Working PaperPDF Available

Experimental signal dissection and method sensitivity analyses reaffirm the potential of fossils and morphology in the resolution of seed plant phylogeny

June 2017

DOI:10.1101/134262

License
CC BY-NC 4.0

State: Under Submission

Authors:

Mario Coiro

Senckenberg Research Institute and Natural History Museum Frankfurt/M

Guillaume Chomicki

Durham University

James A. Doyle

University of California, Davis

The placement of angiosperms and Gnetales in seed plant phylogeny remains one of the most enigmatic problems in plant evolution, with morphological analyses (which have usually included fossils) and molecular analyses pointing to very distinct topologies. Almost all morphology-based phylogenies group angiosperms with Gnetales and certain extinct seed plant lineages, while most molecular phylogenies link Gnetales with conifers. In this study, we investigate the phylogenetic signal present in published seed plant morphological datasets. We use parsimony, Bayesian inference, and maximum likelihood approaches, combined with a number of experiments with the data, to address the morphological-molecular conflict. First, we ask whether the lack of association of Gnetales with conifers in morphological analyses is due to an absence of signal or to the presence of competing signals, and second, we compare the performance of parsimony and model based approaches with morphological datasets. Our results imply that the grouping of Gnetales and angiosperms is largely the result of long branch attraction, consistent across a range of methodological approaches. Thus, there is a signal for the grouping of Gnetales with conifers in morphological matrices, but it was swamped by convergence between angiosperms and Gnetales, both situated on long branches. However, this effect becomes weaker in more recent analyses, as a result of addition and critical reassessment of characters. Even when a clade including angiosperms and Gnetales is still weakly supported by parsimony, model-based approaches favor a clade of Gnetales and conifers, presumably because they are more resistant to long branch attraction. Inclusion of fossil taxa weakens rather than strengthens support for a relationship of angiosperms and Gnetales. Our analyses finally reconcile morphology with molecules in favoring a relationship of Gnetales to conifers, and show that morphology may therefore be useful in reconstructing other aspects of the phylogenetic history of the seed plants.

Statistics for the maximum parsimony analyses of fossil matrices. 972

…

Figures - uploaded by Mario Coiro

Content may be subject to copyright.

Content uploaded by Mario Coiro

Content may be subject to copyright.

Content uploaded by Mario Coiro

Content may be subject to copyright.

Manuscript type: Article 1

Running head: MORPHOLOGY AND SEED PLANT PHYLOGENY 2

Experimental signal dissection and method sensitivity analyses reaffirm 4

the potential of fossils and morphology in the resolution of seed plant 5

phylogeny 6

Mario Coiro1*, Guillaume Chomicki2,3, James A. Doyle4 8

1Department of Systematic and Evolutionary Botany, University of Zurich, 8008 Zurich, 9

Switzerland. 2Department of Plant Sciences, University of Oxford, South Parks Road, 10

Oxford OX1 3RB, UK. 11

3The Queen’s College, University of Oxford, High St, Oxford OX1 4AW, UK. 12

4Department of Evolution and Ecology, University of California, Davis, CA 95616, USA. 13

*Correspondence to be sent to: mario.coiro@systbot.uzh.ch 14

.CC-BY 4.0 International licensepeer-reviewed) is the author/funder. It is made available under a The copyright holder for this preprint (which was not. http://dx.doi.org/10.1101/134262doi: bioRxiv preprint first posted online Jun. 7, 2017;

Abstract [292 words]: 20

The phylogeny of seed plants remains one of the most enigmatic problems in plant 21

evolution, with morphological analyses (which have usually included fossils) and molecular 22

analyses pointing to very distinct topologies. Almost all morphology-based phylogenies 23

support the anthophyte hypothesis, grouping the angiosperms with Gnetales and certain 24

extinct seed plant lineages, while most molecular phylogenies link Gnetales with conifers. 25

In this study, we investigate the phylogenetic signal present in published seed plant 26

morphological datasets. We use maximum parsimony, Bayesian inference, and maximum 27

likelihood approaches, combined with a number of experiments with the data, to address 28

the morphological-molecular conflict. First, we ask whether the lack of association of 29

Gnetales with conifers in morphological analyses is due to an absence of signal or to the 30

presence of competing signals, and second, we compare the performance of parsimony 31

and model-based approaches with morphological datasets. Our results imply that the 32

grouping of Gnetales and angiosperms is largely the result of long branch attraction, 33

consistent across a range of methodological approaches. Thus, there is a signal for the 34

grouping of Gnetales with conifers in morphological matrices, but it was swamped by 35

convergence between angiosperms and Gnetales, both situated on long branches. 36

However, this effect becomes weaker in more recent analyses, as a result of addition and 37

critical reassessment of characters. Even when an anthophyte topology is still weakly 38

supported by parsimony, model-based approaches favor a clade of Gnetales and conifers, 39

presumably because they are more resistant to long branch attraction. Inclusion of fossil 40

taxa weakens rather than strengthens support for the anthophyte hypothesis. Our 41

analyses finally reconcile morphology with molecules in favoring a relationship of Gnetales 42

to conifers, and show that morphology may therefore be useful in reconstructing other 43

aspects of the phylogenetic history of the seed plants. 44

INTRODUCTION 46

The use of morphology as a source of data for reconstructing phylogenetic relationships 47

has lost most of its ground since the advent of molecular phylogenetics, except in 48

paleontology. However, there has recently been renewed interest in morphological 49

phylogenetics (Pyron 2015; Lee and Palci 2015). This is partly because of increased focus 50

on the phylogenetic placement of fossil taxa in trees of living organisms, stimulated by the 51

necessity of accurate calibrations for dating the molecular trees that have become the 52

main basis for comparative evolutionary studies. This has led to the development of 53

methods that integrate phylogenetic placement of fossils in the dating process (Pyron 2011; 54

Ronquist et al. 2012; Zhang et al. 2016). Another focus has been the application of 55

statistical phylogenetics to morphological data on both a theoretical (Wright et al. 2014, 56

2015; O’Reilly et al. 2016) and an empirical level (Lee and Worthy 2012; Godefroit et al. 57

2013; Cau et al. 2015). In paleontology, where only morphological data are available 58

(except in the recent past), questions on the role of morphology in phylogenetics are even 59

more critical. A major issue concerns the value of fossils in reconstructing relationships 60

among living organisms. Early in the history of phylogenetics, there were claims that 61

fossils are incapable of overturning phylogenetic relationships inferred from living taxa 62

(Patterson 1981), but also demonstrations that they can, as for instance in morphological 63

analysis of amniote phylogeny (Gauthier et al., 1988). Whether or not fossils affect the 64

inferred topology of living taxa, there is little doubt that they are often either useful or 65

necessary in elucidating the homologies of novel structures (e.g., the seed plant ovule and 66

eustele) and the order of origin of the morphological synapomorphies of extant (crown) 67

groups (e.g., origin of secondary growth before the ovule in the seed plant line), as 68

discussed in Doyle (2013). This is critical because major groups, such as angiosperms, 69

are often separated from their closest living relatives by major morphological gaps 70

(numbers of character changes), even if the incorporation of fossils does not affect inferred 71

relationships among living taxa (Doyle and Donoghue 1987; Donoghue et al. 1989). 72

Many phylogenies based on morphology have been recently published for important 73

groups with both living and fossil representatives, including mammals (O’Leary et al. 2013), 74

squamate reptiles (Gauthier et al. 2012), arthropods (Legg et al. 2013), and the genus 75

Homo (Dembo et al. 2016). However, the validity and use of morphological data in 76

reconstructing phylogeny have been severely criticized, notably by Scotland et al. (2003), 77

based on supposed diminishing returns in the discovery of new morphological characters 78

and the prevalence of functional convergence. The painstaking acquisition of 79

morphological characters, which requires a relatively large amount of training and time, 80

could turn out to be systematically worthless if the phylogenetic signal present in these 81

data is either insufficient or misleading. Indeed, the number of characters that can be 82

coded for morphological datasets represents a major limit to the use of morphology and its 83

integration with molecular data, especially in the age of phylogenomics, where the ever-84

increasing amount of molecular signal could simply “swamp” the weak signal present in 85

morphological datasets (Doyle and Endress 2000; Bateman et al. 2006). Morphological 86

data may also be afflicted to a higher degree than molecules by functional convergence 87

and parallelism (Givnish and Sytsma 1997), which could lead a morphological dataset to 88

infer a wrong phylogenetic tree. Even though the confounding effect of convergence has 89

been formally tested in only a few studies (Wiens et al. 2003), it seems to be at the base of 90

one of the deepest cases of conflict between molecules and morphology in the 91

reconstruction of evolutionary history, namely the phylogeny of placental mammals (Foley 92

et al. 2016). In this case, the strong effect of selection on general morphology caused by 93

similar lifestyle seems to hinder attempts to use morphology to reconstruct phylogenetic 94

history in this group (Springer et al. 2007), and it affects even large “phenomic” datasets 95

(Springer et al. 2013). 96

Another example of conflict between morphology and molecular data involves the 97

relationships among seed plants. Before the advent of cladistics, some authors proposed 98

that angiosperms were related to the highly derived living seed plant order Gnetales, while 99

others argued that these two groups were strictly convergent and Gnetales were instead 100

related to conifers (for a review, see Doyle and Donoghue 1986). However, the view that 101

angiosperms are related to Gnetales and fossil Bennettitales, called the anthophyte 102

hypothesis, is one of the oldest and seemingly most stable results of morphologically 103

based parsimony analyses of seed plant phylogeny. Since Hill and Crane (1982) and 104

Crane (1985), the grouping of Bennettitales, Gnetales, the fossil Pentoxylon, and 105

angiosperms (sometimes with the fossil Caytonia as the closest outgroup of angiosperms) 106

was retrieved in almost all successive analyses (Doyle and Donoghue 1986, 1992; Nixon 107

et al. 1994; Rothwell and Serbet 1994; Doyle 1996, 2006, 2008; Hilton and Bateman 2006; 108

Friis et al. 2007; Rothwell et al. 2009; Rothwell and Stockey 2016; Fig. 1). Some analyses 109

associated anthophytes with “Mesozoic seed ferns” (glossopterids, corystosperms, and 110

Caytonia), others with “coniferophytes” (conifers, Ginkgo, and fossil cordaites). By contrast, 111

since the advent of molecular phylogenetics, the anthophyte hypothesis has lost most of 112

its support among plant biologists. Although molecular analyses cannot directly evaluate 113

the status of presumed fossil anthophytes, they can address the relationship of 114

angiosperms and Gnetales. Molecular data from different genomes analyzed with different 115

approaches do not yield a Gnetales plus angiosperm clade, with the exception of few 116

maximum parsimony (MP) and neighbor joining analyses of nuclear ribosomal RNA or 117

DNA (Hamby and Zimmer 1992; Stefanovic et al. 1998; Rydin et al. 2002) and one MP 118

analysis of rbcL (Rydin and Källersjö 2002). The majority of molecular analyses retrieve a 119

clade of Gnetales plus Pinaceae (Bowe et al. 2000; Chaw et al. 2000; Gugerli et al. 2001; 120

Qiu et al. 2007; Zhong et al. 2011), conifers other than Pinaceae (cupressophytes) 121

(Nickrent et al. 2000; Rydin and Källersjö 2002), or conifers as a whole (Wickett et al. 122

2014), which we refer to collectively as “gneconifer” trees. In most of these trees 123

angiosperms are the sister group of all other living seed plants (acrogymnosperms). The 124

main exceptions are “Gnetales-basal” trees, in which Gnetales are sister to all other living 125

seed plants (e.g., Albert et al. 1994; Rydin and Källersjö 2002). 126

Several potential issues have been identified with both sorts of data. Regarding 127

molecules, these include limited taxonomic sampling resulting from extinction of the 128

majority of seed plant lineages (Rothwell et al. 2009), loss of phylogenetic signal due to 129

saturation (particularly at third codon positions), strong rate heterogeneity among sites 130

across lineages and conflict between gene trees (Mathews 2009), composition biases 131

among synonymous substitutions (Cox et al. 2014), as well as systematic errors and 132

biases (Sanderson et al. 2000; Magallón and Sanderson 2002; Burleigh and Mathews 133

2007; Zhong et al. 2011), leading to a plethora of conflicting signals. In analyzing datasets 134

that yielded Gnetales-basal trees, studies that have attempted to correct for these biases 135

have generally favored trees in which Gnetales are associated with conifers (Sanderson et 136

al. 2000; Magallón and Sanderson 2002; Burleigh and Mathews 2007). Regarding 137

morphology, in addition to the role of functional convergence in confounding relationships, 138

it has been shown that different taxon sampling strategies, such as choice of the closest 139

progymnosperm outgroup of seed plants (Hilton and Bateman 2006), can lead to different 140

results concerning the rooting of the seed plants. 141

The conflict between molecules and morphology has led to different attitudes 142

toward morphological data within the botanical community (Donoghue and Doyle 2000; 143

Scotland et al. 2003; Bateman et al. 2006; Rothwell et al. 2009). Following suggestions of 144

Donoghue and Doyle (2000), Doyle (2006, 2008) reconsidered several supposed 145

homologies between angiosperms and Gnetales in the light of the molecular results. 146

These studies and the analysis of Hilton and Bateman (2006) also incorporated newly 147

recognized similarities between Gnetales and conifers, for example in wood anatomy 148

(Carlquist 1996), as well as new evidence on the morphology of the seed-bearing cupules 149

in fossil taxa. When building a morphological matrix, dissecting a character into more 150

character states may represent an improvement by distinguishing convergent states and 151

avoiding bias toward particular phylogenetic hypotheses during primary homology 152

assessment (Jenner 2004; Zou and Zhang, 2016), but it may be disadvantageous 153

because it leads to a lack of resolution when the number of states becomes excessive. In 154

seed plants, there are many special factors that complicate character coding. Among living 155

taxa, the assessment of homology is complicated by the plastic and modular nature of 156

plant development (Mathews and Kramer 2012). Among fossil taxa, the mode of 157

preservation of many key fossils has critical consequences for the amount of data 158

available. This affects not only the number of missing characters, but also the process of 159

primary homology assessment and character coding. Although these issues with coding 160

are most severe in fossils preserved as compressions, such as Caytonia (Doyle 2008; 161

Rothwell et al. 2009) and Archaefructus (Sun et al. 2002; Friis et al. 2003; Doyle 2008; 162

Rudall and Bateman 2008; Endress and Doyle 2009), even fossil groups that are 163

exquisitely preserved as permineralizations (e.g., Bennettitales) are not immune to 164

conflicting interpretations (Friis et al. 2007; Rothwell et al. 2009; Crepet and Stevenson 165

2010; Doyle 2012; Pott 2016). 166

Despite careful reconsideration of potentially convergent traits between Gnetales 167

and angiosperms, the conflict between morphological and molecular data appeared to 168

persist, with most morphological parsimony analyses continuing to favor the anthophyte 169

hypothesis (Doyle 2006; Hilton and Bateman 2006; Rothwell et al. 2009). The possibility 170

that morphological data are inadequate to resolve the phylogeny of seed plants would 171

represent a severe hindrance in understanding plant evolution, especially in the light of the 172

small number of extant lineages that survived extinction during the Paleozoic and 173

Mesozoic (Mathews 2009) and the great morphological gaps among these surviving 174

lineages. However, there have been signs that the conflicts with molecular data are 175

weakening: Doyle (2006) found that trees in which Gnetales were nested in conifers were 176

only one step less parsimonious than anthophyte trees, and in Doyle (2008) trees of the 177

two types became equally parsimonious. 178

In this study, we attempt to elucidate the phylogenetic signal present in published 179

morphological datasets of the seed plants. We first test whether the possibility of 180

convergence between angiosperms and Gnetales represents a major problem by 181

reanalyzing the matrices that incorporated earlier homology assumptions concerning 182

characters of the two groups (i.e., the matrices compiled before the incoming of molecular 183

results) and later matrices that revised such assumptions (the matrices of Doyle 2006 and 184

Hilton and Bateman 2006, and datasets derived from them) and testing whether the signal 185

and the relative support for the anthophyte and gneconifer clades changed between these 186

two sets of matrices. After revealing a more coherent signal supporting a gneconifer clade 187

in the more recent matrices, we investigate whether the retrieval of an anthophyte 188

topology by maximum parsimony was at least partly due to methodological biases that 189

could be overcome by using model-based methods. Hopefully these approaches may be 190

useful in resolving cases of conflict between morphological and molecular data in other 191

taxa, particularly those with significant fossil representatives. 192

193

MATERIALS AND METHODS 194

Matrices 195

The Crane (1985, version two), Doyle and Donoghue (1986, 1992), Nixon et al. 196

(1994), Rothwell and Serbet (1994), and Doyle (1996, 2006, 2008) matrices were 197

manually coded from the respective articles. The Hilton and Bateman (2006) matrix was 198

kindly provided by Richard Bateman. The matrices from Analysis 3 of Rothwell et al. (2009) 199

and from Rothwell and Stockey (2016) were downloaded from the supplementary 200

materials of the respective articles. 201

202

Parsimony analyses 203

We performed maximum parsimony analyses of all matrices with PAUP 4.0a136 204

(Swofford 2003), using the heuristic search algorithm with random addition of taxa and 205

1000 replicates. Bootstrap analyses were conducted using 10,000 replicates, using the 206

“asis” addition option and keeping one tree per replicate (Müller 2005). 207

We also conducted analyses with a topological backbone constraint, forcing the 208

Gnetales into a clade with the extant conifers and leaving the position of other living taxa 209

and fossils unconstrained. Significant differences between the constrained and 210

unconstrained topologies were evaluated using the Templeton test (Templeton 1983) as 211

implemented in PAUP v. 4.0a136 (Swofford 2003). We investigated the effects of recoding 212

characters by Doyle (2006, 2008) in more detail by using MacClade (Maddison and 213

Maddison 2003) to compare the number of steps in each character on trees with Gnetales 214

nested in anthophytes and associated with conifers. 215

216

Maximum likelihood (ML) 217

Maximum likelihood analyses were conducted using RaxML ver 8.2.10 (Stamatakis 218

2014). Matrices were modified by recoding all ambiguities (e.g., 0/1 in a three-state 219

character) as missing data, since the method cannot cope with ambiguous characters. 220

Topology is inferred using branch lengths, which are estimated as the expected number of 221

state changes per character on that particular branch. We conducted 1000 bootstrap 222

replicates using a Markov k-states (Mk) model (Lewis 2001) , which assumes an equal 223

probability of changes in both directions between all states, with a gamma-distributed rate 224

variation, which models different rates across characters by employing a multiplier drawn 225

from a discretized gamma distribution. 226

227

Bayesian inference (BI) 228

Bayesian analyses relied on MrBayes v. 3.2.3 (Ronquist et al. 2012), under the 229

Markov k-states (Mk) model (Lewis 2001). For each matrix, we conducted two analyses, 230

one with an equal rate of evolution among characters and another with gamma-distributed 231

rate variation. In both cases, we used the MKpr-inf correction for parsimony informative 232

characters. The analyses were run for 5,000,000 generations, sampling every 1000th 233

generation. The first 10,000 runs were discarded as burn-in. Posterior traces were 234

inspected using Tracer (Rambaut and Drummond 2007). 235

236

Model testing and rate variation 237

We also conducted stepping stone analyses (SS) (Xie et al. 2011; Ronquist et al. 238

2012) in order to evaluate the most appropriate model of rate variation among characters 239

(equal rates vs. gamma-distributed rates). These analyses allow us to estimate the 240

marginal likelihood for different models with better accuracy than other measures (e.g., 241

harmonic mean estimator). We used 4 independent runs with 2 chains with the default 242

MrBayes parameters, run for 5,000,000 generations and sampling every 1000th generation. 243

Using the marginal likelihoods from the SS analysis, we then calculated the support for the 244

two models using Bayes factors (BF) (Kass and Raftery 1995). 245

246

Exploring conflict in the data 247

To explore phylogenetic conflict in the data, we employed the software SplitsTree 4 248

(Huson and Bryant 2006). We used this program to visualize conflicts among the bootstrap 249

replicates from the MP and ML analysis and among the posterior tree samples found with 250

Bayesian inference. The software summarizes the sets of trees using split networks, which 251

allow us to visualize all possible conflicting hypotheses. A consensus network (Holland et 252

al. 2004) was built using the “count” option, with the cut-off for visualizing the splits set at 253

0.05. 254

255

Long branch attraction tests 256

We modified the matrices to perform tests for long branch attraction (LBA), following 257

the suggestions of Bergsten (2005). Two matrices were created to test the potentially 258

destabilizing effect of the two long-branched groups suspected to create this artifact, 259

angiosperms and Gnetales, by alternately removing each of them (long branch extraction 260

analysis, LBE). To test further the hypothesis of an LBA artifact exerted by angiosperms, 261

we followed a similar approach to the sampling experiment in Rota-Stabelli et al. (2010): 262

another matrix was created to elongate the branch subtending angiosperms by removing 263

the three fossil taxa most commonly identified as angiosperm outgroups (Pentoxylon, 264

Bennettitales, and Caytonia) (branch elongation analysis, BE). To test the effect of 265

including fossil data in the matrices, we created a set of matrices in which all fossil taxa 266

were removed (extant experiment, EX). 267

Morphospace analysis 268

To visualize morphological patterns in the different matrices, we conducted 269

principal coordinates (PCO) analyses using the R package Claddis (Lloyd 2016). 270

We employed the maximum observed rescaled distance between all pairs of taxa to 271

generate the ordination. The taxa were then plotted on the first two PCO axes. 272

273

Data availability 274 275 All data are available on figshare: 276 https://figshare.com/s/9a4fc5d4accff8e62084. 277 278

RESULTS 279

280

Our re-analyses of the historical morphological matrices of seed plants with 281

parsimony resulted in trees identical to the published trees (Table 1). The MP trees and the 282

consensus trees always show an anthophyte clade (with or without Caytonia), with the 283

exception of trees based on the Doyle (2008) matrix, in which anthophyte and gneconifer 284

topologies are equally parsimonious. Constraining Gnetales and conifers to form a clade 285

always results in trees longer than the most parsimonious trees, except with the Doyle 286

(2008) matrix (Table 2). The Templeton test of the best trees against the worst constrained 287

trees (i.e., the most parsimonious constrained tree that is statistically most different from 288

the most parsimonious unconstrained tree) does however show that this difference is only 289

significant at the 0.05 level with the Nixon et al. (1994) matrix. 290

Bootstrap analysis shows that the anthophyte clade is not strongly supported by 291

any of the matrices, with the exception of the Nixon et al. (1994) matrix (Fig. 2). In the MP 292

bootstrap analysis of the post-2000 matrices (Fig. 2A), support for an anthophyte topology 293

appears to be lower than support for a gneconifer topology in all matrices except that of 294

Rothwell et al. (2009). The ML bootstrap (Fig. 2B) shows higher support for a gneconifer 295

topology than the MP bootstrap in all post-2000 analyses, as well as in the two pre-2000 296

Doyle and Donoghue (1986, 1992) matrices. In the post-2000 matrices, the support for 297

gneconifers is always higher than the support for anthophytes. 298

The stepping stone analysis shows strong support for rate variation among 299

characters in all matrices except those of Crane (1985) and Doyle and Donoghue (1986) 300

(Table 3), as indicated by ln-Bayes factors higher than 2. 301

The trees obtained from the Bayesian analyses show a much sharper differentiation 302

between early and late matrices (Fig. 2C). With the pre-2000 matrices, support and 303

topology are mostly in agreement with the MP analyses. However, with the post-2000 304

matrices we observe a shift in support from the anthophytes to a clade of Gnetales and 305

conifers. This is illustrated by a split network consensus based on the Rothwell and 306

Stockey (2016) matrix (Fig. 3C), in which Gnetales are linked with conifers, and 307

Glossopteris, Caytonia, and Petriellaea (a Triassic fossil not included in earlier analyses 308

that is now better known vegetatively thanks to work of Bomfleur et al. 2014) are the 309

closest outgroups of angiosperms. 310

Our first test of the hypothesis that the anthophyte topology is the result of long 311

branch attraction consists of long branch extraction (LBE) experiments. These involved 312

separate removal of the two potential long branch taxa: angiosperms and Gnetales. 313

The removal of the angiosperms has different effects on the pre- and post-2000 314

matrices. With the Crane (1985) version two matrix analyzed here, a topology with 315

Bennettitales, Pentoxylon and the Gnetales diverging after Lyginopteris and before the 316

other taxa becomes as parsimonious as the topology with the anthophytes nested among 317

Mesozoic seed ferns that was retrieved with the full matrix. The new tree corresponds to 318

the most parsimonious tree that Crane (1985) found with his version one matrix, in which 319

Bennettitales and Pentoxylon were scored as not having cupules homologous with those 320

of Mesozoic seed ferns. With the Doyle and Donoghue (1986) matrix, Bennettitales, 321

Pentoxylon, and Gnetales are nested within coniferophytes. With the Doyle and Donoghue 322

(1992) and Rothwell and Serbet (1994) matrices, the consensus tree is identical to the 323

trimmed consensus derived from the full matrix. With the Nixon et al. (1994) matrix, 324

Cordaites and Ginkgo are successive outgroups to a conifer plus anthophyte clade, 325

whereas with the full matrix they are equally parsimoniously placed as successive 326

outgroups to the conifers, in a clade that is sister to anthophytes. The inverse happens 327

with the Doyle (1996) matrix, where the position of Ginkgo and cordaites is destabilized by 328

the removal of the angiosperms, with these taxa being either successive outgroups to 329

extant and fossil conifers or sister to a clade composed of anthophytes, conifers, 330

Peltaspermum, and Autunia. The position of the Gnetales in a truncated anthophyte clade 331

(i.e., with Bennettitales and Pentoxylon) is maintained in all matrices. 332

With the post-2000 matrices, the effect of removal of the angiosperms is consistent 333

among different matrices. With the Hilton and Bateman (2006), Doyle (2006), and Doyle 334

(2008) datasets, the resulting trees see the Gnetales nested within the coniferophytes, 335

with or without Bennettitales. With the Rothwell et al. (2009) matrix (Fig. 4D-F), a topology 336

with a clade of Gnetales and conifers that excludes Bennettitales and Pentoxylon 337

becomes most parsimonious (Fig. 4E). With the Rothwell and Stockey (2016) matrix, 338

Gnetales are sister to Tax us in a coniferophyte clade that also includes Doylea, an Early 339

Cretaceous cone-like structure interpreted as consisting of seed-bearing cupules (Stockey 340

and Rothwell 2009; Rothwell and Stockey 2016). 341

The removal of the Gnetales has no impact at all on trees based on the Crane 342

(1985), Doyle and Donoghue (1986), and Doyle and Donoghue (1992) matrices, in which 343

the topology is identical to the trimmed topology of the consensus in the full analysis. With 344

the Nixon et al. (1994) matrix, the removal of the Gnetales results in trees in which 345

coniferophytes form a clade (including Ginkgo and Cordaites), i.e., eliminating most 346

parsimonious trees in which anthophytes are linked with conifers. With the Rothwell and 347

Serbet (1994) matrix, the removal of Gnetales results in a breakup of the Caytonia-348

Glossopteris-corystosperm clade, with the angiosperms still nested within the other 349

anthophytes. With the Doyle (1996) matrix, the only difference lies in the placement of the 350

corystosperms, Autunia, and Peltaspermum, which are sister to a coniferophyte clade in 351

the analysis without Gnetales. 352

With the post-2000 matrices, the removal of the Gnetales results in trees in which 353

the remaining anthophytes (which may or may not include Caytonia) form a clade outside 354

the coniferophytes (e.g., Fig. 4F). With the Doyle (2006) and Doyle (2008) matrices, a 355

clade including Cycadales, glossopterids, and anthophytes (including Caytonia) is sister to 356

a clade of Callistophyton, Peltaspermum, Autunia, and corystosperms plus coniferophytes. 357

The analysis of the Rothwell and Stockey (2016) matrix represents an exception, where 358

the placement of the anthophytes is not affected by the removal of Gnetales. However, the 359

removal of Doylea in addition to Gnetales results in a pattern similar to that found with the 360

other post-2000 matrices. 361

In the branch elongation (BE) experiment, where three fossils commonly associated 362

with angiosperms (Bennettitales, Pentoxylon, Caytonia) were removed, we observed that 363

MP bootstrap support for the angiosperm plus Gnetales clade increases in all matrices 364

(Fig. 4G). This effect is even stronger in the extant (EX) experiment matrices, in which all 365

fossil taxa were removed, where a split including angiosperms plus Gnetales is strongly 366

supported by the MP bootstrap in all matrices. 367

Bayesian analysis (BI) of the BE and EX matrices shows a less linear pattern (Fig. 368

4H, I). In the BE analyses, the signal for the anthophytes decreases with the Doyle and 369

Donoghue (1986, 1992) matrices, reaching less than 0.5 posterior probability (PP) in the 370

analysis with gamma-distributed rate variation. With the Nixon et al. (1994), Rothwell and 371

Serbet (1994), and Doyle (1996) matrices, the PP of the anthophytes in the BE matrices is 372

comparable to that from the full matrices. In the post-2000 BE matrices, BI support for the 373

anthophytes is almost null with the Hilton and Bateman (2006) and Doyle (2006) matrices 374

(<0.07 PP) and increases with the Doyle (2008) and Rothwell et al. (2009) matrices 375

analyzed using gamma-rate variation (0.55 and 0.51 respectively) and with the Rothwell 376

and Stockey (2016) matrix (0.23 for the equal-rate analysis, 0.37 for the gamma analysis). 377

The analyses of the EX matrices all show high to moderate support (1-0.75 PP) for the 378

split containing angiosperms plus Gnetales. With the post-2000 matrices, the use of the 379

gamma-distributed model recovers a higher PP for the anthophytes. 380

The morphospace analyses (Fig. 5) provide a graphic confirmation of the 381

morphological separation of both Gnetales and angiosperms from other seed plants and 382

the perception that Gnetales share competing morphological similarities with both 383

angiosperms and conifers. In the morphospace generated from most of the pre-2000 384

matrices, Gnetales lie closer to angiosperms (data not shown). With the Doyle (1996) 385

matrix and the post-2000 matrices, the first PCO axis appears to separate angiosperm-like 386

and non-angiosperm-like taxa, whereas the second axis seems to represent a tendency 387

from a seed fern-like towards a conifer-like morphology. Gnetales are always placed closer 388

to the conifers than to the angiosperms (Fig. 5). However, in all cases, Gnetales seem to 389

have higher levels of “angiosperm-like” morphology than do conifers, represented by their 390

rightward placement on the first PCO axis. This position on the first axis is shared by 391

Doylea with the Rothwell and Stockey (2016) matrix. Between the analyses of the Doyle 392

(1996) and Doyle (2008) matrices (Fig. 5A, B), there is a modest shift of Gnetales away 393

from angiosperms and towards conifers. 394

395

DISCUSSION 396

The results of our analyses help to resolve some of the main issues regarding the 397

phylogenetic signal for the anthophyte clade in morphological matrices of seed plants. Our 398

meta-analyses of published datasets (Fig. 2) show a two-step trend: first, changes in 399

character sampling and analysis weakened support for the anthophyte hypothesis, and 400

second, the use of model-based methods shifted the balance in favor of a relationship 401

between Gnetales and conifers, bringing the results in line with molecular data. The effect 402

of changes in character analysis is seen in the switch in support between matrices 403

compiled before the main molecular analyses of seed plant phylogeny (pre-2000) and 404

afterwards: i.e., Doyle (2006) and Hilton and Bateman (2006). These two matrices, which 405

both used Doyle (1996) as a starting point but were modified independently, with only 406

limited discussion at later stages of the two projects, and made different choices regarding 407

character coding, taxon sampling, and splitting of higher-level taxa, both show a very 408

similar pattern. Under the MP criterion, an anthophyte topology was still more 409

parsimonious, but with reduced support. By contrast, ML and the Bayesian criterion 410

positively favor a grouping of Gnetales and conifers. The matrices descended from Doyle 411

(2006) (i.e., Doyle 2008) and from Hilton and Bateman (2006) (i.e., Rothwell et al. 2009, 412

2016) exhibit a similar pattern, except that in Doyle (2008) anthophyte and gneconifer 413

trees were equally parsimonious. This phenomenon was already reported by Mathews et 414

al. (2010), who reanalyzed the matrix of Doyle (2008) using BI. 415

416

Critical character reassessment weakened the conflict between morphology and 417

molecules 418

Examination of the behavior of characters on anthophyte and gneconifer trees 419

illustrates how changes in character analysis between the studies of Doyle (1996) and 420

Doyle (2006, 2008) increased support for gneconifer trees. Some of these changes were 421

the result of new discoveries concerning the morphology of Gnetales and other taxa, 422

others of critical reassessment of previous character definitions aimed at reducing bias in 423

favor of the anthophyte hypothesis. The shift of Gnetales away from angiosperms and 424

towards conifers observed in the morphospace analyses based on the datasets of Doyle 425

(1996) and Doyle (2008) (Fig. 5A, B) is presumably the result of these changes. 426

Most changes of the first sort involved previously overlooked conifer-like features of 427

Gnetales. For example, Doyle (2006) added a character for presence of a torus in the pit 428

membranes of xylem elements in conifers and Gnetales, based on observations on 429

Gnetales by Carlquist (1996) and studies of conifers by Bauch et al. (1972). Doyle (2006) 430

also rescored Gnetales as having a tiered proembryo, as in conifers; two tiers of cells were 431

illustrated by Martens (1971) and called “étages,” and by Singh (1978). This similarity may 432

have been overlooked because of other differences related to elimination of a free-nuclear 433

phase in the embryogenesis of Gnetales (Doyle 2006). Both characters undergo one less 434

step on gneconifer trees than on most anthophyte trees (exceptions are some with major 435

rearrangements elsewhere in seed plants). In male “flowers” of Ephedra and Welwitschia, 436

microsynangia are borne in two lateral groups, which Doyle (1996) interpreted as reduced 437

pinnate sporophylls. Because Bennettitales, Caytonia, and “seed fern” outgroups have 438

pinnately organized microsporophylls, this character favored an anthophyte tree by one 439

step. However, developmental studies by Mundry and Stützel (2004) indicated that the two 440

lateral structures are more likely branches (strobili) bearing three or four simple 441

sporophylls. Based on these observations, Doyle (2008) rescored microsporophylls in 442

Gnetales as simple and one-veined, as in conifers, and as a result the character favored 443

the gneconifer topology by one or two steps. 444

Doyle (2006) also made changes based on improved data on a character 445

expressing the position of the ovule or ovules on the sporophylls or “cupules” that bear 446

them, which is not directly relevant to Gnetales but potentially useful for identification of 447

anthophyte outgroups. Ovules are on the abaxial surface of the sporophyll/cupule in 448

corystosperms (Axsmith et al. 2000; Klavins et al. 2002), rather than on the adaxial 449

surface in glossopterids (Taylor and Taylor 1992), probably Caytonia, and angiosperms (if 450

the outer integument is a modified leaf or cupule: Doyle 2006, 2008; Kelley and Gasser 451

2009). Ovules are also adaxial in the cupules of Petriellaea (Taylor et al. 1994; Bomfleur et 452

al. 2014), which was included in the analysis of Rothwell and Stockey (2016). 453

Other changes were the result of doubts concerning the homology of characters 454

that supported the anthophyte hypothesis, along lines suggested by Donoghue and Doyle 455

(2000). For example, in the apical meristem character, Doyle (1996) contrasted the 456

presence of a tunica (an outer layer that maintains its integrity by undergoing only 457

anticlinal cell divisions, i.e., perpendicular to the surface) of Gnetales, angiosperms, and 458

Araucariaceae, vs. its absence in cycads, Ginkgo, and other conifers. This character 459

undergoes two steps when Gnetales are linked with angiosperms (the state in fossils is 460

unknown), three when Gnetales are linked with conifers. However, the tunica consists of 461

one layer of cells in Gnetales, but two layers in angiosperms, suggesting that it may not be 462

homologous in the two groups. To reduce bias in favor of homology of these two 463

conditions, Doyle (2006) split presence of a tunica into two states. The resulting three-464

state character undergoes three steps with Gnetales in both positions. Redefinition of the 465

megaspore membrane character involved a shift in the limit between states, from thick vs. 466

reduced (thin or absent) to present vs. absent; the megaspore membrane is thin in 467

Gnetales, but absent in angiosperms, Caytonia, and probably Bennettitales. In 468

compressions of bennettitalean seeds prepared by oxidative maceration, Harris (1954) 469

observed no megaspore membrane, but Wieland (1916) and Stockey and Rothwell (2003) 470

reported a thin layer around the megagametophyte in permineralized seeds. However, as 471

noted by Harris (1954), there is no evidence that this layer is a true megaspore membrane 472

(i.e., consisting of exinous material). These changes in character definition do involve a 473

subjective element and were doubtless influenced by knowledge of the molecular 474

evidence for a relationship of Gnetales and conifers, but the new definitions represent a 475

shift toward greater caution in evaluating the potential homology of similar but not identical 476

structures. 477

These trends show that recognition of previously overlooked similarities between 478

Gnetales and conifers and reconsideration of potentially convergent characters between 479

angiosperms and Gnetales succeeded in strengthening a morphological signal associating 480

Gnetales with conifers. This result clearly contradicts the view that morphology and 481

molecules are in strong conflict with each other (Bateman et al. 2006; Rothwell et al. 2009) 482

and validates arguments to this effect advanced by Doyle (2006, 2008) on a parsimony 483

basis. Indeed, in all post-2000 matrices a topology with Gnetales linked with conifers 484

requires the addition of only a few steps to the length of anthophyte trees: e.g., four in the 485

case of Hilton and Bateman (2006) and one in Doyle (2006), and in Doyle (2008) both 486

topologies became equally parsimonious. A tendency to focus on the MP consensus tree 487

and lack of exploration of almost equally parsimonious alternatives may have tended to 488

inflate the perceived conflict between molecules and morphology. Among analyses since 489

1994, bootstrap and/or decay values were reported by Doyle (1996, 2006, 2008), Hilton 490

and Bateman (2006), and Rothwell and Stockey (2016), but not by Nixon et al. (1994), 491

Rothwell and Serbet (1994), and Rothwell et al. (2009). Our analyses show that the signal 492

retrieved using MP is more correctly characterized as profoundly ambiguous. 493

494

Contribution of model-based methods 495

By contrast, maximum likelihood and especially Bayesian analyses of all post-2000 496

matrices converge on a similar result, unambiguously favoring placement of Gnetales in a 497

coniferophyte clade that includes Ginkgoales, cordaites, and extant and extinct conifers. 498

Stronger support is obtained in BI analyses in which gamma rate variation among sites is 499

implemented in the model. With ML the difference in relative support for the two 500

hypotheses appears smaller, but a gnetifer arrangement is consistently favored with all 501

datasets. These results of model-based analyses of post-2000 morphological matrices 502

have interesting implications regarding stem relatives of the angiosperms. Indeed, most 503

post-2000 matrices are broadly congruent in attaching Pentoxylon, glossopterids, 504

Bennettitales, and Caytonia to the stem lineage of the angiosperms. To these the analysis 505

of Rothwell and Stockey (2016) adds the Triassic genus Petriellaea (Taylor et al. 1994; 506

Bomfleur et al. 2014) (Fig. 3), which has simple reticulate laminar venation, as in Caytonia, 507

and cupules containing adaxial ovules. This may be consistent with the view that these 508

fossils shed light on evolution of the complex reticulate venation and bitegmic ovules of 509

angiosperms (Doyle 2006, 2008). 510

A cautionary note on the results of our Bayesian analyses is necessary. The 511

differences between bootstrap support values in the MP and ML analyses and posterior 512

probabilities in the BI analyses could be due to the very different nature of these support 513

metrics. It has been shown that the relationship between character support and increase in 514

PP is far from linear, and PP can easily sway results toward a hypothesis that is supported 515

by only a few characters (Zander 2004). The strong PP support for groupings (like 516

Caytonia or Petriellea plus angiosperms) that receive weak or non-existent support on a 517

character basis (MP and ML bootstrap, Fig. 3A, B) could indicate either the ability of 518

Bayesian inference to pick up a significant signal in an otherwise noisy background or the 519

possibility that this method can be led astray by a few potentially unimportant characters. 520

521

The conflict between morphology and molecules is partially due to long branch 522

attraction 523

Our results also add new empirical evidence on debates concerning the strengths 524

and weaknesses of morphological data in reconstructing phylogenetic relationships, the 525

phylogenetic importance of fossils, and the best methods to analyze morphological data 526

(Wright and Hillis 2014; O’Reilly et al. 2016; Puttick et al. 2017a). A well-known cause of 527

phylogenetic conflict is the presence of long branches in the tree, which can lead to LBA 528

phenomena (Felsenstein 1978; Bergsten 2005). Analyses based on simulated matrices 529

and real data have repeatedly shown that probabilistic, model-based approaches are more 530

robust to LBA than parsimony (Swofford et al. 2001; Brinkmann et al. 2005; and references 531

therein). Long branch attraction is most commonly discussed as a confounding factor in 532

molecular studies, as in the case of Gnetales-basal trees found with molecular data 533

(Sanderson et al. 2000; Magallón and Sanderson 2002; Burleigh and Mathews 2007), but 534

here it is morphology that is potentially affected: the BI trees show that both angiosperms 535

and Gnetales are situated on very long morphological branches, especially in the post-536

2000 matrices. 537

After following suggestions by Bergsten (2005) and other methodologies (Rota 538

Stabelli et al. 2011), we conclude that LBA is responsible at least in part for the continuing 539

support for the anthophyte clade in MP analyses of the post-2000 matrices. First, BI 540

recovers a gneconifer topology with higher probability than a topology with Gnetales in 541

anthophytes, thus favoring a topology that separates the long branches over a topology 542

that unites them. Second, more complex and better-fitting models recover a higher 543

posterior probability for the topology in which angiosperms and Gnetales are separated 544

(Fig. 2C). Third, removing Gnetales or angiosperms results in a rearrangement of the MP 545

topologies in which the other long branch “flies away” from its original position. Fourth, 546

support for Gnetales plus angiosperms increases with decreased sampling of fossil taxa 547

on the branch leading to the angiosperms, and still more with the removal of all fossils (Fig. 548

4G-I). It has been suggested that molecular analyses may be incorrect about the 549

relationship of angiosperms and Gnetales because they ignore the great diversity of 550

extinct seed plant taxa (e.g., Rothwell et al. 2009). This reasoning would predict that 551

addition of fossils would strengthen the anthophyte hypothesis, but in fact our results 552

indicate that the opposite is true. 553

To our knowledge, this represents the first reported case of LBA in a morphological 554

analysis that is supported by multiple tests (Bergsten 2005), with much stronger support 555

than previously reported cases (Lockhart and Cameron 2001; Wiens and Hollingsworth 556

2000). These analyses also support the view that model-based methods can overcome the 557

shortcomings of parsimony in such cases. However, it may be significant that relationships 558

in many other parts of the trees obtained with MP and BI are similar, suggesting that MP is 559

not necessarily misleading where long branch effects are lacking. It is also noteworthy that 560

the impact of LBA can be easily visualized with the principal coordinates analysis (Fig. 5), 561

where the presumed close relationship between Gnetales and conifers and the 562

convergence of Gnetales with the angiosperms are effectively congruent with the positions 563

of the three taxa in the plot of the first two PCO axes. This tool could represent an 564

interesting option for exploring the structure of the data in future phylogenetic analyses. 565

566

CONCLUSIONS 567

For students of seed plant phylogeny, the main lesson of our analyses may be that, 568

contrary to previous impressions, morphological data do not present a strong conflict with 569

the results of molecular analyses regarding the position of the Gnetales. This strongly 570

suggests that morphology carries a phylogenetic signal that is consistent with molecular 571

data, and may therefore be useful in reconstructing other aspects of the phylogenetic 572

history of the seed plants, most notably the position of fossils relative to living taxa. The 573

supposed conflict between the two sorts of data on the phylogeny of seed plants seems to 574

be due to a combination of difficult problems in character analysis and limitations of 575

phylogenetic methods. Since data from the fossil record are particularly important for 576

resolving the evolutionary history of seed plants, because of the wide gaps that separate 577

extant groups and the potential biases in analysis of such sparsely sampled taxa (Burleigh 578

and Mathews 2007; Mathews 2009; Rothwell et al. 2009; Magallón et al. 2013), our results 579

give new hope for the possibility of integrating fossils and molecules in a coherent way. 580

This is even more important in light of new fossil discoveries (e.g., Rothwell and Stockey 581

2013, 2016), some of which show similarities to fossils previously associated with 582

angiosperms (e.g., the Triassic Petriellaea plant, which shares leaf and cupule features 583

with Caytonia: Bomfleur et al. 2014). 584

An important general message that emerges from our study is the importance of 585

including an exploration of the signal in all phylogenetic analyses involving morphology. 586

The overreliance on single consensus trees, as discussed in Brown et al. (2017) and 587

Puttick et al. (2017b), has been a major driver of the perceived conflict in seed plant 588

phylogeny; another factor has been the lack of support statistics in many studies. Among 589

methods of signal dissection, consensus networks and distance-based neighbor-nets 590

(Bryant and Moulton 2004) present promising avenues for the exploration of morphological 591

datasets, and have proven their power in understanding the history of different groups of 592

fossil and extant taxa at different taxonomic scales (Denk and Grimm 2009; Bomfleur et al. 593

2017; Grimm 2017). 594

Although most phylogenetic analyses based on morphology are still conducted in a 595

parsimony framework, some authors have already underlined the potential of model-based 596

approaches in this field (Lee and Worthy 2012; Lee et al. 2014). Our analyses show that 597

BI yields more robust results under different taxon sampling strategies, and is particularly 598

promising for correcting errors due to long branch effects. Our study converges with 599

previous work indicating that the use of model-based techniques could allow the 600

successful integration of taxa with a high proportion of missing data (Wiens 2005; Wiens 601

and Tiu 2012), which is a prime consideration when dealing with the paleobotanical record. 602

603

SUPPLEMENTARY MATERIAL 604

The supplementary material is available as an online appendix. 605

ACKNOWLEDGMENTS 606

MC acknowledges H. Peter Linder for his fundamental support to this work, and for 607

important comments on this manuscript. We would like to thank Richard Bateman and Gar 608

Rothwell for making their matrices available, and Omar Rota-Stabelli for useful 609

discussions of long branch attraction. Guido Grimm is thanked for thorough comments on 610

the manuscript and a detailed discussion of the use of phylogenetic networks in the 611

analysis of morphological data. Tanja Stadler, Susanne Renner, Elisabeth Truernit, Gavin 612

George, Frank Anderson, Erika Edwards, and two anonymous reviewers are gratefully 613

acknowledged for comments on a previous version of this manuscript. Guy Atchison,Yanis 614

Bouchenak-Khelladi, and Merten Ehmig are thanked for useful comments on the present 615

version. 616

REFERENCES 617

Albert, V. A., A. Backlund, K. Bremer, M. W. Chase, J. R. Manhart, B. D. Mishler, and K. C. 618

Nixon. 1994. Functional constraints and rbcL evidence for land plant phylogeny. Annals 619

of the Missouri Botanical Garden 81:534-567. 620

Axsmith, B. J., E. L. Taylor, T. N. Taylor, and N. R. Cuneo. 2000. New perspectives on the 621

Mesozoic seed fern order Corystospermales based on attached organs from the 622

Triassic of Antarctica. American Journal of Botany 87:757-768. 623

Bateman, R. M., J. Hilton, and P. J. Rudall. 2006. Morphological and molecular 624

phylogenetic context of the angiosperms: contrasting the 'top-down' and 'bottom-up' 625

approaches used to infer the likely characteristics of the first flowers. Journal of 626

Experimental Botany 57:3471-503. 627

Bauch, J., W. Liese, and R. Schultze. 1972. The morphological variability of the bordered 628

pit membranes in gymnosperms. Wood Science and Technology 6:165-184. 629

Bergsten, J. 2005. A review of long-branch attraction. Cladistics 21:163-193. 630

Bomfleur, B., A.-L. Decombeix, A. B. Schwendemann, I. H. Escapa, E. L. Taylor, T. N. 631

Taylor, and S. McLoughlin. 2014. Habit and ecology of the Petriellales, an unusual 632

group of seed plants from the Triassic of Gondwana. International Journal of Plant 633

Sciences 175:1062-1075. 634

Bomfleur, B., G. W. Grimm, and S. McLoughlin. 2017. The fossil Osmundales (Royal 635

Ferns)—a phylogenetic network analysis, revised taxonomy, and evolutionary 636

classification of anatomically preserved trunks and rhizomes. PeerJ, 5, p.e3433. 637

Bowe, L. M., C. Coat, and C. W. dePamphilis. 2000. Phylogeny of seed plants based on all 638

three genomic compartments: extant gymnosperms are monophyletic and Gnetales' 639

closest relatives are conifers. Proceedings of the National Academy of Sciences USA 640

97:4092-4097. 641

Boyce, C. K. 2005. Patterns of segregation and convergence in the evolution of fern and 642

seed plant leaf morphologies. Paleobiology 31:117–140. 643

Brinkmann, H., M. van der Giezen, Y. Zhou, G. P. de Raucourt, and H. Philippe. 2005. An 644

empirical assessment of long-branch attraction artefacts in deep eukaryotic 645

phylogenomics. Systematic Biology 54:743-757. 646

Brown, J. W., C. Parins-Fukuchi, G. W. Stull, O. M. Vargas, S. A. Smith. 2017. Bayesian 647

and likelihood phylogenetic reconstructions of morphological traits are not discordant 648

when taking uncertainty into consideration: a comment on Puttick et al. Proceedings of 649

the Royal Society of London B. 284:2017.0986. 650

Bryant, D., and V. Moulton. 2004. Neighbor-net: an agglomerative method for the 651

construction of phylogenetic networks. Molecular Biology and Evolution 21:255-265. 652

Burleigh, J. G., and S. Mathews. 2007. Assessing systematic error in the inference of seed 653

plant phylogeny. International Journal of Plant Sciences 168:125-135. 654

Carlquist, S. 1996. Wood, bark, and stem anatomy of Gnetales: a summary. International 655

Journal of Plant Sciences 157(6 Suppl.):S58-S76. 656

Cau, A., T. Brougham, and D. Naish. 2015. The phylogenetic affinities of the bizarre Late 657

Cretaceous Romanian theropod Balaur bondoc (Dinosauria, Maniraptora): 658

dromaeosaurid or flightless bird? PeerJ 3:e1032. 659

Chaw, S.-M., C. L. Parkinson, Y. Cheng, T. M. Vincent, and J. D. Palmer. 2000. Seed plant 660

phylogeny inferred from all three plant genomes: monophyly of extant gymnosperms 661

and origin of Gnetales from conifers. Proceedings of the National Academy of Sciences 662

USA 97:4086-4091. 663

Cox, C. J., B. Li, P. G. Foster, T. M. Embley, and P. Civá . 2014. Conflicting phylogenies 664

for early land plants are caused by composition biases among synonymous 665

substitutions. Systematic Biology 63:272-279. 666

Crane, P. R. 1985. Phylogenetic relationships in seed plants. Cladistics 1:329-348. 667

Crepet, W. L., and D. W. Stevenson. 2010. The Bennettitales (Cycadeoidales): a 668

preliminary perspective on this arguably enigmatic group. Pp. 215-244 in C. T. Gee, ed. 669

Plants in Mesozoic time, morphological innovations, phylogeny, ecosystems. Indiana 670

University Press, Bloomington. 671

Dembo, M., D. Radov i, H. M. Garvin, M. F. Laird, L. Schroeder, J. E. Scott, J. Brophy, R. 672

R. Ackermann, C. M. Musiba, D. J. de Ruiter, and A. Ø.Mooers. 2016. The evolutionary 673

relationships and age of Homo naledi: An assessment using dated Bayesian 674

phylogenetic methods. Journal of Human Evolution 97:17-26. 675

Denk, T. and G.W. Grimm. 2009. The biogeographic history of beech trees. Review of 676

Palaeobotany and Palynology 158:83–100. 677

Donoghue, M. J., and J. A. Doyle. 2000. Seed plant phylogeny: demise of the anthophyte 678

hypothesis? Current Biology 10:R106-R109. 679

Donoghue, M. J., J. A. Doyle, J. Gauthier, A. G. Kluge, and T. Rowe. 1989. The importance 680

of fossils in phylogeny reconstruction. Annual Review of Ecology and Systematics 681

20:431-460. 682

Doyle, J. A. 1996. Seed plant phylogeny and the relationships of the Gnetales. 683

International Journal of Plant Sciences 157(6, Suppl.):S3-S39. 684

Doyle, J. A. 2006. Seed ferns and the origin of the angiosperms. Journal of the Torrey 685

Botanical Society 133:169-209. 686

Doyle, J. A. 2008. Integrating molecular phylogenetic and paleobotanical evidence on 687

origin of the flower. International Journal of Plant Sciences 169:816-843. 688

Doyle, J. A. 2012. Molecular and fossil evidence on the origin of angiosperms. Annual 689

Review of Earth and Planetary Sciences 40:301-326. 690

Doyle, J. A. 2013. Phylogenetic analyses and morphological innovations in land plants. In 691

B. A. Ambrose and M. Purugganan, eds. The evolution of plant form. Annual Plant 692

Reviews 45:1-50. Wiley-Blackwell, Oxford. 693

Doyle, J. A., and M. J. Donoghue. 1986. Seed plant phylogeny and the origin of 694

angiosperms: an experimental cladistic approach. Botanical Review 52:321-431. 695

Doyle, J. A., and M. J. Donoghue. 1987. The importance of fossils in elucidating seed plant 696

phylogeny and macroevolution. Review of Palaeobotany and Palynology 50:63-95. 697

Doyle, J. A., and M. J. Donoghue. 1992. Fossils and seed plant phylogeny revisited. 698

Brittonia 44:89-106. 699

Doyle, J. A., and P. K. Endress. 2000. Morphological phylogenetic analysis of basal 700

angiosperms: comparison and combination with molecular data. International Journal of 701

Plant Sciences 161(Suppl.):S121-S153. 702

Endress, P. K., and J. A. Doyle. 2009. Reconstructing the ancestral angiosperm flower and 703

its initial specializations. American Journal of Botany 96:22-66. 704

Felsenstein. J. 1978. Cases in which parsimony or compatibility methods will be positively 705

misleading. Systematic Zoology 27:401-410. 706

Foley, N. M., M. S. Springer, and E. C. Teeling. 2016. Mammal madness: is the mammal 707

tree of life not yet resolved? Philosophical Transactions of the Royal Society of London 708

B 371:2015.0140. 709

Friis, E. M., P. R. Crane, K. R. Pedersen, S. Bengtson, P. J. C. Donoghue, G. W. Grimm, 710

and M. Stampanoni. 2007. Phase-contrast X-ray microtomography links Cretaceous 711

seeds with Gnetales and Bennettitales. Nature 450:549-552. 712

Friis, E. M., J. A. Doyle, P. K. Endress, and Q. Leng. 2003. Archaefructus – angiosperm 713

precursor or specialized early angiosperm? Trends in Plant Science 8:369-373. 714

Gauthier, J., A. G. Kluge, and T. Rowe. 1988. Amniote phylogeny and the importance of 715

fossils. Cladistics 4:105-209. 716

Gauthier, J., M. Kearney, J. A. Maisano, O. Rieppel, and A. D. B. Behlke. 2012. 717

Assembling the squamate tree of life: perspectives from the phenotype and the fossil 718

record. Bulletin of the Peabody Museum of Natural History 53:3–308. 719

Givnish, T. J., and K. J. Sytsma, eds. 1997. Molecular evolution and adaptive radiation. 720

Cambridge University Press, Cambridge, UK. 721

Godefroit, P., A. Cau, D.-Y. Hu, F. Escuillié, W. Wu, and G. Dyke. 2013. A Jurassic avialan 722

dinosaur from China resolves the early phylogenetic history of birds. Nature 498:359-723

362. 724

Grimm, G. 2017. Should we try to infer trees on tree-unlikely matrices? 725

http://phylonetworks.blogspot.ch/ 726

Gugerli, F., C. Sperisen, U. Biichler, I. Brunner, S. Brodbeck, J. D. Palmer, and Y.-L. Qiu. 727

2001. The evolutionary split of Pinaceae from other conifers: evidence from an intron 728

loss and a multigene phylogeny. Molecular Phylogenetics and Evolution 21:167-175. 729

Hamby, R.K., and E. A. Zimmer. 1992. Ribosomal RNA as a phylogenetic tool in plant 730

systematics. Pp. 50-91 in P. S. Soltis, D. E. Soltis, and J. J. Doyle, eds. Molecular 731

Systematics of Plants. Chapman and Hall, New York. 732

Harris, T. M. 1954. Mesozoic seed cuticles. Svensk Botanisk Tidskrift 48:281-291. 733

Hill, C.R., and P. R. Crane. 1982. Evolutionary cladistics and the origin of angiosperms. In 734

K. A. Joysey, and A. E. Friday, eds. Problems of phylogenetic reconstruction. 735

Systematics Association Special Volume 21:269-361. Academic Press, London. 736

Hilton, J., and R. M. Bateman. 2006. Pteridosperms are the backbone of seed plant 737

phylogeny. Journal of the Torrey Botanical Society 133:119-168. 738

Holland, B., K. T. Huber, V. Moulton, and P. J. Lockhart. 2004. Using consensus networks to 739

visualize contradictory evidence for species phylogeny. Molecular Biology and Evolution 740

21:1459–1461. 741

Huson, D.H., and D. Bryant. 2006. Application of phylogenetic networks in evolutionary 742

studies. Molecular Biology and Evolution 23:254-267. 743

Jenner, R. A. 2004. Accepting partnership by submission? Morphological phylogenetics in 744

a molecular millennium. Systematic Biology 53:333-359. 745

Kass, R. E., and A. E. Raftery. 1995. Bayes factors. Journal of the American Statistical 746

Association 90:773-795. 747

Kelley, D. R., and C. S. Gasser. 2009. Ovule development: genetic trends and evolutionary 748

considerations. Sexual Plant Reproduction 22:229-234. 749

Klavins, S. D., T. N. Taylor, and E. L. Taylor. 2002. Anatomy of Umkomasia 750

(Corystospermales) from the Triassic of Antarctica. American Journal of Botany 89:664-751

676. 752

Lee, M. S., and A. Palci. 2015. Morphological phylogenetics in the genomic age. Current 753

Biology 25:R922-R929. 754

Lee, M. S., and T. H. Worthy. 2012. Likelihood reinstates Archaeopteryx as a primitive bird. 755

Biology Letters 8:299-303. 756

Legg, D. A., M. D. Sutton, and G. D. Edgecombe. 2013. Arthropod fossil data increase 757

congruence of morphological and molecular phylogenies. Nature Communications 758

4:2485. 759

Lewis, P. O. 2001. A likelihood approach to estimating phylogeny from discrete 760

morphological character data. Systematic Biology 50:913-925. 761

Lloyd, G. T. 2016. Estimating morphological diversity and tempo with discrete 762

character taxon matrices: implementation, challenges, progress, and future directions. 763

Biological Journal of the Linnean Society 118:131-151. 764

Lockhart, P. J., and S. A. Cameron. 2001. Trees for bees. Trends in Ecology & Evolution 765

16:84-88. 766

Maddison, D. R., and W. P. Maddison. 2003. MacClade 4: analysis of phylogeny and 767

character evolution, version 4.06. Sinauer Associates, Sunderland, MA. 768

Magallón, S., and M. J. Sanderson. 2002. Relationships among seed plants inferred from 769

highly conserved genes: sorting conflicting phylogenetic signals among ancient lineages. 770

American Journal of Botany 89:1991-2006. 771

Magallón, S., K. W. Hilu, and D. Quandt. 2013. Land plant evolutionary timeline: gene 772

effects are secondary to fossil constraints in relaxed clock estimation of age and 773

substitution rates. American Journal of Botany 100:556-573. 774

Martens, P. 1971. Les Gnétophytes. Encyclopedia of Plant Anatomy, 12(2). Borntraeger, 775

Stuttgart. 776

Mathews, S. 2009. Phylogenetic relationships among seed plants: persistent questions 777

and the limits of molecular data. American Journal of Botany 96:228-236. 778

Mathews, S., M. D. Clements, and M. A. Beilstein. 2010. A duplicate gene rooting of seed 779

plants and the phylogenetic position of flowering plants. Philosophical Transactions of 780

the Royal Society of London B 365:383-395. 781

Mathews, S., and E. Kramer. 2012. The evolution of reproductive structures in seed plants: 782

a re-examination based on insights from developmental genetics. New Phytologist 783

194:910–923. 784

Müller, K. F. 2005. The efficiency of different search strategies for estimating parsimony, 785

jackknife, bootstrap, and Bremer support. BMC Evolutionary Biology 2005;5:58. 786

Mundry, M., and T. Stützel. 2004. Morphogenesis of the reproductive shoots of 787

Welwitschia mirabilis and Ephedra distachya (Gnetales), and its evolutionary 788

implications. Organisms Diversity & Evolution 4:91-108. 789

Nickrent, D. L., C. L. Parkinson, J. D. Palmer, and R. J. Duff. 2000. Multigene phylogeny of 790

land plants with special reference to bryophytes and the earliest land plants. Molecular 791

Biology and Evolution 17:1885-1895. 792

Nixon, K. C., W. L. Crepet, D. W. Stevenson, and E. M. Friis. 1994. A reevaluation of seed 793

plant phylogeny. Annals of the Missouri Botanical Garden 81:484-533. 794

O'Leary, M. A., J. I. Bloch, J. J. Flynn, T. J. Gaudin, A. Giallombardo, N. P. Giannini, S. L. 795

Goldberg, B. P. Kraatz, Z.-X. Luo, J. Meng, X. Ni, M. J. Novacek, F. A. Perini, Z. S. 796

Randall, G. W. Rougier, E. J. Sargis, M. T. Silcox, N. B. Simmons, M. Spaulding, P. M. 797

Velazco, M. Weksler, J. R. Wible, and A. L. Cirranello. 2013. The placental mammal 798

ancestor and the post–K-Pg radiation of placentals. Science 339:662-667. 799

O’Reilly, J. E., M. N. Puttick, L. Parry, A. R. Tanner, J. E. Tarver, J. Fleming, D. Pisani, and 800

P. C. J. Donoghue. 2016. Bayesian methods outperform parsimony but at the expense 801

of precision in the estimation of phylogeny from discrete morphological data. Biology 802

Letters 12:2016.0081. 803

Patterson, C. 1981. Significance of fossils in determining evolutionary relationships. 804

Annual Review of Ecology and Systematics 12:195-223. 805

Pott, C. 2016. Westersheimia pramelreuthensis from the Carnian (Upper Triassic) of Lunz, 806

Austria: more evidence for a unitegmic seed coat in early Bennettitales. International 807

Journal of Plant Sciences 177:771–791. 808

Puttick, M. N., J. E. O'Reilly, A. R. Tanner, J. F. Fleming, J. Clark, L. Holloway, J. Lozano-809

Fernandez, L. A. Parry, J. E. Tarver, D. Pisani, and P. C. J. Donoghue. 2017a. 810

Uncertain-tree: discriminating among competing approaches to the phylogenetic 811

analysis of phenotype data. Proceedings of the Royal Society of London B 812

284:2016.2290. 813

Puttick, M. N., J. E. O'Reilly, D. Oakley, A. R. Tanner, J. F. Fleming, J. Clark, L. Holloway, J. 814

Lozano-Fernandez, L. A. Parry, J. E. Tarver, D. Pisani, P. C. J. Donoghue. 2017b. 815

Parsimony and maximum-likelihood phylogenetic analyses of morphology do not 816

generally integrate uncertainty in inferring evolutionary history: a response to Brown et 817

al. Proceedings of the Royal Society of London B. 284:2017.1636 818

Pyron, R. A. 2011. Divergence time estimation using fossils as terminal taxa and the 819

origins of Lissamphibia. Systematic Biology 60:466-481. 820

Pyron, R. A. 2015. Post-molecular systematics and the future of phylogenetics. Trends in 821

Ecology & Evolution 30:384-389. 822

Qiu, Y.-L., L. Li, B. Wang, Z. Chen, O. Dombrovska, J. Lee, L. Kent, L. Li, R. W. Jobson, T. 823

A. Hendry, D. W. Taylor, C. M. Testa, and M. Ambros. 2007. A nonflowering land plant 824

phylogeny inferred from nucleotide sequences of seven chloroplast, mitochondrial, and 825

nuclear genes. International Journal of Plant Sciences 168:691-708. 826

Rambaut, A., and A. J. Drummond. 2007. Tracer, version 1.5. 827

Ronquist, F., S. Klopfstein, L. Vilhelmsen, S. Schulmeister, D. L. Murray, and A. P. 828

Rasnitsyn. 2012. A total-evidence approach to dating with fossils, applied to the early 829

radiation of the Hymenoptera. Systematic Biology 61:973-999. 830

Rota-Stabelli, O., E. Kayal, D. Gleeson, J. Daub, J. L. Boore, M. J. Telford, D. Pisani, M. 831

Blaxter, and D. V. Lavrov. 2010. Ecdysozoan mitogenomics: evidence for a common 832

origin of the legged invertebrates, the Panarthropoda. Genome Biology and Evolution 833

2:425-440. 834

Rothwell, G. W., W. L. Crepet, and R. A. Stockey. 2009. Is the anthophyte hypothesis alive 835

and well? New evidence from the reproductive structures of Bennettitales. American 836

Journal of Botany 96:296-322. 837

Rothwell, G. W., and R. Serbet. 1994. Lignophyte phylogeny and the evolution of 838

spermatophytes: a numerical cladistic analysis. Systematic Botany 19:443-482. 839

Rothwell, G. W., and R. A. Stockey. 2013. Evolution and phylogeny of Gnetophytes: 840

evidence from the anatomically preserved seed cone Protoephedrites eamesii gen. et 841

sp. nov. and the seeds of several bennettitalean species. International Journal of Plant 842

Sciences 174:511-529. 843

Rothwell, G. W., and R. A. Stockey. 2016. Phylogenetic diversification of Early Cretaceous 844

seed plants: the compound seed cone of Doylea tetrahedrasperma. American Journal 845

of Botany 103:923-937. 846

Rudall, P. J., and R. M. Bateman. 2010. Defining the limits of flowers: the challenge of 847

distinguishing between the evolutionary products of simple versus compound strobili. 848

Philosophical Transactions of the Royal Society of London B 365:397-409. 849

Rydin, C., and M. Källersjö. 2002. Taxon sampling and seed plant phylogeny. Cladistics 850

18:484-513. 851

Rydin, C., and M. Källersjö, and E. M. Friis E.M. 2002. Seed plant relationships and the 852

systematic position of Gnetales based on nuclear and chloroplast DNA: conflicting data, 853

rooting problems, and the monophyly of conifers. International Journal of Plant 854

Sciences 163:197-214. 855

Sanderson, M. J., M. F. Wojciechowski, J.-M. Hu, T. Sher Khan, and S. G. Brady. 2000. 856

Error, bias, and long-branch attraction in data for two chloroplast photosystem genes in 857

seed plants. Molecular Biology and Evolution 17:782-797. 858

Scotland, R. W., R. G. Olmstead, and J. R. Bennett. 2003. Phylogeny reconstruction: the 859

role of morphology. Systematic Biology 52:539-548. 860

Singh, H. 1978. Embryology of gymnosperms. Handbuch der Pflanzenanatomie 10(2). 861

Borntraeger, Berlin. 862

Springer, M. S., A. Burk-Herrick, R. Meredith, E. Eizirik, E. Teeling, S. J. O'Brien, and W. J. 863

Murphy. 2007. The adequacy of morphology for reconstructing the early history of 864

placental mammals. Systematic Biology 56:673-684. 865

Springer, M. S., R. W. Meredith, E. C. Teeling, and W. J. Murphy. 2013. Technical 866

comment on “The placental mammal ancestor and the post–K-Pg radiation of 867

placentals”. Science 341:613. 868

A. Stamatakis. 2014. RAxML version 8: a tool for phylogenetic analysis and post-analysis 869

of large phylogenies. Bioinformatics 30:1312-1313. 870

Stefanovic, S., M. Jager, J. Deutsch, J. Broutin, and M. Masselot. 1998. Phylogenetic 871

relationships of conifers inferred from partial 28S rRNA gene sequences. American 872

Journal of Botany 85:688-697. 873

Stockey, R. A., and G. W. Rothwell. 2003. Anatomically preserved Williamsonia 874

(Williamsoniaceae): evidence for bennettitalean reproduction in the Late Cretaceous of 875

western North America. International Journal of Plant Sciences 164:251-262. 876

Stockey, R. A., and G. W. Rothwell. 2009. Distinguishing angiophytes from the earliest 877

angiosperms: A Lower Cretaceous (Valanginian-Hauterivian) fruit-like reproductive 878

structure. American Journal of Botany 96:323-335. 879

Sun, G., Q. Ji, D. L. Dilcher, S. Zheng, K. C. Nixon, and X. Wang. 2002. Archaefructaceae, 880

a new basal angiosperm family. Science 296:899-904. 881

Swofford, D. L. 2003. PAUP*. Phylogenetic Analysis Using Parsimony (* and other 882

methods), version 4. 883

Swofford, D. L., P. J. Waddell, J. P. Huelsenbeck, P. G. Foster, P. O. Lewis, and J. S. 884

Rogers. 2001. Bias in phylogenetic estimation and its relevance to the choice between 885

parsimony and likelihood methods. Systematic Biology 50:525–539. 886

Taylor, E. L., and T. N. Taylor. 1992. Reproductive biology of the Permian Glossopteridales 887

and their suggested relationship to flowering plants. Proceedings of the National 888

Academy of Sciences USA 89:11495-11497. 889

Taylor, T. N., G. M. Del Fueyo, and E. L. Taylor. 1994. Permineralized seed fern cupules 890

from the Triassic of Antarctica: implications for cupule and carpel evolution. American 891

Journal of Botany 81:666-677. 892

Templeton, A. R. 1983. Phylogenetic inference from restriction endonuclease cleavage site 893

maps with particular reference to the evolution of humans and the apes. Evolution 894

37:221-244. 895

Wickett, N. J., S. Mirarab, N. Nguyen, T. Warnow, E. Carpenter, N. Matasci, S. 896

Ayyampalayam, M. S. Barker, J. G. Burleigh, M. A. Gitzendanner, B. R. Ruhfel, E. 897

Wafula, J. P. Der, S. W. Graham, S. Mathews, M. Melkonian, D. E. Soltis, P. S. Soltis, N. 898

W. Miles, C. J. Rothfels, L. Pokorny, A. J. Shaw, L. DeGironimo, D. W. Stevenson, B. 899

Surek, J. C. Villarreal, B. Roure, H. Philippe, C. W. dePamphilis, T. Chen, M. K. 900

Deyholos, R. S. Baucom, T. M. Kutchan, M. M. Augustin, J. Wang, Y. Zhang, Z. Tian, Z. 901

Yan, X. Wu, X. Sun, G. K. S. Wong, and J. Leebens-Mack. 2014. Phylotranscriptomic 902

analysis of the origin and early diversification of land plants. Proceedings of the National 903

Academy of Sciences USA 111:E4859–E4868. 904

Wieland, G. R. 1916. American fossil cycads. Vol. 2. Taxonomy. Carnegie Institution of 905

Washington, Washington, D.C. 906

Wiens, J. J., P. T. Chippindale, and D. M. Hillis. 2003. When are phylogenetic analyses 907

misled by convergence? A case study in Texas cave salamanders. Systematic Biology 908

52:501-514. 909

Wiens, J. J. 2005. Can incomplete taxa rescue phylogenetic analyses from long-branch 910

attraction? Systematic Biology 54:731-742. 911

Wiens, J. J., and B. D. Hollingsworth. 2000. War of the iguanas: conflicting phylogenies, 912

long-branch attraction, and disparate rates of molecular and morphological evolution in 913

iguanid lizards. Systematic Biology 49:69-85. 914

Wiens, J. J., and J. Tiu. 2012. Highly incomplete taxa can rescue phylogenetic analyses 915

from the negative impacts of limited taxon sampling. PloS One 7:e42925. 916

Wright, A. M., D. M. Hills. 2014. Bayesian analysis using a simple likelihood model 917

outperforms parsimony for estimation of phylogeny from discrete morphological data. 918

PLoS One 9:e109210. 919

Wright, A. M., G. T. Lloyd, and D. M. Hillis. 2015. Modeling character change heterogeneity 920

in phylogenetic analyses of morphology through the use of priors. Systematic Biology 921

65:602-611. 922

Xie, W., P. O. Lewis, Y. Fan, L. Kuo, and M.-H. Chen. 2011. Improving marginal likelihood 923

estimation for Bayesian phylogenetic model selection. Systematic Biology 60:150-160. 924

Zander, R. H. 2004. Minimal values of reliability of bootstrap and jackknife proportions, 925

Decay Index, and Bayesian posterior probability. PhyloInformatics 2:1-13. 926

Zhang, C., T. Stadler, S. Klopfstein, T. A. Heath, and F. Ronquist. 2016. Total-evidence 927

dating under the fossilized birth–death process. Systematic Biology 65:228-249. 928

Zhong, B., O. Deusch, V. V. Goremkin, D. Penny, P. J. Briggs, R. A. Atherton, S. V. 929

Nikiforova, and P. J. Lockhart. 2011. Systematic error in seed plant phylogenomics. 930

Genome Biology and Evolution 3:1340-1348. 931

Zou, Z., and J. Zhang. 2016. Morphological and molecular convergences in mammalian 932

phylogenetics. Nature Communications 7: 12758. 933

934

Figure 1. A, Relationships among extant seed plants. On the left an anthophyte topology, 935

and on the right a gneconifer topology. Relationships between Cycadales and Ginkgo 936

vary among analyses of both sorts. B, Relationships among the matrices reanalyzed in 937

this paper. 938

Figure 2. Support for the anthophytes or gneconifers in the different matrices and using 939

different methods. A, Results from the MP bootstrap analyses; B, results from the ML 940

bootstrap analyses; C, results from the BI analyses. The difference between the pre-941

2000 and post-2000 matrices is clearly underlined by a shift in support from 942

anthophytes to gneconifers in the ML and BI analyses, and a drop in support for the 943

anthophytes in the MP analyses. 944

Figure 3. Split network consensus of A, the posterior tree sample of the MP bootstrap 945

analysis, B, the ML bootsrap analysis, and C, the BI analysis of the Rothwell and 946

Stockey (2016) matrix using gamma-distributed rate variation. Only splits with more 947

than 0.15 PP or 15% boostrap are shown, and support is shown only for splits with 948

more than 0.20 PP or 20% bootstrap. The support for the splits within the angiosperms 949

has been removed for clarity. If the Gnetales-conifer clade is present and supported 950

with all three methods, other relationships (i.e. Caytonia in a clade with angiosperms) 951

are only supported in the BI analysis. 952

Figure 4. A-C, Scheme of the long branch attraction tests; A and B represent the long 953

branch extraction experiment, C represents the branch elongation experiment. Null 954

hypotheses are in the right upper corner. D-F, Results of the LBE experiment on the 955

Rothwell et al. (2009) matrix. All trees are MP consensus trees. Fossil taxa diverging 956

below the most recent common ancestor of extant seed plants removed for ease of 957

comparison. D, Untrimmed matrix, showing an anthophyte topology and paraphyletic 958

conifers. E, Angiosperm removal matrix, showing Gnetales nested in the conifers and 959

other anthophytes removed from the coniferophyte clade. F, Gnetales removal matrix, 960

with monophyletic conifers nested in a large coniferophyte clade. G-I, Results of the BE 961

and EX experiments. G, Results of the MP analyses. H-I, Results of the BI analyses 962

under the Markov k-states (Mk) model with H, equal rates and I, gamma-distributed rate 963

variation. 964

Figure 5. Plot of the first two principal coordinate axes for four of the matrices analyzed. 965

The first PCO axis mainly separates the angiosperms and the other seed plants, while 966

the second PCO axis separates more conifer-like and more fern-like groups. These 967

plots illustrate the effect of the reassessment of gnetalean characters between the two 968

Doyle matrices (A, B), and the similar structure of the data in the Hilton and Bateman 969

(2006) (C) and Rothwell and Stockey (2016) (D) matrices. 970

971

Table 1. Statistics for the maximum parsimony analyses of fossil matrices. 972

Number of

trees

Length Ci Ri

Crane 1985

Doyle and Donoghue

1986

Doyle and Donoghue

1992

Nixon et al. 1994

Rothwell and Serbet

1994

Doyle 1996

Hilton and

Bateman

2006

Doyle 2006

Doyle 2008

Rothwell et al. 2009

Rothwell and Stockey

2016

973

974

975

976

Table 2: Results from the MP analysis of constrained gneconifer trees 977

Length

unconstrained

Length

Gnetales+Conifer

Length

difference

Templeton

Test p-

value

(best

value)

Crane

1985

Doyle and

Donoghue

1986

Doyle and

Donoghue

1992

Nixon et

al. 1994

Rothwell

and

Serbet

1994

Doyle

1996

Hilton and

Bateman

2006

Doyle

2006

Doyle

2008

Rothwell

et al. 2009

Rothwell

and

Stockey

2016

978

979

Table 3. Model-testing statistics for the Bayesian inference analyses. 980

Mkprinf Mk

prinf

lnBF 2xlnBF

Crane 1985

-223.03

-223.01

0.02

0.04

Doyle and Donoghue

1986

-473.68

-473.70

-0.02

-0.04

Doyle and Donoghue

1992

-432.38

-431.00

1.38

2.76

Rothwell and Serbet

1994

-861.53

-854.14

7.39

14.78

Nixon et al. 1994

-1555.76

-1538.27

17.49

34.98

Doyle 2006

-1383.60

-1365.27

18.33

36.66

Hilton and Bateman

2006

-1559.87

-1532.70

27.17

54.34

Doyle 2008

-1481.46

-1455.09

26.37

52.74

Rothwell et al. 2009

-1541.68

-1527.09

14.59

29.18

Rothwell and Stockey

2016

-1511.73

-1493.78

17.95

35.90

981

982

983

984

Key questions and challenges in angiosperm macroevolution

Article

Full-text available

Mar 2018

Contents Summary 1170 I. Introduction 1170 II. Six key questions 1172 III. Three key challenges 1177 IV. Conclusions 1181 Acknowledgements 1182 References 1183 Summary The origin and rapid diversification of angiosperms (flowering plants) represent one of the most intriguing topics in evolutionary biology. Despite considerable progress made in complementary fields over the last two decades (paleobotany, phylogenetics, ecology, evo‐devo, genomics), many important questions remain. For instance, what has been the impact of mass extinctions on angiosperm diversification? Are the angiosperms an adaptive radiation? Has morphological evolution in angiosperms been gradual or pulsed? We propose that the recent and ongoing revolution in macroevolutionary methods provides an unprecedented opportunity to explore long‐standing questions that probably hold important clues to understand present‐day biodiversity. We present six key questions that explore the origin and diversification of angiosperms. We also identify three key challenges to address these questions: (1) the development of new integrative models that include diversification, multiple intrinsic and environmental traits, biogeography and the fossil record all at once, whilst accounting for sampling bias and heterogeneity of macroevolutionary processes through time and among lineages; (2) the need for large and standardized synthetic databases of morphological variation; and (3) continuous effort on sampling the fossil record, but with a revolution in current paleobotanical practice.

Origin of Equisetum: Evolution of horsetails (Equisetales) within the major euphyllophyte clade Sphenopsida

Article

Full-text available

Jul 2018

Premise of the Study Equisetum is the sole living representative of Sphenopsida, a clade with impressive species richness, a long fossil history dating back to the Devonian, and obscure relationships with other living pteridophytes. Based on molecular data, the crown group age of Equisetum is mid‐Paleogene, although fossils with possible crown synapomorphies appear in the Triassic. The most widely circulated hypothesis states that the lineage of Equisetum derives from calamitaceans, but no comprehensive phylogenetic studies support the claim. Using a combined approach, we provide a comprehensive phylogenetic analysis of Equisetales, with special emphasis on the origin of genus Equisetum. Methods We performed parsimony phylogenetic analyses to address relationships of 43 equisetalean species (15 extant, 28 extinct) using a combination of morphological and molecular characters. Key Results We recovered Equisetaceae + Neocalamites as sister to Calamitaceae + a clade of Angaran and Gondwanan horsetails, with the four groups forming a clade that is sister to Archaeocalamitaceae. The estimated age for the Equisetum crown group is mid‐Mesozoic. Conclusions Modern horsetails are not nested within calamitaceans; instead, both groups have explored independent evolutionary trajectories since the Carboniferous. Diverse fossil taxon sampling helps to shed light on the position and relationships of equisetalean lineages, of which only a tiny remnant is present within the extant flora. Understanding these relationships and early character configurations of ancient plant clades as Equisetales provide useful tests of hypotheses about overall phylogenetic relationships of euphyllophytes and foundations for future tests of molecular dates with paleontological data.

The fossil Osmundales (Royal Ferns) - A phylogenetic network analysis, revised taxonomy, and evolutionary classification of anatomically preserved trunks and rhizomes

Article

Full-text available

Jul 2017

The Osmundales (Royal Fern order) originated in the late Paleozoic and is the most ancient surviving lineage of leptosporangiate ferns. In contrast to its low diversity today (less than 20 species in six genera), it has the richest fossil record of any extant group of ferns. The structurally preserved trunks and rhizomes alone are referable to more than 100 fossil species that are classified in up to 20 genera, four subfamilies, and two families. This diverse fossil record constitutes an exceptional source of information on the evolutionary history of the group from the Permian to the present. However, inconsistent terminology, varying formats of description, and the general lack of a uniform taxonomic concept renders this wealth of information poorly accessible. To this end, we provide a comprehensive review of the diversity of structural features of osmundalean axes under a standardized, descriptive terminology. A novel morphological character matrix with 45 anatomical characters scored for 15 extant species and for 114 fossil operational units (species or specimens) is analysed using networks in order to establish systematic relationships among fossil and extant Osmundales rooted in axis anatomy. The results lead us to propose an evolutionary classification for fossil Osmundales and a revised, standardized taxonomy for all taxa down to the rank of (sub)genus. We introduce several nomenclatural novelties: (1) a new subfamily Itopsidemoideae (Guaireaceae) is established to contain Itopsidema, Donwelliacaulis, and Tiania; (2) the thamnopteroid genera Zalesskya, Iegosigopteris, and Petcheropteris are all considered synonymous with Thamnopteris; (3) 12 species of Millerocaulis and Ashicaulis are assigned to modern genera (tribe Osmundeae); (4) the hitherto enigmatic Aurealcaulis is identified as an extinct subgenus of Plenasium; and (5) the poorly known Osmundites tuhajkulensis is assigned to Millerocaulis. In addition, we consider Millerocaulis stipabonettiorum a possible member of Palaeosmunda and Millerocaulis estipularis as probably constituting the earliest representative of the (Todea-)Leptopteris lineage (subtribe Todeinae) of modern Osmundoideae.

Uncertain-tree: Discriminating among competing approaches to the phylogenetic analysis of phenotype data

Article

Full-text available

Jan 2017
Proc Biol Sci

Morphological data provide the only means of classifying the majority of life's history, but the choice between competing phylogenetic methods for the analysis of morphology is unclear. Traditionally, parsimony methods have been favoured but recent studies have shown that these approaches are less accurate than the Bayesian implementation of the Mk model. Here we expand on these findings in several ways: we assess the impact of tree shape and maximum-likelihood estimation using the Mk model, as well as analysing data composed of both binary and multistate characters. We find that all methods struggle to correctly resolve deep clades within asymmetric trees, and when analysing small character matrices. The Bayesian Mk model is the most accurate method for estimating topology, but with lower resolution than other methods. Equal weights parsimony is more accurate than implied weights parsimony, and maximum-likelihood estimation using the Mk model is the least accurate method. We conclude that the Bayesian implementation of the Mk model should be the default method for phylogenetic estimation from phenotype datasets, and we explore the implications of our simulations in rea-nalysing several empirical morphological character matrices. A consequence of our finding is that high levels of resolution or the ability to classify species or groups with much confidence should not be expected when using small datasets. It is now necessary to depart from the traditional parsimony paradigms of constructing character matrices, towards datasets constructed explicitly for Bayesian methods.

Westersheimia pramelreuthensis from the Carnian (Upper Triassic) of Lunz, Austria: More Evidence for a Unitegmic Seed Coat in Early Bennettitales

Article

Full-text available

Oct 2016

Christian Pott

Premise of research. Isolated ovules and dispersed seeds have been obtained from bulk macerations of carbonaceous shales bearing fossil plants from the Carnian (Upper Triassic) flora of Lunz am See, Austria. Through cuticle analysis, these well-preserved specimens were identified as ovules, interseminal scales, and seeds of Westersheimia pramelreuthensis, a peculiar bennettitalean ovuliferous organ. The preservation of several cuticular layers enables for interpretation of the architecture of the ovules and seeds.Methodology. The excellently preserved plant fossils were investigated using LM and epifluorescence microscopy of cuticles. For comparison, the architecture of ovules and seeds of bennettitalean reproductive organs from Scania (Sweden), Jameson Land (Greenland), and Yorkshire (United Kingdom) were reevaluated.Pivotal results. The preserved layers indicate that the nucellus is surrounded by an integument, whose apical end constitutes the micropyle. The single integument constitutes the monolayered seed coat in Westersheimia. The ovules and seeds are surrounded by interseminal scales. Of the latter, the cuticle is preserved that abuts the ovule-seed surface, together with portions of the interseminal scale heads. The seeds provide additional information on the cuticles of the nucellus or embryo. Conclusions. Studies of bennettitalean reproductive organs from Scania (Sweden), Jameson Land (Greenland), and Yorkshire (United Kingdom) facilitated reevaluation of the disputed architecture of bennettitalean seeds. The findings clearly indicate that early bennettitalean seeds can be interpreted as unitegmic.

Morphological and molecular convergences in mammalian phylogenetics

Article

Full-text available

Sep 2016

Phylogenetic trees reconstructed from molecular sequences are often considered more reliable than those reconstructed from morphological characters, in part because convergent evolution, which confounds phylogenetic reconstruction, is believed to be rarer for molecular sequences than for morphologies. However, neither the validity of this belief nor its underlying cause is known. Here comparing thousands of characters of each type that have been used for inferring the phylogeny of mammals, we find that on average morphological characters indeed experience much more convergences than amino acid sites, but this disparity is explained by fewer states per character rather than an intrinsically higher susceptibility to convergence for morphologies than sequences. We show by computer simulation and actual data analysis that a simple method for identifying and removing convergence-prone characters improves phylogenetic accuracy, potentially enabling, when necessary, the inclusion of morphologies and hence fossils for reliable tree inference.

Mammal madness: Is the mammal tree of life not yet resolved?

Article

Full-text available

Jul 2016
PHILOS T R SOC B

Most molecular phylogenetic studies place all placental mammals into four superordinal groups, Laurasiatheria (e.g. dogs, bats, whales), Euarchontoglires (e.g. humans, rodents, colugos), Xenarthra (e.g. armadillos, anteaters) and Afrotheria (e.g. elephants, sea cows, tenrecs), and estimate that these clades last shared a common ancestor 90–110 million years ago. This phylogeny has provided a framework for numerous functional and comparative studies. Despite the high level of congruence among most molecular studies, questions still remain regarding the position and divergence time of the root of placental mammals, and certain ‘hard nodes’ such as the Laurasiatheria polytomy and Paenungulata that seem impossible to resolve. Here, we explore recent consensus and conflict among mammalian phylogenetic studies and explore the reasons for the remaining conflicts. The question of whether the mammal tree of life is or can be ever resolved is also addressed. This article is part of the themed issue ‘Dating species divergences using rocks and clocks’.

Bayesian methods outperform parsimony but at the expense of precision in the estimation of phylogeny from discrete morphological data

Article

Full-text available

Apr 2016
BIOL LETTERS

Different analytical methods can yield competing interpretations of evolutionary history and, currently, there is no definitive method for phylogenetic reconstruction using morphological data. Parsimony has been the primary method for analysing morphological data, but there has been a resurgence of interest in the likelihood-based Mk-model. Here, we test the perfounance of the Bayesian implementation of the Mk-model relative to both equal and implied-weight implementations of parsimony. Using simulated morphological data, we demonstrate that the Mk-model outperforms equal-weights parsimony in terms of topological accuracy, and implied-weights performs the most poorly. However, the Mk-model produces phylogenies that have less resolution than parsimony methods. This difference in the accuracy and precision of parsimony and Bayesian approaches to topology estimation needs to be considered when selecting a method for phylogeny reconstruction.

Molecular Evolution and Adaptive Radiation.

Article

Apr 1998

PHYLOGENETIC INFERENCE FROM RESTRICTION ENDONUCLEASE CLEAVAGE SITE MAPS WITH PARTICULAR REFERENCE TO THE EVOLUTION OF HUMANS AND THE APES

Article

Mar 1983
EVOLUTION

Alan Templeton

Bayes factors

Article

Jan 1995

Phylogenetic diversification of Early Cretaceous seed plants: The compound seed cone of Doylea tetrahedrasperma

Article

May 2016

Premise of the study: Discovery of cupulate ovules of Doylea tetrahedrasperma within a compact, compound seed cone highlights the rich diversity of fructification morphologies, pollination biologies, postpollination enclosure of seeds, and systematic diversity of Early Cretaceous gymnosperms. Methods: Specimens were studied using the cellulose acetate peel technique, three-dimensional reconstructions (in AVIZO), and morphological phylogenetic analyses (in TNT). Key results: Doylea tetrahedrasperma has bract/fertile short shoot complexes helically arranged within a compact, compound seed cone. Complexes diverge from the axis as a single unit and separate distally into a free bract tip and two sporophylls. Each sporophyll bears a single, abaxial seed, recurved toward the cone axis, that is enveloped after pollinaton by sporophyll tissue, forming a closed cupule. Ovules are pollinated by bisaccate grains captured by micropylar pollination horns. Conclusions: The unique combination of characters shown by D. tetrahedrasperma includes the presence of cupulate seeds borne in conifer-like compound seed cones, an ovuliferous scale analogue structurally equivalent to the ovulate stalk of Ginkgo biloba, gymnospermous pollination, and nearly complete enclosure of mature seeds. These features characterize the Doyleales ord. nov., clearly distinguish it from the seed fern order Corystospermales, and allow for recognition of another recently described Early Cretaceous seed plant as a second species in genus Doylea. A morphological phylogenetic analysis highlights systematic relationships of the Doyleales ord. nov. and emphasizes the explosive phylogenetic diversification of gymnosperms that was underway at the time when flowering plants may have originated and/or first began to radiate.

Experimental signal dissection and method sensitivity analyses reaffirm the potential of fossils and morphology in the resolution of seed plant phylogeny

Abstract and Figures

Recommended publications

Research to empower and inspire from Durham University Business School

Research to empower and inspire from the Science Faculty at Durham University

Research to empower and inspire from the Social Sciences Faculty at Durham University

Research to empower and inspire from the Arts and Humanities Faculty at Durham University

Experimental signal dissection and method sensitivity analyses reaffirm the potential of fossils and...

Integrating Molecular Phylogenetic and Paleobotanical Evidence on Origin of the Flower

Mesozoic cupules and the origin of the angiosperm second integument

Seed Ferns and the Origin of Angiosperms