University of ZurichZurich Open Repository and Archive

Winterthurerstr. 190

CH-8057 Zurich

http://www.zora.uzh.ch

Year: 2011

Absence of mammals and the evolution of New Zealand grasses

Antonelli, A; Humphreys, A M; Lee, W G; Linder, H P

Antonelli, A; Humphreys, A M; Lee, W G; Linder, H P (2011). Absence of mammals and the evolution of NewZealand grasses. Proceedings of the Royal Society. Series B, Biological Sciences, 278(1706):695-701.Postprint available at:http://www.zora.uzh.ch

Posted at the Zurich Open Repository and Archive, University of Zurich.http://www.zora.uzh.ch

Originally published at:Antonelli, A; Humphreys, A M; Lee, W G; Linder, H P (2011). Absence of mammals and the evolution of NewZealand grasses. Proceedings of the Royal Society. Series B, Biological Sciences, 278(1706):695-701.

Antonelli, A; Humphreys, A M; Lee, W G; Linder, H P (2011). Absence of mammals and the evolution of NewZealand grasses. Proceedings of the Royal Society. Series B, Biological Sciences, 278(1706):695-701.Postprint available at:http://www.zora.uzh.ch

Posted at the Zurich Open Repository and Archive, University of Zurich.http://www.zora.uzh.ch

Originally published at:Antonelli, A; Humphreys, A M; Lee, W G; Linder, H P (2011). Absence of mammals and the evolution of NewZealand grasses. Proceedings of the Royal Society. Series B, Biological Sciences, 278(1706):695-701.

Absence of mammals and the evolution of New Zealand grasses

Abstract

Anthropogenic alteration of biotic distributions and disturbance regimes has dramatically changed theevolutionary context for the differentiation of species traits. Some of the most striking examples inrecent centuries have been on islands where flightless birds, which evolved in the absence ofmammalian carnivores, have been decimated following the widespread introduction of exotic predators.Until now, no equivalent case has been reported for plants. Here, we make use of robust analytical toolsand an exceptionally well-sampled molecular phylogeny to show that a majority of New Zealanddanthonioid grasses (Poaceae) may have adapted to the relaxed vertebrate herbivore pressure during thelate Cenozoic through the development of a distinctive and unusual habit: abscission of old leaves. Thisfeature occurs in only about 3 per cent of the world's roughly 11 000 grass species and has beenempirically shown to increase plant productivity but to reduce protection against mammal grazing. Thisresult suggests that release from a selective pressure can lead to species radiations. This seeminglyanachronistic adaptation may represent an overlooked factor contributing to the severe decline in thegeographical extent and species diversity of New Zealand's indigenous grasslands following theintroduction of herbivorous terrestrial mammals in the 19th century.

1

Absence of mammals and the evolution of New Zealand 1

grasses 2

Alexandre Antonelli1,3,*

, Aelys M. Humphreys1, William G. Lee

2 & H. Peter Linder

1 3

1Institute of Systematic Botany, University of Zurich, Zollikerstrasse 107, CH 8008, 4

Zurich, Switzerland. 2Landcare Research, Private Bag 1930, Dunedin 9016, New 5

Zealand. 3Current address: Gothenburg Botanical Garden, Carl Skottsbergs gata 22A, 6

413 19, Göteborg, Sweden. 7

*Corresponding author (E-mail: alexandre.antonelli@vgregion.se) 8

9

Running Head: „Ghost‟ adaptation in New Zealand 10

11

SUMMARY 12

Anthropogenic alteration of biotic distributions and disturbance regimes has 13

dramatically changed the evolutionary context for the differentiation of species traits. 14

Some of the most striking examples in recent centuries have been on islands where 15

flightless birds, which evolved in the absence of mammalian carnivores, have been 16

decimated following the widespread introduction of exotic predators. Until now, no 17

equivalent case has been reported for plants. Here we make use of robust analytical 18

tools and an exceptionally well sampled molecular phylogeny to show that a majority of 19

New Zealand danthonioid grasses (Poaceae) may have adapted to the relaxed vertebrate 20

herbivore pressure during the late Cenozoic through the development of a distinctive 21

2

and unusual habit: abscission of old leaves. This feature occurs in only about 3% of the 22

world‟s ca. 11,000 grass species, and has been empirically shown to increase plant 23

productivity but to reduce protection against mammal grazing. This result suggests that 24

release from a selective pressure can lead to species radiations. This seemingly 25

anachronistic adaptation may represent an overlooked factor contributing to the severe 26

decline in the geographical extent and species diversity of New Zealand‟s indigenous 27

grasslands following the introduction of herbivorous terrestrial mammals in the 19th

28

century. 29

30

KEY WORDS: evolution; plant-animal interactions; species radiations. 31

32

33

3

1. INTRODUCTION 34

Plant species show many adaptations caused by specific interactions with past and 35

present herbivorous animals, including leaf toughness and spines. However, 36

distinguishing current and historical pressures for different plant traits is hampered by 37

the Late Pleistocene and Holocene extinctions of herbivores associated with human 38

colonisation of continents and islands (1-5). A possible solution to this is to utilise 39

natural evolutionary experiments by comparing plants in regions with different 40

herbivore histories (6). 41

Diversification and expansion of grasslands during the late Cenozoic has been 42

linked to the evolution of large herbivorous mammals (4). Various common features of 43

grasses (family Poaceae), e.g. phytoliths and rhizomatous growth form, may decrease 44

vulnerability to grazers. New Zealand constitutes an ideal system for investigating the 45

evolution of herbivores and grasses because the animals that lived in the archipelago 46

prior to human colonisation ca. 750 years ago are well documented. During prehistoric 47

times, no terrestrial mammals existed – except for a mouse-sized species that went 48

extinct (7) – and birds were the dominant herbivores, including moas, diverse 49

waterfowl, and rails (8). Mammals and birds feed differently on grasses. Mammals 50

remove and ingest entire leaves by manipulating forage with their lips and tongue and 51

cutting material with their teeth; birds graze tussock grasses by pulling and cutting 52

foliage, and several (notably rails) use clamping and tugging to remove tillers and 53

access basal meristematic tissue (9). The historical absence of mammals could thus have 54

resulted in different grass adaptations in New Zealand compared to the mammal 55

dominated savannahs of Africa and pampas of South America. 56

4

One possible candidate for such an adaptation is the ability to shed dead leaves. 57

Accumulation of dead leaves in grasses is known to reduce light availability and CO2 58

uptake, convert immediately usable inorganic nitrogen in rainwater to less readily 59

available organic nitrogen in microbial biomass, inhibit nitrogen fixation, decrease soil 60

temperatures and reduce root productivity (10). Despite the deleterious effects of 61

retaining dead leaves, only about 3% of the world‟s ca. 11,000 grass species are able to 62

abscise old leaves (11) (Table 1). Leaf abscission typically occurs at a fracture zone at 63

the base of the leaf blade (Fig. 1). Experimental evidence has led to a recent possible 64

explanation of this paradox. Leaf abscission (mimicked by manual removal of dead 65

leaves) increases biomass production, but it also makes grasses more palatable (12). 66

Under pressure by grazing mammals, leaf loss to herbivory becomes so severe that it 67

outweighs the benefits of increased biomass production (12). 68

New Zealand grasslands are dominated by long-lived tussock species, a major 69

component of which are the snow grasses (Chionochloa) and allied short tussock 70

species (Rytidosperma). Both genera belong to the grass subfamily Danthonioideae 71

(“pampas grasses” and allies), a clade of ca. 280 species of temperate grasses 72

distributed on all continents except Antarctica (13) (Fig. 2), and comprising both leaf-73

abscising species and species in which old leaves are retained. If New Zealand grasses 74

have experienced different selection pressures to their African and American relatives 75

we would expect these differences to have left a detectable morphological signature in 76

the extant species. We explored this hypothesis by analysing the occurrence of leaf 77

abscission in danthonioid grasses present in New Zealand compared to those occurring 78

elsewhere. 79

80

5

2. MATERIAL AND METHODS 81

(a) Phylogenetic and dating analyses 82

Recent studies have investigated relationships (14, 15) and hybridization (16) within the 83

grass subfamily Danthonioideae, leading to a new taxonomic revision of its genera (17). 84

To take into account phylogenetic and branch length uncertainty in our analyses, we 85

generated here a set of optimal trees by combining two recently published data sets (14, 86

15). The final matrix comprised 299 accessions (representing ca. 81% of the 280 87

described species), including 8 genera outside the Danthonioideae (Amphipogon, 88

Andropogon, Aristida, Arundo, Hordeum, Micraira, Setaria and Stipagrostis) and 89

contained 14,425 aligned nucleotide positions from 8 plastid and 2 nuclear sequence 90

regions: trnL–trnF, rpl16, rbcL, ndhF, matK, atpB-rbcL, trnT-trnL, trnC-trnD, ITS and 91

26S rDNA. 92

We inferred phylogenetic relationships in MrBayes v. 3.12 (18, 19) by 93

performing 10 topologically unconstrained runs of 1.2 x 106 generations each, under the 94

GTR++I model as selected by the Akaike Information Criterion in ModelTest v. 3.7 95

(20), sampling every 500th

generation, with 1 cold and 3 heated chains. All runs started 96

from the tree with Maximum Likelihood (ML) score obtained during 50 independent 97

ML analyses in GARLI v. 0.96b (21). We estimated absolute divergence times in 98

BEAST v.1.5.2 (22), by calibrating the root of the Danthonioideae with a normally 99

distributed prior (mean 26.1 Ma, standard deviation 0.5) following the results of a large 100

analysis of Poaceae based on fossil constraints (23), assuming a Yule model of 101

speciation under an uncorrelated lognormal clock, and following other standard settings. 102

We performed 2 independent runs of 107 generations each, sampling every 5000

th 103

6

generation. We assessed convergence of runs and effective sample sizes for all MCMC 104

parameters in Tracer v. 1.5 (22) and AWTY (24). An initial burn in of 500 trees was 105

excluded from each independent MrBayes runs, leaving 11406 trees for calculating a 106

50% majority-rule consensus. For the BEAST runs, 1500 trees were excluded, leaving 107

8232 time-calibrated trees (see also Humphreys et al., submitted). 108

To minimize sampling biases, where multiple DNA accessions were available 109

for a species and the species was shown to be monophyletic, we subsequently pruned all 110

but one of the sequences from the MrBayes and BEAST post burn-in tree samples, 111

while keeping the remaining branch lengths unaltered. In few cases where plastid and 112

nuclear DNA partitions for the same species appeared in conflict during pilot runs, we 113

duplicated these taxa in the matrix such that each duplicate was represented by one 114

partition only, following the approach and rationale described by Pirie et al. (16). 115

(b) Ancestral state optimisation 116

We used a carefully verified DELTA database (17) to code for species distributions 117

(absent in New Zealand=0; present in New Zealand=1) and leaf type (leaves 118

persistent=0, leaves abscising=1). We used both Fitch parsimony and ML (under two 119

models: the Markov k-state 1 parameter and the Asymmetrical Markov k-state 2 120

parameters) implemented in Mesquite v. 2.72 (25) to reconstruct ancestral states for all 121

nodes in a sample of 5,000 Bayesian trees, randomly selected from all independent runs 122

(burn-in excluded). For each node of the 50% majority-rule consensus tree from the 123

Bayesian analysis, we computed the relative frequency of uniquely best states across the 124

tree sample. These results were used to identify clades where shifts to abscission most 125

probably occurred and compute the number of species descending from these shifts. 126

7

(c) Directionality and conservativeness of shifts 127

We tested whether shifts to leaf abscission occurred more times than shifts into the other 128

direction by comparing the distribution of shifts into each direction across our sample of 129

optimal trees. We tested whether shifts in leaf type were phylogenetically conservative 130

by calculating the distribution of observed shifts with number of shifts across a sample 131

of trees in which terminals had been randomly shuffled (10 x 1,000 randomly selected 132

Bayesian trees). To evaluate whether any pattern of conservativeness depended on 133

presence in New Zealand, we performed this test for trees containing all sampled 134

accessions of Danthonioideae and trees in which New Zealand terminals had been 135

pruned while retaining branch lengths of unpruned lineages. These analyses were 136

performed in Mesquite (25). 137

(d) Correlation analysis 138

To test whether leaf retention and leaf abscission have evolved depending on 139

geographic distribution, we compared the fit of models of dependent and independent 140

evolution to our data. Correlation analyses were carried out using the Discrete (ML) and 141

BayesDiscrete (mcmc) commands (26, 27) in BayesTraits (available from: 142

www.evolution.rdg.ac.uk/BayesTraits.html). Eight rate parameters constitute the 143

dependent model, which assumes that each character evolves (forward and backward 144

shifts) at different rates depending on the state of the second character. In the 145

independent model forward and backward shifts in one character occur at the same rate 146

regardless of the state of the second parameter (coefficients q12=q34, q13=q24, 147

q21=q43 and q31=q42; Table S1). Hence, a model of independent evolution has four 148

parameters. Definitions of rate coefficients and the results for the correlation analyses 149

8

are provided in Table S1. Since the number of reversals inferred seemed unrealistically 150

large we did not use these values for testing directionality of shifts. Inflated rate 151

coefficients are to be expected when shifts are inferred to have occurred along short 152

branches (28), therefore we only used information on reversals from the parsimony 153

analysis, as this test ignores differences in branch lengths. 154

Fit of dependent and independent models using ML was compared using a 155

likelihood ratio test over a sample of 5,000 randomly selected Bayesian trees (burn-in 156

excluded) and with ten likelihood iterations per tree. Fit of dependent and independent 157

models in a Bayesian framework were compared using Bayes factors, calculated as 158

twice the difference in log harmonic mean of the worst fitting model and the better 159

fitting model (29). Multiple, long Markov runs were performed to ensure that the 160

harmonic mean remained stable within and among runs. We used the reversible jump 161

MCMC method (26, 30), which allows sampling of the various possible models of 162

evolution in proportion to their posterior probabilities (26) as opposed to only the rate 163

parameters being sampled in this way, as in conventional MCMC (28). We used an 164

exponentially distributed hyperprior (28) with its mean value seeded from a uniform 165

distribution in the interval specified, to ensure that posterior values were contained 166

within, but not determined by, the prior range. We varied the amount by which the rate 167

parameters are allowed to change between iterations of the Markov chain, by varying 168

the „ratedev‟ value, so that acceptance rates averaged 20-40%. This should avoid 169

autocorrelation while ensuring adequate exploration of parameter space. To improve 170

initially low acceptance rates we used a modified version of the code that accepts either 171

a move to a new model or a move to a different tree (courtesy of A. Meade). We ran 1.5 172

x 108 generations, sampling every 1,000 generations, yielding a sample of 145,000 173

9

iterations after 5 x 106 iterations were removed as burnin. In addition we ran a separate 174

series of analyses restricted to sampling only independent models. 175

In total, 5 sampled species are polymorphic for leaf type, and 3 for distribution. 176

All analyses above were repeated coding polymorphic species with (0,1) followed by 177

coding only for presence (1), with the rationale of retaining information about all those 178

species able to abscise their leaves and being native to New Zealand even if they show 179

polymorphism in one or both traits (except ML optimisations in Mesquite, unable to 180

handle polymorphisms). All statistical tests yielded the same level of significance in 181

both cases. 182

3. RESULTS AND DISCUSSION 183

The 50% majority-rule consensus tree of the Bayesian analysis in MrBayes, together 184

with posterior probabilities of all resolved clades, is provided in Electronic 185

Supplementary Material, Fig. S1. Figure 3 shows a simplified version of that tree, with 186

duplicated species pruned and indicating the presence of leaf abscission and distribution 187

in New Zealand. Relationships obtained are congruent with previous estimates of the 188

danthonioid phylogeny (14, 15). The time-calibrated tree of the subfamily, estimated in 189

BEAST and showing 95% highest posterior densities of node ages, is provided in Fig. 190

S2. 191

Results from the parsimony and ML optimisations for leaf abscission and 192

distribution, coding only for presence, produced nearly identical results with respect to 193

ancestral states (the main difference being a higher level of ambiguity for the 194

reconstruction of early diverging nodes under the 2-rate ML model, as compared to 195

parsimony and the 1-rate ML model). Given these similarities, and since ML in 196

10

Mesquite cannot handle polymorphic taxa and thus requires an arbitrary simplification 197

of reality, we report here only the parsimony results for leaf type and distribution (Figs. 198

S3 and S4, respectively). The optimisations over 5,000 trees resulted in a mean of 29 199

shifts in leaf abscission and 12 shifts in distribution. 200

Leaf abscission occurs in the majority of danthonioid species native to New 201

Zealand, a proportion that is markedly higher than found in other regions (Table 1; 202

Fisher‟s Exact test, one tailed P<0.001). This striking imbalance could be the result of 203

differences in the rates of shifts between leaf abscission and leaf retention, differences 204

in the retention of one state over the other, differences in speciation rates between 205

lineages with abscising and persistent leaves, or a combination of more than one of 206

these factors. 207

(a) Evolution and retention of leaf abscission 208

Shifts in leaf type (persistentabscising) are strongly correlated with geographical 209

distribution (present in New Zealandabsent in New Zealand; Figs. 3 and 4a–b; 210

average Likelihood Ratio 16.1, average logBayes Factor 14.6; in both cases 4 d.f.). The 211

rate of transition to leaf abscission is higher in New Zealand than elsewhere (Wilcoxon 212

rank sum test, based on Bayesian posterior rate coefficients, n=1.8x108, one tailed 213

P<0.001). On all continents, shifts from persistent to abscising leaves have been more 214

common than shifts in the opposite direction (Fig. S5; parsimony ancestral state 215

reconstruction, t test for independent samples, n=5000, one tailed P<0.001). However, 216

the ultimate consequences of evolving abscission have been very different in New 217

Zealand compared to other regions: lineages that evolved this ability outside New 218

Zealand have rarely diversified, producing at most 6 leaf-abscising species in genus 219

11

Merxmuellera in the southern African Drakensberg – a region that, interestingly and in 220

accordance with the patterns obtained for New Zealand, historically has only sustained a 221

very low density of small-sized mammal herbivores (31, 32). Within New Zealand, on 222

the other hand, lineages that evolved leaf abscission have become significantly more 223

species rich, as indicated by a Wilcoxon rank sum test (n1=16, n2=3, one tailed P=0.001; 224

Fig. S3 shows which clades were included in the test). The same test remains significant 225

even after correcting for incomplete species sampling in Merxmuellera (4 out of 7 226

species sampled) and the potential effect of time in species accumulation. Time was 227

taken into account by calculating speciation rates under a pure birth [Yule] model for 228

each leaf abscising clade, based on the number of descendant species S and the median 229

stem age t obtained in the BEAST analysis, such that = ln[S]/t (33) (P=0.004 in both 230

cases). Retention of leaf abscission over more diversification events in New Zealand is 231

also evident by the fact that in all mammal dominated regions the phylogenetic 232

distribution of leaf-abscising species is not distinguishable from a random distribution, 233

whereas in New Zealand leaf-abscising species show a phylogenetically conserved 234

pattern (Figs. S6 and S7; t test for independent samples, n=5000, one tailed P<0.001). 235

These results show that the higher proportion of leaf-abscising danthonioid grass 236

species native to New Zealand compared to species absent from New Zealand is due to 237

two factors: a higher frequency of evolution of leaf abscission in new Zealand lineages 238

and greater net diversification rates of these lineages. 239

(b) A “ghost” adaptation? 240

12

There has been a long debate on the factors shaping the evolution of New Zealand 241

plants. New Zealand enjoys a rich endemic flora with many peculiar adaptations, such 242

as the „divaricate‟ life form, which characterizes ca. 20% of the endemic woody species 243

(9). Juveniles of these species produce densely intertangled branches no taller than 2–3 244

m, forming cage-like barriers to predation. In adult individuals, branches are acute-245

angled, often with larger leaves and reach above 3 m in height. Although the divaricate 246

habit has long been interpreted as a response to climatic conditions (34), recent studies 247

have provided overwhelming evidence that it instead served as an adaptation against 248

browsing by the now extinct moas (9, 35). However, it is unlikely that leaf abscission 249

evolved as a response to evolutionary pressures exerted by browsing birds. From 250

detailed analyses of fossilized dung and gizzards, moas appear to have eaten tussock 251

grasses only rarely (35). Similarly, there is no indication that the extant native grassland 252

rail (Porphyrio hochstetteri) would have influenced the evolution of leaf abscising 253

grasses, since it has been documented to access leaf bases of certain tussock species 254

irrespective of dehiscent features (36). The absence of mammals in New Zealand 255

therefore appears to have been the major factor conferring a strong evolutionary 256

advantage to leaf abscising grasses in New Zealand in the past. The results presented 257

here support this hypothesis and even if they do not provide evidence against other 258

possible causative factors (e.g. soil nutrient differences) they do provide insight into a 259

possibly important factor in the shaping of the New Zealand grass flora. 260

(c) A current disadvantage? 261

If leaf abscission in New Zealand grasses evolved as a response to the absence of 262

grazing mammals, native grasses would be expected to be poorly adapted to the 263

13

introduction of sheep, cattle and rabbits during European settlement after ca. 1850. 264

There is strong evidence that endemic grasses are inherently susceptible to mammalian 265

grazers (37-39), but grassland deterioration at large scales is complex and usually 266

involves multiple causes including burning, stocking rates, feral animals and invasive 267

weeds (40). To date no studies have experimentally tested whether adult leaf-abscising 268

grasses native to New Zealand are more susceptible to grazing than grasses with 269

persistent leaves, because leaf abscission has never been considered a relevant 270

parameter in ecological studies. There is however evidence that dominance (m2/ha) of a 271

leaf-abscising species (Chionochloa flavescens) increased significantly during the 21 272

years that followed the removal of mammal grazers in north facing plots of New 273

Zealand‟s South Island (38). A comparable significant increase in dominance was not 274

observed in the same plots for C. macra, a closely related species with persistent leaves. 275

Although further studies are clearly needed to test this prediction, these differences 276

reinforce the idea that leaf abscission confers higher productivity and thus a competitive 277

advantage in the absence of mammal grazers (12) – an advantage that may have been 278

turned into a disadvantage in many areas of New Zealand today. 279

280

281

ACKNOWLEDGEMENTS 282

We thank J. Wood, C. Carbutt and T. O‟Connor for discussion, R. Wüest for help with 283

GIS, and the Computational Biology Service Unit hosted by Cornell University, USA 284

(http://cbsuapps.tc.cornell.edu) where the MrBayes and BEAST analyses were 285

performed. We are very thankful to the constructive comments by the associated editor 286

14

Susanne Renner and two anonymous reviewers. This project was financed by a grant 287

from the Swiss Science Foundation to HPL. 288

289

FIGURE CAPTIONS 290

Fig 1. Leaf abscission in grasses. a) Herbarium specimen of the New 291

Zealand grass Chionochloa rigida, a leaf-abscising species. b) Detail of fracture 292

zone where old leaves are shed. 293

Fig 2: Distribution of grass subfamily Danthonioideae. GIS-based map 294

based on 22,025 occurrences from GBIF (www.gbif.org) and 19,372 from 295

various sources. Occurrences coded as New Zealand are marked in red, all 296

others in green. 297

Fig 3. Phylogeny of Danthonioideae. Bayesian consensus cladogram 298

showing leaf type and distribution for 270 accessions (representing ca. 81% of 299

all described species; only presence is coded). Relationships were derived from 300

analysis of 14,425 aligned nucleotide positions from 8 plastid and 2 nuclear 301

DNA sequence regions. See Fig. S1 for the fully annotated MrBayes consensus 302

reporting posterior probabilities for all clades. 303

Fig 4. Results from the phylogeny-based correlation analyses. a) 304

distribution and leaf type, using Maximum Likelihood implemented in 305

BayesTraits (26); b) Bayes factor (calculated as twice the difference of the log 306

harmonic mean between dependent and independent analyses) plotted against 307

iteration in the reversible-jump Markov chain Monte Carlo analysis in 308

15

BayesTraits (26). The analysis stabilizes after an initial burn-in phase. The 309

dashed lines indicate significance levels, under which there is “very strong 310

evidence” for correlation (29). 311

312

16

313

REFERENCES 314

1. Temple SA (1979) The dodo and the tambalacoque tree. Science 315

203(4387):1364. 316

2. Guimarães Jr PR, Galetti M, & Jordano P (2008) Seed dispersal 317

anachronisms: Rethinking the fruits extinct megafauna ate. PLoS ONE 318

3(3). 319

3. Janzen DH & Martin PS (1982) Neotropical Anachronisms: The Fruits the 320

Gomphotheres Ate. Science 215(4528):19-27. 321

4. Johnson CN (2009) Ecological consequences of Late Quaternary 322

extinctions of megafauna. Proceedings of the Royal Society B: Biological 323

Sciences 276(1667):2509-2519. 324

5. Bond WJ & Silander JA (2007) Springs and wire plants: Anachronistic 325

defences against Madagascar's extinct elephant birds. Proceedings of 326

the Royal Society B: Biological Sciences 274(1621):1985-1992. 327

6. Mack RN & Thompson JN (1982) Evolution in steppe with few large 328

hooved mammals. American Naturalist 119(6):757-773. 329

7. Worthy TH, et al. (2006) Miocene mammal reveals a Mesozoic ghost 330

lineage on insular New Zealand, southwest Pacific. Proceedings of the 331

National Academy of Sciences 103(51):19419-19423. 332

8. Lee WG, Wood JR, & Rogers GM (2010) Legacy of avian-dominated 333

plant-herbivore systems in New Zealand. N. Z. J. Ecol 34:28-47. 334

17

9. Bond WJ, Lee WG, & Craine JM (2004) Plant structural defences against 335

browsing birds: A legacy of New Zealand's extinct moas. Oikos 336

104(3):500-508. 337

10. Knapp A & Seastedt T (1986) Detritus accumulation limits productivity of 338

tallgrass prairie. Bioscience 36(10):662-668. 339

11. Clayton WD, Harman KT, & Williamson H (2002) World Grass Species: 340

Descriptions, Identification, and Information Retrieval. 341

12. Mingo A & Oesterheld M (2009) Retention of dead leaves by grasses as 342

a defense against herbivores. A test on the palatable grass Paspalum 343

dilatatum. Oikos 118(5):753-757. 344

13. Linder HP & Barker NP (2000) Biogeography of the Danthonieae. Grass 345

Systematics and Evolution - Vol. 2 of Proceedings of the Second 346

International Conference on the Comparative Biology of the Monocots, 347

Sydney, September 1998., eds Jacobs SWL & Everett J (CSIRO, 348

Melbourne). 349

14. Pirie MD, et al. (2008) A novel supermatrix approach improves resolution 350

of phylogenetic relationships in a comprehensive sample of danthonioid 351

grasses Molecular Phylogenetics and Evolution 48:1106-1119. 352

15. Humphreys AM, Pirie MD, & Linder HP (2010) A plastid tree can bring 353

order to the chaotic generic taxonomy of Rytidosperma Steud. s.l. 354

(Poaceae). Molecular Phylogenetics and Evolution 55(3):911-928. 355

16. Pirie MD, Humphreys AM, Barker NP, & Linder HP (2009) Reticulation, 356

Data Combination, and Inferring Evolutionary History: An Example from 357

Danthonioideae (Poaceae). Syst Biol 58(6):612-628. 358

18

17. Linder HP, et al. (accepted) A Classification of the Danthonioideae 359

(Poaceae). Annals of the Missouri Botanic Garden. 360

18. Huelsenbeck JP & Ronquist F (2001) MRBAYES: Bayesian inference of 361

phylogenetic trees. Bioinformatics 17(8):754-755. 362

19. Ronquist F & Huelsenbeck JP (2003) MrBayes 3: Bayesian phylogenetic 363

inference under mixed models. Bioinformatics 19(12):1572-1574. 364

20. Posada D & Crandall KA (1998) MODELTEST: Testing the model of 365

DNA substitution. Bioinformatics 14(9):817-818. 366

21. Zwickl DJ (2006) Genetic algorithm approaches for the phylogenetic 367

analysis of large biological sequence datasets under the maximum 368

likelihood criterion. Ph.D. dissertation, The University of Texas at Austin. 369

Software available at 370

www.bio.utexas.edu/faculty/antisense/garli/Garli.html. 371

22. Drummond AJ, Ho SY, Phillips MJ, & Rambaut A (2006) Relaxed 372

phylogenetics and dating with confidence. PLoS Biology 4(5). 373

23. Christin PA, et al. (2008) Oligocene CO2 Decline Promoted C4 374

Photosynthesis in Grasses. Current Biology 18(1):37-43. 375

24. Nylander JAA, Wilgenbusch JC, Warren DL, & Swofford DL (2008) 376

AWTY (are we there yet?): a system for graphical exploration of MCMC 377

convergence in Bayesian phylogenetics. Bioinformatics 24(4):581-583. 378

25. Maddison WP & Maddison DR (2009) Mesquite: a modular system for 379

evolutionary analysis. Version 2.71 http://mesquiteproject.org. 380

19

26. Pagel M & Meade A (2006) Bayesian analysis of correlated evolution of 381

discrete characters by reversible-jump Markov chain Monte Carlo. 382

American Naturalist 167(6):808-825. 383

27. Pagel M (1994) Detecting correlated evolution on phylogenies: A general 384

method for the comparative analysis of discrete characters. Proceedings 385

of the Royal Society B: Biological Sciences 255(1342):37-45. 386

28. Pagel M, Meade A, & Barker D (2004) Bayesian estimation of ancestral 387

character states on phylogenies. Systematic Biology 53(5):673-684. 388

29. Raftery AE (1996) Hypothesis testing and model selection. Markov chain 389

Monte Carlo in practice, eds Gilks WR, Richardson S, & Spiegelhalter DJ 390

(Chapman & Hall, London). 391

30. Green PJ (1995) Reversible jump Markov chain Monte Carlo 392

computation and Bayesian model determination. Biometrika 82(4):711-393

732. 394

31. O'Connor TG (2005) Influence of land use on plant community 395

composition and diversity in Highland Sourveld grassland in the southern 396

Drakensberg, South Africa. Journal of Applied Ecology 42(5):975-988. 397

32. Rowe-Rowe D & Scotcher J (1986) Ecological carrying capacity of the 398

Natal Drakensberg for wild ungulates. South African Journal of Wildlife 399

Research 16(1):12-16. 400

33. Nee S, Mooers AO, & Harvey PH (1992) Tempo and mode of evolution 401

revealed from molecular phylogenies. Proceedings of the National 402

Academy of Sciences of the United States of America 89(17):8322-8326. 403

20

34. Howell CJ, Kelly D, & Turnbull MH (2002) Moa ghosts exorcised? New 404

Zealand's divaricate shrubs avoid photoinhibition. Functional Ecology 405

16(2):232-240. 406

35. Wood JR, et al. (2008) Coprolite deposits reveal the diet and ecology of 407

the extinct New Zealand megaherbivore moa (Aves, Dinornithiformes). 408

(Translated from English) Quaternary Science Reviews 27(27-28):2593-409

2602 (in English). 410

36. Mills JA, Lavers RB, Lee WG, & Mara MK (1991) Food selection by 411

takahe Notornis mantelli in relation to chemical composition. Ornis 412

Scandinavica 22(2):111-128. 413

37. Lee WG, Fenner M, Loughnan A, & Lloyd KM (2000) Long-term effects 414

of defoliation: Incomplete recovery of a New Zealand alpine tussock 415

grass, Chionochloa pallens, after 20 years. Journal of Applied Ecology 416

37(2):348-355. 417

38. Rose AB & Platt KH (1992) Snow tussock (Chionochloa) population 418

responses to removal of sheep and European hares, Canterbury, New 419

Zealand. New Zealand Journal of Botany 30:373-382. 420

39. Mark AF & McLennan B (2005) The conservation status of New 421

Zealand's indigenous grasslands. New Zealand Journal of Botany 422

43(1):245-270. 423

40. Mark AF (1994) Effects of burning and grazing on sustainable utilisation 424

of upland snow tussock (Chionochloa spp.) rangelands for pastoralism in 425

South Island, New Zealand). Australian Journal of Botany 42(2):149-161. 426

427