Hydrobiologia (2017) 785:47–59 DOI 10.1007/s10750-016-2902-8

PRIMARY RESEARCH PAPER

Dressing down: convergent reduction of the mental disc in (Teleostei: ) in the Middle East

I. Hashemzadeh Segherloo . A. Abdoli . S. Eagderi . H. R. Esmaeili . G. Sayyadzadeh . L. Bernatchez . E. Hallerman . M. F. Geiger . M. O¨ zulug . J. Laroche . J. Freyhof

Received: 3 February 2016 / Revised: 21 June 2016 / Accepted: 22 June 2016 / Published online: 4 August 2016 Ó Springer International Publishing Switzerland 2016

Abstract In the Middle East, species of Garra are While several species have such a mental disc, others believed to have invaded the area in two independent completely lack a mental disc, being adapted to slow- waves from the Indo-Malayan biogeographic region. moving water or to subterranean life. In this study, the This hypothesis is based on the structure of the mental phylogenetic relationships of Middle Eastern Garra disc, a unique specialization of the lower lip, which is species, including 16 described and 4 undescribed believed to be an adaptation to fast-flowing waters. species, were analysed using mitochondrial cytochrome c oxidase I sequences. The results are concordant with traditional hypotheses on two invasion events; however, Handling editor: Christian Sturmbauer these invasion events are independent from the pres- Electronic supplementary material The online version of ence, absence or shape of the mental disc. We postulate this article (doi:10.1007/s10750-016-2902-8) contains supple- mentary material, which is available to authorized users.

I. Hashemzadeh Segherloo (&) E. Hallerman Department of Fisheries and Environmental Sciences, Department of Fish and Wildlife Conservation, College of Faculty of Natural Resources and Earth Sciences, Shahr- Natural Resources and Environment, Virginia Polytechnic e-Kord University, P.B. 115, Shahr-e-Kord, Iran Institute and State University, Blacksburg, VA, USA e-mail: [email protected] M. F. Geiger A. Abdoli Zoological Research Museum A. Koenig - Leibniz Department of Biodiversity and Ecosystem Management, Institute for Biodiversity, Foundation under Environmental Science Research Institute, Shahid Public Law, Bonn, Germany Beheshti University, Tehran, Iran M. O¨ zulug S. Eagderi Department of Biology, Faculty of Science, Istanbul Department of Fisheries, Faculty of Natural Resources, University, Istanbul, Turkey University of Tehran, Karaj, Alborz, Iran J. Freyhof H. R. Esmaeili Á G. Sayyadzadeh German Centre for Integrative Biodiversity Research Department of Biology, College of Sciences, Shiraz (iDiv) Halle-Jena-Leipzig, Deutscher Platz 5a, University, Shiraz, Fars, Iran 04103 Leipzig, Germany

L. Bernatchez Á J. Laroche Institut de Biologie Inte´grative et des Syste`mes (IBIS), Pavillon Charles-Euge`ne-Marchand, Universite´ Laval, Quebec City, QC, Canada 123 48 Hydrobiologia (2017) 785:47–59 convergent reduction of the mental disc in 5–6 inde- Nemacheilidae, and Sisoridae. All of these lineages pendent lineages of Garra in the Middle East. have species that are endemic to the Middle East, most in the Persian Gulf basin, which is adjacent to the Indo- Keywords COI Á Cyprinidae Á Labeonin Á Mental Malayan biogeographical realm. This region has been disc reduction Á Middle East proposed to be a transition zone between the Palearctic and the Indo-Malayan realms (Coad, 2010). One of the freshwater fish genera believed to have invaded the Middle East from the Indo-Malayan Introduction biogeographical realm is the cyprinid genus Garra Hamilton,1822(Menon,1964).Menon (1964) proposed Biogeographical realms and their subunits, such as two independent ‘‘waves’’ of migration from a hypo- ecoregions, capture the uneven distribution of species thetical area of origin in Yunnan (southern China) to the and communities on Earth more accurately than do Middle East, with one of the waves also reaching Africa units based on gross biophysical features, such as (Tang et al., 2009; Yang & Mayden, 2010). Menon’s rainfall, temperature (Holdridge, 1967; Walter & Box, (1964) hypothesis of two invasion events is based on 1976; Schultz, 1995; Bailey & Ropes, 1998), or morphological differences in the mental disc among two vegetation structure (Ellenberg & Mueller-Dombois, groups of Garra species in the Middle East. The mental 1969; De Laubenfels, 1975; Schmidthu¨sen, 1976). The disc is an expansion of the lower lip in Garra and some overwhelming importance of biogeography and the related genera such as Discocheilus Zhang, 1997; role of tectonic and climatic changes on the distribution Discogobio Lin, 1931; and Placocheilus Wu, 1977 of biodiversity have long been appreciated by biolo- (Zhang, 2005). Species of the Garra variabilis (Heckel, gists (Wallace, 1876; Shermer, 2002). The re-connec- 1843) group have a small, poorly differentiated mental tion of formerly separated continents by tectonic plate disc, which Menon (1964) interpreted as an ancestral movements led to large-scale exchanges of species. character state, while species of the Garra rufa (Heckel, Such exchanges occurred after the crash of the Indian 1843) group have a large, highly differentiated mental subcontinent into Asia in the Paleogene (around 50 disc, a character state that he interpreted as a derived MYA; Karanth, 2006), the connection of Africa to Asia state. However, Menon did not comment on the fact that by the rotation of the Arabian plate in the late Miocene no species with the presumed ancestral character state (Hedges, 2001), and the connection of South America occurred in Yunnan, the proposed area of origin of the to North America in the Pliocene (Marshall, 1988). genus (Chu & Province, 1989). Only recently, Zheng The Western Palearctic (Europe and the Middle East) et al. (2012) postulated that the mental disc in labeonin became connected to the Asian continent as soon as the cyprinids has independently become reduced in several early Oligocene period when the Turgai Strait dried, lineages. This view gains additional support by checking allowing East Asian freshwater fishes to move west the position of the respective character states within (Ba˘na˘rescu, 1992; Briggs, 1995;Ro¨gl, 1999). The phylogenetic tree reconstructions presented by Geiger process of freshwater fish faunal exchange between et al. (2014) and Yang et al. (2012). All species lacking a Europe and Siberia lasted until the very recent post- mental disc found to be nested within Garra by Zheng glacial past (Durand et al., 2003). In contrast, the Indo- et al. (2012), Yang et al. (2012), Geiger et al. (2014)had Malayan or Oriental biogeographical realm was, and been described as belonging to different genera, still is, isolated from Europe and northern Asia, most although the presence of the mental disc is the diagnostic likely due to mountain ranges including the character state for the genus (Menon, 1964). In the uplifted by the northward-pushing Indian plate. As a Middle East, labeonin cyprinid species without a mental consequence, none of the species of primary freshwater disc that formerly had been placed in the genera fishes found in the Indus River occur further west than Hemigrammocapoeta Pellegrin, 1927, Crossocheilus Baluchistan in the Iran-Pakistan border area (Talwar, Kuhl & van Hasselt, 1823, Gonorynchus McClelland, 1991). From the great diversity of Asian freshwater 1838, Iranocypris Bruun & Kaiser, 1944 or Tylognathus fishes, only a few genera are hypothesized to have Heckel, 1843 (Hashemzadeh Segherloo et al., 2012; entered the Palearctic region, namely some lineages of Farashi et al., 2014; Behrens-Chapuis et al., 2015) have families Bagridae, Cyprinidae, Mastacembelidae, later been moved to the genus Garra (Hamidan et al., 123 Hydrobiologia (2017) 785:47–59 49

2014). Also, Yang et al. (2012) found G. variabilis,a analyses. Voucher specimens were preserved in 5% species with a small mental disc, to be nested among formaldehyde, and stored in 70% ethanol. other species of the genus having a large and fully DNA extraction was performed using the Chelex developed mental disc. Based on this, Yang et al. (2012) 100 method (Estoup et al., 1996) or NucleoSpinÒ questioned the interpretation of a small and less Tissue kits (Macherey & Nagel GmbH). Either the structured mental disc as the ancestral character state complete mtDNA cytochrome c oxidase subunit I in Garra species. The hypothesis that species of Garra (COI) gene was amplified using primers FCOI20 (50- having a small, or even absent mental disc might AACCTCTGTCTTCGGGGCTA-30) and RCOI20 (50- represent a derived rather than an ancestral character AGTGGTTATGYGGCTGGCTT-30) (Hashemzadeh state is inconsistent with Menon’s (1964) hypothesis and Segherloo et al., 2012), or the 50 end of COI was leads to different biogeographical conclusions. amplified using primers FishF2_t1 (50-TGTAAAAC In this context, our main objective was to assess the GACGGCCAGTCGACTAATCATAAAGATATCG phylogenetic relationships of all Garra species in the GCAC-30), FishR2_t1 (5-CAGGAAACAGCTAT Middle East, to test the hypothesis that a small or GACACTTCAGGGTGACCGAAGAATCAGAA- absent mental disc has independently evolved multiple 30), VF2_t1 (50-TGTAAAACGACGGCCAGTCAAC times in Garra, and that species without a mental disc CAACCACAAAGACATTGGCAC-30), and FR1d_t1 nest within species groups having a normally devel- (50-CAGGAAACAGCTATGACACCTCAGGGTGT oped mental disc. To do so, we analysed a full set of CCGAARAAYCARAA-30) (Ivanova et al., 2007). species of Garra species found in the Middle East, Polymerase chain reaction (PCR) conditions were as excluding those from the Arabian Peninsula, which follows: a 25-ll final reaction volume containing were recently studied by Hamidan et al. (2014). A 2.5 ll of 10X Taq polymerase buffer, 0.5 llof second objective was to shed light on the potential (50 mM) MgCl2, 0.5 ll of (10 mM) dNTPs, 0.5 ll ecological drivers for the reduction of the mental disc (10 lm) of each primer, 0.5 llofTaq polymerase (5u -1 in species of Garra. Many species of Garra are highly ll ), 2 ll of total DNA, and 18 llofH2O. Ampli- rheophilic, inhabiting rapids and fast-flowing rivers fication cycles were as follows: denaturation for (our own field data). The mental disc in Garra is a 10 min at 94°C; 35 cycles at 94°C for 1 min, 52°C structure which is interpreted as an adhesive structure for 1 min, and 72°C for 1 min; and a final extension allowing fish to hold a stationary position in high water for 10 min at 72°C. The 50 end of the COI gene was velocities (Zhang, 2005; Zhou et al., 2005). Thus, it is sequenced in both direction using primer FCOI20 on plausible that species with a reduced mental disc an ABI-3130xl DNA sequencer using the manufac- inhabit different habitats than species with a fully turer’s protocol (Life Technologies, www. developed mental disc. If the mental disc functions as lifetechnologies.com), or at Macrogen Europe Labo- an adhesive structure, stagnophilic species should lack ratories with the forward sequencing primer M13F (50 or show a reduced mental disc. To test this hypothesis, GTAAAACGACGGCCAGT) and reverse sequencing we compared general habitat information collected primer M13R-pUC (50 CAGGAAACAGCTATGAC). through our own fieldwork in the Middle East. DNA sequence data were aligned and processed using ClustalX (Thompson et al., 1997) and MEGA6 (Tamura et al., 2013). To determine the haplotype Materials and methods groups and the related frequencies, a parsimony network was constructed using TCS 1.21 (Thompson et al., Fish specimens (n = 150) of 16 described and 4 1997) with parsimony probability of 90%; haplotype undescribed species were collected by electrofishing frequencies were extracted manually from the resulting (Supplementary Table 1). The of all species groups. All generated sequences were deposited in the studied here follows strictly the latest version of the NCBI GenBank database (Supplementary Table 1). Catalogue of Fishes (Eschmeyer et al., 2016). Fish The COI haplotype sequences generated were com- were killed by over-anaesthesia using clove powder, pared to published cyprinid sequences following chlorobutanol or tricaine methanesulfonate (MS222). BLAST searches (Altschul et al., 1997), identifying The right pectoral fin was clipped and fixed in 96% similar sequences for use in phylogenetic analysis ethanol for subsequent DNA extraction and genetic (Table 1). Kimura two-parameter (K2P) distances 123 50 Hydrobiologia (2017) 785:47–59

Table 1 List of DNA sequences included from the NCBI GenBank database and associated accession numbers Species Accession no. Species Accession no.

Bangana sp. JX074149.1 Garra lamta JX074158.1 Cirrhinus mrigala (Hamilton, 1822) GU195083.1 Garra longipinnis Banister & Clarke, 1977 KM214701.1 Garra barreimiae Fowler & Steinitz, 1956 KM214783.1 Garra mullya (Sykes, 1839) JX074155.1 Garra bicornuta Narayan Rao, 1920 JX074156.1 Garra nasuta (McClelland, 1838) JX074219.1 Garra bourreti (Pellegrin, 1928) JQ864601.1 Garra nasuta JX074219.1 Garra congoensis Poll, 1959 HM418168.1 Garra orientalis Nichols, 1925 GU086602.1 Garra cyrano Kottelat, 2000 JX074214.1 Garra orientalis JQ864603.1 Garra cyrano JX074214.1 Garra orientalis HM536884.1 Garra dembeensis (Ru¨ppell, 1835) KT192819.1 Garra ornate (Nichols & Griscom, 1917) JX074202.1 Garra dembeensis KF929909.1 Garra sahilia Krupp, 1983 KM214682.1 Garra flavatra Kullander & Fang, 2004 JF915607.1 Garra salweenica Hora & Mukerji, 1934 KM610651.1 Garra gotyla (Gray, 1830) FJ459493.1 Garra tengchongensis Zhang & Chen, 2002 JQ864607.1 Garra gotyla JF915613.1 Garra turcica Karaman, 1971 KM214696.1 Garra gravelyi (Annandale, 1919) JF915614.1 Garra waterloti (Pellegrin, 1935) JX074212.1 Garra hughi Silas, 1955 HQ219117.1 Garra waterloti JX074212.1 Garra jordanica KJ553525.1 Labeo bata (Hamilton, 1822) KC757216.1 Garra kempi Hora, 1921 JX074161.1 Labeo chrysophekadion (Bleeker, 1849) AP011199.1 Garra lamta (Hamilton, 1822) JX074157.1 Labeo rohita (Hamilton, 1822) JN412817.1

(Kimura, 1980) were computed at the intra- and inferences of phylogeny were selected using Parti- interspecific levels using MEGA6 (Tamura et al., 2013). tionFinder (Lanfear et al., 2012). Five different com- For the phylogenetic analyses, the sequences were binations of codon positions were checked using BIC treated in two different ways. In the first approach, no criterion, and the best combination of model/s for each evolutionary difference among the codon positions partitioning scheme was selected among the models. was assumed and hence all codon positions were The Bayesian phylogenetic analysis was performed for exposed to regular non-partitioned phylogenetic anal- 2,600,000 generations, sampling the Markov chain ysis. In the second approach, the sequences were every 100 generations, with the first 25% of the exposed to codon partitioning and analysed as differ- resulting trees discarded as burn-in. Codon-partitioned ent partitions to test for possible differences in Maximum Likelihood analysis was performed using phylogenetic inferences. RAxML (Stamatakis, 2006) with 1,000 bootstrap For the non-partitioned phylogenetic analysis, the replicates as branch support. Five labeonin species— best-fit model of base substitution was selected based including Labeo rohita, Cirrhinus mrigala, Labeo on the Akaike information criterion (AIC) and bata, sp., and Labeo chrysophekadion—were Bayesian information criterion (BIC) using jModeltest used as out-group species based on other studies of 2.1.8 (Darriba et al., 2012). Bayesian inference was labeonin phylogeny (Yang et al., 2012) and our used to analyse the sequences with MrBayes 3.1 preliminary analysis of data. In addition to the (Huelsenbeck & Ronquist, 2001; Ronquist & Huelsen- sequences produced in this study, sequences from 28 beck, 2003). The MCMC search was run for 5,000,000 species from GenBank (NCBI) were included in the generations, sampling the Markov chain every 1000 analysis (Table 1). Substitution saturation was generations. The first 25% (1,250) of the resulting trees checked using the substitution test (Xia et al., 2003; were discarded as burn-in. The Maximum Likelihood Xia & Lemey, 2009) implemented in DAMBE soft- analysis was performed using PhyML program (Guin- ware (Xia, 2013). don et al., 2010). In order to analyse the codon positions The molecular clock test was performed by com- under different partitioning schemes, the best schemes paring the ML value for the given topology with and for both the Bayesian and Maximum likelihood without the molecular clock constraints under the 123 Hydrobiologia (2017) 785:47–59 51

Tamura-Nei (1993) model as implemented in MEGA Phylogenetic analysis 6 (Tamura et al., 2013). The null hypothesis of equal evolutionary rate throughout the tree was rejected The best mutational model based on AIC and BIC (P \ 0.05). Hence, to check for the divergence times values was the GTR?C?I model with an a value of among the Middle Eastern and Afro-Asian species of 1.49. Phylogenetic trees reconstructed by Bayesian the genus, a molecular clock was calibrated using and Maximum Likelihood methods resulted in similar BEAST v 1.7 (Drummond et al., 2012). The time of topologies. Under the codon partitioning approach, the the most recent common ancestor of the Middle best partitioning scheme for both the Bayesian infer- Eastern G. rufa and its African sister clade has been ence of phylogeny and Maximum Likelihood methods estimated to be around 8–10 Ma (the possible con- was the scheme in which each codon position consid- nection period between the Africa and Asia via the ered as a separate partition. The best mutational Red Sea, Tang et al., 2009) and 23.4 Ma for the models to be used for Bayesian inference of phylogeny divergence of Labeo Cuvier, 1816 from Garra (Tang were SYM?I?G (for the 1st codon position), F81?I et al., 2009). The analyses were performed using the (for the 2nd codon position), and GTR?G (for the 3rd SRD06 model of sequence evolution, the relaxed codon position). For the codon-partitioned Maximum uncorrelated lognormal (UCLN) molecular clock Likelihood analysis of phylogeny, the 3-partitioned model, a Yule prior set on the tree, and other priors scheme was selected as the best scheme, but only one left as default (Tsigenopoulos et al., 2010). The model (GTR?G) was used. The phylogenetic trees analysis was performed for 107 generations. Parame- resulting from different approaches and models were ters were logged every 1,000 generations, and the first mostly similar in topology, and the only observable 10% was discarded as burn-in. Because our molecular difference between the non-partitioning and partition- clock could not be calibrated with fossils, time ing approaches was changes in support values related estimates must be interpreted cautiously. to each approach. Under the Bayesian method, in some cases, the posterior probability values were lower in the phylograms reconstructed via the codon partition- Results ing approach, and for the Maximum Likelihood method the bootstrap support values were improved COI sequence variation in some cases. Here the phylogram reconstructed via the codon partitioning approach is presented; in cases Among the 150 individuals sequenced, 60 haplotypes where the support values or posterior probabilities were recovered. A total of 232 bp of the 652 bp were were different between the partitioned and non- variable, and of these 199 sites were parsimony partitioned analyses, we indicate that in the text as informative. The base composition over all sites was npBS and npBI. on average 29.2% (T), 26.2% (C), 26.1% (A), and In the phylogram (Fig. 1), the Middle Eastern 18.4% (G). There was no significant saturation in species of Garra form two major clades that are called codon positions (ISS \ ISSC; P \ 0.01). The mean here clade I (BS: 58; npBS: 59, npBI: 0.95) and clade interspecies K2P distances ranged from 1.31% II (BS: 95; npBS: 89, npBI: 1.00). In our analysis, the between Garra mondica Sayyadzadeh, Esmaeili & African Garra clade is the sister to clade II (BS: 76; Freyhof, 2015 and G. sp. (Kol) to 19.75% between npBS: 65; npBI: 0.96). Clade I is composed of Garra Garra kemali (Hanko´, 1925) and Garra ghorensis rossica (Nikolskii, 1900), G. kemali, Garra mendere- Krupp, 1982 (mean ± SD; 9.55 ± 4.84). The mean sensis (Ku¨c¸u¨k, Bayc¸elebi, Gu¨c¸lu¨ &Gu¨lle, 2015), and intraspecies genetic distances ranged between 0.1% in G. varibilis. Within clade II, there are a number of sub- Garra thyphlops (Bruun & Kaiser, 1944) and 2.05% in clades, including sub-clades A (Garra culiciphaga Garra gymnothorax Berg, 1949 (mean ± SD; (Pellegrin, 1927); BS: 99; BI: 1.00; npBS:96), B 0.33 ± 0.48). The mean between-group K2P distance (Garra lorestanensis Mousavi-Sabet & Eagderi, 2016, between the two major phylogenetic groupings (Clade Garra typhlops (Bruun & Kaiser, 1944), G. gym- I and Clade II, Fig. 1) was 15.9%. The mean within- nothorax; BS: 68; BI: 0.98; npBS:55; npBI:0.96), C group distances among the members of the clade I and (Garra jordanica Hamidan, Geiger & Freyhof, 2014 clade II were 8.4 and 5.4%, respectively. and G. ghorensis; BS: 93; BI: 0.99), D (G. sp. 123 52 Hydrobiologia (2017) 785:47–59

Nahralkabir and Orontes; BS: 78), E (Garra persica from G. kemali and G. menderesensis occurred around Berg, 1914; BS: 84; BI: 0.99), F (Garra widdowsoni 5.5 Ma (95% HPD: 2.96–8.46). The divergence of clade (Trewavas, 1955), Garra turcica, Garra elegans II from the Asian and African Garra clades probably (Gu¨nther, 1868), G. mondica, G. sp. Tigris, G. sp. occurred around 10.5 Ma (95% HPD: 8.81–12.59) and Kol; BS: 70; npBS:64), and G (G. rufa; BS: 54; 9.3 Ma (95% HPD: 8.21–10.47), respectively. npBS:58), with Garra festai (Tortonese, 1939) as the sister group of other sub-clades of clade II. Discussion Molecular clock calibration Subdivisions and dispersal of Garra clades Based on the molecular clock calibration applied here, in the Middle East the mean divergence rate (substitution per site in million years) over all taxa included in the analysis was Middle Eastern Garra forms two major clades 0.008 ± 0.00007 (mean ± SE). According to the supported with low to high bootstrap values of molecular clock calibration, clade I probably diverged 58–95% (npBS: 59–89). The two clades concord with from some Asian species of Garra about 12 Ma (95% the two-invasion-wave hypothesis proposed by HPD: 8.09–16.75), and from the other group that Menon (1964). Menon (1964) even postulated that includes clade II around 15.7 Ma (95% HPD: G. variabilis and G. rossica are closely related as both 12.07–19.90). The divergence of G. rossica from other have a small and poorly structured mental disc, a result members of clade I probably occurred around 10 Ma confirmed by our study, although with low likelihood (95% HPD: 6.46–14.36). The divergence of G. variabilis support (BS: 58; npBS: 59) (Fig. 1). In addition to G.

Fig. 1 Bayesian phylogram Garra flavatra reconstructed based on the Garra bicornuta 50 end of the mitochondrial Garra kempi COI gene using a codon Garra tengchongensis Garra cf. nana partitioning approach. The Garra hughi values on the branches are 100/1.00 Garra mullya the maximum likelihood 100/1.00 69 Garra rossica_Sistan_Iran_2004_Ex82B8 bootstrap values and Garra rossica_Kavir_Iran_M318_Ex90F5 99/1.00 Garra cf. rossica_Karvander_Iran_1999_Ex82C1

Bayesian posterior C 58 98 Garra cf. rossica_Karvander_Iran_1999_Ex82B12 probability values (after l 88/0.99 Garra kemali_Tuz_Turkey_1076_Ex15C11 a slashes) calculated as node Garra menderesensis_Iskili_Turkey_74_Ex13C11 d e I supports for the Bayesian 92/0.95 Garra menderesensis_Iskili_Turkey_74_Ex13C10 phylogram (higher than 99/1.00 Garra variabilis_Orontes_Syria_1166 Green branches Garra variabilis_Orontes_Syria_1168_Ex15G10 0.95). = no 100/1.00 mental disc, blue Garra variabilis_Orontes_Syria_1168_Ex15G9 66 Garra variabilis_Orontes_Syria_1159_Ex15G4 branches = reduced mental Garra gravelyi disc, and black Garra salweenica branches = fully developed 100/1.00 Garra cyrano mental disc. The triangle Garra cyrano Garra orientalis and circles denote the 67 Garra bourreti African and non-Middle 97/1.00 Garra orientalis Eastern/Asian haplotypes, Garra orientalis 68 respectively. The codes Garra nasuta Garra nasuta presented after the names 0.1 100/1.00 are unique voucher IDs Garra gotyla Garra lamta 100/1.00 containing the species name, Garra gotyla 96/0.99 main river basin, country of Garra lamta origin of the specimen, and 100/1.00 Garra waterloti the unique voucher code for Garra waterloti Garra dembeensis the specimens used in this 98/1.00 study. The out-groups are Garra dembeensis AF Garra ornata not shown 68/0.98 Garra congoensis

123 Hydrobiologia (2017) 785:47–59 53

Fig. 1 continued 100/1.00 Garra festai_Litani_Lebanon_2153_Ex25F8 Garra festai_Litani_Lebanon_2153_Ex25F6

75 Garra festai_Litani_Lebanon_2153_Ex25F7 Garra longipinnis 76 Garra sahilia 99/1.00 Garra nana _Jordan_Syria_Ex16A Garra nana_Jordan_Jordan_NHJO_068_Ex91E9 100/1.00 Garra culiciphaga_Seyhan_Turkey_400Ex14H3 Garra culiciphaga_Ceyhan_Turkey_944_Ex15A6 99/1.00 Garra culiciphaga_Orontes_Syria_1143_Ex18E9 99/1.00 Garra culiciphaga_Orontes_Turkey_297Ex13H9 A 95 100/0.99 Garra culiciphaga_Orontes_Turkey_297_Ex13H10

68/0.98 100/1.00 Garra lorestanensis_Tigris_Iran _2022d Garra lorestanensis_Tigris_Iran Garra typhlops_Tigris_Iran 100/1.00 Garra typhlops_Tigris_Iran Garra typhlops_Tigris_Iran B 50 Garra gymnothorax_Tigris_Iran_M744_Ex90G12 Garra gymnothorax_Tigris_Iran_1662 75/0.95 Garra gymnothorax_Tigris_Iran_M848Ex90C6 100/1.00 Garra gymnothorax_Tigris_Iran_M845_Ex90C5 76 77 Garra gymnothorax_Tigris_Iran_M846_Ex90D3 Garra barreimiae 93/0.99 Garra jordanica Garra ghorensis_Dead Sea_Jordan_1225 C 78 Garra cf. persica_Makran_Iran_10 Garra persica_Kol_Iran_M701_Ex90H6 84/0.99 Clade II Garra persica_Makran_Iran_M700_Ex90F1 D Garra sp._NahralKabir_Syria_1147_Ex82C6 Garra sp._Orontes_Turkey_331_Ex14B8 Garra sp._Orontes_Turkey_331_Ex14B10 E 97/0.99 71 Garra sp._Orontes_Turkey_331_Ex14B9 Garra widdowsoni_Euphrates_2301_Ex72B7 Garra turcica 70 1.00 Garra elegans_Iraq_2300_Ex72B5

90 Garra sp._Tigris_Iran_M747_Ex90D5 Garra mondica_Mond_Iran_M793_Ex86D11 91/0.99 Garra mondica_Mond_Iran_M699_Ex90E7 F Garra mondica_Mond_Iran_1594 67 Garra sp._Kol_Iran_M301_Ex90G8 Garra sp._Kol_Iran_M299_Ex90H5 0.99 Garra sp._Kol_Iran_M302_Ex90H8 Garra rufa_Tigris_Iran_M745_Ex90E4 Garra rufa_Euphrates_Turkey_2414_Ex63H5 Garra rufa_Tigris_Iran_452 Garra rufa _ Mond_Iran_1762 Garra rufa_Zohreh_Iran_M711_Ex90G9 Garra rufa_Maharlu_Iran _M842_Ex90B5 Garra rufa_Helleh_Iran _M836_Ex90H11 Garra rufa_Tigris_Iran_2234_Ex72A11 54 Garra rufa_Tigris_Iran_461 G Garra rufa_ Tigris_Iran_M847_Ex90C10 54 Garra rufa_Helleh_Iran_M704_Ex90E10 Garra rufa_ Tigris_Iraq_2234_Ex72A12

80/0.99 _Garra rufa_Tigris_Iran_M844_Ex90C12 Garra rufa_Tigris_Iran_581 Garra rufa_ Dalaki_Iran_1721 Garra rufa_ Dalaki_Iran_1724 68 Garra rufa_ Dalaki_Iran_1725

0.1

123 54 Hydrobiologia (2017) 785:47–59 variabilis and G. rossica, G. kemali and G. mendere- divergence is around 4 Ma, which may be related to sensis that lack a mental disc are nested in this clade. river confluences that provided the possibility of The inclusion of these species in the G. variabilis invasion from the Orontes drainage to the Seyhan/ clade also is supported by the phylogeny reconstructed Ceyhan drainages. However, one must keep in mind based on the rhodopsin gene (Behrens-Chapuis et al., that the molecular clock calibrated for the Garra 2015). species may not be precise. G. kemali and G. mendrensis are endemic to lakes Sub-clade B includes G. gymnothorax, G. typhlops, Tuz and Iskili in central Anatolia, and G. variabilis is and G. lorestanensi. Garra typhlops lacks the mental distributed in the Orontes and Tigris–Euphrates sys- disc, but the other two species have a fully developed tems. According to the geographic proximity of the mental disc. These three species inhabit the Karun and Anatolian lakes and the Orontes system, the species Karkheh river drainages in the Iranian Tigris catch- may have their origin in the Orontes system. The ment. They show a high level of mtDNA divergence molecular clock calibrated based on the dates proposed that may be due to the habitat isolation of G. in Tang et al. (2009) suggests that the divergence of G. lorestanensi and G. typhlops from G. gymnothorax. kemali and G. menderesensis from G. variabilis Based on our data, G. lorestanensis and G. typhlops, occurred around 5.5 Ma (95% HPD: 2.96–8.46). This two sympatric subterranean species, probably resulted period is concordant with the Mesenian salinity crisis from two different colonization events of the subter- of the Mediterranean Sea, during which confluence of ranean habitat. This justification also can be the case the freshwater habitats may have occurred. Based on for G. widdowsoni of sub-clade F, which is a the molecular clock, the divergence between G. kemali subterranean species diverged from its surface-dwell- of the Lake Tuz and G. menderesensis of Lake Iskili ing relatives. The members of sub-clade C, including would have been occurred around 2 Ma (95% HPD: G. jordanica and G. ghorensis—both with well- 0.90–4.10), which may be related to local-scale developed mental disc—occur in the Dead Sea basin geodynamics in central Anatolia. It should be noted and show close relationships to one another. that these molecular clock dates may not be precise, The non-disc-bearing G. cf. persica also nests since they are based only on COI sequences and among members of the sub-clade D, which includes geological data. As mentioned above, no Garra fossil the disc-bearing G. persica from the Persian Gulf data with relevant extant species exist from the Middle basin. Garra cf. persica is highly diverged from G. East to be used for molecular calibration. The only persica and inhabits the Kesh River (Makran Basin) fossil data in the region are reported from Armenia draining to the Sea of Oman. This pronounced level of (Vasilyan & Carnevale, 2013) where there is no extant divergence has resulted from the historic isolation of Garra species and so no molecular data for clock the basins in the Persian Gulf region and the Sea of calibration. The relationship of the clades I and II is not Oman, a region which was not as strongly affected as supported (BS \ 50). The African species are sister to the Persian Gulf during the last ice age. Hence, any clade II (BS: 76; npBS: 65; npBI: 0.96). This inference interchange of the freshwater ichthyofauna between is concordant with the Asia-to-Africa movement of the the two basins should relate to earlier periods. species of Garra proposed by Tang et al. (2009) and Sub-clade F includes G. widdowsoni (Euphrates), Yang et al. (2012). This relationship can be supported G. turcica (Ceyhan drainage), G. elegans (Tigris), G. by the land bridge created between Africa and Asia sp. (Tigris), G. mondica (Mond), and G. sp. (Kol). All around 8–10 Ma when sea levels decreased to about the members of this sub-clade except G. elegans have 60 m below the current level and connection of Asia to a fully developed mental disc. Colonization events for Africa became possible via the Red Sea (Tang et al., members of this sub-clade, spanning the region from 2009). the Tigris to the Mond and Kol Basins in the Persian Based on the COI phylogeny, seven sub-clades can Gulf region, may have occurred during the last ice age be defined within clade II (Fig. 1). Sub-clade A via the watershed confluence events in the Persian includes G. culiciphaga, a species lacking the mental Gulf. The members of this sub-clade in the Mond disc, of the Seyhan, Ceyhan and Orontes River Basin were recently described as Garra mondica drainages, all located in the northeastern Mediter- (Sayyadzadeh et al., 2015). The centre of origin of ranean basin. The approximate time of their sub-clade F based on the haplotype network produced 123 Hydrobiologia (2017) 785:47–59 55 using the present haplotypes cannot be resolved relationship may indicate recent dispersal of the sub- (Fig. 2), since none of the available haplotypes show clade during the last ice age (18,000 years bp). At that to be ancestral. This question may be resolved via time, the Persian Gulf was dry, and the smaller Iranian more intensive sampling to obtain more ancestral rivers, including the Mond, were part of the greater haplotypes. Tigris drainage (Fagan, 2014). Hence, colonization Sub-clade G includes populations of G. rufa, all events among the respective rivers were possible. found in the Persian Gulf basin. Their close According to the haplotype network, it seems that

G. G. G. rufa rufa_ persic Tigris Helleh a_Kol _Iran _Iran _Iran

1 2 1 G. b G. G. G. G. rufa 10 persic rufa rufa_ rufa Tigris a_Mak Tigris Helleh Tigris _Iran ran_Ir _Iran _Iran _Iraq an 1 1 1 1 a G. G. G. rufa rufa G. G. sp. rufa 2 Euphrate 1 Tigris s_Mahar rufa Tigris Tigris _Iran 1 lu_Zohr Dalaki Iran _Iran eh 1 1 1 c 3 1 G. rufa_ G. G. G. 2 Tigris elegan rufa_ rufa_ 1 _Iran s Mond Helleh Iraq _Iran _Iran 5 7 11 4 G. mondi G. sp. G. sp. ca Oronte 1 Oronte 4 1 G. 35 Mond s s Iran Turke Turke turcica 4 y y 2 G. 3 mondi ca 1 Mond G. sp. Iran Nahral 3 Kabir 2 Syria G. G. sp. 1 G. sp. mondi Kol Kol ca Iran Iran Mond Iran

Fig. 2 Haplotype network showing mutational connections of ancestral haplotypes. The blue lines connecting the haplotypes different haplotypes in sub-clades D, E, F, and G. The large denote the mutational links and the numbers beside each line circles denote different haplotypes, the related species and the indicate the number of point-mutational differences between geographic locality. The size of the circles is not related to each pair of neighbouring haplotypes. The smaller black circles haplotype frequencies. The colours of the circles denote: black show the hypothetical haplotypes that lead to observed sub-clade D; grey: sub-clade E; green: sub-clade F; and blue: haplotypes sub-clade G. The blue circles with black outline are the possible

123 56 Hydrobiologia (2017) 785:47–59 haplotype a observed in the Euphrates (Turkey), the reduced disc character as the ancestral state and Maharlu (Iran), and Zohreh (Iran) systems is appar- the disc-bearing state as a derived state. We argue, ently one of the ancestral haplotypes due to its wide however, that the mental disc may have been fully geographic distribution and more connections to other reduced independently in 6–7 lineages in the Middle haplotypes. Among the other haplotypes, haplotypes East (Fig. 1). This reduction may have occurred b and c also seem to be old haplotypes, since they are once in clade I, in which G. kemali and G. mendere- intermediate haplotypes located on the mutational sensis lack a mental disc, while G. variabilis and G. path between haplotype a and the G. persica and G. rossica have a small and partly reduced mental disc, mondica groups, respectively. Other haplotypes respectively. Within clade II, 5–6 independent within sub-clade G are probably newer haplotypes lineages apparently have reduced the mental disc, with fewer connections and terminal positions in the depending upon whether G. festai is or is not related network (Fig. 2). to Garra nana (Heckel, 1843) (Fig. 3). Tree Because our G. rufa samples are mostly from Iran topologies presented by Behrens-Chapuis et al. with only a few specimens from other regions, samples (2015) place G. culiciphaga, G. nana, and G. festai of this species may not be sufficiently informative in one group, while our data separate G. festai from derive an inference regarding the centre of origin of the other two species. As disc evolution in labeonins sub-clade G. However, since an east–west direction of had been shown to be bidirectional by Zheng et al. invasion is accepted (Tang et al., 2009) for labeonins, (2012), the question is whether the mental disc was the direction of the G. rufa colonization also may be reduced in convergence or developed in conver- inferred from easterly drainages to the Tigris– gence. As the structure of the mental disc is Euphrates. This also may be confirmed by observation identical or differs in very minor aspects in all of intermediate haplotypes connecting the eastern G. disc-bearing species studied here, as well as Garra persica (sub-clade D) and G. mondica (sub-clade F) to species from Africa (Stiassny & Getahun, 2007) and G. rufa (sub-clade G) (Fig. 2). Clearly, this inference Asia (Menon, 1964), we see convergent reduction as cannot be confirmed until more specimens and addi- the most parsimonious explanation of the phenom- tional molecular data are collected and analysed. ena described here.

Disc shape Possible character displacement

Menon (1964) synthesized his hypothesis regarding Based on our own field data in which general habitat two waves of invasion of Garra from East Asia to and microhabitats were noted, up to three Garra the Middle East based on morphological attributes, species with different disc forms can be found including characters of the mental disc. He assumed sympatrically (Table 2). This is especially true in

Fig. 3 Ventral views of heads showing the mental disc. From the left: G. rufa, FSJF 3368, 102 mm SL, Nalparez River, Iraq; G. variabilis, FSJF 2845, 95 mm SL, Tigris River, Turkey; G. festai, FSJF 3268, 65 mm SL, Ammiq Marshes, Lebanon. Heads are not shown in scale

123 Hydrobiologia (2017) 785:47–59 57

Table 2 Middle Eastern Drainage studied Mental disc state Garra species with different mental disc states and their Fully developed Absent Reduced sympatric species Ceyhan G. rufa G. culiciphaga – Euphrates (epigean) G. rufa – G. variabilis Jordan G. sp. G. nana – Karun (subterranean) G. lorestanensi G. typhlops – Kol G. persica & G. sp. – Mond G. rufa & G. mondica –– Orontes G. sp. G. culiciphaga G. variabilis Tigris G. rufa G. elegans G. variabilis

springs, where open gravel fields with outflowing of this hypothesis will require more data collected both water occur closely adjacent to densely vegetated, in situ and used for more detailed phylogenetic stagnant microhabitats. Species with a fully developed assessment. Our observations indicating a possible mental disc are generally the only species in rapids and relationship between mental disc shape and habitat riffle habitats with fast-flowing water. Species with a preferences should be checked using a more quantita- reduced or absent mental disc are absent from fast- tive approach recording habitat parameters and habitat flowing waters and instead occur in slow-flowing selectivity for fish with different mental disc forms. streams, marshes and densely vegetated parts of springs. Different field collection notes reveal that Acknowledgments The authors thank Guillaume Coˆte´ and sympatric G. rufa and G. variabilis show remarkable Janine Schmidt for DNA extraction and sequence library preparation. This study is a product of the FREDIE project, differences in microhabitat choice, with G. rufa being financed in the SAW program by the Leibniz Association strongly connected to gravel and flowing water, and G. (SAW-2011-ZFMK-3; http://www.leibniz-gemeinschaft.de), an variabilis associated with submerged vegetation and NSERC (Canada) Discovery grant (http://www.nserc-crsng.gc. standing water. Other sympatric species, for example, ca) to L.B., grant no. 688M1GRD94 from Shahr-e-Kord University (http://www.sku.ac.ir) to I.H., a grant from Shiraz G. typhlops and G. lorestanensis collected at the University (www.shirazu.ac.ir) to H.R.E., a grant from Shahid opening of the subterranean water layer in Iran, seem Beheshti University (http://sbu.ac.ir) to A.A., and a grant from to differ in their ecological niches. Based on several Istanbul University (www2.istanbul.edu.tr) to M.O. observations during different periods of the year, the disc-bearing species appear at the opening during the pluvial periods of the year when there is water flow, but References the non-disc-bearing species (G. typhlops) can be observed at the locality all year-round. These obser- Altschul, S. F., T. L. Madden, A. A. Scha¨ffer, J. Zhang, Z. vations may indicate niche isolation driven by disc Zhang, W. Miller & D. J. Lipman, 1997. Gapped BLAST and PSI-BLAST: a new generation of protein database form and water flow. This situation might be inter- search programs. Nucleic Acids Research 25(17): preted as indicating character displacement (Brown & 3389–3402. Wilson, 1956; Gray et al., 2005; Pfennig & Pfennig, Bailey, R. G. & L. Ropes, 1998. Ecoregions: the ecosystem 2010, 2009; Robinson & Pfennig, 2013; Robinson & geography of the oceans and continents. Springer, Berlin. Ba˘na˘rescu, P., 1992. Zoogeography of Fresh Waters. Vol. 2. Wilson, 1994; Schluter, 2000), allowing the species to Distribution and Dispersal of Freshwater in North reduce competition in the subterranean habitat. Char- America and Eurasia. Aula-Verlag, Wiesbaden: 520–1091. acter displacement leading different species to avoid Behrens-Chapuis, S., F. Herder, M. Geiger, H. Esmaeili, N. competition in sympatry may have allowed the eco- Hamidan, M. O¨ zulug˘ &R.Sˇanda, 2015. Adding nuclear Garra rhodopsin data where mitochondrial COI indicates dis- logical co-occurrence of other species after their crepancies – can this marker help to explain conflicts in secondary contact. Admittedly, our data are not cyprinids? DNA Barcodes 3(1): 187–199. sufficient to rigorously support this hypothesis. Testing Briggs, J. C., 1995. Global Biogeography. Elsevier.

123 58 Hydrobiologia (2017) 785:47–59

Brown, W. L. & E. O. Wilson, 1956. Character displacement. Iran cave barb Iranocypris typhlops. Journal of Fish Biol- Systematic Zoology 5(2): 49–64. ogy 81(5): 1747–1753. Chu, X. & Y. Province, 1989. The Fishes of Yunnan, China, Part Hedges, S. B., 2001. Afrotheria: plate tectonics meets genomics. I. Cyprinidae. Science Press, Beijing. Proceedings of the National Academy of Sciences USA Coad, B. W., 2010. Freshwater Fishes of Iraq. Pensoft, Sofia and 98(1): 1–2. Moscow. Holdridge, L. R., 1967. Life zone ecology, rev ed. Centro Darriba, D., G. L. Taboada, R. Doallo & D. Posada, 2012. Cientı´fico Tropical, San Jose´. jModelTest 2: more models, new heuristics and parallel Huelsenbeck, J. P. & F. Ronquist, 2001. MRBAYES: Bayesian computing. Nature Methods 9(8): 772. inference of phylogenetic trees. Bioinformatics 17(8): De Laubenfels, D. J., 1975. Mapping the world’s vegetation. 754–755. Regionalization of formations and flora. Syracuse Geo- Ivanova, N. V., T. S. Zemlak, R. H. Hanner & P. D. Hebert, graphical Series (USA) no 4. 2007. Universal primer cocktails for fish DNA barcoding. Drummond, A. J., M. A. Suchard, D. Xie & A. Rambaut, 2012. Molecular Ecology Notes 7(4): 544–548. Bayesian phylogenetics with BEAUti and the BEAST 1.7. Karanth, P. K., 2006. Out-of- Gondwanan origin of some Molecular Biology and Evolution 29(8): 1969–1973. tropical Asian biota. Current Science 90(6): 789–792. Durand, J.-D., P. Bianco, J. Laroche & A. Gilles, 2003. Insight Kimura, M., 1980. A simple method for estimating evolutionary into the origin of endemic Mediterranean ichthyofauna: rates of base substitutions through comparative studies of phylogeography of Chondrostoma genus (Teleostei, nucleotide sequences. Journal of Molecular Evolution Cyprinidae). Journal of Heredity 94(4): 315–328. 16(2): 111–120. Ellenberg, H. & D. Mueller-Dombois, 1969. A framework for a Ku¨c¸u¨k, F. A., E. S. Bayc¸elebi, S. S. Gu¨c¸lu¨ &I.Gu¨lle, 2015. classification of world vegetation. UNESCO report SCWS/ Description of a new species of Hemigrammocapoeta 269, Paris. (Teleostei: Cyprinidae) from Lake Isıklı, Turkey. Zootaxa Eschmeyer, W. N., R. Fricke & R. van der Laan. 2016. Catalog 4052(3): 359–365. of Fishes. http://www.calacademy.org/scientists/projects/ Lanfear, R., B. Calcott, S. Y. Ho & S. Guindon, 2012. Parti- catalog-of-fishes. tionFinder: combined selection of partitioning schemes Estoup, A., C. Largiader, E. Perrot & D. Chourrout, 1996. Rapid and substitution models for phylogenetic analyses. one-tube DNA extraction for reliable PCR detection of fish Molecular Biology and Evolution 29(6): 1695–1701. polymorphic markers and transgenes. Molecular Marine Marshall, L. G., 1988. Land mammals and the great American Biology and Biotechnology 5(4): 295–298. interchange. American Scientist 76(4): 380–388. Fagan, B., 2014. The Attacking Ocean: The Past, Present, and Menon, A. G. K., 1964. Monograph of the cyprinid fishes of the Future of Rising Sea Levels. Bloomsbury Publishing, New genus Garra Hamilton. Memoirs of the Indian Museum York. 14(4): 173–260. Farashi, A., M. Kaboli, H. R. Rezaei, M. R. Naghavi, H. Rahi- Pfennig, D. W. & K. S. Pfennig, 2010. Character displacement mian & B. Coad, 2014. Reassessment of the taxonomic and the origins of diversity. The American Naturalist position of Iranocypris typhlops Bruun & Kaiser, 1944 176(Suppl 1): S26. (, Cyprinidae). ZooKeys 374: 69–77. Pfennig, K. S. & D. W. Pfennig, 2009. Character displacement: Geiger, M., F. Herder, M. Monaghan, V. Almada, R. Barbieri, ecological and reproductive responses to a common evo- M. Bariche, P. Berrebi, J. Bohlen, M. Casal-Lopez & G. lutionary problem. The Quarterly Review of Biology Delmastro, 2014. Spatial heterogeneity in the Mediter- 84(3): 253. ranean biodiversity hotspot affects barcoding accuracy of Robinson, B. W. & D. W. Pfennig, 2013. Inducible competitors and its freshwater fishes. Molecular Ecology Resources 14(6): adaptive diversification. Current Zoology 59(4): 537–552. 1210–1221. Robinson, B. W. & D. S. Wilson, 1994. Character release and Gray, S., B. Robinson & K. Parsons, 2005. Testing alternative displacement in fishes: a neglected literature. American explanations of character shifts against ecological charac- Naturalist 144: 596–627. ter displacement in brook sticklebacks (Culaea inconstans) Ro¨gl, F., 1999. Mediterranean and Paratethys: facts and that coexist with ninespine sticklebacks (Pungitius pungi- hypotheses of an Oligocene to Miocene paleogeography tius). Oecologia 146(1): 25–35. (short overview). Geologica Carpathica 50(4): 339–349. Guindon, S., J. F. Dufayard, V. Lefort, M. Anisimova, W. Ronquist, F. & J. P. Huelsenbeck, 2003. MrBayes 3: Bayesian Hordijk & O. Gascuel, 2010. New algorithms and methods phylogenetic inference under mixed models. Bioinfor- to estimate Maximum-Likelihood phylogenies: assessing matics 19(12): 1572–1574. the performance of PhyML 3.0. Systematic Biology 59(3): Sayyadzadeh, G., H. R. Esmaeili & J. Freyhof, 2015. Garra 307–321. mondica, a new species from the Mond River drainage with Hamidan, N. A., M. F. Geiger & J. Freyhof, 2014. Garra jor- remarks on the genus Garra from the Persian Gulf basin in danica, a new species from the Dead Sea basin with Iran (Teleostei: Cyprinidae). Zootaxa 4048(1): 75–89. remarks on the relationship of G. ghorensis, G. tibanica Schluter, D., 2000. The Ecology of Adaptive Radiation. Oxford and G. rufa (Teleostei: Cyprinidae). Ichthyolology and University Press, Oxford. Exploration of Freshwaters 25: 223–236. Schmidthu¨sen, J., 1976. Atlas zur Biogeographie. Bibli- Hashemzadeh Segherloo, I., L. Bernatchez, K. Golzarianpour, ographisches Institut, Zu¨rich. A. Abdoli, C. Primmer & M. Bakhtiary, 2012. Genetic Schultz, J., 1995. The Ecozones of the World: The Ecological differentiation between two sympatric morphs of the blind Divisions of the Geosphere. Springer, Berlin.

123 Hydrobiologia (2017) 785:47–59 59

Shermer, M., 2002. In Darwin’s Shadow: The Life and Sci- Surface; with Maps and Illustrations; in Two Volumes. ence of Alfred Russel Wallace: A Biographical Study on Cambridge University Press, Cambridge. the Psychology of History. Oxford University Press, Walter, H. & E. Box, 1976. Global classification of natural Oxford. terrestrial ecosystems. Vegetatio 32(2): 75–81. Stamatakis, A., 2006. RAxML-VI-HPC: maximum likelihood- Xia, X., 2013. DAMBE5: a comprehensive software package for based phylogenetic analyses with thousands of taxa and data analysis in molecular biology and evolution. Molec- mixed models. Bioinformatics 22(21): 2688–2690. ular Biology and Evolution 30(7): 1720–1728. Stiassny, M. L. & A. Getahun, 2007. An overview of labeonin Xia, X. & P. Lemey, 2009. Assessing Substitution Saturation relationships and the phylogenetic placement of the Afro- with DAMBE. In Salemi, M. & A.-M. Vandamme (eds), Asian genus Garra Hamilton, 1922 (Teleostei: Cyprinidae), The phylogenetic handbook: a practical approach to DNA with the description of five new species of Garra from and protein phylogeny. Cambridge University Press, Ethiopia, and a key to all African species. Zoological Cambridge: 615–630. Journal of the Linnean Society 150(1): 41–83. Xia, X., Z. Xie, M. Salemi, L. Chen & Y. Wang, 2003. An index Talwar, P. K., 1991. Inland Fishes of India and Adjacent of substitution saturation and its application. Molecular Countries, Vol. 2. CRC Press, Boca Raton, FL. Phylogenetics and Evolution 26(1): 1–7. Tamura, K., S. Glen, P. Daniel, F. Alan & K. Sudhir, 2013. Yang, L. & R. L. Mayden, 2010. Phylogenetic relationships, MEGA6: molecular evolutionary genetics analysis version subdivision, and biogeography of the cyprinid tribe 6.0. Molecular Biology and Evolution 30: 2725–2729. Labeonini (sensu) (Teleostei: ), with com- Tang, Q., A. Getahun & H. Liu, 2009. Multiple in-to-Africa ments on the implications of lips and associated structures dispersals of labeonin fishes (Teleostei: Cyprinidae) in the labeonin classification. Molecular Phylogenetics and revealed by molecular phylogenetic analysis. Hydrobi- Evolution 54(1): 254–265. ologia 632(1): 261–271. Yang, L., M. Arunachalam, T. Sado, B. A. Levin, A. S. Gol- Thompson, J. D., T. J. Gibson, F. Plewniak, F. Jeanmougin & D. ubtsov, J. Freyhof, J. P. Friel, W.-J. Chen, M. Vincent Hirt G. Higgins, 1997. The CLUSTAL_X windows interface: & R. Manickam, 2012. Molecular phylogeny of the flexible strategies for multiple sequence alignment aided cyprinid tribe Labeonini (Teleostei: Cypriniformes). by quality analysis tools. Nucleic Acids Research 25(24): Molecular Phylogenetics and Evolution 65(2): 362–379. 4876–4882. Zhang, E., 2005. Phylogenetic relationships of labeonine Tsigenopoulos, C. S., P. Kasapidis & P. Berrebi, 2010. Phylo- cyprinids of the disc-bearing group (Pisces: Teleostei). genetic relationships of hexaploid large-sized barbs (genus Zoological Studies 44(1): 130–143. Labeobarbus, Cyprinidae) based on mtDNA data. Molec- Zheng, L., Y. Junxing & C. Xiaoyong, 2012. Phylogeny of the ular Phylogenetics and Evolution 56(2): 851–856. (Teleostei, Cypriniformes) based on nuclear Vasilyan, D. & G. Carnevale, 2013. The Afro-Asian labeonine DNA sequences and implications on character evolution genus Garra Hamilton, 1822 (Teleostei, Cyprinidae) in the and biogeography. Current Zoology 58(6): 837–850. Pliocene of Central Armenia: Palaeoecological and Zhou, W., X.-F. Pan & M. Kottelat, 2005. Species of Garra and palaeobiogeographical implications. Journal of Asian Discogobio (Teleostei: Cyprinidae) in the Yuanjiang Earth Sciences 62: 788–796. (Upper Red River) drainage of Yunnan Province, China Wallace, A. R., 1876. The Geographical Distribution of Ani- with a description of a new species. Zoological Studies mals, with a Study of the Relations of Living and Extinct Taipei 44(4): 445. Faunas as Elucidating the Past Changes of the Earth’s

123