Accepted Manuscript

Molecular phylogenetic reconstructions identify East Asia as the cradle for the

evolution of the cosmopolitan genus Myotis (Mammalia, Chiroptera)

Manuel Ruedi, Benoît Stadelmann, Yann Gager, Emmanuel J.P. Douzery,

Charles M. Francis, Liang-Kong Lin, Antonio Guillén-Servent, Alice Cibois

PII: S1055-7903(13)00333-3

DOI: http://dx.doi.org/10.1016/j.ympev.2013.08.011

Reference: YMPEV 4692

To appear in: Molecular Phylogenetics and Evolution

Please cite this article as: Ruedi, M., Stadelmann, B., Gager, Y., Douzery, E.J.P., Francis, C.M., Lin, L-K.,

Guillén-Servent, A., Cibois, A., Molecular phylogenetic reconstructions identify East Asia as the cradle for the

evolution of the cosmopolitan genus Myotis (Mammalia, Chiroptera), Molecular Phylogenetics and Evolution

(2013), doi: http://dx.doi.org/10.1016/j.ympev.2013.08.011

This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers

we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and

review of the resulting proof before it is published in its final form. Please note that during the production process

errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.

1

Molecular phylogenetic reconstructions identify East Asia as the cradle for the evolution

of the cosmopolitan genus Myotis (Mammalia, Chiroptera)

Manuel Ruedia, Benoît Stadelmanna, b, Yann Gagerc, d, Emmanuel J. P. Douzery c, d, Charles

M. Francise, Liang-Kong Linf, Antonio Guillén-Serventg, and Alice Ciboisa

a Department of Mammalogy and Ornithology, Natural History Museum of Geneva, Route de

Malagnou 1, BP 6434, 1211 Geneva 6, Switzerland

b Department of Zoology and Animal Biology, University of Geneva, 30 quai Ernest-

Ansermet, 1211 Geneva 4, Switzerland

c Université Montpellier II, Place Eugène Bataillon, 34 095 Montpellier Cedex 5, France

d CNRS, Institut des Sciences de l’Evolution (UMR 5554), CC064- Place Eugène Bataillon,

34095 Montpellier Cedex 5, France

e Canadian Wildlife Service, Environment Canada, Ottawa, Ontario, K1A 0H3, Canada

f Laboratory of Wildlife Ecology, Department of Biology, Tunghai University, Taichung,

Taiwan 407, R.O.C.

g Instituto de Ecología A.C., Carretera Antigua a Coatepec 351, El Haya, Xalapa 91070,

Veracruz, México

* Corresponding author. Address: Department of Mammalogy and Ornithology, Natural

History Museum of Geneva, Route de Malagnou 1, BP 6434, 1211 Geneva (6), Switzerland.

Tel.: +41 22 418 6320; Fax: +41 22 418 6301.

E-mail address: Manuel.Ruedi@ville-ge.ch (Manuel Ruedi)

SUGGESTED RUNNING TITLE: Evolutionary origins of Myotis bats.

2

ABSTRACT

___________________________________________________________________________ 1

Sequences of the mitochondrial cytochrome b (1140 bp) and nuclear Rag 2 (1148 bp) genes 2

were used to assess the evolutionary history of the cosmopolitan bat genus Myotis, based on a 3

worldwide sampling of over 88 named species plus 7 species with uncertain nomenclature. 4

Phylogenetic reconstructions of this comprehensive taxon sampling show that most radiation 5

of species occurred independently within each biogeographic region. Our molecular study 6

supports an early divergence of species from the New World, where all Nearctic and 7

Neotropical species plus a lineage from the Palaearctic constitute a monophyletic clade, sister 8

to the remaining Old World taxa. The major Old World clade includes all remaining Eurasian 9

taxa, most Oriental species, one Oceanian, and all Ethiopian species. Another lineage, 10

including M. latirostris from Taiwan, appears at the base of these two major biogeographic 11

clades and, because it bears nyctalodont molars, could be considered as a distinct genus. 12

However, this molar configuration is also found in crown-group species, indicating that these 13

dental characters are variable in the genus Myotis and may confound interpretation of the 14

fossil record . Molecular datings suggest an origin of all recent Myotis in the early Miocene 15

(about 21 MYA with 95% highest posterior density interval 23-20 MYA). This period was 16

characterized by a global climatic cooling that reduced the availability of tropical habitats and 17

favoured the development of more temperate vegetation. This sharp climatic change might 18

have triggered the evolution of Myotis in the Northern continents, because Myotis ancestors 19

seem to have been well adapted and successful in such temperate habitats. Ancestral area 20

reconstructions based on the molecular phylogeny suggest that the eastern portion of the 21

Asian continent was an important centre of origin for the early diversification of all Myotis 22

lineages, and involved relatively few subsequent transcontinental range expansions. 23

24

Keywords: Myotis, worldwide radiation, Vespertilionidae, Lagrange, cytochrome b, Rag 2, 25

molecular dating, biogeography 26

27

28

1. Introduction 29

The evolutionary history of species is the outcome of a complex balance between 30

processes of extinction and speciation. At any time, some groups may thus be highly 31

diversified and successful while sister clades are depauperate (Purvis et al., 2000). The causes 32

that may promote or inhibit these processes in a particular clade and not in others are still 33

highly debated, but may include intrinsic key innovations; extrinsic processes linked to plate 34

3

tectonics, climatic fluctuations or vegetation changes; differential rates of gene evolution; and 35

niche competition. One difficulty in identifying the common processes underlying the 36

evolution of species diversification and natural history of species is the need for accurate 37

knowledge of the phylogenetic relationships of the group considered (Paradis, 1998; 38

Sanderson and Donoghue, 1996). In particular complete or nearly complete phylogenies 39

including all major branches (i.e. an exhaustive phylogenetic hypothesis for the evolution of a 40

whole group of extant species), allow for robust analysis of the factors that influenced the 41

rates of speciation and extinction (Purvis et al., 2000). Moreover, these complete phylogenies 42

can help to elaborate conservation plans aimed to preserve most of the current evolutionary 43

history of these groups (Bininda-Emonds et al., 2000; Sechrest et al., 2002). 44

Bats from the order Chiroptera are part of the Laurasiatheria clade of mammals, and 45

have diversified in all continents except Antarctica, with over 1100 currently recognized 46

species worldwide (Simmons, 2005), a number that continues to increase with new taxonomic 47

work. Within the superfamily Vespertilionoidea (Teeling et al., 2005), the genus Myotis 48

belongs to the family Vespertilionidae, and is classified in a distinct subfamily, the Myotinae 49

(Hoofer and Van den Bussche, 2003; Simmons, 2005). Because these bats have a rather 50

“generalist morphology” (Horácek et al., 2000) and are very speciose, their taxonomy is 51

difficult. In a phenetic study based on external and cranial morphology, Findley (1972) 52

estimated that the current fauna included about 60 species of Myotis. This number has 53

increased through morphological and genetic studies and recent revisions now exceed 100 54

species (Koopman, 1994; Simmons, 2005). Indeed, new species continue to be described 55

regularly, even in well-known areas such as Europe (Helversen et al., 2001; Ibáñez et al., 56

2006). Thus, Myotis is one of the most diverse genera of mammals, second only to Crocidura 57

shrews, and offers an exceptional substrate for studying speciation and diversification at a 58

worldwide scale. Myotis bats have colonised most terrestrial habitats, except polar regions, 59

and represents the only mammal genus that is naturally distributed on every continent except 60

Antarctica. However, the species diversity is unequal by region, with the maximum diversity 61

found in the northern continents (see inset of Fig. 1). 62

To date, few molecular studies have approached the evolution of the genus Myotis. It 63

has been demonstrated that the prevailing morphology-based subdivision of the genus Myotis 64

into four or more subgenera (Findley, 1972; Koopman, 1993; Tate, 1941) does not reflect 65

phylogenetic history (Hoofer and Van den Bussche, 2003; Kawai et al., 2003; Lack et al., 66

2010; Ruedi and Mayer, 2001). Instead, it appears that adaptive convergences have produced 67

similar ecomorphs independently through some deterministic processes (Fenton and 68

Bogdanowicz, 2002). By contrast, several independent biogeographic radiations emerged 69

4

from phylogenetic reconstructions, including one that unites all Ethiopian taxa (Stadelmann et 70

al., 2004b) and another that comprises all New World species (Lack et al., 2010; Stadelmann 71

et al., 2007). However, Myotis from the Palaearctic and the Oriental regions have been 72

underrepresented in previous studies, impeding phylogenetic inference for the whole group 73

(Kawai et al., 2003; Zhang et al., 2009). 74

By expanding the Old World sampling to include over 80% of all named Myotis 75

species (Simmons, 2005) as well as several unnamed taxa, we produced a comprehensive and 76

robust phylogeny of extant species covering all major branches. With this almost complete 77

picture of the Myotis radiation, we first investigate phylogenetic relationships within Old 78

World taxa. Second, we evaluate the timing and biogeographic evolution of Myotis worldwide 79

by using a Bayesian relaxed molecular clock approach, as well as likelihood reconstructions 80

of ancestral geographic distribution to identify the possible area of origin of this striking 81

evolutionary radiation. 82

83

2. Materials and Methods 84

2.1 Taxon and geographic sampling 85

According to the last available revision of the genus (Simmons, 2005), 103 species of 86

Myotis are recognized worldwide. Unless indicated hereafter, we follow Simmons‟ taxonomic 87

arrangement to refer to the nominal species analysed in this study (see Table S1). Additional 88

taxa recognized here include M. aurascens from Central and East Asia [distinct from M. 89

mystacinus (Tsytsulina et al., 2012)], M. cf. browni from the Philippines and M. latirostris 90

from Taiwan [both distinct from M. muricola (Lack et al., 2010; Stadelmann et al., 2007)], 91

M. escalerai from the Iberian peninsula [distinct from M. nattereri (Ibáñez et al., 2006)], M. 92

formosus flavus [specifically distinct from M. formosus (Jiang et al., 2010)], M. gracilis from 93

the Far East [distinct from M. brandtii though M. sibiricus may be a senior synonym and 94

hence the appropriate name for this species (Horácek et al., 2000; Kruskop et al., 2012)], M. 95

petax from the Eastern Palaearctic [distinct from M. daubentonii and a senior synonym of M. 96

abei (Kruskop et al., 2012; Matveev et al., 2005; Tsytsulina, 2004)], M. phanluongi from 97

Vietnam [distinct from M. siligorensis (Borisenko et al., 2008)], and M. taiwanensis [distinct 98

from M. adversus (Han et al., 2010)]. In turn the European taxon oxygnathus is maintained as 99

a subspecies of M. blythii (Evin et al., 2008), while the Japanese hosonoi, ozensis and 100

yesoensis are now all considered as synonyms of M. ikonnikovi (Abe et al., 2005; Kawai et al., 101

2003). Species for which no tissue or GenBank entries were available include M. adversus 102

(Oriental), M. aelleni (Neotropical), M. australis (Oceanian), M. bucharensis (Eastern 103

Palaearctic), M. cobanensis (Neotropical), M. csorbai (Oriental), M. findleyi (Neotropical), M. 104

5

fortidens (Neotropical), M. gomantongensis (Oriental), M. hajastanicus (Western Palaearctic), 105

M. hermani (Oriental), M. insularum (Oceanian), M. melanorhinus (Nearctic), M. 106

moluccarum (Oceanian), M. morrisi (Ethiopian), M. nipalensis (Eastern Palaearctic), M. 107

oreias (Oriental), M. peninsularis (Nearctic), M. planiceps (Neotropical) and M. stalkeri 108

(Oriental). We also included 7 additional species of Myotis that diverge genetically and 109

morphologically from other currently recognized taxa and await proper taxonomic description 110

(see e.g. Francis et al., 2010; Larsen et al., 2012b; Salicini et al., 2011). Although not 111

complete, we feel that this taxonomic coverage (Table S1) represents most, if not all, major 112

lineages in the Myotis radiation, as most of the missing taxa appear morphologically or 113

genetically to be part of clades that are represented in our sampling, a conclusion confirmed 114

for New World taxa by recently published molecular surveys (Larsen et al., 2012a; Larsen et 115

al., 2012b). 116

In total, 105 taxa representing 88 nominal species, 10 distinct subspecies and 7 117

additional unnamed/unidentified species were analysed (see Table S1). Because of potential 118

concerns about species assignments and nomenclature, especially for the Asian Myotis (Bates 119

et al., 1999), most sequences were derived from vouchered specimens that can be examined in 120

public institutions. In addition to the newly sequenced specimens, some taxa were represented 121

by published sequences (Bickham et al., 2004; Ibáñez et al., 2006; Jones et al., 2006; Kawai et 122

al., 2003; Lack et al., 2010; Larsen et al., 2012b; Ruedi and Mayer, 2001; Stadelmann et al., 123

2004a; Stadelmann et al., 2004b; Stadelmann et al., 2007; Tsytsulina et al., 2012; Zhang et al., 124

2009) available in GenBank (see Table S1). We refrained from considering several other 125

unpublished Myotis sequences available in public databases because they were of uncertain 126

taxonomic origin or had experimental problems. For instance, the sequence of “Myotis 127

hajastanicus” AY665138 was discarded because it proved to be a chimera between partial 128

cytochrome b sequences of M. gracilis and of Bos indicus. 129

The Kerivoulinae and Murininae are the closest relatives of Myotis (e. g. Hoofer and 130

Van den Bussche, 2003; Kawai et al., 2002). Therefore, six species representing the genera 131

Harpiocephalus, Kerivoula and Murina from the Oriental region (Hoofer et al., 2003; Lack et 132

al., 2010; Ruedi et al., 2012; Stadelmann et al., 2004b; Stadelmann et al., 2007) were used as 133

a composite outgroup (Table S1). 134

135

2.2 DNA amplification and sequencing 136

Tissue samples preserved in ethanol or DMSO were soaked for 1 hour in sterile water 137

before extraction. Total genomic DNA was then isolated following a salting out protocol 138

(Miller et al., 1988), as detailed in Castella et al. (2001) and rediluted into 200 µl of low TE 139

6

buffer. The complete mitochondrial cytochrome b gene (Cyt b; 1140 bp) and the partial 140

nuclear Recombination Activating Gene 2 (Rag 2; 1148 bp) were amplified, and sequenced in 141

both directions following the procedures described in previous papers (Stadelmann et al., 142

2004b; Stadelmann et al., 2007). Sequences were checked for possible stop codons or indels, 143

which would indicate the presence of pseudogenes. As none of the coding sequences 144

presented such anomalies, they were assumed to be the functional gene and alignment of 145

orthologs was thus straightforward. Heterozygous individuals at the nuclear Rag 2 gene were 146

identified by double peaks on both sequencing strands. These heterozygous positions were not 147

phased out, but replaced by „N‟. Fragments shorter than the target length were also completed 148

by „N‟s to replace missing data. 149

150

2.3 Phylogenetic analyses 151

Phylogenetic reconstructions were conducted on the Cyt b (141 sequences of 1140 bp; 152

601 variable positions in the ingroup) and Rag 2 datasets (122 sequences, 1148 bp, 220 153

variable positions) analysed separately and in combination (141 lineages of 2288 bp) with 154

strings of Ns replacing the missing sequences. These missing data represented 1.4% and 3.5% 155

of the character matrix for the Cyt b and Rag 2 datasets, respectively. Probabilistic methods 156

were used to reconstruct phylogenetic trees with the Maximum Likelihood (ML) approach 157

implemented in RAxML (Stamatakis, 2006) and Bayesian inferences (BA) in MrBayes v3.2.0 158

(Ronquist and Huelsenbeck, 2003). All analyses were done on a fully partitioned model, 159

where each gene and/or codon partition was allowed to have its own set of model parameters. 160

The most appropriate model of nucleotide substitution for each partition was evaluated using 161

MrModeltest 2.3 (Nylander, 2004) and the Akaike Information Criterion (AIC; data 162

summarized in Table S2). The General Time Reversible (GTR) model with rate variation 163

among sites (Γ) and a proportion of invariable sites (I) represented the best fitting model of 164

nucleotide substitution for the Cyt b and the combined Cyt b + Rag 2 dataset (Table S2). 165

Topological searches were initiated from random trees. 166

Bayesian analyses were performed with a Markov chain Monte-Carlo technique and 167

run for 10x106 generations, with a sampling every 1000 generations. The chains were checked 168

for convergence and appropriate effective sample size (ESS>200) with Tracer v.1.5 (Rambaut 169

and Drummond, 2009). Chains converged and the log-likelihood reached an asymptote after 170

about 1x106 (Cyt b or Rag 2) or 3.5 x106 (combined) generations. These initial trees were 171

discarded as burn-in. Posterior probabilities (PP) were subsequently computed from the 172

consensus of the remaining sampled trees. Reliability of nodes in the ML analyses was 173

assessed by 100 standard bootstraps (BP) with RAxML. 174

7

175

2.3 Molecular dating analyses 176

We estimated the mean divergence times between taxa by a relaxed uncorrelated 177

lognormal molecular clock as implemented in BEAST v1.7.4 (Drummond and Rambaut, 178

2007). Similarly to the MrBayes analyses, we used the combined dataset partitioned into 6 179

categories (3 codon positions each for Cyt b and Rag 2) with the most appropriate model of 180

nucleotide substitution for each codon partition determined with MrModeltest (Table S2). 181

Tree searches started from a random tree and assumed a constant lineage birth rate for each 182

branch (Yule tree prior; Drummond et al., 2006); all other parameters were kept at default 183

values. Chains were sampled every 1000 generations over 10x106 generations, with a burn-in 184

of 10%. 185

Fossil calibrations were used to place temporal constraints on two nodes, as minimum 186

and maximum soft bounds. The choice of these calibrations was difficult as extant Myotis 187

species show a combination of ancestral and derived character states and are difficult to 188

discriminate in the fossil record (Gunnell et al., 2012; Horácek, 2001). This is especially true 189

for the different dental remains that constitute the fossils of ancient Myotis-like taxa (Horácek, 190

2001). The first calibration constraint was the split of M. daubentonii and M. bechsteinii, 191

estimated to have diverged between 5 and 11.6 MYA (Topál, 1983). We used an exponential 192

prior distribution (offset 5.0, mean 2.5) to encapsulate this calibration in the 95% CI. The 193

second calibration involved the most recent common ancestor of Myotis estimated to have 194

diverged in the Late Oligocene - Early Miocene (Horácek, 2001; Gunnell et al., 2012), some 195

20-31 MYA. If Khonsunycteris is also part of the early Myotinae radiation, as hypothesized 196

by Gunnell et al. (2012), this lower boundary would be slightly older (34 MYA), but still 197

included in the lognormal prior distribution entered in our BEAST analysis (offset 20.0, S.D. 198

1.6). 199

200

2.4 Biogeographic reconstructions 201

Species distributions were obtained from the geographic information available in 202

Simmons (2005) and were coded in seven biogeographic regions (bioregions): Nearctic, 203

Neotropical, Western and Eastern Palaearctic, Oriental (= Indomalayan, Corbet and Hill, 204

1992), Ethiopian and Oceanian regions (see inset of Fig. 1 and Table S1 for details of 205

geographic assignations). The dispersal-extinction-cladogenesis (DEC) likelihood model 206

implemented in LAGRANGE (Ree and Smith, 2008) was used to investigate the 207

biogeographic evolution of the genus Myotis. Following a probabilistic approach, this 208

program traces from the tips to the root of a phylogenetic tree changes in distribution and 209

8

reconstructs the ancestral state by giving probabilities of each state at each node. We used the 210

ultrametric tree issued from the BEAST analysis as a template topology for all biogeographic 211

reconstructions. Although Myotis bats can be strong flyers, areas separated by large stretches 212

of open sea represent important, if not impassable barriers to dispersal (Castella et al., 2000; 213

García-Mudarra et al., 2009; Larsen et al., 2012b), while adjacent land masses can be reached 214

more easily. Hence, we performed biogeographic reconstructions assuming two dispersal 215

models, one without any constraint on dispersal across the planet (all regions can be reached 216

from any area; H0), and one where possible dispersals were restricted to adjacent land masses 217

(H1). In this constrained model, an adjacency matrix of connections between regions was 218

implemented as following: any dispersal was allowed within the Old World, or within the 219

New World, but between those continents, only dispersals from/to the Eastern Palaearctic and 220

the Nearctic were allowed (i.e. across the Beringian Strait). Similarly, connections with the 221

Oceanian region was only allowed to/from the Oriental region (i.e. across Wallacea), as 222

suggested by previous genetic studies of Myotis species living in this region (Kitchener et al., 223

1995). Because of the global scale of these analyses, we assumed that ancestral areas could 224

not cover more than two bioregions, and set the analyses accordingly. The best fitting results 225

were chosen for interpretation by comparing their log-likelihoods (see Ree and Smith, 2008). 226

The influence of the biogeographic origin of the outgroup species (here they were all sampled 227

within the Oriental region, see Table S1) may introduce a bias in the reconstruction of the 228

deepest nodes of the ingroup (personal observation). To avoid this potential problem, we 229

artificially assigned both the Ethiopian and Oriental regions to the geographic range of 230

Kerivoula, and the Oriental and Eastern Palearctic regions to the distribution of Murina, in 231

order to reflect the global distribution of these outgroup genera (Simmons, 2005). 232

233

3. Results 234

3.1 mtDNA and nuclear diversity 235

Thirty-nine sequences of the Cyt b gene were newly obtained, which represent 32 236

Oriental, 6 Palaearctic and one Ethiopian lineages of Myotis (Table S1). These sequences 237

were deposited in GenBank under accession numbers KF312497-KF312535. The final dataset 238

for the Cyt b gene thus consisted of 141 sequences of 1140 bp (135 ingroup and 6 outgroup 239

taxa). Whenever possible, we ensured that the same individual was sequenced for the Cyt b 240

and Rag 2 genes, but this was not possible for all species. Furthermore, because tissue sample 241

was not available for all species used in the Cyt b dataset, the taxon sampling of the Rag 2 242

dataset was reduced to 122 sequences (including 6 outgroups). The 48 new partial Rag 2 243

sequences obtained in this study were deposited in GenBank under accession numbers 244

9

KF312536-KF312583. A complete list of all specimens and sequences is provided in Table 245

S1. 246

247

3.2 Phylogenetic reconstructions 248

Phylogenetic trees based on ML or BA reconstructions were very similar, although 249

nodes tended to be less well supported by bootstraps (BP) when compared to posterior 250

probabilities (PP, Table 1), a well-known bias inherent to the different methods used to 251

estimate nodal support (Alfaro et al., 2003; Douady et al., 2003). The overall levels of support 252

were also higher towards the terminal clades than for deeper nodes, indicating that additional 253

mitochondrial and nuclear markers are required to resolve the entire Myotis radiation with 254

high statistic support. Deep nodes that were well supported in most analyses include the 255

monophyly of all current Myotis species and of the New World clade (Table 1). As expected 256

at low taxonomic levels, trees reconstructed with the more variable Cyt b gene (Fig. S1) were 257

much more resolved when compared with those generated with the Rag 2 gene alone (Fig. 258

S2). Phylogenetic incongruences between gene partitions include the position of latirostris or 259

the clade mystacinus-ikonnikovi, both of which appear within the New World clade (albeit 260

with low bootstrap) in the nuclear tree (Fig. S2), whereas it is basal in reconstructions 261

including the Cyt b data set (Figs. 1 and S1). Another study including many more nuclear 262

characters (2352 variable positions; Lack et al., 2010) also places with strong support 263

latirostris as the sister-group relative to the remaining Myotis, not within the New World 264

clade, suggesting that inference from the nuclear gene tree of Fig. S2 might require many 265

more characters or a larger sampling of nuclear genes to be more reliable. The combination of 266

both gene datasets resulted in significant improvement of most nodal supports (Table 1). We 267

therefore report only phylogenetic reconstructions resulting from this combined Cyt b + Rag 2 268

dataset and representing 141 species or lineages (Fig. 1). Phylogenetic tree can be accessed in 269

TreeBase (http://purl.org/phylo/treebase/phylows/study/TB2:S14471). 270

Three major monophyletic groupings are statistically supported (PP and BP 72-100 %) 271

in the early radiation of Myotis: the latirostris lineage, the New World clade, and the Old 272

World clade (Fig. 1). The latirostris lineage comprises M. latirostris from Taiwan and an 273

unidentified taxon (M. sp 4) from South China. This lineage is clearly basal to the Myotis 274

radiation. The monophyletic and strongly supported New World clade includes all sequenced 275

New World species of Myotis plus the Palaearctic brandtii lineage. The latter is represented 276

by the Western Palaearctic M. brandtii and the Eastern Palaearctic M. gracilis. The New 277

World clade is comprised of two further, well-supported subclades, one containing 278

10

exclusively Nearctic species, while the other is composed of both Nearctic and Neotropical 279

species, as detailed in Stadelmann et al. (2007). 280

The Old World clade encompasses the greatest species diversity with over 60 distinct 281

taxa, and is structured in two major groups: the Ethiopian clade, and the Eurasian assemblage, 282

the latter with low nodal support (Fig. 1, Table 1). Such low support can be due to a 283

succession of short internodes or to the uncertain position of two species, M. alcathoe and M. 284

dasycneme (called here “floating” species) that branch close to the base of this Eurasian 285

radiation, but with little statistical support. The Ethiopian clade is highly supported with all 286

methods of reconstruction, and has been described in Stadelmann et al. (2004b). It unites all 287

species from sub-Saharan Africa and islands of the western Indian Ocean (Weyeneth et al., 288

2011), plus the circum-Mediterranean M. emarginatus, and two Asian species, M. cf. 289

formosus and M. formosus flavus. Finally, the Eurasian assemblage includes all remaining 290

species from the Palaearctic, Oriental and Oceanian regions (Fig. 1). This assemblage can be 291

divided in several clades already described in previous studies (Clade II, III and IV, Ruedi and 292

Mayer 2001) and in five new ones, as indicated by continued Roman numerals (Fig. 1 and 293

Table 1). Some of these clades are composed exclusively of lineages from the Oriental region 294

(e.g. Clade VI or IX), while others, such as Clade II, are geographically more heterogeneous. 295

The European M. capaccinii and the South-East Asian M. annectans are monotypic lineages 296

with phylogenetic positions unsettled within this Eurasian assemblage. 297

Interestingly, species endemic to the Western Palaearctic region, such as M. 298

bechsteinii, M. escalerai, M. capaccinii, M. alcathoe or those from the Far East (e.g. M. 299

pruinosus, M. macrodactylus, M. yanbarensis, M. ikonnikovi) are largely unrelated and 300

dispersed across the entire phylogenetic tree (Fig. 1), while those from Africa or from the 301

Americas are much more closely related to their respective geographic congeners. Species 302

from the Oriental region are either closely related within an endemic clade (e.g. Clade VII or 303

IX) or part of geographically more widespread assemblages (e.g. Clade II or V). The most 304

dramatic endemic radiation involves species from the Nearctic and Neotropical subclades 305

(Stadelmann et al., 2007) that largely speciated within their biogeographic regions. 306

307

3.3 Divergence times 308

Divergence time estimates obtained with a relaxed molecular clock on the combined 309

dataset are reported on Fig. 2 and suggest that all Myotis diverged from their closest sister 310

group (represented here by several Murininae and Kerivoulinae species) about 26 MYA (HPD 311

31-21 MYA). The latirostris lineage diverged about 21 MYA from the lineage leading to all 312

other modern Myotis. The early divergence between the New World clade and the Old World 313

11

clade represents the second major split within the Myotis radiation and occurred about 19 314

MYA. Then, the Nearctic subclade started diverging about 12 MYA, while the brandtii 315

lineage split from the Neotropical subclade 11 MYA. The most recent common ancestor 316

shared by the Ethiopian clade and the Eurasian assemblage is about 17 MYA. The lineages 317

that have further diversified in the seven corresponding clades of the Eurasian assemblage 318

originated within a narrow time frame in the middle Miocene, between 16 and 11 MYA (Fig. 319

2). 320

Compared to our previous study based on the same markers but on a reduced taxon 321

sampling of Myotis species and on a different method to calibrate nodes (see Stadelmann et 322

al., 2007), the dates inferred here are largely concordant for the recent nodes (< 8 MYA), but 323

for the deeper ones, the dates inferred here are much older. For instance, the basal node for all 324

modern Myotis is estimated here at about 20.9 MYA (with 95% highest posterior density 325

interval, HPD = 23-20 MYA; Table 1), while it was at 13.0 ± 2.2 MYA in Stadelmann et al. 326

(2007). Another example is the origin of the New World clade, which was estimated at 9.9 ± 327

1.7 MYA by Stadelmann et al. (2007), while it appears about two million years older in the 328

present estimates (12.3 MYA, HPD 15-10 MYA). Other dated molecular trees aimed to 329

resolve the Vespertilionidae radiation, based on different genes, fossil calibration and taxon 330

sampling (Lack et al., 2010), are also more concordant with the present estimates. We 331

therefore feel that divergence times presented here are consistent with the best available 332

information, even if largely imprecise due to the usual uncertainty associated with molecular 333

tree calibrations. This calibrated phylogeny served as the basis for the following Lagrange 334

biogeographic reconstructions. 335

336

3.4 Biogeographic inference 337

The likelihood value of biogeographic reconstructions of the unconstrained-dispersal 338

model (H0 -lnL = 195.3) was significantly worse than the constrained one (H1 -lnL = 180.3; 339

delta lnL = 15; p < 0.001), and hence only this second model with dispersals limited to 340

adjacent bioregions will be discussed further. Overall, 111 out of the 134 ingroup nodes had 341

robust ancestral area reconstructions (AAR) with over 70% relative likelihood. For the 342

remaining nodes some uncertainties existed due to alternative, albeit less likely, AAR. For 343

instance, the AAR for the deepest node of all modern Myotis was not clearly identified, with 344

the co-occurrence of the Oriental and Eastern Palaearctic regions being more likely (35% 345

relative probability, as shown in Fig. 3), but with either of those two bioregions alone only 346

slightly less likely (23 and 21 %, respectively). Likewise, there is considerable uncertainty 347

about the most likely scenario for the area occupied by the ancestor leading to the Ethiopian 348

12

clade. This AAR was either identified as being Oriental (46% relative probability) or Western 349

Palaearctic (45%). To ease interpretation, we recorded only the ancestral area with the highest 350

relative likelihood on Fig. 3, but acknowledge that some of those AAR are uncertain. 351

Globally, the biogeographic scenario for the Myotis evolution inferred under the DEC 352

model (Fig. 3) suggests relatively few intercontinental colonisation events (noted as range 353

expansions on the tree). These are 21 in total, 13 of which stem from the Oriental, four from 354

the Western Palaearctic, two from the Eastern Palearctic, and one each from the Ethiopian and 355

Nearctic regions (see inset of Fig. 3). With only eight occurrences recorded on the tree, 356

vicariant events leading to daughter lineages were much rarer than range expansions (Fig. 3). 357

One such event occurred in the early ancestor of most modern Myotis that occupied the 358

Eastern Palaearctic plus Nearctic regions, and which subsequently split into an American 359

branch (leading eventually to all extant New World species (Clade I) and an Oriental branch 360

(leading to the Eurasian assemblage, clades II to X). Another major vicariant event was 361

identified in the Neotropical subclade (Clade I, Fig. 3) where the common ancestor of the 362

clade comprising M. grisescens, M. austroriparius, M. velifer and M. yumanensis (all are 363

Nearctic species) was split by vicariance from the remaining Neotropical taxa. According to 364

the dated phylogeny of Fig. 3, these inferred colonisation and vicariance events apparently 365

occurred sporadically during the entire evolution of the Myotis, with no temporal aggregation. 366

367

4. Discussion 368

369

4.1 Morphological convergences and taxonomy 370

371

The first major dichotomy in molecular reconstructions is the basal split of the 372

latirostris lineage that represent the sister group of all other modern Myotis species (Fig. 1). 373

This phylogenetic position of the latirostris lineage also appeared in analyses involving a 374

much larger character sampling (over 6000 aligned nucleotide positions) and wider taxonomic 375

coverage (Lack et al., 2010), confirming that this outlier position is not an artefact of our 376

sampling. Species representing this lineage (M. latirostris from Taiwan and an unnamed 377

species from China, M. sp4; table S1) have a lower molar configuration (nyctalodoncy, Fig. 378

S3) that would exclude them from members of the Myotinae, as this subfamily is supposed to 379

contain species with exclusively myotodont molars (Gunnell et al., 2012; Menu and Sigé, 380

1971). Based on those genetic and morphologic lines of evidences, the latirostris lineage 381

could be treated as a distinct genus (Lack et al., 2010; Stadelmann et al., 2007). However the 382

presence of other fully nyctalodont species among crown-group Myotis (e.g. M. siligorensis 383

13

alticraniatus; Fig. S3) suggest that this character state was acquired several times within the 384

group and that other truly synapomorphic characters are needed to diagnose the unique 385

latirostris lineage. Remarkably, the overall conserved morphology and very close 386

resemblance to other small-sized, Asian Myotis led most authors to include latirostris in the 387

synonymy of M. muricola (Corbet and Hill, 1992; Simmons, 2005), which is clearly 388

inappropriate owing to their paraphyletic position (Fig. 1). 389

The second dichotomy within the Myotis radiation separates all New World (Clade I) 390

from most Old World species (clades II to X), the only exceptions being two Palaearctic taxa 391

embedded within Nearctic species (Fig. 1). These two boreal species, M. brandtii and M. 392

gracilis [for which M. sibiricus may be the appropriate name according to Kruskop et al. 393

(2012)] are morphologically almost indistinguishable from each other and were long 394

considered as synonyms of M. mystacinus. They differ from mystacinus by subtle penial and 395

dental characteristics, yet appear as distantly related on the Myotis tree (Fig. 1). This further 396

illustrates that close morphological resemblance in this difficult genus does not necessarily 397

imply close phylogenetic relationships (Ruedi and Mayer, 2001). This observation also holds 398

for other chiropteran taxa, for example the Phyllostomidae genera (Davalos et al., 2012). 399

The evolution of the New World clade was already described in a previous paper 400

(Stadelmann et al., 2007). Since the current sampling only adds a few more species (M. 401

nesopolus and M. nyctor) which do not alter the main phylogenetic relationships within this 402

clade, we will not discuss it in more detail here. We do note that the multiple new lineages or 403

cryptic species recently evidenced in South America (Larsen et al., 2012) are all nested within 404

the Nearctic or the Neotropical subclades, as predicted by their geographic origins. This 405

supports our supposition that our sampling of Myotis species, although not complete, is 406

representative of the major lineages existing in this continent. 407

The Old World clade is fairly well supported in our reconstructions (Table 1) and 408

includes all species from the remaining continents in the Western Hemisphere (clades II to X, 409

Fig. 1). This large assemblage unites morphologically disparate species, including the 410

smallest (M. siligorensis, less than 4 grams) and the largest species (M. chinensis, over 30 411

grams), orange (M. anjouanensis) and solid black species (M. macropus), short-eared (M. 412

daubentonii) and long-eared species (M. bechsteinii) or short-footed (M. frater) and long-413

footed species (M. pilosus). Conversely, examples of morphologically very similar species 414

that appear distantly related within this large clade are also numerous. For instance the 415

Eastern Palaearctic M. petax was long considered a single species with the Western 416

Palaearctic M. daubentonii, but multiple lines of evidence (Datzmann et al., 2012; Kruskop et 417

al., 2012; Matveev et al., 2005) clearly show that they are not even sister-species (they appear 418

14

in clades X and III, respectively, on Fig. 1). Even species sharing a remarkable black-and-419

orange pigmentation (e.g. M. welwitschii, M. cf. formosus and M. formosus flavus) are not 420

each other‟s closest relatives, although they do all belong to the Ethiopian clade (Clade V, 421

Fig. 1). Phylogenetic relationships rather indicate that the striking black-and-orange 422

patterning of these Myotis is either a plesiomorphic feature among members of the Ethiopian 423

clade, or is a convergent character. These examples confirm that convergence in several 424

morphological characters can mislead the systematics of Myotis bats. 425

Strong, misleading resemblances of unrelated taxa may also explain the difficulties in 426

classifying properly many species. In some cases, species that clearly fit within the phylogeny 427

of Myotis were originally classified in different genera [e.g. M. annectans and M. ridleyi were 428

originally classified as Pipistrellus and M. rosseti as Glischropus because they often lack the 429

middle upper and lower premolars, although other morphological and genetic characters 430

clearly indicate they belong in Myotis (Hill and Topál, 1973; Topál, 1970)]. In other cases, 431

multiple unrelated lineages with similar morphology have been classified as the same species 432

(e.g. M. nigricans Larsen et al., 2012a). Hence several of the specimens included in our 433

molecular sampling, although keying out to known species, bear “cf.” in their name to warn 434

that they are not assigned confidently to a particular taxonomic name. This is particularly 435

evident in the M. horsfieldii, M. montivagus and M. muricola species groups that are certainly 436

composed of several cryptic taxa (Francis et al., 2010; Görföl et al., 2013; Volleth and Heller, 437

2012). Other unnamed taxa are more divergent, both morphologically and genetically, and 438

await formal description (e.g. unknown species 1 to 7). Such descriptions will require 439

integrated approaches combining molecular and morphological characters to ensure that they 440

are accurately described in relation to currently recognized taxa. 441

The next major phylogenetic subdivision of the Myotis tree separates the Ethiopian 442

Clade V from the Eurasian clade (clades II to IV and VI to X, Fig. 1). The former is well 443

supported in all analyses and was described by Stadelmann et al. (2004b). We add to this 444

clade M. anjouanensis from the Comoros and both Oriental species of the formosus species 445

complex (Jiang et al., 2010). Interestingly, the latter are not each other closest relatives as M. 446

formosus flavus is sister to the circum-Mediterranean M. emarginatus, while M. cf. formosus 447

is more closely related to the African M. welwitschii. The Ethiopian clade is thus composed of 448

all sampled sub-Saharan species (M. morrisi remains unsampled) plus at least three Eurasian 449

and Oriental species. 450

The Eurasian assemblage is the most speciose and includes at least 57 recognized 451

species and a number of undescribed taxa (Fig. 1). The monophyly of this Eurasian 452

assemblage is not supported statistically (Table 1) probably due to the four “floating” species 453

15

that are variably placed near the base or in a more internal position within this assemblage. 454

These “floating” species include M. alcathoe, M. dasycneme, and M. capaccinii, all 455

distributed mainly in Europe, and the Asian species M. annectans (Fig. 1). It is unclear if the 456

phylogenetic uncertainty reflects hard polytomies (i.e. episodes of rapid radiation), or merely 457

soft polytomies that could be better resolved with increased character sampling (Lack and 458

Van Den Bussche, 2010). Published DNA barcode sequences for M. gomantongensis suggest 459

that it may be allied to M. annectans (Francis et al., 2010) and hence to this Eurasian 460

assemblage, but we were not able to include it in our analysis as we lacked the matching 461

genes. 462

Our molecular results further demonstrate the existence of at least eight strongly 463

supported clades within this Eurasian assemblage. Clade IX is composed exclusively of 464

species endemic to the Oriental region. Clade II is geographically more heterogeneous as it 465

includes species endemic to Europe, to Central or to Eastern Asia, all within Eurasia. Clade 466

III is a composite of geographically and ecomorphologically very distinct Myotis. It includes 467

the surface gleaner M. bechsteinii, the aerial feeder M. frater, the trawling bat M. daubentonii, 468

as well as species living in mountain forests of the Himalayas (M. sicarius) or found in 469

lowland bamboo forests in Taiwan (M. sp3). Similarly, the Oriental Clade IV includes several 470

trawling species with large feet (e.g., macrotarsus, horsfieldii, and hasseltii) as well as some 471

unusual small species with small feet and only two premolars (M. ridleyi and M. rossetti). 472

This diversity illustrates the rapid morpho-anatomical and/or ecological changes that can 473

occur within these clades. 474

475

4.2 Biogeographic evolution of the genus Myotis 476

477

Among placental mammals, molecular characters have shown that there is a 478

widespread pattern of phylogenetic relationships matching with the geographical origin of the 479

members of the corresponding major clades (Murphy et al., 2001). At lower taxonomic scales 480

within laurasiatherians, we here recover a similar pattern for Myotis bats as continental 481

assemblages of species, e.g. those found in Africa, in North and South America, or in Eurasia 482

tend to appear in monophyletic groups (Fig. 1). This suggests that most speciation events and 483

diversification occurred within each continent independently. The strong influence of the 484

biogeographic origin of the Myotis species in shaping global phylogenetic relationships 485

corroborate one earlier study based on less complete taxonomic sampling (Stadelmann et al., 486

2007). Indeed, the vast majority of species segregate in clades with a strong geographic 487

component, especially if we consider that the Eurasian continent is a huge landmass dissected 488

16

by rather loose biogeographic boundaries. For instance, the northeastern portion of the 489

Oriental region has no major topological discontinuity that would separate its faunal elements 490

from those in the Eastern Palaearctic region. This is reflected by a broad transition zone 491

comprising faunal elements typical of both regions (Corbet and Hill, 1992), and in the 492

phylogenetic relationships of Myotis by a mixture of endemics from both regions in the same 493

clades (Fig. 1). The same is true between the Western and the Eastern Palaearctic regions 494

where the limits are fuzzy and not bounded to major geographic discontinuities. The evolution 495

of ancestral areas reconstructed onto the phylogenetic relationships of Myotis (Fig. 3) reflect 496

these fuzzy biogeographic boundaries within the Eurasian continent, as the vast majority (16 497

out of 21) of ancestral range expansions recorded worldwide occurred within this huge land 498

mass (see inset of Fig. 3). 499

Barriers between the other biogeographic regions, such as mountain ranges, large 500

deserts or sea channels, appear to be much more effective at reducing dispersal of Myotis 501

species (Castella et al., 2000; García-Mudarra et al., 2009). For instance, none of the species 502

found on either side of the Sahara seem to have a common ancestor that lived across this 503

desert. Likewise no current Myotis species have a Holarctic distribution, and AAR suggest 504

that a single expansion across the Beringian Strait in the Middle Miocene could explain the 505

occurrence of Myotis species in the New World (Fig. 3). This scenario is strongly supported 506

by the Lagrange analysis (relative probability >98%) and contradicts the alternative scenario 507

of a back colonization of a putative common ancestor of brandtii/gracilis lineage coming 508

from the Nearctic region. 509

If we assume that all current species living on the Australian continent form a 510

monophyletic assemblage (Cooper et al., 2001), including M. macropus represented here in 511

Clade IV, the colonisation of this southern landmass can also be traced back to a single event 512

that occurred at the end of the Miocene epoch (Fig. 3). Interestingly, Myotis species from the 513

Japanese Archipelago (Kawai et al., 2003) appear in various independent clades (e.g. M. 514

macrodactylus, M. yanbarensis or M. ikonnikovi), suggesting that these islands were 515

colonized independently by several lineages before evolving eventually into endemic taxa. 516

The high proportion of expansion/colonisation events originating from the Oriental 517

region clearly highlights the pivotal importance of this region in the diversification of the 518

Myotis species worldwide. AAR (Fig. 3) also suggests that only eight ancestral ranges were 519

split by vicariance into descendent lineages, whereas most other diversification events took 520

place within the bioregions. However, as the geographic scale of each bioregion is very large 521

(i.e. continent-level) and probably many more vicariant events may have occurred within each 522

of these large and heterogeneous regions, our data are not sufficient to infer more precisely 523

17

which mode of speciation (sympatric versus allopatric) prevailed during the evolution of 524

Myotis. 525

The global number of species contained in each biogeographic region is very unequal, 526

with the northern continents much more diversified than the southern ones. For instance, the 527

number of species currently known to inhabit the Ethiopian region (up to 6 species, excluding 528

Madagascar and the Comoros; Happold and Happold, 2013) is in sharp contrast with the 529

species diversity found on other continents (except Australia; see inset of Fig. 1). This 530

suggests that Myotis species from the African continent did not radiate into many endemic 531

species despite the fact that genuine Myotis apparently entered into this region relatively early 532

(Middle Miocene; Fig. 3) during the evolution of modern species. Given the poor record of 533

Tertiary fossils of vespertilionids in Africa (Gunnell et al., 2012), it is not possible to know if 534

this paucity of species is due to an increased rate of extinction or a decreased rate of 535

diversification. By contrast, the Oceanian region was likely the last part of the world that was 536

reached by Myotis species, about 5.0 MYA (Fig. 3), suggesting that a late arrival might have 537

contributed to low current diversity of Myotis species found in Australia. 538

539

4.3 Fossils and molecules in the Myotinae 540

541

The Myotis species are rather abundant and already diverse in the European fossil 542

record of the Miocene through the Pleistocene, and recent studies (Gunnell et al., 2012; 543

Horácek, 2001) suggest that their ancestry might date back to the Late Oligocene. The 544

difficulty in assigning the most ancient fossils, often represented by isolated tooth (Menu et 545

al., 2002) or dentary fragments, is that the genus Myotis itself is defined by a combination of 546

plesiomorphic characters (e.g. three upper and lower premolars), as well as more derived ones 547

shared with other genera (e.g. single-rooted third premolars). The most prominent feature 548

believed to be characteristic of the genus Myotis is the myotodont configuration of lower 549

molars (Menu and Sigé, 1971). According to these dental characters, several of the more 550

ancient fossils initially attributed to Myotis (e.g. the Early Oligocene “Myotis” misonnei) are 551

now assigned to other genera (e.g. Quietia or Stehlinia) (Horácek, 2001) that are not related to 552

the Myotinae. By relying mainly on these dental characters, Gunnel et al. (2012) recently 553

revised the fossil origins of the Myotinae and stressed that there is a large gap between the 554

divergence times estimated from the molecules and the first appearance of fossil Myotis on 555

the different continents. The dated molecular phylogeny of Fig. 2 and divergence dates 556

independently obtained by Lack et al. (2010) now suggest that the most recent common 557

ancestor of all modern Myotis (including latirostris) is older than previously estimated, with a 558

18

divergence of 21 MYA (95% HPD 23-20 MYA, Table 1) or 18 MYA (HPD 23-13 MYA, 559

Lack et al., 2010). As a lower bound these molecular reconstructions also suggest that the 560

closer outgroups to the Myotinae (the Kerivoulinae-Murininae subfamilies) diverged in the 561

Late Oligocene (about 26 MYA, HPD 31-21 MYA; Fig. 3 and Lack et al., 2010). These 562

molecular estimates are still over 10 million years younger than the presumed oldest Myotinae 563

fossil (Khonsunycteris, 34 MYA, Gunnel et al., 2012). A plausible explanation for these large 564

temporal discrepancies between molecular and fossil interpretation is that the dental 565

characteristics used to assign fossil fragments to the Myotinae are convergent characters that 566

do not identify unambiguously members of this subfamily. For instance, it is now clear that 567

species in the siligorensis group are nyctalodont or semi-nyctalodont (Borisenko et al., 2008; 568

Tiunov et al., 2011), including all specimens of M. s. alticraniatus analyzed here (Fig. S3 and 569

Table S1). These specimens are nested within the Asian clade IX (Fig. 1), but on a 570

paleontological point of view, their cranial characteristics would have excluded them from the 571

Myotinae. Myotis latirostris and the unknown species 4 both have entirely nyctalodont molars 572

(Fig. S3) and are basal to, but clearly part of the early Myotinae radiation (Fig. 1), yet have 573

diverged only about 21 MYA from other recent species of Myotis (Table 1). These modern 574

taxa share striking dental and cranial similarities with a fossil genus, Submyotodon that was 575

described recently from the Upper part of the Middle Miocene deposits in Europe, but 576

classified away from Myotis lineages owing to its dental features (Ziegler, 2003). Finally, 577

members of the genus Cistugo have all the dental characteristics of Myotinae, including 578

myotodont molars (Fig. S3), yet they differ by the presence of unique wing glands and a 579

distinctive karyotype (Rautenbach et al., 1993), and are also genetically so divergent from any 580

other Vespertinionidae that they are now classified in their own family (Lack et al., 2010). 581

Again, paleontologically, Cistugo species would have been classified within the genus Myotis 582

as all those distinguishing features would have been invisible on fossils. Given these 583

morphological ambiguities, we therefore postulate that the most ancient, Oligocene fossils of 584

Myotis-like bats (including Khonsunycteris or Leuconoe = Myotis) are stem taxa not directly 585

related to extant Myotis, nor even part of the subfamily Myotinae. Their exact phylogenetic 586

affinities should therefore be reevaluated. This problematic paleontological assignation of 587

dental remains might also explain the great apparent gap between the Early Miocene (about 588

20 MYA) presence of Myotis–like bats in Australia, as suggested by Menu et al. (2002), and 589

our molecular reconstructions that date their earliest arrival on this continent in the Pliocene 590

(Fig. 3), about 5 MYA. Such a late, Pliocene, arrival of Myotinae in Australia is also 591

compatible with a scenario based on chromosomal evolution as shown for Australian 592

Vespertilionini (Volleth and Tidemann, 1991). 593

19

Molecular divergence times of younger nodes are more concordant with 594

paleontological data. Indeed, fossil deposits show a burst of diversification during the Middle 595

Miocene, in particular in Europe (Ziegler, 2003), which corroborates the emergence of most 596

basal lineages within the Eurasian assemblage (Fig. 2). 597

598

4.4 Paleoenvironmental changes and Myotis evolution 599

600

The Middle Miocene is characterized by a major climatic transition where global 601

cooling provoked a drop in sea level, and the development of more extensive temperate 602

habitats. These major environmental changes correspond to faunal turnovers (Janis, 1993) that 603

coincide with the basal radiation of crown-group Myotis, including the divergence of the Old 604

World versus New World clades (Fig. 2) that presumably occurred after a range expansion of 605

an ancestral Eastern Palaearctic taxon into North America , some 12.3 MYA (HPD 15-10 606

MYA, Table 1). 607

The northward drift of the African-Arabian plate resulted about 18-19 MYA in the 608

gradual closure of the Tethyan Seaway which had previously separated Africa from Eurasia. 609

Some mammals experienced faunal interchanges at that early time, but Myotis (and other 610

groups such as horses) apparently dispersed much later between these two continents, as 611

indicated by the colonization date of ancestors of the Ethiopian clade (12.3 MYA, HPD 15-10 612

MYA, Table 1, which corresponds to the Astracian age). It seems that besides the creation of 613

the land bridge, a change in habitat, probably driven by climatic factors, was necessary to the 614

establishment of dispersal routes suitable for Myotis. Our biogeographic reconstructions 615

suggest that the Ethiopian clade is most probably linked to an Oriental stock (Fig. 3). This 616

supposed Oriental origin together with the fact that the Sahara and its adjacent regions have 617

gradually shifted from a tropical to an arid environment since the Neogene warmth climax, 618

suggest that Myotis took advantage of these paleoenvironmental changes to reach Africa from 619

Asia (Fig. 3). The Asian M. cf. formosus represents the sister lineage of the African M. 620

welwitschii and colonized secondarily the Oriental region out of Africa (Fig. 3). Likewise the 621

biogeographic interpretation of the AAR of Fig. 3 suggests that M. emarginatus, now 622

confined to the Mediterranean region, descended from widespread ancestors that ranged 623

across Asia and Africa, then became isolated in the Oriental region, and finally reached 624

Europe through a range expansion (Fig. 3). 625

The Late Miocene epoch coincides with further cooling of the climate and 626

aridification, resulting in a continuing reduction of both tropical forests, and expansion of 627

more open and temperate habitats. This might have triggered the diversification of the Old 628

20

World Myotis lineages, especially the species-rich assemblage of Eurasian taxa. Indeed, 629

temperate forests, whether present in high latitude regions or in mountain ranges of 630

intertropical regions, characterize the habitats occupied today by most species of Myotis (La 631

Val, 1973). 632

633

4.5 Geographic origins of the Myotinae 634

635

The fossil record of Tertiary vespertilionids in tropical Asia (Mein and Ginsburg, 636

1997) or Africa (Butler, 1984) is poorly documented compared to that found in Europe (e.g. 637

Horácek, 2001; Ziegler, 2003) or North America (Gunnell and Simmons, 2005). Hence any 638

firm biogeographic inference based solely on the relative abundance or on the first occurrence 639

of Myotis fossils in sediments is difficult. The presumed African origin for the subfamily 640

Myotinae proposed by Gunnell and colleagues (Gunnell et al., 2012) strongly relies on the 641

dentary fragments of Khonsunycteris (a Late Eocene fossil found in Egypt) being part of that 642

subfamily. We however showed that this interpretation is questionable, as taxa unrelated to 643

the Myotinae, such as the endemic African Cistugo, can share all dental characteristics 644

believed to be unique to the genus Myotis. Contrary to the current interpretation of the fossil 645

record, our molecular phylogeny (Fig. 1) and likelihood biogeographic reconstructions (Fig. 646

3) identify eastern Asia as the cradle of Myotis evolution. There is some uncertainty whether 647

the Oriental region alone or in combination with the Eastern Palaearctic (as shown on Fig. 3) 648

is the origin of early Myotinae, but alternative bioregions are clearly less likely. In addition, 649

13 of the 21 intercontinental range expansions originated from the Oriental region (inset of 650

Fig. 3) suggesting that this area is a center of origin for the radiation of Myotis species. The 651

deepest lineage identified in molecular reconstructions, the latirostris lineage, that is sister to 652

all remaining modern Myotis (Fig. 1 and Lack et al., 2010), is also endemic to this region, 653

further supporting an ancient, East Asian origin for this radiation. Finally, the Myotis species 654

diversity in the Oriental region (inset of Fig. 1) is already the highest in the world based on 655

currently recognized taxa, and is likely strongly underestimated (Francis et al., 2010), again 656

pointing to this area as a cradle of evolution for the genus. 657

658

4.6 Conclusions 659

660

Reconstruction of the biogeographic origin of Myotis (Fig. 3) suggests that the East 661

Asian region is the center of origin for the radiation of modern Myotis and is thus probably 662

also the cradle for the entire Myotinae. The Indo-Malayan, Indo-Chinese, Sino-Himalayan, 663

21

and East Asiatic floras and faunas (Corbet and Hill, 1992) meet in this region and engendered 664

the highest biodiversity hotspots on earth (Myers et al., 2000). This geographical area, which 665

has been subjected to gradual uplifting of the Tibetan plateau (the trans-Himalayan range) 666

during the last several million years, includes several high mountain ranges exhibiting sharp 667

topographical complexity, and shows climates ranging from tropical to arctic. Such high 668

environmental and topographic complexity could have promoted speciation. 669

Considering the potential vagility of bats, and their current worldwide distribution, the 670

overall number of transcontinental migrations in the Myotis radiation is relatively low (Fig. 3). 671

This illustrates the relative inability of Myotis species to cross some physical barriers. In turn, 672

Myotis species can occupy a variety of ecological niches, and it is not exceptional to find up 673

to 12 species coexisting in sympatry. This and the high ecological and morphological 674

diversity found within each Myotis assemblage at a worldwide scale support the central 675

impact of biogeography on the Myotis evolutionary history. Climatic and vegetation changes 676

since the Middle Miocene might also have favored speciation during the evolution of Myotis. 677

Recent molecular surveys of Myotis taxa from different parts of the world identified 678

that several species currently diagnosed by morphological characters certainly contain species 679

complexes: M. muricola, M. siligorensis, M. horsfieldii (Francis et al., 2010), M. montivagus 680

(Görföl et al., 2013; Volleth and Heller, 2012), M. nigricans (Larsen et al., 2012a), and even 681

European species such as M. nattereri (Salicini et al., 2011). Many more cryptic species are 682

present in the Oriental region, as suggested by the presence of numerous unnamed taxa in our 683

reconstructions (Table S1). Here, phylogenetic tools based on the comparison of polymorphic 684

mitochondrial and nuclear molecular markers will be of central interest to identify cryptic 685

species, refine the Myotis systematics, and help choosing among priorities for biological 686

conservation purposes. Unfortunately, this region of high biodiversity is also facing high rates 687

of habitat destruction, and is in urgent need of protection to maintain this evolutionary 688

diversity (Sodhi et al., 2004). 689

690

Acknowledgements 691

We are deeply indebted to those individuals and institutions who generously donated 692

tissues for genetic analyses. These samples provide the foundation of such broad taxonomic 693

surveys and would not be possible without their help. These individuals are: P. Benda 694

(Natural History Museum, Prague), L. R. Heaney and W. Stanley (Field Museum of Natural 695

History, Chicago, USA), G. Jones (University of Bristol, UK), K. Kawai (Institute of Low 696

Temperature Science, Hokkaido University, Sapporo, Japan), T. Kingston (Boston 697

University, Boston, USA), S. Kruskop and A. V. Borisenko (Zoological Museum of Moscow 698

22

State University), T. H. Kunz (Boston University, Boston, USA), J. Ma (Beijing, China), V. 699

Matveev (Moscow Lomonosov State University, Russia), late O. von Helversen, and F. 700

Mayer (University of Erlangen, Germany). This research was supported by grants from the 701

Swiss National Funds for Scientific research to M. Ruedi and A. Cibois (3100A0-105588), 702

and from Bat Conservation International to B. Stadelmann. This publication is contribution No 703

2013-XXX of the Institut des Sciences de l‟Evolution de Montpellier (UMR 5554 – CNRS - 704

IRD). 705

706

References 707

Abe, H., Ishii, N., Itoo, T., Kaneko, Y., Maeda, K., Miura, S., Yoneda, M., 2005. A guide to 708 the mammals of Japan. Tokai University Press, Kanagawa, Japan. 709 Alfaro, M.E., Zoller, S., Lutzoni, F., 2003. Bayes or bootstrap? A simulation study comparing 710 the performance of Bayesian Markov Chain Monte Carlo sampling and bootstrapping in 711 assessing phylogenetic confidence. Mol. Biol. Evol. 20, 255-266. 712 Bates, P.J.J., Hendrichsen, D.K., Walston, J.L., Hayes, B., 1999. A review of the mouse-eared 713 bats (Chiroptera: Vespertilionidae: Myotis) from Vietnam with significant new records. Acta 714 Chiropterol. 1, 47-74. 715 Bickham, J.W., Patton, J.C., Schlitter, D.A., Rautenbach, I.L., Honeycutt, R.L., 2004. 716 Molecular phylogenetics, karyotypic diversity, and partition of the genus Myotis (Chiroptera: 717 Vespertilionidae). Mol. Phylogenet. Evol. 33, 333-338. 718 Bininda-Emonds, O.R.P., Vazquez, D.P., Manne, L.L., 2000. The calculus of biodiversity: 719 integrating phylogeny and conservation. Trends Ecol. Evol. 15, 92-94. 720 Borisenko, A.V., Kruskop, S.V., Ivanova, N.V., 2008. A new mouse-eared bat (Mammalia: 721 Chiroptera: Vespertilionidae) from Vietnam. Russian J. Theriol. 7, 57-69. 722 Butler, P.M., 1984. Macrocelidea, Insectivora and Chiroptera from the Miocene of East 723 Africa. Paleovertebrata 14, 117-200. 724 Castella, V., Ruedi, M., Excoffier, L., 2001. Contrasted patterns of mitochondrial and nuclear 725 structure among nursery colonies of the bat Myotis myotis. J. Evolution. Biol. 14, 708-720. 726 Castella, V., Ruedi, M., Excoffier, L., Ibáñez, C., Arlettaz, R., Hausser, J., 2000. Is the 727 Gibraltar Strait a barrier to gene flow for the bat Myotis myotis (Chiroptera: 728 Vespertilionidae)? Mol. Ecol. 9, 1761-1772. 729 Cooper, S.J.B., Day, P.R., Reardon, T.B., Schulz, M., 2001. Assessment of species 730 boundaries in Australian Myotis (Chiroptera : Vespertilionidae) using mitochondrial DNA. J. 731 Mammal. 82, 328-338. 732 Corbet, G.B., Hill, J.E., 1992. The mammals of the Indomalayan region: a systematic review. 733 Oxford University Press, Oxford, UK. 734 Datzmann, T., Dolch, D., Batsaikhan, N., Kiefer, A., Helbig-Bonitz, M., phel, U., Stubbe, M., 735 Mayer, F., 2012. Cryptic diversity in Mongolian vespertilionid bats (Vespertilionidae, 736 Chiroptera, Mammalia). Results of the Mongolian-German biological expeditions since 1962, 737 No. 299. Acta Chiropterol. 14, 243-264. 738 Davalos, L.M., Cirranello, A.L., Geisler, J.H., Simmons, N.B., 2012. Understanding 739 phylogenetic incongruence: lessons from phyllostomid bats. Biol. Rev. 87, 991-1024. 740 Douady, C., Delsuc, F., Boucher, Y., Doolittle, W., Douzery, E., 2003. Comparison of 741 Bayesian and maximum likelihood bootstrap measures of phylogenetic reliability. Mol. Biol. 742 Evol. 20, 248-254. 743 Drummond, A.J., Ho, S.Y.W., Phillips, M.J., Rambaut, A., 2006. Relaxed phylogenetics and 744 dating with confidence. Plos Biol. 4, 699-710. 745

23

Drummond, A.J., Rambaut, A., 2007. BEAST: Bayesian evolutionary analysis by sampling 746 trees. Bmc Evolutionary Biology 7, 214. 747 Evin, A., Baylac, M., Ruedi, M., Mucedda, M., Pons, J.M., 2008. Taxonomy, skull diversity 748 and evolution in a species complex of Myotis (Chirotpera: Vespertilionidae): a geometric 749 morphometric appraisal. Biol. J. Linn. Soc. 95, 529-538. 750 Fenton, M.B., Bogdanowicz, W., 2002. Relationships between external morphology and 751 foraging behaviour: bats in the genus Myotis. Can J. Zool. 80, 1004-1013. 752 Findley, J.S., 1972. Phenetic relationships among bats of the genus Myotis. Syt. Zool. 21, 31-753 52. 754 Francis, C., Borisenko, A., Ivanova, N., Eger, J., Lim, B., Guillen-Servent, A., Kruskop, S., 755 Mackie, I., Hebert, P., 2010. The role of DNA barcodes in understanding and conservation of 756 mammal diversity in Southeast Asia. Plos One 5, e12575. 757 García-Mudarra, J.L., Ibañez, C., Juste, J., 2009. The Straits of Gibraltar: barrier or bridge to 758 Ibero-Moroccan bat diversity? Biol. J. Linn. Soc. 96, 434-450. 759 Görföl, T., Estók, P., Csorba, G., 2013. The subspecies of Myotis montivagus - taxonomic 760 revision and species limits (Mammalia: Chiroptera: Vespertilionidae). Acta Zool. Hung. 59, 761 41-59. 762 Gunnell, G.F., Eiting, T.P., Simons, E.L., 2012. African Vespertilionoidea (Chiroptera) and 763 the antiquity of Myotinae. In: Gunnell, G.F., Simmons, N.B. (Eds.), Evolutionary history of 764 bats. Fossils, molecules and morphology. Cambridge University Press, New York, pp. 252-765 266. 766 Gunnell, G.F., Simmons, N.B., 2005. Fossil evidence and the origin of bats. J. Mamm. Evol. 767 12, 209-246. 768 Han, N.-J., Zhang, J.-S., Reardon, T., Lin, L.-K., Zhang, J.-P., Zhang, S.-Y., 2010. 769 Revalidation of Myotis taiwanensis Ärnbäck-Christie-Linde 1908 and its molecular 770 relationship with M. adversus (Horsfield 1824) (Vespertilionidae, Chiroptera). Acta 771 Chiropterol. 12, 449-456. 772 Happold, M., Happold, D.C.D., 2013. The Mammals of Africa: Hedgehogs, Shrews and Bats. 773 Bloomsbury Publishing, London, UK. 774 Helversen, v.O., Heller, K.-G., Mayer, F., Nemeth, A., Volleth, M., Gombkötö, P., 2001. 775 Cryptic mammalian species: a new species of whiskered bat (Myotis alcathoe n. sp.) in 776 Europe. Naturwissenschaften 88, 217-223. 777 Hill, J.E., Topál, G., 1973. The affinities of Pipistrellus ridleyi Thomas 1898 and Glischropus 778 rosseti Oey, 1951 (Chiroptera: Vespertilionidae). Bull.br.Mus.nat.Hist.Zool. 24, 447-454. 779 Hoofer, S.R., Reeder, S.A., Hansen, E.W., Van den Bussche, R.A., 2003. Molecular 780 phylogenetics and taxonomic review of noctilionoid and vespertilionoid bats (Chiroptera: 781 Yangochiroptera). J. Mammal. 84, 809-821. 782 Hoofer, S.R., Van den Bussche, R.A., 2003. Molecular phylogenetics of the chiropteran 783 family Vespertilionidae. Acta Chiropterol. 5 (supplement), 1-63. 784 Horácek, I., 2001. On the early history of vespertilionid bats in Europe: the Lower Miocene 785 record from the Bohemian Massif. Lynx 32, 123-154. 786 Horácek, I., Hanák, V., Gaisler, J. (Eds.), 2000. Bats of the Palearctic region: a taxonomic 787 and biogeographic review. Platan publishing house, Warsaw, Poland. 788 Ibáñez, C., García-Mudarra, J.L., Ruedi, M., Stadelmann, B., Juste, J., 2006. The Iberian 789 contribution to cryptic diversity in European bats. Acta Chiropterol. 8, 277-297. 790 Janis, C.M., 1993. Tertiary mammal evolution in the context of changing climates, vegetation, 791 and tectonic events. Annu. Rev. Ecol. Syst. 24, 467-500. 792 Jiang, T.L., Sun, K.P., Chou, C.H., Zhang, Z.Z., Feng, J., 2010. First record of Myotis flavus 793 (Chiroptera: Vespertilionidae) from mainland China and a reassessment of its taxonomic 794 status. Zootaxa 2414, 41-51. 795 Jones, G., Parsons, S., Zhang, S.Y., Stadelmann, B., Benda, P., Ruedi, M., 2006. Echolocation 796 calls, wing shape, diet and phylogenetic diagnosis of the endemic Chinese bat Myotis 797 pequinius. Acta Chiropterol. 8, 451-463. 798

24

Kawai, K., Nikaido, M., Harada, M., Matsumura, S., Lin, L.K., Wu, Y., Hasegawa, M., 799 Okada, N., 2002. Intra- and interfamily relationships of Vespertilionidae inferred by various 800 molecular markers including SINE insertion data. J. Mol. Evol. 55, 284-301. 801 Kawai, K., Nikaido, M., Harada, M., Matsumura, S., Lin, L.K., Wu, Y., Hasegawa, M., 802 Okada, N., 2003. The status of the Japanese and East Asian bats of the genus Myotis 803 (Vespertilionidae) based on mitochondrial sequences. Mol. Phylogenet. Evol. 28, 297-307. 804 Kitchener, D.J., Cooper, N., Maryanto, I., 1995. The Myotis adversus (Chiroptera: 805 Vespertilionidae) species complex in Eastern Indonesia, Australia, Papua New Guinea and the 806 Solomon Islands. Rec. West. Aust. Mus. 17, 191-212. 807 Koopman, K.F., 1993. Order Chiroptera. In: Wilson, D.E., Reeder, D.M. (Eds.), Mammal 808 species of the World. A taxonomic and geographic reference. Smithsonian Institution Press, 809 Washington, pp. 137-242. 810 Koopman, K.F., 1994. Chiroptera: Systematics. In: Niethammer, J., Schliemann, H., Starck, 811 D. (Eds.), Handbuch der Zoologie. de Gruyter, Berlin, Germany, pp. 100-109. 812 Kruskop, S.V., Borisenko, A.V., Ivanova, N.V., Lim, B.K., Eger, J.L., 2012. Genetic diversity 813 of northeastern Palaearctic bats as revealed by DNA barcodes. Acta Chiropterol. 14, 1-14. 814 La Val, R.K., 1973. A revision of the Neotropical bats of the genus Myotis. Natural History 815 Museum of Los Angeles County, Sciences Bulletin 15, 1-54. 816 Lack, J.B., Roehrs, Z.P., Stanley Jr, C.E., Ruedi, M., Van den Bussche, R.A., 2010. Molecular 817 phylogenetics of Myotis indicate familial-level divergence for the genus Cistugo (Chiroptera). 818 J. Mammal. 91, 976-992. 819 Lack, J.B., Van Den Bussche, R.A., 2010. Identifying the confounding factors in resolving 820 phylogenetic relationships in Vespertilionidae. J. Mammal. 91, 1435-1448. 821 Larsen, R.J., Knapp, M.C., Genoways, H.H., Khan, F.A.A., Larsen, P.A., Wilson, D.E., 822 Baker, R.J., 2012a. Genetic diversity of Neotropical Myotis (Chiroptera: Vespertilionidae) 823 with an emphasis on South American species. Plos One 7. 824 Larsen, R.J., Larsen, P.A., Genoways, H.H., Catzeflis, F.M., Geluso, K., Kwiecinski, G.G., 825 Pedersen, S.C., Simal, F., Baker, R.J., 2012b. Evolutionary history of Caribbean species of 826 Myotis, with evidence of a third Lesser Antillean endemic. Mamm. Biol. 77, 124-134. 827 Matveev, V.A., Kruskop, S.V., Kramerov, D.A., 2005. Revalidation of Myotis petax Hollister, 828 1912 and its new status in connection with M. daubentonii (Kuhl, 1817) (Vespertilionidae, 829 Chiroptera). Acta Chiropterol. 7, 23-37. 830 Mein, P., Ginsburg, L., 1997. Les mammifères du gisement miocène inférieur de Li Mae 831 Long, Thaïlande: systématique, biostratigraphie et paléoenvironnement. Geodiversitas 19, 832 783-844. 833 Menu, H., Hand, S., Sigé, B., 2002. Oldest Australian vespertilionid (Microchiroptera) from 834 the early Miocene of Riversleigh, Queensland. Archeringa 26, 319-331. 835 Menu, H., Sigé, B., 1971. Nyctalodontie et myotodontie, importants caractères de grades 836 évolutifs chez les chiroptères entomophages. Comptes Rendus de l'Académie des Sciences de 837 Paris 272, 1735-1738. 838 Miller, S.A., Dykes, D.D., Polesky, H.F., 1988. A simple salting procedure for extracting 839 DNA from human nucleated cells. Nucleid Acids Res. 16, 215. 840 Murphy, W.J., Eizirik, E., O'Brien, S.J., Madsen, O., Scally, M., Douady, C.J., Teeling, E., 841 Ryder, O.A., Stanhope, M.J., de Jong, W.W., Springer, M.S., 2001. Resolution of the early 842 placental mammal radiation using Bayesian phylogenetics. Science 294, 2348-2351. 843 Myers, N., Mittermeier, R.A., Mittermeier, C.G., da Fonseca, G.A.B., Kent, J., 2000. 844 Biodiversity hotspots for conservation priorities. Nature 403, 853-858. 845 Nylander, J.A.A., 2004. MrModeltest v. 2.3. Program distributed by the author. . Uppsala 846 University, Sweden 847 Paradis, E., 1998. Detecting shifts in diversification rates without fossils. Am. Nat. 15, 176-848 187. 849 Purvis, A., Agapow, P.-M., Gittleman, J.L., Mace, G.M., 2000. Nonrandom extinction and the 850 loss of evolutionary history. Science 288, 328-330. 851

25

Rambaut, A., Drummond, A.J., 2009. Tracer v1.5. Available from 852 http://beast.bio.ed.ac.uk/Tracer 853 Rautenbach, I.L., Bronner, G.N., Schlitter, D.A., 1993. Karyotypic data and attendant 854 systematic implications for the bats of southern Africa. Koedoe 36, 87-104. 855 Ree, R.H., Smith, S.A., 2008. Maximum likelihood inference of geographic range evolution 856 by dispersal, local extinction, and cladogenesis. Syst. Biol. 57, 4-14. 857 Ronquist, F., Huelsenbeck, J.P., 2003. MRBAYES 3: Bayesian phylogenetic inference under 858 mixed models. Bioinformatics 19, 1572-1574. 859 Ruedi, M., Biswas, J., Csorba, G., 2012. Bats from the wet: two new species of Tube-nosed 860 bats (Chiroptera: Vespertilionidae) from Meghalaya, India. Rev. suisse Zool. 119, 111-135. 861 Ruedi, M., Mayer, F., 2001. Molecular systematics of bats of the genus Myotis 862 (Vespertilionidae) suggests deterministic ecomorphological convergences. Mol. Phylogenet. 863 Evol. 21, 436-448. 864 Salicini, I., Ibáñez, C., Juste, J., 2011. Multilocus phylogeny and species delimitation within 865 the Natterer‟s bat species complex in the Western Palearctic. Mol. Phylogenet. Evol. 61, 888-866 898. 867 Sanderson, M.J., Donoghue, M.J., 1996. Reconstructing shifts in diversification rates on 868 phylogenetic trees. Trends Ecol. Evol. 11, 15-20. 869 Sechrest, W., Brooks, T.M., da Fonseca, G.A., Konstant, W.R., Mittermeier, R.A., Purvis, A., 870 Rylands, A.B., Gittleman, J.L., 2002. Hotspots and the conservation of evolutionary history. 871 Proc. Natl. Acad. Sci. USA 99, 2067-2071. 872 Simmons, N.B., 2005. Order Chiroptera. In: Wilson, D.E., Reeder, D.M. (Eds.), Mammal 873 species of the World. A taxonomic and geographic reference. Johns Hopkins University Press, 874 Washington, pp. 312-529. 875 Sodhi, N.S., Koh, L.P., Brook, B.W., Ng, P.K.L., 2004. Southeast Asian biodiversity: an 876 impending disaster. Trends Ecol. Evol. 19, 654-660. 877 Stadelmann, B., Herrera, G., Arroyo-Cabrales, J., Ruedi, M., 2004a. Molecular systematics of 878 the piscivorous bat Myotis (Pizonyx) vivesi. J. Mammal. 85, 133-139. 879 Stadelmann, B., Jacobs, D., Schoeman, C., Ruedi, M., 2004b. Phylogeny of African Myotis 880 bats (Chiroptera, Vespertilionidae) inferred from cytochrome b sequences. Acta Chiropterol. 881 6, 177-192. 882 Stadelmann, B., Kunz, T.H., Lin, L.K., Ruedi, M., 2007. Molecular phylogeny of New World 883 Myotis (Chiroptera, Vespertilionidae) inferred from mitochondrial and nuclear DNA genes. 884 Mol. Phylogenet. Evol. 43, 32-48. 885 Stamatakis, A., 2006. RAxML-VI-HPC: Maximum likelihood-based phylogenetic analyses 886 with thousands of taxa and mixed models. Bioinformatics 22, 2688-2690. 887 Tate, G.H., 1941. A review of the genus Myotis (Chiroptera) of Eurasia, with special 888 reference to species occurring in the East Indies. B. Am. Mus. Nat. Hist. 78, 537-565. 889 Teeling, E.C., Springer, M.S., Madsen, O., Bates, P., O'Brien, S.J., Murphy, W.J., 2005. A 890 molecular phylogeny for bats illuminates biogeography and the fossil record. Science 307, 891 580-584. 892 Tiunov, M.P., Kruskop, S.V., Feng, J., 2011. A new mouse-eared bat (Mammalia: Chiroptera, 893 Vespertilionidae) from South China. Acta Chiropterol. 13, 271-278. 894 Topál, G., 1970. On the systematic status of Pipistrellus annectans Dobson, 1871 and Myotis 895 primula Thomas, 1920 (Mammalia). Annals hist.-nat. Mus. nat. hung. 62, 373-379. 896 Topál, G., 1983. New and rare fossil mouse-eared bats from the Middle Pliocene of Hungary 897 (Mammalia, Chiroptera). Fragm. Min. Palaeont. 11, 43-54. 898 Tsytsulina, K., 2004. On the taxonomical status of Myotis abei Yoshikura, 1944 (Chiroptera, 899 Vespertilionidae). Zool. Sci. 21, 963-966. 900 Tsytsulina, K., Dick, M.H., Maeda, K., Masuda, R., 2012. Systematics and phylogeography of 901 the steppe whiskered bat Myotis aurascens Kuzyakin, 1935 (Chiroptera, Vespertilionidae). 902 Russian J. Theriol. 11, 1-20. 903

26

Volleth, M., Heller, K.G., 2012. Varations on a theme: Karyotype comparison in Eurasian 904 Myotis species and implications for phylogeny. Vespertilio 16, 329-350. 905 Volleth, M., Tidemann, C.R., 1991. The origin of the Australian Vespertilioninae bats, as 906 indicated by chromosomal studies. Z. Saugetierkd. 56, 321-330. 907 Weyeneth, N., Goodman, S.M., Ruedi, M., 2011. Do diversification models of Madagascar‟s 908 biota explain the population structure of the endemic bat Myotis goudoti (Chiroptera: 909 Vespertilionidae)? J. Biogeogr. 38, 44-54. 910 Zhang, Z.Z., Tan, X.Y., Sun, K.P., Liu, S., Xu, L.J., Feng, J., 2009. Molecular systematics of 911 the Chinese Myotis (Chiroptera, Vespertilionidae) inferred from cytochrome-b sequences. 912 Mammalia 73, 323-330. 913 Ziegler, R., 2003. Bats (Chiroptera, Mammalia) from Middle Miocene karstic fissure fillings 914 of Petersbuch near Eichstätt, Southern Franconian Alb (Bavaria). Geobios 36, 447-490. 915 916 917

918

27

Table 1 Clade name, level of support (over 50%) and inferred age estimated from the 919

different molecular analyses. Supports were recovered from a maximum likelihood (ML) or a 920

Bayesian approach (BA), and data sets consisted of either Cyt b or Rag 2 sequences analyzed 921

separately or in combination. Values are expressed as percent bootstraps (for ML) or posterior 922

probabilities (for BA). Node age is expressed in million years ago (MYA) followed by the 923

95% highest posterior density interval (HPD). 924

925

Table S1 Origin and biogeographic assignation of the specimens analyzed for the Cyt b and 926

Rag 2 genes. Species names not listed in Simmons (2005) are marked with a star (*) and their 927

taxonomy discussed in the material and methods section. The symbol ø denotes information 928

unavailable. Vouchers (if any) are deposited in the following institutions or belong to personal 929

collections: Instituto Politécnico Nacional in Mexico (CDR), Estacion Biologica de Doñana 930

(EBD), Frieder Mayer (FM), Field Museum of Natural History in Chicago (FMNH), Gareth 931

Jones (GJ), Jean-François Maillard, DIREN, Martinique (JFM), Kuniko Kawai (KK), Kishi 932

Maeda (KM), Natural History Museum of Geneva (MHNG), National Natural History 933

Museum of Paris (MNHN), Manuel Ruedi (MR), Museum of Vertebrate Zoology at Berkeley 934

(MVZ), National Museum Prague (NMP), Osaka City University Graduate School of 935

Medicine (OCUMS), Royal Ontario Museum (ROM), Sumiko Matsumura (SM), 936

Senckenberg Museum of Frankfurt (SMF), T.H. Kunz (THK), Tigga Kingston (TiK), 937

Museum of Texas Tech University (TK), Transvaal Museum, South Africa (TM), University 938

of Alaska Museum (UAM), UKM (Universiti Kebangsaan Malaysia), Zoological Museum of 939

Moscow State University (ZMMU), Z. Zhang (ZZ). 940

941

Figure 1 Maximum likelihood tree for 135 sequences of Myotis and six outgroup species 942

based on analyses of the combined mitochondrial and nuclear data set (1140 bp of Cyt b and 943

1148 bp of Rag 2). Nodal support is represented as standard bootstrap value (BP) obtained 944

with RAxML or posterior probabilities (PP) obtained with MrBayes. Empty circles represent 945

28

support values comprised between 50 and 70 %, grey circles values between 71 and 90% and 946

filled circles values above 90% in both ML and BA reconstructions. Species with uncertain 947

phylogenetic position (“floating species”) are indicated with a filled diamond. The various 948

clades discussed in the text are highlighted by boxes or broken lines. The inset represents the 949

geographic distribution of the 103 Myotis species listed in Simmons (2005) (right value) 950

versus the numbers sampled for the molecular reconstructions (left values) in the different 951

biogeographic regions of the world. 952

953

Figure 2 Chronogram of Myotis taxa based on Bayesian dating analysis using BEAST and the 954

combined mitochondrial and nuclear DNA gene dataset. Mean divergence values (expressed 955

as million year ago, MYA) are given at each node and horizontal bars represent the 95% 956

highest posterior density ranges. Clade names correspond to those given in Fig. 1. 957

958

Figure 3 Biogeographic evolution of ancestral areas of Myotis species reconstructed in the 959

chronogram of Fig. 2. The most likely ancestral areas determined with Lagrange under a 960

model with dispersals limited to adjacent bioregions are given in boxes at each node. The 961

inset map illustrates the seven biogeographic regions included for this study – A: the Western 962

Palaearctic, B: the Eastern Palaearctic, C: the Oriental, D: the Oceanian, E: the Ethiopian, F: 963

the Nearctic and G: the Neotropical region. Inferred range expansion events are highlighted 964

by black boxes (on the chronogram) or by arrows (on the world map), and vicariance events 965

by filled diamonds. The bottom scale represents time (in million years before present, MYA) 966

and the approximate subdivision of paleontological epochs. 967

968

Figure S1 Majority rule (50%) consensus Bayesian tree with posterior probability values 969

obtained with the analysis of 135 mitochondrial Cyt b sequences of Myotis taxa. Six outgroup 970

taxa from the Murininae and Kerivoulinae were used to root the tree. 971

972

29

Figure S2 Majority rule (50%) consensus Bayesian tree with posterior probability values 973

obtained with the analysis of 122 nuclear Rag 2 sequences of Myotis taxa. Six outgroup taxa 974

from the Murininae and Kerivoulinae were used to root the tree. 975

976

Figure S3 Occlusal views of right dentaries showing the nyctalodont lower molars (where the 977

postcristid links the hypoconulid) of Myotis latirostris (A.) and M. siligorensis alticraniatus 978

(B.), and the myotodont molars (where the postcristid links the entoconid) of Cistugo seabrae 979

(C.). See Menu and Sigé (1971) for a definition of lower molar configurations. 980

981

Table 1 : Clade name, level of support (over 50%) and inferred age estimated from the different molecular analyses. Supports were recovered from a maximum likelihood (ML) or a Bayesian approach (BA), and data sets consisted of either Cyt b or Rag 2 sequences analyzed separately or in combination. Values are expressed as percent bootstraps (for ML) or posterior probabilities (for BA). Nodes age are expressed in million years ago (MYA) followed by the 95% highest posterior density interval (HPD).

Clade name Cyt b Rag 2 combined Age (HPD)

ML BA ML BA ML BA MYA Myotis (incl. latirostris) 96 100 75 100 99 100 20.94 (23-20)

Myotis (excl. latirostris) 76 100 - - 80 100 18.66 (21-16) I New World clade 93 100 - - 97 100 12.34 (15-10) II-X Old World clade 62 100 - - 72 100 16.96 (20-14) II-III + VI-X Eurasian clade - 51 - - - 65 16.22 (19-14) II Large Myotis clade 92 100 - - 98 100 9.93 (12- 8) III Myotis clade 100 100 - - 100 100 9.33 (12-7) IV Oriental Myotis clade 92 100 - - 90 100 11.92 (14-10) V Ethiopian clade 94 100 - - 98 100 12.33 (15-10) VI Whiskered Myotis clade 85 100 - - 88 100 9.06 (11-7) VII muricola clade 72 100 - - 76 100 10.24 (13-8) VIII Eastern Myotis clade 100 100 - - 97 100 9.55 (12-7) IX Asian Myotis clade 87 100 52 96 87 100 8.77 (11-6) X Trawling Myotis clade 95 100 - - 99 100 7.30 (9-5)

50 < BP & PP < 70

70 < BP & PP < 90

90 < BP & PP

"floating" species

blythii omari

annamiticus

taiwanensis

cf. montivagus 4

cf. montivagus 2cf. montivagus 1

cf. montivagus 3

h. horsfieldii 2horsfieldii deignani

h. horsfieldii 1

cf. muricola 4

Murina annamiticus

latirostris lineage

Ne

w W

orld

Eth

iop

ian

wh

iskere

dm

uri

co

laO

rie

nta

lA

sia

nla

rge

My

oti

s

I

V

VI

VII

VIII

IV

IX

III

X

II

Eura

sia

n

Mu. pluvialis

daubentonii nathalinae

Nearc

tic

subcla

de

Neotr

op

ical

subcla

de

brandtii lineage

Old

World

pilosus

tra

wli

ng

cf. browni

cf. formosus 2cf. formosus 1

formosus flavus

WesternPalaearctic

EasternPalaearctic

Ethiopian

Oriental

13/14

1/4

6/7 21/26

8/14 17/19

13/19

Nearctic

0.0 MYA5.010.015.020.025.0

nesopolus

tricolor 2tricolor 1

brandtii

23.1

26.2

20.9

18.7

12.3

7.8

16.4

12.3

12.3

17.0

12.3

16.2

15.5

15.2

12.9

10.2

9.1

12.1

14.8 13.4

12.68.8

4.5

3.1

14.2

12.5

13.1

7.811.9

9.8

9.9

8.1

9.4

0.6

0.3

1.6

1.3

3.2

4.44.9

4.2

1.4

2.8

2.9

4.0

5.64.1

2.34.4

0.58

0.35

0.55

1.1

1.4

1.4

0.8

0.6

0.6

0.5

1.2

0.7

pilosus

b. oxygnathusb. omari

b. blythiib. ancilla

b. amurensis

n. tschuliensis

d. nathalinae

taiwanensis

s. alticraniatus 3s. alticraniatus 2

s. alticraniatus 1s. alticraniatus 4

lucifugus

I

V

VI

VII

VIII

IX

IV

X

III

II

cf. montivagus 4

formosus flavus

cf. formosus 1cf. formosus 2

6.6

6.0

4.46.01.0

2.7

1.1

8.5

10.9

9.4

6.43.8

8.5

6.8

6.0

5.58.1

9.3

8.2

11.3

11.6 9.53.1

1.7

7.8

5.9

2.5

2.6

2.9

2.4

9.6

8.8

5.5

4.9 2.2

3.4

3.0

2.2

8.6 2.0

6.34.9

7.3

6.0

5.11.7

9.38.4

6.8

7.1

1.0

8.98.2

6.42.7

1.34.4

5.94.3

3.0

1.8

1.9

1.0

0.9

1.1

1.5

1.9

1.71.1

1.6

0.9

0.5

1.3

0.8

brandtii

pilosus

Miocene Pliocene Plei

s. alticraniatus

FF

cf.muricola1-3cf.muricola 4

Harpiocephalus

l. carissima

n. larensis

cf. browni

taiwanensis

b. ancillab. blythii

n. tschuliensis

Oligocen.

I

V

VI

VII

VIII

IX

IV

X

III

II

f. flavus

cf. formosus

C

C

C

CCC

C

C

C

C

C

C

C

C

CC

CC

C

C

C C

C

CC

CC

C

C

C C

C

C

CC

C

C

C

F

F

F

F

F

F

F

FF

F

F

F

FF

F

F

GG

G

G

G

G

G

G

G

G

G

GG

G

GG

G

B

B

B

B

BB

BB

B

E

E

E

E

E

BF

A

FG

FG

FG

CE

CA

E

E

A

AC

A

AC

AC

A A

C

BC

A

C

AC

A

C

AC

A

B

AB

A

ABA

F

A

AA A

CB

D

A

C

vicariance (8x)

expansions (21x)

0.0 MYA5.010.015.020.025.0

AB

E ECC CE

C AC

A ABC BC

A AB

C BC

C BC

C CB

C CD

B BF

B BA

F FG

C CA

C AC

C AC

C AC

A AC

A AB

C BC

C BC

Ethiopian

clade

latirostris lineage

New World Myotis

Nearctic

subclade

brandtii lineage

Neotropical

subclade

Eurasian

clade

25 15 10 5 MYA 20 0

Old World Myotis

BC

C

B => BF

C => CE

C => CA

B

C

Nearly complete phylogeny of Myotis distributed worldwide

New World versus Old World clade

Likelihood models of range evolution identify Eastern Asia for their origins

Ancestral area reconstructions suggest 21 intercontinental dispersals during 21 MYA of

evolution

Discrepancies in molecular dating and interpretation of fossil record due to homoplasies

in dental evolution