Accepted Manuscript

Phylogeny and Polyploidy: Resolving the Classification of Cyprinine Fishes (Teleostei: Cypriniformes)

Lei Yang, Tetsuya Sado, M. Vincent Hirt, Emmanuel Pasco-Viel, M. Arunachalam, Junbing Li, Xuzhen Wang, Jörg Freyhof, Kenji Saitoh, Andrew M. Simons, Masaki Miya, Shunping He, Richard L. Mayden

PII: S1055-7903(15)00028-7 DOI: http://dx.doi.org/10.1016/j.ympev.2015.01.014 Reference: YMPEV 5112

To appear in: Molecular Phylogenetics and Evolution

Received Date: 12 April 2014 Revised Date: 29 January 2015 Accepted Date: 30 January 2015

Please cite this article as: Yang, L., Sado, T., Vincent Hirt, M., Pasco-Viel, E., Arunachalam, M., Li, J., Wang, X., Freyhof, J., Saitoh, K., Simons, A.M., Miya, M., He, S., Mayden, R.L., Phylogeny and Polyploidy: Resolving the Classification of Cyprinine Fishes (Teleostei: Cypriniformes), Molecular Phylogenetics and Evolution (2015), doi: http://dx.doi.org/10.1016/j.ympev.2015.01.014

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 2 Phylogeny and Polyploidy: Resolving the Classification of Cyprinine Fishes 3 (Teleostei: Cypriniformes) 4 5 Lei Yang 1,†,*, Tetsuya Sado2, M. Vincent Hirt3, Emmanuel Pasco-Viel4, 6 M. Arunachalam5, Junbing Li6, Xuzhen Wang6, Jörg Freyhof7, Kenji Saitoh8, 7 Andrew M. Simons9, Masaki Miya2, Shunping He6, Richard L. Mayden 1,*

8 9 1Department of Biology, Saint Louis University, St. Louis, MO 63103, USA 10 11 2Department of Zoology, Natural History Museum and Institute, Chiba, 955-2 Aoba-cho, 12 Chuo-ku, Chiba 260-8682, Japan 13 14 3Graduate Program in Ecology, Evolution, and Behavior, University of Minnesota, St. 15 Paul, MN 55108, USA 16 17 4 - 18 19 , Lyon, 69007, France 20 21 5Sri Paramakalyani Centre for Environmental Sciences, Manonmaniam Sundaranar 22 University, Alwarkurichi - 627 412, Tamil Nadu, India 23 24 6Laboratory of Fish Phylogenetics and Biogeography, Institute of Hydrobiology, Chinese 25 Academy of Sciences, Wuhan 430072, China 26 27 7German Centre for Integrative Biodiversity Research (iDiv), Halle-Jena-Leipzig, 28 Deutscher Platz 5e, 04103 Leipzig, Germany 29 30 8National Research Institute of Fisheries Science, Aquatic Genomics Research Center, 31 Fukuura, Kanazawa, Yokohama 236-8648, Japan 32 33 9Department of Fisheries, Wildlife, and Conservation Biology and Bell Museum of 34 Natural History, University of Minnesota, St. Paul, MN 55108, USA 35 36 †Present address: Hollings Marine Laboratory, College of Charleston, Charleston, SC 37 29401, USA 38 39 40 41 42

1 43 *To whom correspondence should be addressed: 44 45 Lei Yang & Richard L. Mayden 46 Department of Biology, 47 3507 Laclede Ave, 48 Saint Louis University, 49 St. Louis, MO 63103, USA. 50 Phone: +001-314-977-3910

51 Fax: +001-314-977-3658 52 Emails: [email protected] (LY); [email protected] (RLM) 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73

2 74 Abstract: 75 Cyprininae is the largest subfamily (>1300 species) of the family Cyprinidae and 76 contains more polyploid species (~400) than any other group of fishes. We examined the 77 phylogenetic relationships of the Cyprininae based on extensive taxon, geographical, and 78 genomic sampling of the taxa, using both mitochondrial and nuclear genes to address the 79 phylogenetic challenges posed by polyploidy. Four datasets were analyzed in this study: 80 two mitochondrial gene datasets (465 and 791 taxa, 5604bp), a mitogenome dataset (85 81 taxa, 14,771bp), and a cloned nuclear RAG1 dataset (97 taxa, 1497 bp). Based on

82 resulting trees, the subfamily Cyprininae was subdivided into 11 tribes: Probarbini (new; 83 Probarbus + Catlocarpio), Labeonini Bleeker, 1859 (Labeo & allies), Torini Karaman, 84 1971 (Tor, Labeobarbus & allies), Smiliogastrini Bleeker, 1863 (Puntius, Enteromius & 85 allies), Poropuntiini (Poropuntius & allies), Cyprinini Rafinesque, 1815 (Cyprinus & 86 allies), Acrossocheilini (new; Acrossocheilus & allies), Spinibarbini (new; Spinibarbus), 87 Schizothoracini McClelland, 1842 (Schizothorax & allies), Schizopygopsini Mirza, 1991 88 (Schizopygopsis & allies), and Barbini Bleeker, 1859 (Barbus & allies). Phylogenetic 89 relationships within each tribe were discussed. Two or three distinct RAG1 lineages were 90 identified for each of the following tribes Torini, Cyprinini, Spinibarbini, and Barbini, 91 indicating their hybrid origin. The hexaploid African Labeobarbus & allies and Western 92 Asian Capoeta are likely derived from two independent hybridization events between 93 their respective maternal tetraploid ancestors and Cyprinion. 94 95 Keywords: Biogeography; Cyprinidae; Evolution; Hexaploids; Taxonomy; Tetraploids 96 97 98 99 100 101 102 103 104

3 105 1. Introduction

106 Within vertebrates, phylogenetic resolution of the largest group, Actinopterygii or 107 ray-finned fishes, lags behind other taxa, due to the large number of species and the 108 difficulties obtaining and identifying specimens. Another challenge in actinopterygian 109 phylogeny is the evolution of genome duplications producing polyploids. Genome 110 duplications have occurred multiple times in actinopterygian evolution but are 111 particularly prevalent in Cypriniformes, the largest clade of freshwater fishes (Amores, et 112 al. 1998; Leggatt & Iwama, 2003; Taylor et al. 2003). Within Cypriniformes, the 113 Cyprininae contains more than 1300 freshwater species in over 120 genera, accounting 114 for nearly 4% of bony fish diversity (Eschmeyer, 2015). Most species of this subfamily 115 inhabit waters of southern Eurasia and Africa. Some are well known, such as the common 116 carp (Cyprinus carpio) and Goldfish (Carassius auratus). The Cyprininae contains 117 around 400 closely related polyploid species (Arai, 2011; Froese & Pauly, 2015), more 118 than other polyploid fish lineages, such as Acipenseriformes, Salmoniformes, and 119 Catostomidae (all <250 species; Eschmeyer, 2015). Over 30 genera in this subfamily 120 consist only of polyploid species (Leggatt & Iwama, 2003; Arai, 2011). Most polyploids 121 in this subfamily are either tetraploids (2n = ca.100) or hexaploids (2n = ca.150). The 122 species Ptychobarbus dipogon has an amazingly large number of chromosomes (2n = 123 446, Cui et al. 1991; 2n= 424-432, Wu et al. 1999). 124 The classification of the subfamily Cyprininae is still under discussion as various 125 numbers of formal or informal groups have been recognized within it (Tables 1&S1). 126 Many molecular studies have been conducted on the phylogenetic relationships of the 127 Cyprininae; however, most were limited by restricted taxon or character sampling or 128 limited geographical sampling. Most importantly, these studies either did not use nuclear 129 genes or used, but ignored the issue of paralogy associated with polyploid taxa. If a 130 nuclear gene has only one copy in diploids, it is expected to have two copies in 131 tetraploids, three copies in hexaploids, and four copies in octoploids. In polyploids, 132 especially allopolyploids, the different nuclear gene copies could be quite divergent and 133 belong to distinct clades in a gene tree (e.g. Evans et al. 2005; Saitoh et al. 2010). Direct 134 amplification of these nuclear genes with PCR (polymerase chain reaction) without using 135 specially designed paralog-specific primers will likely result in a mixture of gene copies.

4 136 If nuclear copies are not sorted appropriately, homology cannot be confidently 137 established, potentially misleading phylogenetic studies. In the present study, extensive 138 DNA cloning was performed and multiple alleles were used in phylogeny reconstruction 139 of all major diploid and polyploid lineages of the subfamily Cyprininae. These data and 140 molecular phylogenies inferred from mitochondrial genes made it possible for us to 141 explore the phylogenetic relationships and subdivisions of this subfamily as well as the 142 evolution of polyploidy. 143 The major objectives of this study are: 1) to investigate the phylogenetic

144 relationships and subdivisions of the subfamily Cyprininae based on the largest taxon, 145 geographical, and genomic sampling to date; and 2) to propose a classification for this 146 group. The evolution of polyploid lineages; the distribution of some morphological 147 characters important in classification; and biogeography of some taxa will be discussed. 148 149 2. Materials and Methods 150 2.1 Taxon sampling and datasets 151 Four datasets were analyzed in this study: 1) a mitochondrial gene dataset with 465 152 taxa (465-taxon mt dataset); 2) an expanded mitochondrial gene dataset with 791 taxa 153 (791-taxon mt dataset); 3) a mitochondrial genome dataset with 85 taxa (mitogenome 154 dataset); and 4) a cloned RAG1 dataset with 97 taxa (RAG1 dataset). 155 In the 465-taxon mt dataset, most taxa (97.4%) were represented by sequences from 156 five mitochondrial genes: cytochrome oxidase subunit I (COI), Cytochrome b (Cyt b), 157 16S ribosomal RNA (16S rRNA), NADH dehydrogenase subunit 4 (ND4), and NADH 158 dehydrogenase subunit 5 (ND5); remaining taxa (2.6%) were represented by sequences 159 from at least three genes. This dataset has good representation for major lineages of 160 cyprinines with only 9.2% missing data and thus should provide a good estimate of the 161 mitochondrial phylogeny (Wiens, 2006; Roure et al., 2013). 162 The 791-taxon mt dataset was constructed by adding to the 465-taxon mt dataset, 163 326 taxa represented by sequences for Cyt b or Cyt b plus another gene. All taxa in the 164 791-taxon mt dataset were represented by Cyt b sequences, providing comparable data 165 for all included taxa. This dataset was assembled to expand taxon sampling to include 166 most of the generic diversity, nearly half of the species diversity, and important

5 167 biogeographic diversity for the subfamily Cyprininae. Analysis of this dataset was 168 constrained using results from analysis of the 456-taxon mt dataset. 169 The mitogenome dataset was assembled and used to create a strongly supported 170 estimate of mitochondrial phylogeny that could be used to place five enigmatic species 171 whose phylogenetic position could not be established using the other mitochondrial 172 datasets. Of the five enigmatic species, Semiplotus semiplotum, Luciocyprinus striolatus, 173 Eirmotus octozona, and Oreichthys cosuatis were represented by five mitochondrial 174 genes, while Aaptosyax grypus was represented only by Cyt b.

175 The RAG1 dataset was assembled to create a phylogeny based on a nuclear locus 176 and used to assess ploidy and the pattern and impact of gene duplication on phylogeny. 177 Nuclear RAG1 gene (recombination activating gene 1, exon 3) is widely used in 178 phylogenetic studies. This gene is single copy in diploids (Evans et al. 2005; Saitoh et al. 179 2010) and our results corroborated this observation (see below). 180 See Table 2 for the composition of each dataset and Table S2 for detailed sample 181 information. Nomenclature for genera and species generally follows Eschmeyer (2015). 182 To avoid confusion, the classification system proposed by us is adopted throughout the 183 paper (Table 3). 184 185 2.2 DNA extraction, PCR, Cloning, and Sequencing 186 Genomic DNA was isolated from ethanol-preserved muscle or fin clips using 187 DNeasy tissue extraction kits (Qiagen, USA). The five mitochondrial genes were 188 amplified and sequenced using primers and protocols from Yang et al. (2010). Complete 189 mitochondrial genomic sequences used in this study were downloaded from GenBank 190 except for three sequences provided by M.M. (Table S2). 191 The RAG1 sequences of 34 diploid ingroup and 26 outgroup taxa were either 192 downloaded from GenBank or obtained following Yang and Mayden (2010). We cloned 193 RAG1 for all 30 polyploids sampled, including tetraploids and hexaploids, as well as six 194 possible diploids, to represent all major ingroup clades recovered in analyses of 195 mitochondrial data (Arai, 2011; Table S2). Cloning was also performed for Catostomus 196 commersoni, a tetraploid outgroup taxon. Conditions for PCR followed Yang and 197 Mayden (2010) using primers R1_2533F, R1_4090R, and R1_4078R and 30

6 198 amplification cycles rather than 35. The PCR products were ligated into pGEM-T vector 199 (Promega, USA) and cloned into JM109 competent Escherichia coli (Promega, USA). 200 Standard blue-white colony screening was performed following cloning and incubation. 201 We picked 12 or more positive clones for diploids and tetraploids, and 24 positive clones 202 for hexaploids. Those clones were then used as templates for second-round PCR 203 amplifications, also using 30 cycles. 204 The PCR products of mitochondrial and nuclear genes were purified using one of 205 three methods: AMPure (Agencourt Bioscience), QIAquick Gel Extraction Kits (Qiagen),

206 or outsourced commercial purification with ExoSAP-IT (USB/Affymetrix). Sequencing 207 was conducted at two facilities: htSEQ High-Throughput Genomics Unit (University of 208 Washington, USA) and Macrogen (South Korea). Primers used for PCR amplifications 209 were also used for sequencing. All novel mitochondrial sequences and RAG1 sequences 210 were deposited in GenBank, see Table S2 for accession numbers for all sequences 211 included in this study. Vouchered specimens were deposited at Saint Louis University or 212 other museums associated with the Cypriniformes Tree of Life (CToL) Project. 213 214 2.3 Sequence alignment and phylogenetic analyses 215 MT DATASETS: Multiple alignment of the mitochondrial gene datasets followed 216 Yang et al. (2012b, 2012c) and the final alignment was 5604 bp in length. Partitioned 217 Maximum Likelihood (ML) analysis of the 465-taxon mt dataset was conducted using 218 RAxML v.8.0.26 (Stamatakis, 2006, 2014). The optimal partitioning scheme and models 219 of nucleotide substitution for this dataset as well as other datasets in this study were 220 analyzed using PartitionFinder v1.1.1 (Lanfear et al., 2012, 2014). See Table S3 for 221 parameter settings and results. A total of 200 distinct runs were performed based on 200 222 random starting trees using the default algorithm of the program. The tree with the best 223 likelihood score was chosen as the final tree. In analysis of the 791-taxon mt dataset, the 224 optimal ML tree was constructed through constraint searching (200 runs) using as the 225 constraint the best ML tree from analysis of the 465-taxon mt dataset. Maximum 226 Likelihood searches were not conducted directly on the 791-taxon mt dataset due to the 227 large amount of missing data. Our strategy was designed to minimize the impact of 228 missing data on tree topology. During constraint searches, the relationships of nearly 60%

7 229 of taxa in the 791-taxon dataset were stabilized and other taxa were placed with respect to 230 the constraint tree. The inclusion of Cyt b for all taxa was expected to facilitate this 231 process. Maximum likelihood bootstrap analyses were conducted for both mitochondrial 232 datasets using RAxML (Felsenstein, 1985; Stamatakis et al., 2008) using the same 233 partitioning strategy and nucleotide substitution model as in the initial maximum 234 likelihood search. The number of nonparametric bootstrap replications was set as 1000 235 and other parameters were set as default. The resulting trees were imported into 236 PAUP*4.0.b10 (Swofford, 2002) to obtain the 50% majority rule consensus tree and

237 bootstrap values (BP). 238 MITOGENOME DATASET: Mitochondrial genomic sequences were aligned 239 following Saitoh et al. (2006, 2011). The complementary strand sequences were used for 240 L-strand-encoded genes (ND6 and eight tRNA genes). The final alignment had 14,771 241 nucleotide sites in total, including 11,337 sites of 13 protein-coding genes, 2089 sites of 242 two rRNA genes and 1345 sites of 22 tRNA genes. All gaps and ambiguous sites were 243 excluded during alignment and the final dataset contained no missing data. Gaps were 244 excluded because they are frequently adjacent to or within ambiguously aligned regions. 245 Some parsimony-informative sites may have been pruned in this process, but these 246 represent a small percentage of the entire alignment. The partitioning scheme and model 247 of nucleotide substitution used is included in Table S3. Partitioned Maximum Likelihood 248 analysis and bootstrap analysis were conducted using RAxML with the same settings as 249 other mitochondrial analyses. Five independent ML analyses and bootstrap analyses were 250 then performed to determine the placement of each of the five enigmatic species 251 described above. For each of these five species, the mitogenome tree built previously was 252 used as a constraint and 100 independent ML and bootstrap analyses were performed 253 using RAxML. Other settings for the ML analyses and bootstrap analyses were the same 254 as previous analyses. 255 For the 465-taxon mt dataset, 791-taxon mt dataset, and mitogenome dataset, 256 corresponding RY-coding datasets were built by coding “A” and “G” as “R”, and “C” 257 and “T” as “Y” for the third codon positions of all protein-coding mitochondrial genes, 258 thus taking only transversions into account. According to Phillips et al. (2001, 2004), 259 RY-coding can remove noise from third codon positions effectively while retaining all

8 260 available positions in the dataset. Phylogenetic analyses were conducted on each of these 261 datasets with the same settings as used for their corresponding normally coded datasets. 262 See Table S3 for partitioning schemes and nucleotide substitution models used. 263 RAG1 DATASET: For each species, RAG1 sequences of all sequenced clones were 264 aligned. Consensus sequences of each putative allele of each species were phased 265 following Saitoh & Chen (2008). Chimeric sequences, sequences with private sites 266 (unique variant nucleotide appeared in a single clone), and sequences with private 267 combinations (unique nucleotide combinations of variable sites appeared in a single

268 clone), were discarded during the process to reduce cloning artifacts for recovery of real 269 allelic sequences of RAG1. Alleles of cloned species and sequences from other species 270 were then aligned, producing the RAG1 dataset (1497 bp in length). The dataset was 271 partitioned by codon positions and GTR+I+G model was used according to results of 272 PartitionFinder (Table S3). A total of 1000 distinct runs were performed based on 1000 273 random starting trees using the default algorithm of RAxML. The tree with the best 274 likelihood score was chosen as the final tree. 275

276 3. Results 277 MT DATASETS: The subfamily Cyprininae was resolved as monophyletic in both 278 ML trees derived from the 465-taxon mt dataset (BP=100%; full tree embedded in Fig. 2) 279 and the 791-taxon mt dataset (BP=99%) (Figs. 1&2). Eleven major clades were resolved: 280 Probarbini, Acrossocheilini, Spinibarbini, Poropuntiini, Labeonini, Torini, Smiliogastrini, 281 Cyprinini, Schizothoracini, Schizopygopsini, and Barbini. The first three clades represent 282 tribes described herein (see discussion). Support for these clades was strong with the 283 exception of Smiliogastrini (BP ≤ 64%), Poropuntiini (BP ≤ 72%), and Cyprinini (BP < 284 50%). The position of some genera such as Sinocyclocheilus, Procypris, Chagunius, and 285 Aaptosyax were weakly supported (BP < 50%). To facilitate discussion of relationships 286 and paralogy of RAG1, we identified two unnamed clades: Clade B (BP ≥ 98%) 287 ((Barbini, Schizopygopsini), ((Schizothoracini, Spinibarbini), Acrossocheilini)) and 288 Clade A (BP ≥ 94%) (Smilogasterini, (Poropuntiini, (Cyprinini, Clade B))). Clade A and 289 three other clades constituted the subfamily Cyprininae with the following relationships: 290 (((Clade A, Torini), Labeonini), Probarbini). Relationships within each tribe were

9 291 illustrated in Figs. 2b-h. For Labeonini, we refer reader to Yang et al. (2012b) for a full 292 description of relationships. 293 MITOGENOME DATASET: The subfamily Cyprininae was robustly supported 294 (BP=100%; Fig. 3). The eleven major clades recovered by the analyses of the two mt 295 datasets were also recovered here and nearly all clades were strongly supported, 296 including Clades A and B (BP=100%). Relationships of the major clades are largely 297 consistent with the analyses of the mt dataset, except in Clade B where the following 298 relationships were supported: (((Acrossocheilini, Spinibarbini), Schizothoracini),

299 (Schizopygopsini, Barbini)), and the position of the genera Chagunius and Oreichthys 300 which fell within the Smiliogastrini (BP=80%). The phylogenetic position of the five 301 enigmatic genera were identified as follows: Aaptosyax and Luciocyprinus in Cyprinini 302 (BP=92%); Eirmotus in Poropuntiini (BP=98%), and Semiplotus confirmed as part of the 303 Barbini clade (BP=75%). Phylogenetic trees resulting from RY-coded datasets were 304 consistent with other analyses (see Figs. S1-2). 305 RAG1 DATASET: The number of copies recovered by cloning was less than or 306 equal to the number predicted by ploidy level of each species (Fig. 4, Table S2). The two 307 unnamed clades, A and B, were recognized, corresponding to Clade A and Clade B from 308 the mt analyses (Figs. 1-3). Most major clades recovered in analyses of mitochondrial 309 data were also recovered by the RAG1 phylogeny with the polyploid taxa Cyprinini, 310 Spinibarbini, Barbini, and Torini represented by two or three clades in the tree (Fig. 4). 311 312 4. Discussion 313 4.1 Polyploid taxa 314 Polyploid taxa pose significant challenges for phylogenetic systematists. This is 315 particularly true in taxa such as the Cyprininae, which contains large numbers of diploid, 316 tetraploid, and hexaploid species. Our analyses, comparing trees based on mitochondrial 317 data with trees based on cloned nuclear data, allows the construction and comparison of 318 complex gene trees including paralogous allele sequences. Our data provide the best 319 estimate of phylogeny of the Cyprininae and allow development of a classification 320 system that reflects the phylogenetic history of these organisms. 321 The RAG1 analysis demonstrates that polyploid lineages are embedded in branches

10 322 of paralogs (Fig. 4). Some polyploid tribes have two distant (inter-tribal) RAG1 lineages. 323 For example, Cyprinini has two RAG1 paralogs, one close to the mostly diploid 324 Poropuntiini + Smiliogastrini lineage and the other close to the tetraploid Probarbini. 325 Polyploids are in many cases of hybrid origin (Wu et al., 2001; David et al., 2003; Saitoh 326 et al., 2010). As a probable explanation for our results, the cyprinin common ancestor 327 established itself via hybridization between the diploid common ancestor of Poropuntiini 328 and/or Smiliogastrini and a diploid common ancestor of Probarbini (paternal source; Figs. 329 3&4). Likewise, some other tribes (Spinibarbini, Torini and Barbini) have distant gene

330 lineages indicating their hybrid origin. Alleles of Probarbini, Schizothoracini and 331 Schizopygopsini fall in a single tribal lineage respectively. Both probarbin species show 332 2 + 2 allelic configuration, and we postulated hybridization of closely related ancient 333 diploids within the tribe. On the other hand, alleles of Schizothoracini and 334 Schizopygopsini do not show clear 2 + 2 relationships, and therefore alternative 335 mechanisms such as autopolyploidy, gene conversion between paralogs, or counter- 336 diploidization in which tetrasomy evolves from amphidiploidy through incidental onset 337 of paralogous (homeologous) synapsis and crossing-over (Saitoh, 2003) may better 338 explain polyploidy in these taxa. 339 The hexaploid members of Torini (Labeobarbus, Varicorhinus, and Carasobarbus) 340 and Barbini (Capoeta) formed a clade with the diploid Cyprinion (Fig. 4). The genus 341 Capoeta did not group with any other torin lineage (i.e. Torini 1&3; Fig. 4). A probable 342 explanation for these results is that these hexaploid lineages were derived from two 343 independent hybridization events between their respective tetraploid ancestors and 344 Cyprinion. For the tribe Torini, this hexaplodization event could be ascribed to 345 hybridization between tetraploid torins (maternal source; Fig. 2b) and Cyprinion (paternal 346 source; Fig. 4). Because all hexaploid members of Torini are found in Africa and West 347 Asia but all tetraploid members (Tor etc.) are found in southern Asia and Cyprinion are 348 mainly distributed in West Asia, we hypothesized that this hexaplodization event 349 happened during their dispersal towards Africa but predated their arrival in Africa (see 350 discussion below). For the tribe Barbini, because Capoeta is placed within the tetraploid 351 Luciobarbus in the mitochondrial tree (Fig. 2h), the hexaplodization event of Capoeta 352 might be due to ancient hybridization between a Luciobarbus species (maternal source;

11 353 see also Levin et al. 2012) and a Cyprinion species (paternal source). All discussion 354 above was based only on mitochondrial trees and the RAG1 tree. Further analyses based 355 on more nuclear loci and inheritance test may clarify the state and origin of 356 Schizothoracini and Schizopygopsini and test our observations on other polyploid tribes 357 as well. 358 359 4.2 Phylogenetic relationships among tribes and the revised classification 360 Multiple lines of evidence provided in this study supported the subdivision of

361 Cyprininae into eleven major clades/tribes. Relationships among these clades were shown 362 in our trees (Figs. 1-4). The following seven clades consistently received strong support 363 in all analyses: Probarbini, Labeonini, Torini, Acrossocheilini, Spinibarbini, 364 Schizothoracini, and Schizopygopsini, except for Torini 2 & 3, and Schizothoracini in the 365 RAG1 tree (Figs. 1-4). The following three clades Smiliogastrini (less Oreichthys and 366 Chagunius), Poropuntiini (less Eirmotus, ‗P ’ , and ‗P.’ ), and 367 Barbini (less Cyprinion, Semiplotus, and Scaphiodonichthys) received strong support in 368 most analyses (Figs. 1-4). Oreichthys and Chagunius are likely members of 369 Smiliogastrini, while Eirmotus and the two ‗P ’ species are likely members of 370 Poropuntiini. Analyses based on mitochondrial data indicate that Cyprinion, Semiplotus, 371 and Scaphiodonichthys are members of Barbini (Figs. 1-3); however, it was difficult to 372 sort RAG1 paralogs of barbins with confidence. We chose to be conservative and 373 included these taxa in the tribe Barbini. The tribe Cyprinini was supported in the 374 mitogenome tree (BP = 92%; Fig. 3), and the RAG1 tree (Cyprinini 1, BP = 99%; 375 Cyprinini 2, BP = 53%; Fig. 4). 376 See Table 3 for the new classification we proposed for the subfamily Cyprininae. 377 Species-level relationships depicted in our phylogenetic trees should be interpreted with 378 caution as they may change when more species are added to the analyses. Sequences 379 from GenBank are also more likely to be problematic at the species-level, although we 380 have tried our best to ensure the sequences we used were from trusted institutions and are 381 authentic. 382 383 4.3 Nomenclature and phylogenetic relationships within tribes

12 384 Tribe Probarbini (new tribe) This clade includes two genera native to Southeast 385 Asia, Catlocarpio and Probarbus (Figs. 2b, 3&4). These genera were previously placed 386 in two distinct tribes or subtribes (Rainboth, 1991, 1996a; Rainboth et al. 2012) but were 387 resolved as sister taxa in our mitochondrial trees (Figs. 2b&3); their RAG1 alleles mixed 388 with each other in the nuclear tree (Fig. 4). A formal description of this new tribe is 389 provided at the end of the discussion. 390 Tribe Labeonini Bleeker, 1859 This clade contains all species of Labeo, Garra, 391 and their allies widely distributed in tropical Asia and Africa (Figs. 3&4). The oldest

392 available family-group name for this clade is Labeonini Bleeker, 1859. Our tribe 393 Labeonini is essentially equivalent to the Labeines of Reid (1982, 1985), the subfamily 394 Labeoninae of Chen et al. (1984) and Yue et al. (2000), and the tribe Labeonini of 395 Rainboth (1981, 1991, 1996a) and Rainboth et al. (2012). See Yang et al. (2012b) for 396 details of phylogenetic relationship within the Labeonini. 397 Tribe Torini Karaman, 1971 Large-sized barbs of Asia (e.g. Tor) and Africa (e.g. 398 Labeobarbus) are included in this clade (Figs. 2b, 3&4). Torinae Karaman, 1971 is the 399 oldest family-group name available for this clade. Our tribe Torini is generally equivalent 400 to the subtribe Tores of Rainboth (1981). 401 Large-sized barbs of southern Asia constituted the basal clades. They may all be 402 tetraploids and share the same evolutionary history despite difference in external 403 morphology (Fig. 4). Type species of both Osteochilichthys and Lepidopygopsis are 404 found within Hypselobarbus (Fig. 2b). Further taxonomic revision on this clade is 405 warranted. 406 The genus Barbus formerly included approximately 800 species distributed in 407 Eurasia and Africa. Many of these species and species groups were later reclassified into 408 other, presumably monophyletic, genera based on diagnostic morphological characters 409 (e.g. Tor, Puntioplites, Pseudobarbus). Currently, there remain around 400 species in 410 Barbus s.l. (Barbus s.s. + ‘Barbus’). In our trees, members of „Barbus‟ were located in 411 Torini (hexploids) and Smiliogastrini (diploids and tetraploids). Our phylogeny on 412 hexaploid barbs of the Torini is similar to that of Tsigenopoulos et al. (2010) except for 413 the position of several weakly supported small clades. Here we suggest that all hexaploid 414 ‗Barbus‘ be allocated to the genus Labeobarbus. The genus name Pterocapoeta Günther,

13 415 1902 should be revalidated to accommodate its type species Varicorhinus maroccana. 416 We tentatively recognize three informal groups within the Labeobarbus clade: 417 ―Labeobarbus lineage‖, ―Carasobarbus lineage‖, and ―Pterocapoeta lineage‖. According 418 to Borkenhagen & Krupp (2013), Mesopotamichthys, ‗Barbus‘ grypus, and ‗B.‘ reinii of 419 the Carasobarbus lineage should not be simply moved to Carasobarbus, because they 420 seem more similar in some morphological characters to Labeobarbus or Varicorhinus 421 than to Carasobarbus. Borkenhagen (2014) moved ‗Barbus‘ grypus to the new genus 422 Arabibarbus. In our present study, we put ‗Barbus‘ reinii under the genus-group name

423 ‗Labeobarbus‘ to clearly separate them from Barbus s.s. We tested the position of the 424 monotypic African genus Sanagia based on a 633bp CO1 fragment of S. velifera from 425 GenBank (HM418112) and found that it is a member of the ―Labeobarbus lineage‖ sister 426 to Labeobarbus sp. Kongou (results not shown). 427 Tribe Smiliogastrini Bleeker, 1863 This clade includes small-sized barbs of Asia 428 (Puntius and allies) and Africa (‗Barbus‘ and allies) (Figs. 2c, 2d, 3&4). At least five 429 family-group names are available for this clade: Barbini Bleeker, 1859, Systomi Bleeker, 430 1863, Osteobramae Bleeker, 1863, Smiliogastrini Bleeker, 1863 (Smiliogaster Bleeker, 431 1860 is a junior synonym of Osteobrama Heckel, 1843), and Puntii Karaman, 1971. The 432 name Barbini Bleeker, 1859 was used to designate another major clade resolved in this 433 study, because the type species of Barbus, B. barbus, fell within that clade. The name 434 Puntii appeared later than other names, and thus could not be considered. The names 435 Systomi, Osteobramae, and Smiliogastrini were established simultaneously, but the last 436 name was proposed at a higher rank and takes precedence (Art. 24.1, ICZN, 1999). 437 Therefore, we name this clade as the tribe Smiliogastrini Bleeker, 1863. No equivalent 438 family-group has been proposed before. 439 Sixteeen lineages are recognized from what was Puntius and its Asian allies 440 (Puntius s.l.). Besides the genera redefined or recently described, i.e. Puntius s.s. 441 Systomus, Dawkinsia, Haludaria, Pethia, Barbodes, Oliotius, Puntigrus, Striuntius, 442 Desmopuntius, and Sahyadria (Pethiyagoda et al. 2012; Pethiyagoda, 2013; Kottelat, 443 2013; Raghavan et al. 2013), there are five more lineages: Rohtee (including Osteobrama 444 belangeri and O. sp.), Hampala, Osteobrama, Oreichthys, and Chagunius. Most of these 445 lineages received strong bootstrap support (Oliotius is monotypic), except for several

14 446 large lineages, Systomus, Pethia, Puntius, and Barbodes. Relationships among these 447 lineages are, in general, unresolved. More studies are warranted for Puntius s.l. and 448 nuclear genes with appropriate rate variation should be exploited to further explore 449 relationships among these diploid lineages. 450 Our results show that Asian Puntius and allies are closely related to African small 451 ‗Barbus‘, confirming previous hypotheses based on scale morphology (Rainboth, 1981) 452 and karyology (Golustov & Krysonov, 1993; Ráb et al. 1995). We propose a revalidation 453 of the generic name Enteromius Cope, 1867 (type species: Barbus potamogalis = B.

454 ablabes) to accommodate all African diploid ‗Barbus‘, as Enteromius is the oldest 455 available genus-group name for these fishes. Further studies are required to more fully 456 resolve the phylogenetic relationships among species of the non-monophyletic 457 Enteromius (as resolved herein). We also suggest moving all southern African tetraploid 458 barbs into ‗Pseudobarbus‘, if they cannot be placed in the genus Pseudobarbus. 459 Tribe Poropuntiini This clade is comprised of genera primarily distributed in 460 Southeast Asia (Figs. 2e, 3&4). Poropuntii Rainboth, 1991 and Puntioplitini Nguyen & 461 Ho, 2003 are the only family-group names proposed for various members of this clade. 462 However, neither is available because no diagnoses were provided (art. 13.1.1, ICZN, 463 1999) and it is impossible to make any of them available (art. 13.2, ICZN, 1999; see also 464 Van der Laan et al. 2014). For convenience, here we refer this clade as the tribe 465 Poropuntiini. Most members of this tribe are diagnosed by the combination of the 466 following characters (Rainboth, 1981, 1996a, 1996b; Shan et al. 2000; Kottelat et al. 467 1993; Kottelat, 2001a, 2001b; Arai, 2011): 1) diploid, with a chromosome number of 2n 468 = ca.50; 2) serrated dorsal fin-ray; 3) scales with a moderate number of parallel to 469 converging posterior radii and no or few lateral radii; and 4) body color plain, without 470 bars, blotches or spots, may have a stripe along the body, sometimes with a reticulate 471 pattern. 472 Species in this tribe exhibit great diversity in pharyngeal teeth morphology, which 473 has recently been addressed by Pasco-Viel et al. (2014). The genus Barbonymus is not 474 monophyletic. A taxonomic revision on B. gonionotus is warranted, as this species did 475 not group with the type species of the genus, B. schwanenfeldii. The former species 476 should be a member of Hypsibarbus as they are quite similar in many morphological

15 477 characters (Chaiwut Grudpan pers. comm.) and form a clade (despite BP = 51%; Fig. 2e). 478 Rainboth (1996b) argued that the genus Hypsibarbus is most closely related to 479 Poropuntius and Barbodes (= Barbonymus). Our results suggest that Hypsibarbus is most 480 closely related to „Barbonymus’ gonionotus, Sikukia, and Mystacoleucus. 481 Tribe Cyprinini Rafinesque, 1815 This clade contains Cyprinus and allies found 482 in Eurasia (Figs. 2e, 3&4). This tribe was named based on the oldest available family- 483 group name of this clade, Cyprinia Rafinesque, 1815. Our tribe Cyprinini is most similar 484 to the Cyprinini of Rainboth (1981) but contains more genera.

485 Sinocyclocheilus has usually been considered a barbin genus (e.g. Shan et al. 2000). 486 Yang et al. (2010) was the first to hypothesize that Sinocyclocheilus was a member of 487 Cyprinini and we are now highly confident that Sinocyclocheilus is a member of that 488 tribe. Shan et al. (2000) treated Luciocyprinus as a member of their Barbinae while 489 Rainboth et al. (2012) placed this genus in their subfamily Cyprininae, tribe Oreinini. 490 Multiple lines of evidence provided in this study strongly support Luciocyprinus as a 491 member of the tribe Cyprinini. The monotypic genus Aaptosyax is very rare (Tomoda, 492 2011) and our efforts to obtain specimens and tissue of Aaptosyax failed. The 493 Cytochrome b gene sequence used in He et al. (2004) is the only known available 494 molecular data for this taxon (Tomoda, 2011). Using these data the genus was grouped 495 with Cyprinus and Carassius in the molecular trees of He et al. (2004) and Rüber et al. 496 (2007). Rainboth et al. (2012) allocated this species to their subfamily Danioninae, tribe 497 Chedrini, without justifiable evidence. Our analyses suggest that Aaptosyax is likely a 498 member of the tribe Cyprinini but further studies are certainly needed (Fig. 3). Even if 499 Aaptosyax is not considered, it will be challenging to find morphological synapomorphies 500 for our Cyprinini, because both Sinocyclocheilus and Luciocyprinus are quite different 501 from other members of the Cyprinini (Cyprinus etc.) in external morphology (see Yang et 502 al. 2010). 503 The possession of both a serrated anal and dorsal spine (spinous fin-ray) has been 504 the most important, and sometimes only, characters used to determine membership of a 505 fossil species in the tribe Cyprinini. Our study, however, illustrates that some taxa (e.g. 506 Puntioplites falcifer) have these characters but do not belong to Cyprinini, whereas other 507 members of Cyprinini lack these characters (e.g. Sinocyclocheilus, Luciocyprinus).

16 508 Therefore caution is advised when identifying cyprinin fossils. The lack of explicit 509 morphological synapomorphies suggests that diversity of cyprinin fossils and the age of 510 the clade are likely underestimated. 511 Tribe Acrossocheilini (new tribe) This clade includes three genera: 512 Acrossocheilus, Onychostoma, and Folifer (Figs. 2f, 3&4), distributed in Southeast Asia 513 and China. Members of this tribe have been previously placed in one or more 514 tribes/subtribes together with many other cyprinines (Rainboth, 1981, 1991; Chen et al. 515 1984; Yue et al. 2000). In our analyses, neither Acrossocheilus nor Onychostoma are

516 monophyletic (see also Wang et al. 2007; Li et al. 2008). A thorough taxonomic revision 517 of Acrossocheilus and Onychostoma is highly warranted. 518 The Asian genus Onychostoma was once considered a synonym (but a valid 519 subgenus) of the African genus Varicorhinus (e.g. Bănărescu, 1971; Wu, 1977; Chu & 520 Chen, 1989) but our results demonstrate that Onychostoma and Varicorhinus fall in two 521 distinct major clades. These two genera are also very distinct in chromosome numbers 522 (Arai, 2011), species of Onychostoma are usually diploid (2n = ca.50), whereas species of 523 Varicorhinus are hexaploid (2n = ca.150). The genera Acrossocheilus and Folifer have 524 the same chromosome number (2n = ca.50) as Onychostoma. The genus Folifer was 525 treated as a junior synonym of Tor by some researchers (e.g. Shan et al. 2000) because 526 this genus possesses a median lobe on its lower lip, a character shared by species of Tor. 527 However, Tor is tetraploid (2n = ca.100), while Folifer is diploid (Arai, 2011). They also 528 exhibit differences in scale size, Folifer has relatively small scales, whereas Tor usually 529 has large scales. Zhou & Cui (1996) thought that these genera belonged to two different 530 lineages. Our results indicate that they belong to distinct clades (Acrossocheilini vs. 531 Torini). 532 Tribe Spinibarbini (new tribe) This clade currently contains only the genus 533 Spinibarbus, which is distributed in Laos, northern Vietnam and southern China (Figs. 2f, 534 3&4). No available family-group name was found for this group. The tribe is formally 535 described later in the discussion. The genus Spinibarbus was usually included in Barbini 536 or Barbinae (Rainboth, 1981; Chen et al. 1984; Yue et al. 2000). Rainboth (1991) put it in 537 his tribe Cyprinini subtribe Tores. Wang et al. (2007) also placed it in his tribe Cyprinini. 538 Species of Spinibarbus are tetraploids and have a chromosome number of 2n = 100

17 539 (Arai, 2011). It remains to be seen whether other genera should be assigned to this tribe. 540 The genus Spinibarbus and several other cyprinine genera, including Mystacoleucus, 541 Paraspinibarbus, Rohtee, and Parator (P. zonatus), share one morphological character: 542 the presence of a procumbent predorsal spine. The genera Spinibarbus, Mystacoleucus, 543 and Rohtee were included in our analyses and the latter two were found in Poropuntiini 544 and Smiliogastrini, respectively. 545 Tribe Schizothoracini McClelland, 1842 This clade includes Schizothorax and 546 allies, as well as Percocypris, and included species are mainly found on Asian plateaus

547 and adjacent regions (Figs. 2f, 3&4). At least three available family-group names could 548 be applied to this clade of fishes, Schizothoracinae McClelland, 1842, Oreini Bleeker, 549 1863, and Opistocheili Bleeker, 1863. There is some debate as to whether 550 Schizothoracinae in McClelland (1842) is a family-group name (e.g. Kullander et al. 551 1999; Kottelat, 2013). Here we follow Kottelat (2013) and name this clade as the tribe 552 Schizothoracini McClelland, 1842. It is equivalent to Mirza‟s (1991) tribe 553 Schizothoracini plus the genus Percocypris. 554 Our results illustrate that Percocypris is sister to a clade mainly formed by 555 Schizothorax, consistent with previous studies (Wang et al. 2007, 2013; Kong et al. 2007; 556 Li et al. 2008). Two species of Schizopyge including the type species (S. curvifrons) 557 formed a small distinct lineage within Schizothorax. The monotypic Aspiorhynchus (A. 558 laticeps) constituted a sister group with Schizothorax biddulphi. As discussed below, it is 559 too early to make taxonomic revisions for these taxa. We analyzed the Cytochrome b 560 sequences of Schizocypris altidorsalis from GenBank (JN790240). It fell with Cyprinion 561 and Semiplotus in our tree; however the branch to this taxon was quite long and we doubt 562 the result. Hence, it is not shown (but see Rainboth, 1991:200). More discussion on our 563 tribe Schizothoracini can be found below. 564 Tribe Schizopygopsini Mirza, 1991 This clade contains a group of genera native 565 to Asian Plateaus and adjacent regions (Figs. 2g, 3&4). There are two available family- 566 group names Diptychini and Schizopygopsini both proposed by Mirza (1991). Here, we 567 name this clade as the tribe Schizopygopsini Mirza, 1991, because Diptychini (now 568 Diptychinae) proposed by Janse (1933) is currently used for a group of insects 569 (Lepidoptera: Geometridae). This tribe contains all members of Mirza‟s (1991) tribes

18 570 Schizopygopsini and Diptychini. 571 Fishes of our Schizothoracini (less Percocypris), Schizopygopsini, and sometimes 572 the genus Lepidopygopsis (Torini) are referred to as “snowtrouts.” They share a character 573 called the anal shield, “a peculiar cleft in the ventral side of the body in front of the anal 574 fin, which is bounded laterally with scales of a peculiar form placed vertically like eave- 575 tiles” (Mirza, 1991) which may be correlated with spawning behavior (Cao et al. 1981). 576 In our analyses, these fishes were found in three different tribes suggesting multiple, 577 independent evolution of anal shield. Cao et al. (1981) divided snowtrouts (except

578 Lepidopygopsis) into three groups. Members of the “primitive group” (Schizothorax, 579 Schizocypris, and Aspiorhynchus) usually possess 1 or 2 pairs of barbels, 3 or 4 rows of 580 pharyngeal teeth, and scales usually covering the whole body except for the belly area; 581 members of the “specialized group” (Diptychus, Ptychobarbus, and Gymnodiptychus) 582 usually possess 1 pair of barbels, 2 rows of pharyngeal teeth, and moderately degenerated 583 scales (degenerated entirely in Gymnodiptychus); and members of the “highly specialized 584 group” (Gymnocypris, Oxygymnocypris, Schizopygopsis, Platypharodon, Chuanchia, and 585 Herzensteinia) usually do not have barbels, possess 1 or 2 rows of pharyngeal teeth, and 586 entirely degenerate scales. Mirza‟s (1991) tribes Schizothoracini, Diptychini, and 587 Schizopygopsini are equivalent to the three groups of Cao et al. (1981), respectively. In 588 the present study, the primary group (Schizothorax) is found in the tribe Schizothoracini, 589 whereas the specialized and highly specialized groups are located in the tribe 590 Schizopygopsini, with the former being more basal in the trees than the latter. 591 Recent studies demonstrate that the alpha taxonomy and nomenclature of 592 snowtrouts, especially the former, are problematic (e.g. Kullander et al. 1999; He & 593 Chen, 2006; Yang et al. 2012a; Kottelat, 2013). Similarity in morphological characters 594 widely used in their taxonomy, e.g. mouth and lip structure, may be the result of 595 convergence or parallelism, a situation similar to that in the tribe Labeonini (Yang & 596 Mayden, 2010). This may explain the non-monophyly of several taxa of Schizothoracini 597 and Schizopygopsini in our analyses. Until issues associated with the taxonomy and 598 nomenclature of snowtrouts are resolved, there can be no reasonable phylogenetic 599 discussion on the intra-relationships of either Schizothoracini (less Percocypris) or 600 Schizopygopsini.

19 601 Tribe Barbini Bleeker, 1859 All Barbus s.s. and allies, Cyprinion, Semiplotus, and 602 Scaphiodonichthys are contained in this clade (Figs. 2h, 3&4). These taxa are distributed 603 in Eurasia and northwestern Africa. The name Barbini Bleeker, 1859 is the oldest 604 available family-group name for this clade. Our tribe Barbini contains fewer genera than 605 many previously proposed Barbini or Barbinae (Tables 1&S1). 606 In our phylogenetic trees, the relationships between Luciobarbus, Barbus, Capoeta, 607 and Aulopyge are largely the same as depicted in previous studies (Machordom & 608 Doadrio, 2001; Tsigenopoulos & Berrebi, 2000; Tsigenopoulos et al. 2003) likely due to

609 the fact that many of our sequences were retrieved from these studies. The genus Capoeta 610 was monophyletic and fell within Luciobarbus. The same result was hypothesized in 611 previous studies (Machordom & Doadrio, 2001; Tsigenopoulos et al. 2003; Levin et al. 612 2012). Taki (1974:135) found that the Southeast Asian genus Scaphiodonichthys shows a 613 close affinity to Scaphognathops in many respects. However, in our trees, the former 614 genus was found in the Barbini, whereas the latter in the Poropuntiini. The “Cyprinion- 615 Onychostoma lineage” considered by Chen (1989) to be a monophyletic group, contains 616 Onychostoma, Cyprinion, Semiplotus, and Scaphiodonichthys. However, the present 617 results show that this lineage is not monophyletic, as Onychostoma was located in the 618 Acrossocheilini but the other three genera are in the Barbini. 619 620 4.4 Implications for biogeography 621 Biogeography is not the focus of this study, but we briefly discuss several 622 interesting points. The current distribution of cyprinines in Eurasia and Africa may not 623 have been shaped by the breakup of Gondwana, because their diversification began 55-80 624 Mya (million years ago; Saitoh et al. 2011; Chen et al. 2013; Pasco-Viel et al. 2014), after 625 the separation of the Indian landmass from Africa (130 Mya; Smith et al. 1994). 626 We hypothesize that there are at least five independent dispersal events from 627 Eurasia to Africa. Besides the two (one in Labeonina, the other in Garraina) in the 628 Labeonini (Yang & Mayden, 2010; Yang et al. 2012b), three more such events are 629 identified in Torini, Smiliogastrini, and Barbini, respectively. In Torini, all large-sized 630 African and Western Asian barbs (Labeobarbus, Carasobarbus etc.) are located at the 631 apical clade of the tree, whereas large-sized southern Asian barbs (Hypselobarbus, Tor,

20 632 etc.) are located in the basal clades (Fig. 2b). In Smiliogastrini, the small-sized African 633 barbs and allies and small-sized Asian Puntius and allies are located at the apical clade 634 and basal clade of the tree, respectively (Fig. 2c, 2d). In Barbini, all northwestern African 635 species constituted a monophyletic group in the apical clade of the tree (from 636 Luciobarbus moulouyensis to L. massaensis, Fig. 2h) whereas the rest species of this tribe 637 occur in Eurasia. 638 Because all African cyprinines can be allocated to the four tribes mentioned above, 639 we hypothesize that ancestors of all these species invaded Africa via these dispersal

640 events. The oldest Labeo-like and Barbus-like fossils found in Africa are around 17 and 641 12-13 million years old, respectively (Van Couvering, 1977). These fossils may represent 642 some of the earliest cyprinines in Africa, although we have no evidence to which tribe the 643 Barbus-like fossil belongs because it was described based on pharyngeal teeth only. 644

645 4.5 Description of three new tribes 646 Probarbini, new tribe 647 Type genus 648 Probarbus Sauvage, 1880 649 650 Diagnosis 651 Members of this tribe are diagnosed by combination of the following characters (Smith, 652 1945; Arai, 2011; Roberts, 1992; Rainboth, 1996a): 1) pharyngeal teeth arranged in a 653 single row with 4 teeth on each side; 2) tetraploid, with a chromosome number of 2n = 654 98; 3) scales covering most of the body; and 4) last simple anal fin-ray segmented and 655 flexible. 656 657 Composition 658 Probarbus Sauvage, 1880 and Catlocarpio Boulenger, 1898. Currently, Probarbus 659 includes three valid species, while Catlocarpio has one valid species. 660 661 Distribution

21 662 Mekong, Chao Phraya and Maeklong basins of Indo-China, and the Pahang and Perak 663 basins of Malaysia. 664 665 Acrossocheilini, new tribe 666 Type genus 667 Acrossocheilus Oshima, 1919 668 669 Diagnosis

670 Members of this tribe are diagnosed by combination of the following characters (Chen, 671 1989; Shan et al. 2000; Kottelat, 2001a, 2001b; Arai, 2011): 1) diploid, with a 672 chromosome number of 2n = ca.50; 2) last simple ray of dorsal fin strong, with a serrated 673 or smooth posterior margin; 3) 8 branched dorsal-fin rays; 4) pharyngeal teeth arranged 674 in three rows; and 5) horny sheath on lower jaw. 675 676 Composition 677 Includes over 40 valid species in three genera Acrossocheilus Oshima, 1919, 678 Onychostoma Günther, 1896, and Folifer Wu, 1977. 679 680 Distribution 681 Mainland China, Taiwan, Viet Nam, Laos, Thailand, and Cambodia. 682 683 Spinibarbini, new tribe 684 Type genus 685 Spinibarbus Oshima, 1919 686 687 Diagnosis 688 Members of this tribe are diagnosed by combination of the following characters (Yang & 689 Chen, 1994; Shan et al. 2000; Kottelat, 2001a, 2001b; Arai, 2011): 1) the presence of a 690 procumbent predorsal spine (i.e. an external anterior pointed process of the first proximal 691 radial); 2) tetraploid, with a chromosome number of 2n = 100; and 3) lower lip not 692 developed, confined to side of lower jaw.

22 693 694 Composition 695 Spinibarbus Oshima, 1919 and around eight valid species. Further studies are needed to 696 determine if there are other genera to be included into this tribe. 697 698 Distribution 699 Laos, northern Viet Nam and southern China. 700

701 Acknowledgements 702 This study was developed from the Ph.D. Dissertation of LY (Saint Louis University, 703 2010; Advisor: RLM). LY would like to express his gratitude to Larry Page, Jonathan 704 Armbruster, Milton Tan, Jarungjit Grudpan, and Chaiwut Grudpan and his students for 705 either organizing or participating in the fieldwork in Thailand. Preserved specimens in 706 Saint Louis University, University of Kansas, and Auburn University and some type 707 materials stored at Institute of Hydrobiology (Wuhan), Kunming Institute of Zoology, 708 Institute of Zoology (Beijing) were examined for this study. We thank Wei-Jen Chen, 709 Mary Agnew and Qiu Ren for help with data collection. Kevin Tang, Lanping Zheng, and 710 Leyang Yuan are thanked for helpful discussion. David Neely is thanked for providing 711 samples. We are grateful to Hsin-Hui Wu and Joe Besser for help using the Cluster 712 computing system. Maurice Kottelat and Carl Ferraris are thanked for discussion on the 713 availability of several family-group names. Maurice Kottelat kindly commented on the 714 discussion part of this paper. LY sincerely appreciates the enormous support from Gavin 715 Naylor. We are very grateful to the associate editor and two anonymous reviewers for 716 their valuable comments and suggestions that helped improve this paper. This research 717 was supported in part by personal funds of RLM, by the USA National Science 718 Foundation grants including The Cypriniformes Tree of Life initiative (CToL) (EF 719 0431326 to RLM; EF 0431132 to AMS) and the Collaborative PBI All Cypriniformes 720 Species – Phase II of An Inventory of the Otophysi (ACSI-II) (DEB1021840 to RLM), 721 by the Japan Society for the Promotion of Science (17207007 and 22370035) and by 722 Saint Louis University, St. Louis, Missouri. 723

23 724 725 References 726 Arai, R., 2011. Fish Karyotypes: A Check List. Springer, Japan. 727 Amores, A., Force, A., Yan, Y.L., Joly, L., Amemiya, C., Fritz, A., Ho, R.K., Langeland, 728 J., Prince, V., Wang, Y.L., Westerfield, M., Ekker, M., Postlethwait, J.H., 1998. 729 Zebrafish hox clusters and vertebrate genome evolution. Science 282, 1711–1714. 730 Bănărescu, P.M., 1971. A review of the species of the subgenus Onychostoma s. str. with 731 description of a new species (Pisces, Cyprinidae). Rev. Roum. Biol., Ser. Zool. 16,

732 241–248. 733 Borkenhagen, K., 2014. A new genus and species of cyprinid fish (Actinopterygii, 734 Cyprinidae) from the Arabian Peninsula, and its phylogenetic and zoogeographic 735 affinities. Environ. Biol. Fish. 97, 1179–1195. 736 Borkenhagen, K., Krupp, F., 2013. Taxonomic revision of the genus Carasobarbus 737 Karaman, 1971 (Actinopterygii, Cyprinidae). ZooKeys, 339, 1–53. 738 Cao, W. X., Chen, Y. Y., Wu, Y. F., Zhu, S.Q., 1981. Origin and evolution of 739 schizothoracine fishes in relation to the upheaval of the Xizang Plateau, in Collection 740 in Studies on the Period, Amplitude and Type of the Uplift of the Qinghai-Xizang 741 Plateau (in Chinese) (ed. The Team of the Comprehensive Scientific Expedition to 742 the Qinghai-Xizang Plateau, Chinese Academy of Sciences) pp. 118–130. Beijing: 743 Science Press. 744 Cavender, T.M., Coburn, M.M., 1992. Phylogenetic relationships of North American 745 Cyprinidae. In ―Systematics, Historical ecology, and North American Freshwater 746 Fishes‖ (R. L. Mayden, Ed.), pp. 293–327. Stanford University Press, Stanford, CA. 747 Chen, W.-J., Lavoué, S., Mayden, R.L., 2013. Evolutionary origin and early 748 biogeography of otophysan fishes (Ostariophysi: Teleostei). Evolution 67, 2218– 749 2239. 750 Chen, X.L., Yue, P.Q., Lin, R.D., 1984. Major groups within the family Cyprinidae and 751 their phylogenetic relationships. Acta Zootaxon. Sin. 9, 424–440 (in Chinese with 752 English summary). 753 Chen, Y.Y., 1989. Anatomy and phylogeny of the cyprinid fish genus Onychostoma 754 Günther, 1896. Br. Mus. Nat. Hist (Zool.) 55, 109–121.

24 755 Chu, X.L., Chen, Y.R., 1989. (eds.) The Fishes of Yunnan, China. Part I. Cyprinidae. 756 Science Press, Beijing, China. 757 Van Couvering, J.A.H., 1977. Early records of freshwater fishes in Africa. Copeia, 1977, 758 163–166. 759 Cui, J.X., Ren, X.H., Yu, Q.X., 1991. Nuclear DNA content variation in fishes. Cytologia 760 56, 425–429. 761 David, L., Blum, S., Feldman, M.W., Lavi, U., Millel, J., 2003. Recent duplication of the 762 common carp (Cyprinus carpio L.) genome as revealed by analyses of microsatellite

763 loci. Mol. Biol. Evol. 20, 1425–1434. 764 Donsakul, T., Magtoon, W., Rangsiruji, A., 2012 Karyotypes of Five Cyprinid Fishes 765 (Family Cyprinidae): Macrochirichthys macrochirus, Scaphiodonichthys 766 acanthopterus, Epalzeorhynchos munensis, Opsarius koratensis and Parachela sp. 767 from Thailand. Proceedings of the 50th Kasetsart University Annual Conference, 768 Kasetsart University, Thailand, 31 January - 2 February 2012. Volume 1. Subject: 769 Animals, Veterinary Medicine, Fisheries (pp. 439–446). 770 Eschmeyer, W. N. (ed.), 2015. Catalog of Fishes electronic version. 771 http://research.calacademy.org/ichthyology/catalog/fishcatmain.asp. 772 Evans, B.J., Kelley, D.B., Melnick, D.J., Cannatella, D.C., 2005. Evolution of RAG-1 in 773 polyploid clawed frogs. Mol. Biol. Evol. 22, 1193–1207. 774 Felsenstein, J., 1985. Confidence limits on phylogenies: an approach using the bootstrap. 775 Evolution 39, 783–791. 776 Froese, R., Pauly, D. (Editors), 2015. FishBase. World Wide Web electronic publication. 777 www.fishbase.org, version (1/2015). 778 Ganai, F.A., Yousuf, A.R., Dar, S.A., Tripathi, N.K., Wani, S.U., 2011. Cytotaxonomic 779 status of Schizothoracine fishes of Kashmir Himalaya (Teleostei: Cyprinidae). 780 Caryologia 64, 435–445. 781 Golubtsov, A.S., Krysanov, E.Y., 1993. Karyological study of some cyprinid species 782 from Ethiopia. The ploidy differences between large and small Barbus of Africa. J. 783 Fish Biol. 42, 445–455.

25 784 He, D.K., Chen, Y.F., 2006. Biogeography and molecular phylogeny of the genus 785 Schizothorax (Teleostei: Cyprinidae) in China inferred from cytochrome b sequences. 786 J. Biogeogr. 33, 1448–1460. 787 He, S.P., Liu, H.Z., Chen, Y.Y., Kuwahara, M., Nakajima, T., Zhong, Y., 2004. 788 Molecular phylogenetic relationships of Eastern Asian Cyprinidae (Pisces: 789 Cypriniformes) inferred from cytochrome b sequences. Sci. China Ser. C-Life Sci. 47, 790 130–138. 791 Howes, G.J., 1987. The phylogenetic position of the Yugoslavian cyprinid fish genus

792 Aulopyge Heckel, 1841, with an appraisal of the genus Barbus Cuvier and Cloquet, 793 1816 and the subfamily Cyprininae. Bull. Br. Mus. nat. Hist. (Zool.) 52, 165–196. 794 Howes, G.J., 1991. Systematics and biogeography: an overview. In: Winfield, I.J., 795 Nelson, J.S. (eds) Cyprinid Fishes: Systematics, Biology and Exploitation. Chapman 796 and Hall, London, pp 1–33. 797 ICZN (International Commission on Zoological Nomenclature), 1999. International Code 798 of Zoological Nomenclature, fourth ed. The International Trust for Zoological 799 Nomenclature, London. 800 Janse, A.J.T., 1933–35. The Moths of South Africa 2: Geometridae. E. P. & Commercial 801 Printing, Durban. 448 p. 802 Kong, X.H., Wang, X.Z., Gan, X.N., Li, J.B., He, S.P., 2007. Phylogenetic relationships 803 of Cyprinidae (Teleostei: Cypriniformes) inferred from the partial S6K1 gene 804 sequences and implication of indel sites in intron 1. Sci. China Ser. C-Life Sci. 50, 805 780–788. 806 Kottelat, M., 2001a. Freshwater fishes of northern Vietnam. A preliminary check–list of 807 the fishes known or expected to occur in northern Vietnam with comments on 808 systematics and nomenclature. Environment and Social Development Unit, East Asia 809 and Pacific Region. The World Bank, Washing, D.C. (June): i–iii + 1–123 + 1–18. 810 Kottelat, M., 2001b. Fishes of Laos. Wildlife Heritage Trust Publications, Colombo. 811 Kottelat, M., 2013. The Fishes of the Inland Waters of Southeast Asia: A Catalogue and 812 Core Bibliography of the Fishes Known to Occur in Freshwaters, Mangroves and 813 Estuaries. Raff. Bull. Zool. 27 (Suppl.), 1–663. 814 Kottelat, M., Whitten, A.J., Kartikasari, S.N., Wirjoatmodjo, S., 1993. Freshwater Fishes

26 815 of Western Indonesia and Sulawesi. Periplus Editions, Hong Kong, 259pp., 84 pls. 816 Kullander, S. O., Fang, F., Delling, B., Åhlander, E., 1999. The fishes of the Kashmir 817 Valley. pp. 99–167. In: Nyman, L. (ed). River Jhelum, Kashmir Valley. Impacts on 818 the aquatic environment. Swedmar, Göteborg. 819 Lanfear, R., Calcott, B., Ho, S.Y.M., Guindon, S., 2012. PartitionFinder: Combined 820 selection of partitioning schemes and substitution models for phylogenetic analyses. 821 Mol. Biol. Evol. 29, 1695–1701. 822 Lanfear, R., Calcott, B., Kainer, D., Mayer, C., Stamatakis, A., 2014. Selecting optimal

823 partitioning schemes for phylogenomic datasets. BMC Evol. Biol. 14, 82. 824 Leggatt, R.A., Iwama, G.K., 2003. Occurrence of polyploidy in the fishes. Rev. Fish 825 Biol. Fisher. 13, 237–246. 826 Levin, B.A., Freyhof, J., Lajbner, Z., Perea, S., Abdoli, A., Gaffaro lug, M., Özulu h, M., 827 Rubenyan, H.R., Salnikov, V.B., Doadrio, I., 2012. Phylogenetic relationships of the 828 algae scraping cyprinid genus Capoeta (Teleostei: Cyprinidae). Mol. Phylogenet. 829 Evol. 62, 542–549. 830 Li, J.B., Wang, X.Z., Kong, X.H., Zhao, K., He, S.P., Mayden, R.L., 2008. Variation 831 patterns of the mitochondrial 16S rRNA gene with secondary structure constraints 832 and their application to phylogeny of cyprinine fishes (Teleostei: Cypriniformes). 833 Mol. Phylogenet. Evol. 47, 472–487. 834 Machordom, A., Doadrio, I., 2001. Evolutionary history and speciation modes in the 835 cyprinid genus Barbus. Proc. R. Soc. Lond. B 268, 1297–1306. 836 McClelland, J., 1842. On the fresh-water fishes collected by William Griffith, Esq., F. L. 837 S. Madras Medical Service, during his travels under the orders of the Supreme 838 Government of India, from 1835 to 1842. Calcutta J. Nat. Hist. 2, 560–589. 839 Mirza, M.R., 1991. A contribution to the systematics of the Schizothoracine fishes 840 (Pisces: Cyprinidae) with the description of three new tribes. Pakistan J. Zool. 23, 841 339–341. 842 Nelson, J.S., 1994. Fishes of the World, third ed. Wiely, New York. 843 Nelson, J.S., 2006. Fishes of the World, fourth ed. Wiely, New York. 844 Pasco-Viel, E., Yang, L., Veran, M., Balter, V., Mayden, R.L., Laudet, V., Viriot, L., 845 2014. Stability versus diversity of the dentition during evolutionary radiation in

27 846 cyprinine fish. Proc. R. Soc. B. 281, 20132688. 847 http://dx.doi.org/10.1098/rspb.2013.2688 848 Pethiyagoda, R., 2013. Haludaria, a replacement generic name for Dravidia (Teleostei: 849 Cyprinidae). Zootaxa 3646, 199–199. 850 Pethiyagoda, R., Meegaskumbura, M., Maduwage, K., 2012. A synopsis of the South 851 Asian fishes referred to Puntius (Pisces: Cyprinidae). Ichthyol. Explor. Freshwaters. 852 23, 69–95. 853 Phillips, M.J., Delsuc, F., Penny, D., 2004. Genome-Scale Phylogeny and the Detection

854 of Systematic Biases. Mol. Biol. Evol. 21, 1455–1458. 855 Phillips, M.J., Lin, Y.-H., Harrison, G. L., Penny, D., 2001. Mitochondrial genomes of a 856 bandicoot and a brushtail possum confirm the monophyly of australidelphian 857 marsupials. Proc. R. Soc. Lond. B Biol. Sci. 268, 1533–1538. 858 Ráb, P., Machordom, A., Perdices, A. and Guegan, J.-F., 1995. Karyotypes of three 859 « small » Barbus species (Cyprinidae) from Republic of Guinea (Western Africa) 860 with a review on karyology of African small Barbus. Caryologia, 48, 299–307. 861 Raghavan, R., Philip, S., Ali, A., Dahanukar, N., 2013. Sahyadria, a new genus of barbs 862 (Teleostei: Cyprinidae) from western Ghats of India. Journal of Threatened Taxa. 5, 863 4932–4938. 864 Rainboth, W. J., 1981. Systematics of the Asiatic barbins (Pisces, Cyprinidae). Unpubl. 865 Ph.D. dissertation, The University of Michigan, Ann Arbor. 866 Rainboth, W.J., 1991. Cyprinid fishes of Southeast Asia. In IJ Winfield, JS Nelson, eds. 867 Cyprinid Fishes: Systematics, Biology and Exploitation. London: Chapman and Hall, 868 PP. 156–210. 869 Rainboth, W.J., 1996a. Fishes of the Cambodian Mekong. FAO species identification 870 field guide for fishery purposes. Rome: FAO. 871 Rainboth, W.J., 1996b. The taxonomy, systematics, and zoogeography of Hypsibarbus, a 872 new genus of large barbs (Pisces, Cyprinidae) from the rivers of southeastern Asia. 873 Univ. Calif. Publ. Zool. v. 129: i–xiii, 1–199. 874 Rainboth, W.J., Vidthayanon, C., Mai, D.Y., 2012. Fishes of the Greater Mekong 875 ecosystem with species list and photographic atlas. Miscellaneous Publications, 876 Museum of Zoology, University of Michigan, 201: i–vi, 1–173, 121 pls.

28 877 Reid, G.Mc.G., 1982. The form, function and phylogenetic significance of the 878 vomeropalatine organ in cyprinid fishes. J. Nat. Hist. 16, 497–510. 879 Reid, G.Mc.G., 1985. A revision of African species of Labeo (Pisces: Cyprinidae) and 880 a re-definition of the genus. Theses Zool. Cramer, Braunschweig. 6, 1–322. 881 Roberts, T.R., 1992. Revision of the Southeast Asian cyprinid fish genus Probarbus, with 882 two new species threatened by proposed construction dams on the Mekong River. 883 Ichthyol. Explor. Freshwaters 3, 37-48. 884 Roure, B., Baurain, D., Philippe, H., 2013. Impact of missing data on phylogenies

885 inferred from empirical phylogenomic data sets. Mol. Biol. Evol. 30, 197–214. 886 Rüber, L., Kottelat, M., Tan, H.H., Ng, P.K.L., Britz, R., 2007. Evolution of 887 miniaturization and the phylogenetic position of Paedocypris, comprising the world‘s 888 smallest vertebrate. BMC Evol. Biol. 7, 38. 889 Sahoo, P.K., Nanda, P., Barat, A., 2009. Chromosomal studies on a threatened fish 890 Cyprinion semiplotus (Teleostei: Cyprinidae) from Arunachal Pradesh. Asian Fish. 891 Sci. 22, 501–504. 892 Saitoh, K., 2003. Mitotic and meiotic analyses of the 'large race' of Cobitis striata, a 893 polyploid spined loach of hybrid origin. Folia Biol. (Krakow) 51(Suppl.), 101–105. 894 Saitoh, K., Chen, W.-J., 2008. Reducing cloning artifacts for recovery of allelic 895 sequences by T7 endonuclease I cleavage and single re-extension of PCR products — 896 A benchmark. Gene 423, 92–95. 897 Saitoh, K., Chen, W.-J., Mayden, R.L., 2010. Extensive hybridization and tetrapolyploidy 898 in spined loach fish. Mol. Phylogenet. Evol. 56, 1001–1010. 899 Saitoh, K., Sado, T., Doosey, M.H., Bart, Jr. H.L., Inoue, J.G., Nishida, M., Mayden, 900 R.L., Miya, M., 2011. Evidence from mitochondrial genomics supports the lower 901 Mesozoic of South Asia as the time and place of basal divergence of cypriniform 902 fishes (Actinopterygii: Ostariophysi). Zool. J. Linn. Soc. 161, 633–662. 903 Saitoh, K., Sado, T., Mayden, R.L., Hanzawa, N., Nakamura, K., Nishida, M., Miya, M., 904 2006. Mitogenomic evolution and interrelationships of the Cypriniformes 905 (Actinopterygii: Ostariophysi): The first evidence toward resolution of higher-level 906 relationships of the World‘s largest freshwater fish clade based on 59 whole 907 mitogenome sequences. J. Mol. Evol. 63, 826–841.

29 908 Shan, X.H., Lin, R.D., Yue, P.Q., Chu, X.L., 2000. Barbinae. In: Yue, P.Q. et al. (Ed.) 909 Fauna Sinica. Osteichthyes: Cypriniformes III. Science Press. Beijing. pp. 3–170. 910 Smith, H.M., 1945. The freshwater fishes of Siam, or Thailand. Bull. Am. Mus. Nat. Hist. 911 188, 194–195. 912 Smith, A.G., Smith, D.G., Funnel, B.M., 1994. Atlas of Mesozoic and Cenozoic 913 coastlines. Cambridge: Cambridge University Press, 778–826. 914 Stamatakis, A., 2006. RAxML–VI–HPC: maximum likelihood–based phylogenetic 915 analyses with thousands of taxa and mixed models. Bioinformatics 22, 2688–2690.

916 Stamatakis, A., 2014. RAxML Version 8: A tool for phylogenetic analysis and post- 917 analysis of large phylogenies. Bioinformatics, doi: 10.1093/bioinformatics/btu033. 918 Stamatakis, A., Hoover, P., Rougemont, J., 2008. A rapid bootstrap algorithm for the 919 RAxML web–servers. Syst. Biol. 57, 758–771. 920 Swofford, D.L., 2002. PAUP*: Phylogenetic Analysis Using Parsimony (*and other 921 methods), Version 4.0b10. Sinauer Associates, Sunderland, MA. 922 Taki, Y., 1974. New species of the genus Scaphognathops, Cyprinidae, from the Lao 923 Mekong River system. Jpn. J. Ichthyol. 21, 129–136. 924 Taylor, J.S., Braasch, I., Frickey, T., Meyer, A., Van de Peer, Y., 2003. Genome 925 duplication, a trait shared by 22000 species of ray-finned fish. Genome Res. 13, 382– 926 390. 927 Tomoda, Y., 2011. A note on Aaptosyax grypus, a very rare, unique cyprinid fish. 928 Ichthyol. Res. 58, 288–290. 929 Tsigenopoulos, C.S., Berrebi, P., 2000. Molecular phylogeny of North Mediterranean 930 Barbs (Genus Barbus: Cyprinidae) inferred from Cytochrome b sequences: 931 biogeographic and systematic implications. Mol. Phylogenet. Evol. 14, 165–179. 932 Tsigenopoulos, C.S., Durand, J.D., ünlü, E., Berrebi, P., 2003. Rapid radiation of the 933 Mediterranean Luciobarbus species (Cyprinidae) after the Messinian salinity crisis of 934 the Mediterranean Sea, inferred from mitochondrial phylogenetic analysis. Biol. J. 935 Linn. Soc. 80, 207–222. 936 Tsigenopoulos, C.S., Kasapidis, P., Berrebi, P., 2010. Phylogenetic relationships of 937 hexaploid large-sized barbs (genus Labeobarbus, Cyprinidae) based on mtDNA data. 938 Mol. Phylogenet. Evol. 56, 851–856.

30 939 Van der Laan, R., Eschmeyer, W.N., Fricke, R., 2014. Family-group names of recent 940 fishes. Zootaxa Monograph 3882, 1–230. 941 Wang, M., Yang, J.X., Chen, X.Y., 2013. Molecular phylogeny and biogeography of 942 Percocypris (Cyprinidae, Teleostei). PLoS ONE 8, e61827. doi:10.1371/ 943 journal.pone.0061827 944 Wang, X.Z., Li, J.B., He, S.P., 2007. Molecular evidence for the monophyly of East 945 Asian groups of Cyprinidae (Teleostei: Cypriniformes) derived from the nuclear 946 recombination activating gene 2 sequences. Mol. Phylogenet. Evol. 42, 157–170.

947 Wiens, J.J., 2006. Missing data and the design of phylogenetic analyses. J. Biomed. 948 Inform. 39, 34–42. 949 Wu, H.-W., 1977. (ed.) The Cyprinid Fishes of China. Part 2. Technical Printing House, 950 Shanghai, pp. 229–598 (In Chinese). 951 Wu, R., Gallo-Meagher, M., Littell, R.C., Zeng, Z.B., 2001. A general polyploid model 952 for analyzing gene segregation in outcrossing tetraploid species. Genetics 159, 869– 953 882. 954 Wu, Y.F., Kang, B., Men, Q., 1999. Chromosome diversity of Tibetan fishes. Zool. Res. 955 20, 258–264. 956 Xiao, H., Zhang, R.D., Feng, J.G., Ouyang, M., Li, W.X., Chen, S.Y., Zan, R.G., 2002. 957 Nuclear DNA content and ploidy of seventeen species of fishes in Sinocyclocheilus. 958 Zool. Res. 23, 195–199. 959 Yang, J., Yang, J.X., Chen, X.Y., 2012a. A re-examination of the molecular phylogeny 960 and biogeography of the genus Schizothorax (Teleostei: Cyprinidae) through 961 enhanced sampling, with emphasis on the species in the Yunnan–Guizhou Plateau, 962 China. J. Zool. Syst. Evol. Res. 50, 184–191. 963 Yang, J.X., Chen, Y.R., 1994. Systematic revision of Spinibarbus fishes (Cyprinifromes: 964 Cyprinidae). Zool. Res. 15, 1–10 (In Chinese with English abstract). 965 Yang, L., Arunachalam, M., Sado, T., Levin, B.A., Golubtsov, A.S., Freyhof, J., Friel, 966 J.P., Chen, W.-J., Hirt, M.V., Manickam, R., Agnew, M.K., Simons, A.M., Saitoh, 967 K., Miya, M., Mayden, R.L., He, S.P., 2012b. Molecular phylogeny of the cyprinid 968 tribe Labeonini (Teleostei: Cypriniformes). Mol. Phylogenet. Evol. 65, 362–379. 969 Yang, L., Mayden, R.L., 2010. Phylogenetic relationships, subdivision, and biogeography

31 970 of the Cyprinid tribe Labeonini (sensu Rainboth, 1991) (Teleostei: Cypriniformes), 971 with comments on the implications of lips and associated structures in the labeonin 972 classification. Mol. Phylogenet. Evol. 54, 254–265. 973 Yang, L., Mayden, R.L., Sado, T., He, S.P., Saitoh, K., Miya, M., 2010. Molecular 974 phylogeny of the fishes traditionally referred to Cyprinini sensu stricto (Teleostei: 975 Cypriniformes). Zool. Scr. 39, 527–550. 976 Yang, L., Hirt, M.V., Sado, T., Arunachalam, M., Manickam, R., Tang, K.L., Simons, 977 A.M., Wu, H.-H., Mayden, R.R., Miya, M., 2012c. Phylogenetic placements of the

978 barbin genera Discherodontus, Chagunius, and Hypselobarbus in the subfamily 979 Cyprininae (Teleostei: Cypriniformes) and their relationships with other barbins. 980 Zootaxa 3586, 26–40. 981 Yue, P.Q., et al. (Ed.) 2000. Fauna Sinica. Osteichthyes: Cypriniformes III. Science 982 Press. Beijing. pp. 1–654. 983 Zhou, W., Cui, G.H., 1996. A review of Tor species from the Lancangjiang River (Upper 984 Mekong River), China (Teleostei: Cyprinidae). Ichthyol. Explor. Freshwater 7, 131– 985 142.

32 Table 1. Brief summary of some previous hypotheses regarding the classifications of the subfamily Cyprininae Author Classification Rainboth, 1981 Subfamily Cyprininae 1. Cyprinini 2. Barbini Chen et al. 1984 Series Barbini Tribe Barbines 1. Barbinae (includes Schizothoracinae) 2. Cyprininae 3. Labeoninae Tribe Tincanes 4. Tincinae Howes, 1987, 1991 Subfamily Cyprininae 1. Squaliobarbin 2. Barbin 3. Labein 4. Schizothoracin 5. Cyprinion-Onychostoma (Howes, 1991) 6. Other taxa Rainboth, 1991, 1996a; Subfamily Cyprininae Rainboth et al. 2012 1. Cyprinini 2. Labeonini 3. Systomini 4. Oreinini (Rainboth, 1991; Rainboth et al. 2012) 5. Catlini (Rainboth, 1996a; Rainboth et al. 2012) 6. Semiplotini (Rainboth, 1991) 7. Squaliobarbini (Rainboth, 1991) Cavender & Coburn, 1992 Subfamily Cyprininae 1. barbins 2. cyprinins 3. labeonins Nelson, 1994 Subfamily Cyprininae 1. Barbus, Carassius, etc. 2. Bangana, Cirrhinus etc. 3. Ctenopharyngodon etc. 4. Balantiocheilos, Cyclocheilichthys etc. 5. Cyprinion, Semiplotus 6. Puntius 7. Acrossocheilus, Hampala, Poropuntius 8. Pseudobarbus, Varicorhinus Yue et al. 2000 Series Barbini 1. Barbinae 2. Labeoninae 3. Schizothoracinae 4. Cyprininae Wang, et al. 2007 Subfamily Cyprininae 1. Cyprinini 2. Labeonini Note: In Nelson (2006), members of the Cyprininae scatter within the Cyprinidae Table 2 Statistics of the four datasets used in the present study. Numbers separated by two slashes are (left to right) number of genera, species, and individuals, respectively. Taxon/Character 465-taxon mt 791-taxon mt Mitogenome RAG1* Cyprininae 102/348/395 117/612/721 27/56/58 53/68/68 Probarbini 2/2/2 2/3/3 1/1/1 2/2/2 Labeonini 32/135/139 34/141/158 5/9/9 8/10/10 Torini 7/30/38 13/83/103 1/2/2 7/7/7 Smiliogastrini 19/72/77 20/132/153 5/5/5 8/12/12 Poropuntiini 17/34/37 17/37/42 2/3/3 9/11/11 Cyprinini 6/17/17 7/48/50 4/9/9 6/9/9 Acrossocheilini 3/18/23 3/19/26 2/9/9 2/3/3 Spinibarbini 1/3/6 1/4/7 1/2/3 1/2/2 Schizothoracini 3/15/26 4/42/54 2/6/7 2/3/3 Schizopygopsini 6/11/15 9/26/35 2/7/7 3/3/3 Barbini 6/11/15 7/77/90 3/3/3 7/7/8 Outgroup 63/70/70 63/70/70 26/27/27 26/27/27 Total 165/418/465 180/682/791 53/83/85 80/97/97 Nucleotides (bp) 5604 5604 14771 1497 Missing data (%) 9.2 37.3 0.0 0.6 Variable characters (bp) 3775 3869 8132 728 Parsimony-informative 3392 3429 6787 599 characters (bp) A% 30.5 30.3 28.8 26.0 C% 27.1 27.2 26.4 23.5 G% 15.3 15.3 17.9 26.1 T% 27.1 27.2 26.9 24.4 * A total of 31 polyploids (includes tetraploids and hexaploids) and six diploids were cloned for RAG1 (see Table S2). Cloned species usually have multiple RAG1 sequences and the RAG1 dataset contains a total of 162 sequences (see Fig. 4).

Table 3 Revised classification of the subfamily Cyprininae. Type genus of each tribe is underlined. Classification Taxon Author/Year Genus/Group Family Cyprinidae Rafinesque, 1815 Subfamily Cyprininae Rafinesque, 1815 Tribe Probarbini -This study- Catlocarpio , Probarbus Tribe Labeonini Bleeker, 1859 ~ 40 genera (see Yang et al. 2012b) Tribe Torini Karaman, 1971 Hypselobarbus Clade (Hypselobarbus, Osteochilichthys, Lepidopygopsis), Naziritor, Neolissochilus, Tor, Labeobarbus Clade (Labeobarbus, Varicorhinus, Carasobarbus, Mesopotamichthys, Arabibarbus, Pterocapoeta, ‘Labeobarbus’) Tribe Smiliogastrini Bleeker, 1863 Chagunius, Oreichthys, Osteobrama, Sahyadria, Dawkinsia, Striuntius, Oliotius, Barbodes, Hampala, Puntigrus, Desmopuntius, Rohtee, Puntius, Haludaria, Pethia, Systomus, Enteromius, Barboides, 'Pseudobarbus', Pseudobarbus, Clypeobarbus Tribe Poropuntiini* Eirmotus, 'Puntius' snyderi, 'P.' semifasciolatus, Discherodontus, Cyclocheilos, Balantiocheilos, Cosmochilus, Amblyrhynchichthys, Puntioplites, Albulichthys, Cyclocheilichthys, Barbonymus, Scaphognathops, Sawbwa, Poropuntius, 'Barbonymus' gonionotus, Hypsibarbus, Sikukia, Mystacoleucus Tribe Cyprinini Rafinesque, 1815 Procypris, Luciocyprinus, Cyprinus, Carassioides, Carassius, Sinocyclocheilus, Aaptosyax Tribe Acrossocheilini -This study- Onychostoma, Acrossocheilus, Folifer Tribe Spinibarbini -This study- Spinibarbus Tribe Schizothoracini McClelland, 1842 Percocypris, Schizothorax, Schizopyge, Aspiorhynchus Tribe Schizopygopsini Mirza, 1991 Diptychus, Gymnodiptychus, Ptychobarbus, Oxygymnocypris, Platypharodon, Gymnocypris, Schizopygopsis, Herzensteinia, Chuanchia Tribe Barbini Bleeker, 1859 Scaphiodonichthys, Cyprinion, Semiplotus, Aulopyge, Barbus, Luciobarbus, Capoeta Incertae sedis Schizocypris, Barbopsis, Caecobarbus, Coptostomabarbus, Kalimantania, Laocypris, Neobarynotus, Parasikukia, Paraspinibarbus, Parator, Rohteichthys, Typhlobarbus Note: Sanagia velifera (CO1: HM418112) belongs to the Labeobarbus lineage of the tribe Torini. The following genera probably belong to the subfamily Cyprininae: Troglocyclocheilus Kottelat & Bréhier 1999, Xenobarbus Norman, 1923. *We think that the name Poropuntii Rainboth, 1991 is not available (see discussion). Here we use Poropuntiini only for convenience.

Fig. 1. Cladogram showing topologies of Maximum Likelihood trees resulting from 465-taxon mt dataset (full tree embedded in Fig. 2) and 791-taxon mt dataset (full tree shown as Fig. 2). Outgroups are not shown. Numbers above and below branches are the bootstrap support values for each node in trees resulting from the 465-taxon dataset and the 791-taxon dataset, respectively. Values lower than 50% are indicated by “-”. Values that are not applicable, because some taxa in the 791- taxon dataset are either not included or have only one individual in the 465-taxon dataset, are indicated by “n.a.”. Two strongly supported large clades are labeled “A” and “B” to facilitate discussion of the phylogeny. See discussion for nomenclature of each major clade (tribe).

Fig. 2. The optimal Maximum Likelihood tree (-lnL = 546800.981288) resulting from the 791-taxon mt dataset. Bootstrap support values are shown beside each node. Values lower than 50% are indicated by “-”. Chromosomes numbers (2n), if known, were shown in parentheses following each species (Sahoo et al., 2009; Arai, 2011; Ganai et al., 2011; Donsakul et al., 2012). Some species of Sinocyclocheilus were denoted as “tetraploids” following Xiao et al. (2002). The type species of each genus was also indicated. (a) Outgroups; (b) Probarbini, Labeonini (see Yang et al. 2012b), and Torini; (c) Smiliogastrini (Part I); (d) Smiliogastrini (Part II); (e) Poropuntiini and Cyprinini; (f) Acrossocheilini, Spinibarbini, and Schizothoracini; (g) Schizopygopsini; (h) Barbini. The name Xenocypridinae rather than Oxygastrinae is used, following Kottelat (2013: 69).

Fig. 3. The optimal Maximum Likelihood tree (-lnL = 313101.148476) resulting from the mitogenome dataset. Bootstrap support values are shown beside each node. Values lower than 50% are indicated by “-” .Two strongly supported large clades are labeled “A” and “B” to facilitate discussion of the phylogeny. The phylogenetic position of the five enigmatic species listed on the left are indicated using dashed lines and relevant bootstrap values are shown. Fig. 4. The optimal Maximum Likelihood tree (-lnL = 20257.574486) resulting from the RAG1 dataset. Bootstrap support values are shown beside each node. Values lower than 50% are indicated by “-”. Different putative RAG1 alleles of each species are denoted by A1-n, B1-n, or C1-n following species names. The letters A, B, or C were randomly assigned to alleles. Numbers following family- group names denote different clades (e.g. Cyprinini 1 and Cyprinini 2) of each polyploid lineage. From bottom to top of this figure, smaller numbers were assigned to clades of the same lineage that appeared earlier. Ploidy levels (2n, 4n, or 6n) of tribes and some genera are shown in brackets. Figure 1

Figure 2a

Figure 2b

Figure 2c

Figure 2d

Figure 2e

Figure 2f

Figure 2g

Figure 2h

Figure 3

Figure 4

Highlights:

► Most (117) genera and nearly half (612) of species of the Cyprininae were sampled.

► Five mitochondrial genes (or mitogenomes) and cloned nuclear RAG1 were analyzed.

► Eleven tribes (includes 3 described herein) were recovered from the Cyprininae.

► Each tribe has 1-3 RAG1 copies; origin of tetraploids and hexaploids was discussed.

► Five Eurasia to Africa dispersal events were hypothesized.