo de Assis, M., Meerow, A.W., Renner, S.S., 2012. From EastGondwana to Central America: historical biogeography of the Alstroemeriaceae.J. Biogeogr. 39, 1806–1818.

Chitaley, S.D., Yawale, N.R., 1980. On Notothylites nirulai gen. et sp. nov. a petrifiedsporogonium from the Deccan-Intertrappean beds of Mohgaonkalan M. P.(India). Botanique 9, 111–118.

Cooper, E.E., Henwood, M.J., Brown, E.A., 2012. Are the liverworts really that old?Cretaceous origins and Cenozoic diversifications in Lepidoziaceae reflect arecurrent theme in liverwort evolution. Biol. J. Linn. Soc. 107, 425–441.

Crisp, M.D., Burrows, G.E., Cook, L.G., Thornhill, A.H., Bowman, D.M.J.S., 2011.Flammable biomes dominated by eucalypts originated at the Cretaceous-Palaeogene boundary. Nat. Commun. 2, 193, doi, DOI: 10.1038/ncomms1191.

Devos, N., Vanderpoorten, A., 2009. Range disjunctions, speciation, andmorphological transformation rates in the liverwort genus Leptoscyphus.Evolution 63, 779–792.

Drummond, A.J., Suchard, M.A., Xie, D., Rambaut, A., 2012. Bayesian phylogeneticswith BEAUti and the BEAST 1.7. Mol. Biol. Evol. 29, 1969–1973.

Duff, R.J., Villarreal, J.C., Cargill, D.C., Renzaglia, K.S., 2007. Progress and challengestoward developing a phylogeny and classification of the hornworts. Bryologist110, 214–243.

J.C. Villarreal, S.S. Renner / Molecular Phylogenetics and Evolution 78 (2014) 25–35 35

Feldberg, K., Heinrichs, J., Schmidt, A.R., Vána, J., Schneider, H., 2013. Exploring theimpact of fossil constraints on the divergence time estimates of derivedliverworts. Plant Syst. Evol. 299, 585–601.

Frascaria, N., Maggia, L., Michaud, M., Bousquet, J., 1993. The rbcL gene sequencefrom chestnut indicates a slow rate of evolution in the Fagaceae. Genome 36,668–671.

Graham, A., 1987. Miocene communities and paleoenvironments of Southern CostaRica. Am. J. Bot. 74, 1501–1518.

Hartmann, F.A., Wilson, R., Gradstein, S.R., Schneider, H., Heinrichs, J., 2006.Testing hypotheses on species delimitations and disjunctions in theliverwort Bryopteris (Jungermanniopsida, Lejeuneaceae). Int. J. Plant Sci. 167,1205–1214.

Hasegawa, J., 1983. Taxonomical studies on Asian Anthocerotae. III. Asian species ofMegaceros. J. Hatt. Bot. Lab. 54, 227–240.

Heinrichs, J., Lindner, M., Groth, H., Hentschel, J., Feldberg, K., Renker, C., Engel, J.J.,von Konrat, M., Long, D.G., Schneider, H., 2006. Goodbye or welcomeGondwana? Insights into the phylogenetic biogeography of the leafyliverwort Plagiochila with a description of Proskauera, gen. nov.(Plagiochilaceae, Jungermanniales). Plant Syst. Evol. 258, 227–250.

Heinrichs, J., Hentschel, J., Wilson, R., Feldberg, K., Schneider, H., 2007. Evolution ofleafy liverworts (Jungermanniidae, Marchantiophyta), estimating divergencetimes from chloroplast DNA sequences using penalized likelihood withintegrated fossil evidence. Taxon 56, 31–44.

Heinrichs, J., Hentschel, J., Feldberg, K., Bombosch, A., Schneider, H., 2009a.Phylogenetic biogeography and taxonomy of disjunctly distributedbryophytes. J. Syst. Evol. 47, 497–508.

Heinrichs, J., Klugmann, F., Hentschel, J., Schneider, H., 2009b. DNA taxonomy,cryptic speciation and diversification of the Neotropical-African liverwort,Marchesinia brachiata (Lejeuneaceae, Porellales). Mol. Phylogen. Evol. 53, 113–121.

Heinrichs, J., Vitt, D.H., Schäfer-Verwimp, Ragazzi, E., Marzaro, G., Grimaldi, D.A.,Nascimbene, P.C., Feldberg, K., Schmidt, A.R., 2013. The moss Macromitriumrichardii (Orthotrichaceae) with sporophyte and calyptra enclosed in Hymenaearesin from the Dominican Republic. Polish Bot. J. 58, 221–230.

Herrera, W., 2005. El clima de los Páramos de Costa Rica. In: Kapelle, M., Horn, S.P.(Eds.), Páramos de Costa Rica. Costa Rica, Editorial InBIo, pp. 113–128.

Huttunen, S., Hedenäs, L., Ignatov, M.S., Devos, N., Vanderpoorten, A., 2008. Originand evolution of the northern hemisphere disjunction in the moss genusHomalothecium (Brachytheciaceae). Am. J. Bot. 95, 720–730.

Kay, M.K., Whittall, J.B., Hodges, S.A., 2006. A survey of nuclear ribosomal internaltranscribed spacer substitution rates across angiosperms, an approximatemolecular clock with life history effects. BMC Evol. Biol. 6, 36.

Korber, B., Muldoon, M., Theiler, J., Gao, F., Gupta, R., Lapedes, A., Hahn, B.H.,Wolinsky, S., Bhattacharya, T., 2000. Timing the ancestor of the HIV-1 pandemicstrains. Science 288, 1789–1795.

Lavin, M., Herendeen, P.S., Wojciechowski, M.F., 2005. Evolutionary rates analysis ofLeguminosae implicates a rapid diversification of lineages during the Tertiary.Syst. Biol. 54, 575–594.

Les, D.H., Crawford, D.J., Kimball, R.T., Moody, M.L., Landolt, E., 2003. Biogeographyof discontinuously distributed hydrophytes, a molecular appraisal ofintercontinental disjunctions. Int. J. Plant Sci. 164, 917–932.

Lewis, L.A., Mishler, B.D., Vilgalys, R., 1997. Phylogenetic relationships of theliverworts (Hepaticae), a basal embryophyte lineage, inferred from nucleotidesequence data of the chloroplast gene rbcL. Mol. Phylogen. Evol. 7, 377–393.

Li, F.-W., Villarreal, J.C., et al., 2014. Horizontal gene transfer of a chimericphotoreceptor, neochrome, from bryophytes to ferns. Proc. Natl. Acad. Sci. USA111, 6672–6677. http://dx.doi.org/10.1073/pnas.1319929111.

Linder, H.P., Antonelli, A., Humphreys, A.M., Pirie, M.D., Wüest, R.O., 2013. Whatdetermines biogeographical ranges? Historical wanderings and ecologicalconstraints in the danthonioid grasses. J. Biogeogr. 40, 821–834. http://dx.doi.org/10.1111/jbi.12070.

Lockwood, J.D., Aleksic, J., Zou, J., Wang, J., Liu, J., Renner, S.S., 2013. A newphylogeny for the genus Picea from plastid, mitochondrial, and nuclearsequences. Mol. Phylogen. Evol. 96, 717–727.

McDaniel, S.F., Shaw, A.J., 2003. Phylogeographic structure and cryptic speciation inthe trans–Antartic moss Pyrrhobryum mnioides. Evolution 57, 205–215.

Meißner, K., Frahm, J.-P., Stech, M., Frey, W., 1998. Molecular divergence patternsand infrageneric relationship of Monoclea (Monocleales, Hepaticae). Studies inaustral temperate rain forest bryophytes 1. Nova Hedwigia 67, 289–302.

Nees Von Esenbeck, C.G.D., 1833. Hepaticae. In: Martius, C.F.P. (Ed.), FloraBrasiliensis, vol. I. Pars prior, Stuttgart, pp. 304–305.

Newton, A.E., Wikstrom, N., Bell, N., Forrest, L., Ignatov, M.S., Tangney, R.S., 2007.Dating the diversification of the pleurocarpous mosses. In: Newton, A.E.,Tangney, R.S. (Eds.), Pleurocarpous Mosses, Systematics and Evolution. CRCPress, Boca Raton, pp. 337–366.

Nylander, J.A.A., 2004. MrModeltest v2. Program Distributed by the Author.Evolutionary Biology Centre, Uppsala University, Program distributed by theauthor.

Palmer, J.D., 1991. Plastid chromosome, structure and evolution. In: Bogorad, L.,Vasil, I.K. (Eds.), The Molecular Biology of Plastids. Academic Press, San Diego,California, USA, pp. 5–53.

Patiño, J., Medina, R., Vanderpoorten, A., González-Mancebo, J.M., Werner, O., Devos,N., Mateo, R.G., Lara, F., Ros, R.M., 2013. Origin and fate of the single-islandendemic moss Orthotrichum handiense. J. Biogeogr. 40, 857–868.

Pokorny, L., Oliván, G., Shaw, A.J., 2011. Phylogeographic patterns in two southernhemisphere species of Calyptrochaeta (Daltoniaceae, Bryophyta). Syst. Bot. 36,542–553.

Rambaut, A., 2007. FigTree v.1.4.0. [Online]. Website last modified on December 5,2012 (accessed on 12.06.13). Available at <http://beast.bio.ed.ac.uk/FigTree>.

Rambaut, A., Suchard, M.A., Xie, D., Drummond, A.J., 2013. Tracer v1.5, Availablefrom <http://beast.bio.ed.ac.uk/Tracer>.

Renner, S.S., 2005. Relaxed molecular clocks for dating historical plant dispersalevents. Trends Plant Sci. 10, 550–558.

Richardson, J.E., Pennington, R.T., Pennington, T.D., Hollingsworth, P.M., 2001. Rapiddiversification of a species-rich genus of Neotropical rain forest trees. Science293, 2242–2245.

Ronquist, F., Teslenko, M., van der Mark, P., Ayres, D.L., Darling, A., Höhna, S., Larget,B., Liu, L., Suchard, M.A., Huelsenbeck, J.P., 2012. MrBayes 3.2, Efficient Bayesianphylogenetic inference and model choice across a large model space. Syst. Biol.61, 539–542.

Rothfels, C.J., Schuettpelz, E., 2014. Accelerated rate of molecular evolution forvittarioid ferns is strong and not driven by selection. Syst. Biol. 63, 31–54.

Sanderson, M.J., 2002. Estimating absolute rates of molecular evolution anddivergence times, a penalized likelihood approach. Mol. Biol. Evol. 19, 101–109.

Sanderson, M.J., 2003. R8s, inferring absolute rates of evolution and divergencetimes in the absence of a molecular clock. Bioinformatics 19, 301–302.

Schaefer, H., Heibl, C., Renner, S.S., 2009. Gourds afloat, a dated phylogeny revealsan Asian origin of the gourd family (Cucurbitaceae) and numerous overseadispersal events. Proc. R. Soc. B 276, 843–851.

Schnabel, A., Wendel, J., 1998. Cladistic biogeography of Gleditsia (Leguminosae)based on ndhF and rpl16 chloroplast gene sequences. Am. J. Bot. 85, 1753–1765.

Schneider, H., Schuettpelz, E., Pryer, K.M., Cranfill, R., Magallón, S., Lupia, R., 2004.Ferns diversified in the shadow of angiosperms. Nature 428, 553–557.

Schuettpelz, E., Pryer, K.M., 2009. Evidence for a Cenozoic radiation of ferns in anangiosperm-dominated canopy. Proc. Natl. Acad. Sci. USA 106, 11200–11205.

Shaw, A.J., Devos, N., Cox, C.J., Boles, S.B., Shaw, B., Buchanan, A.M., Cave, L., Seppelt,R., 2010. Peatmoss (Sphagnum) diversification associated with MioceneNorthern Hemisphere climatic cooling? Mol. Phylogen. Evol. 55, 1139–1145.

Smith, S.A., Donoghue, M.J., 2008. Rates of molecular evolution are linked to lifehistory in flowering plants. Science 322, 86–89.

Stamatakis, A., Hoover, P., Rougemont, J., 2008. A rapid bootstrap algorithm for theRAxML Web-Servers. Syst. Biol. 75, 758–771. http://dx.doi.org/10.1080/10635150802429642.

Stenøien, H.K., 2008. Slow molecular evolution in 18S rDNA, rbcL and nad5 genes ofmosses compared with higher plants. J. Evol. Biol. 21, 566–571.

Stephani, F., 1917. Species Hepaticarum, vol. 5. George and Cie, Geneva.Stephani, F., 1923. Species Hepaticarum, vol. 6. George and Cie, Geneva.Swofford, D.L., 2002. PAUP*. Phylogenetic Analysis Using Parsimony (*and Other

Methods). Sinauer Associates, Sunderland, Massachusetts.Villarreal, J.C., Renner, S.S., 2012. Hornwort pyrenoids, a carbon-concentrating

mechanism, evolved and were lost at least five times during the last 100 millionyears. Proc. Natl. Acad. Sci. USA 109, 18873–18878.

Villarreal, J.C., Renner, S.S., 2013. Correlates of monoicy and dioicy in hornworts, theapparent sister group to vascular plants. BMC Evol. Biol. 13, 239.

Villarreal, J.C., Goffinet, B., Duff, R.J., Cargill, D.C., 2010a. Phylogenetic delineation ofthe genera Nothoceros and Megaceros. Bryologist 113, 106–113.

Villarreal, J.C., Cargill, D.C., Hagborg, A., Söderström, L., Renzaglia, K.S., 2010b.Hornwort diversity, patterns, causes and future work. Phytotaxa 9, 150–166.

Villarreal, J.C., Campos, L.V., Uribe, J., Goffinet, B., 2012. Parallel evolution ofendospory in hornworts, Nothoceros renzagliensis, sp. nov. Syst. Bot. 37, 31–37.

Villarreal, J.C., Forrest, L.L., Wickett, N., Goffinet, B., 2013. The plastid genome of thehornwort Nothoceros aenigmaticus, Phylogenetic signal in inverted repeatexpansion, pseudogenization and intron gain. Am. J. Bot. 100, 467–477.

Wall, D.P., 2005. Origin and rapid diversification of a tropical moss. Evolution 59,1413–1424.

Wilson, R., Heinrichs, J., Hentschel, J., Gradstein, S.R., Schneider, H., 2007. Steadydiversification of derived liverworts under Tertiary climatic fluctuations. Biol.Lett. 3, 566–569.

Wilson, D.S., Jamieson, S.S.R., Barret, P.J., Leitchenkov, G., Gohl, K., Larter, R.D., 2012.Antarctic topography at the Eocene-Oligocene boundary. Palaeogeog. Palaeocl.Palaeoecol. 335 (336), 24–34.

Wolfe, K.H., Li, W.-H., Sharp, P.M., 1987. Rates of nucleotide substitution varygreatly among plant mitochondrial, chloroplast, and nuclear DNAs. Proc. Natl.Acad. Sci. USA 84, 9054–9058.

Yoder, A.D., Yang, Z., 2000. Estimation of primate speciation dates using localmolecular clocks. Mol. Biol. Evol. 17, 1081–1090.