Journal of Biology (2002) 61, 929–944 doi:10.1006/jfbi.2002.2105, available online at http://www.idealibrary.com on

Phylogenetic relationships of Eurasian and American cyprinids using cytochrome b sequences

C. C*, N. M*, T. E. D†, A. G‡  M. M. C*§ *Centro de Biologia Ambiental, Departamento de Zoologia e Antropologia, Faculdade de Cieˆncia de Lisboa, Campo Grande, Bloco C2, 3 Piso. 1749-016 Lisboa, Portugal, †Department of Biology, Arizona State University, Tempe, Arizona 85287-1501, U.S.A. and ‡Laboratoire d’Hydrobiology, Universite´ de Provence, 1 Place Victor Hugo, 1331 Marseille, France

(Received 30 January 2002, Accepted 6 August 2002)

Neighbour-joining and parsimony analyses identified five lineages of cyprinids: (1) European leuciscins (including Notemigonus)+North American phoxinins (including Phoxinus phoxinus); (2) European gobionins+; (3) primarily Asian groups [cultrins+acheilognathins+ gobionins (excluding Abbotina)+xenocyprinins]; (4) ++; (5) cyprinins [excluding Sinocyclocheilus and Acrossocheilus]+barbins+labeonins. Relationships among these lineages and the enigmatic taxa Rhodeus were not well-resolved. Tests of mono- phyly of subfamilies and previously proposed relationships were examined by constraining cytochrome b sequences data to fit previous hypotheses. The analysis of constrained trees indicated that sequence data were not consistent with most previously proposed relationships. Inconsistency was largely attributable to Asian taxa, such as and Xenocyprioides. Improved understanding of historical and taxonomic relationships in will require further morphological and molecular studies on Asian cyprinids and taxa representative of the diversity found in Africa.  2002 The Fisheries Society of the British Isles. Published by Elsevier Science Ltd. All rights reserved.

Key words: Cyprinidae; molecular phylogeny; cytochrome b; monophyly of subfamilies.

INTRODUCTION The Cyprinidae family is one of the most widespread, diverse families of freshwater fishes. It consists of >340 genera and 2000 (Banarescu & Coad, 1991) and naturally occurs in almost all types of habitats on all continents except for Australia and South America (Banarescu & Coad, 1991; Howes, 1991). These features make it an excellent group for a diversity of biological studies. Previous morphological studies failed to reach a consensus regarding the number and monophyly of subfamilies (Fig. 1). Morphological studies (Chen et al., 1984; Howes, 1991; Cavender & Coburn, 1992) supported the division of Cyprinidae into two major lineages, but there was disagreement over taxonomic alignment. Chen et al. (1984) identified two lineages: the series Leuciscini (containing the subfamilies Danioninae, , Cultrinae, , and Acheilognathinae) and series Barbini (consisting of the §Author to whom correspondence should be addressed. Tel.: +351 217500000 (ext.: 22314); fax: +351 217500028; email: [email protected] 929 0022–1112/02/010929+16 $35.00/0  2002 The Fisheries Society of the British Isles. Published by Elsevier Science Ltd. All rights reserved. 930 .   . Cyprininae Gobioninae Rasborinae Cultrinae Leuciscinae Acheilognathinae Barbinae Cyprininae Tincinae Danioninae Leuciscinae Cultrinae Xenocyprinae Gobioninae Acheilognathinae Barbelled Barbel- lacking Cyprinines Leuciscines Series Barbini Series Leuciscini (a) (b) Barbins Cyprinins Labeonins Rasborins Gobionins Acheilognathins Xenocyprins Cultrins Leuciscins Phoxinins Tincins Subfamily Cyprininae Subfamily Leuciscinae

(c) F. 1. Hypothesized relationships among cyprinid groups based on morphological characters: (a) Chen et al. (1984);(b)Howes (1991); (c) Cavender & Coburn (1992).

Tincinae, Barbinae, Cyprininae and Labeoninae). Howes (1991) identified the cyprinines (Cyprininae, Gobioninae and Rasborinae) and the leuciscines (Leuciscinae, Alburninae, Cultrinae and Acheilognathinae) based on the presence and absence (sporadic in some groups) of barbels. It was unclear, however, whether the presence of this character is plesiomorphic or apomorphic. Results presented by Cavender & Coburn (1992) partially agreed with those of Chen et al. (1984). They proposed that cyprinids fell in two subfamilies, Leuciscinae (including the phoxinins, leuciscins, cultrins, xenocyprins, acheilognathins, gobionins, rasborins and tincins) and Cyprininae (including the barbins, cyprinins and labeonins). The leuciscins included most Eurasian cyprinids and the monotypic North American Notemigonus with the remaining North American forms placed in the phoxinin lineage.     931

Cyprinids have recently been the focus of molecular studies addressing the evolutionary history of the family and the validity of its subfamilies (Zardoya & Doadrio, 1999; Tsigenopulos & Berrebi, 2000; Gilles et al., 2001; Machordom & Doadrio, 2001). Recently, phylogenetic studies of mainly European cyprinids based on the mitochondrial genes cytochrome b (cytb), D-loop and 16S rRNA sequences (Briolay et al., 1998; Gilles et al., 1998, 2001; Zardoya & Doadrio, 1998, 1999) presented results inconsistent with recognition of several of these subfamilies. Briolay et al. (1998) found a radiation of six subfamilies [Leuciscinae, including Alburninae sensu Howes (1991), Rasborinae, Gobioninae, Tincinae, Acheilognathinae, Cyprininae] rather than an early split between barbelled and barbel-lacking lineages, consistent with several origins of barbels during cyprinid evolution. In all morphological and molecular studies, the relative positions of and Tinca were unresolved, with these taxa considered as Leuciscinae, or referred to the subfamilies Gobioninae and Tincinae, respectively. Zardoya & Doadrio (1999) considered cyprinids to contain two subfamilies: Cyprininae (including barbins) and Leuciscinae (includ- ing cultrins, tincins, gobionins, phoxinins and alburnins+leuciscins). The results obtained by Gilles et al. (2001) are mostly in agreement with those of Briolay et al. (1998) and Zardoya & Doadrio (1999). This study, however, considered the Rasborinae as the most basal cyprinid subfamily and did not consider the Tincinae and the Acheilognathinae as a sister group of the Cyprininae. Some authors (Briolay et al., 1998; Gilles et al., 1998) included a few North American and Asian species, however, the restricted availability of sequences limited their ability to examine historical relationships. The present study is the first attempt to infer evolutionary relationships of cyprinids based on a consider- able number and diversity of representatives from genera representing diversity in Asia, Europe and North America. Sequences from the cyt b gene were used to estimate phylogenetic relationships and examine previously hypothesized relationships presented by Chen et al. (1984), Howes (1991) and Cavender & Coburn (1992), based on morphology.

MATERIAL AND METHODS Several specimens of European (33), Asian (38), North American (18) cyprinids and outgroups [Crossostoma lacustre Steindachner, Balitoridae; Myxocyprinus asiaticus (Bleeker), ] were sampled or were taken from published databases (Table I). Total DNA was extracted from fins preserved in absolute ethanol or from frozen muscle tissue using standard methods (Sambrook et al., 1989). Cyt b was amplified using primers reported elsewhere (Schmidt & Gold, 1993; Alves et al., 1997; Brito et al., 1997; Dowling & Naylor, 1997). Amplification conditions varied, but generally fit the following profile: 20–25 cycles of 1 min at 94 C, 1 min 48–50 C; 2 min at 72 C. Reactions were performed in 25–50 l volumes containing 25–50 ng of template DNA, and 0·5  of each primer. PCR products were purified with QIAquick PCR Purification Kit (Quiagen) or Millipore centrifugation tubes. Sequences were generated with an automated sequencer (Genome Express or ASU). In European sequences, both strands were sequenced to control sequence accuracy. In American sequences only one strand was sequenced, however, there was several hundred bases pairs overlap between fragments. DNA sequences were aligned manually using DNASIS (Macintosh ver. 2.0) and translated into amino acids using McClade v.3.08a (Maddison & Maddison, 1992). No T I. List of cyprinid species analysed in the present study. EMBL access numbers (*present study), specimens’ geographic origin (**data not available) and morphological classification considered by Chen et al. (1984), Howes (1991) and Cavender & Coburn (1992) are presented in this table

Cavender & Chen et al. (1984) Howes (1991) Taxa EMBL no. Drainage Origin Coburn (1992) subfamilies subfamilies lineages

Abbottina rivularis AF051856 ** Asia Gobioninae Gobioninae Gobionins Abramis brama Y10441 Saoˆne Europe Leuciscinae Leuciscinae Leuciscins Acheilognathus chankaensis AF051854 ** Asia Acheilognathinae Acheilognathinae Acheilognathins Acrocheilus alutaceus AF452076* Columbia North America Leuciscinae Leuciscinae Phoxinins Acrossocheilus yunnanensis AF051857 ** Asia Barbinae Cyprininae Cyprinins bipunctatus Y10445 Saoˆne Europe Leuciscinae Alburninae Leuciscins alburnus Y10443 Rhoˆne Europe Leuciscinae Alburninae Leuciscins Alburnus filippi AF095602 Samur Europe Leuciscinae Alburninae Leuciscins AJ439881* Guadiana Europe Leuciscinae Leuciscinae Leuciscins Aulopyge hyegelii AF112133 Krka Europe Barbinae Cyprininae Barbins brachycephalus AF095603 Terek Europe Barbinae Cyprininae Barbins Barbus caninus AF112124 Astico Europe Barbinae Cyprininae Barbins Barbus graellsii AF045973 Ebro Europe Barbinae Cyprininae Barbins Barbus meridionalis AF112130 Aubaygue Europe Barbinae Cyprininae Barbins Y10442 Saoˆne Europe Leuciscinae Leuciscinae Leuciscins anomalum AF452079* Mississipi North America Leuciscinae Leuciscinae Phoxinins auratus AF051858 ** Asia Cyprininae Cyprininae Cyprinins belvica AF090746 Prespa Greece, Europe Leuciscinae Alburninae Leuciscins toxostoma AJ439831* Nive Europe Leuciscinae Leuciscinae Leuciscins Couesius plumbeus AF452083* Yukon North America Leuciscinae Leuciscinae Phoxinins Ctenopharyngodon idella AF051860 ** Asia Leuciscinae Cyprininae Xenocyprins Cultrichthys erythropterus AF051859 ** Asia Cultrinae Cultrinae Cultrins spiloptera U66605 ** North America Leuciscinae Leuciscinae Phoxinins carpio X61010 ** Asia Cyprininae Cyprininae Cyprinins tumirostris AF036173 ** Asia Xenocyprinae Leuciscinae Xenocyprins T I. Continued

Cavender & Chen et al. (1984) Howes (1991) Taxa EMBL no. Drainage Origin Coburn (1992) subfamilies subfamilies lineages

Gila cypha AF452074* Colorado North America Leuciscinae Leuciscinae Phoxinins Gobio banarescui AF090751 Aliakmon Europe Gobioninae Gobioninae Gobionins Gobio ciscaucasicus AF095607 Uluchaj Europe Gobioninae Gobioninae Gobionins AF045996 Tagus Europe Gobioninae Gobioninae Gobionins abbreviata AF051861 ** Asia Gobioninae Gobioninae Gobionins Gobiobotia longibarba AF051863 ** Asia Gobioninae Gobioninae Gobionins Gobiototia ichangensis AF051862 ** Asia Gobioninae Gobioninae Gobionins przewalskii AF051864 ** Asia Barbinae Cyprininae not defined leucisculus AF051865 ** Asia Cultrinae Alburninae Cultrins molitrix AF051866 ** Asia Xenocyprinae Leuciscinae Xenocyprins Hypophthalmichthys nobilis AF051855 ** Asia Xenocyprinae Leuciscinae Xenocyprins Lepidomeda vittata AF452088 Colorado North America Leuciscinae Leuciscinae Phoxinins Y10447* Rhoˆne Europe Leuciscinae Leuciscinae Leuciscins leuciscus Y10449 Rhoˆne Europe Leuciscinae Leuciscinae Leuciscins cerasinus U66599 Roanoke North America Leuciscinae Leuciscinae Phoxinins ardens U17268 Ohio North America Leuciscinae Leuciscinae Phoxinins Meda fulgida AF45209* Colorado North America Leuciscinae Leuciscinae Phoxinins amblycephala AF051867 ** Asia Cultrinae Cultrinae Cultrins AF051869 ** Asia Cultrinae Cultrinae Cultrins Megalobrama skolkovii AF051871 ** Asia Cultrinae Cultrinae Cultrins Megalobrama terminalis AF051872 ** Asia Cultrinae Cultrinae Cultrins Mylopharyngodon piceus AF051870 ** Asia Leuciscinae Cyprininae Xenocyprins micropogon AF452077* Lake Erie North America Leuciscinae Leuciscinae Phoxinins Notemigonus crysoleucas U01318 ** North America Leuciscinae Leuciscinae Leuciscins Notropis atherinoides AY096008* Lake Erie North America Leuciscinae Leuciscinae Phoxinins Orthodon microlepidotus AF452073* Sacramento North America Leuciscinae Leuciscinae Phoxinins T I. Continued

Cavender & Chen et al. (1984) Howes (1991) Taxa EMBL no. Drainage Origin Coburn (1992) subfamilies subfamilies lineages

Pachychilon pictus AF090762 Aoos Europe Leuciscinae Leuciscinae Leuciscins Parabramis pekinensis AF051874 ** Asia Cultrinae Cultrinae Cultrins prespensis AF090763 Prespa Europe Leuciscinae Leuciscinae Leuciscins Phoxinus phoxinus Y10448 Rhoˆne Europe Leuciscinae Leuciscinae Phoxinins notatus U66606 Lake Erie North America Leuciscinae Leuciscinae Phoxinins macrolepidotum AY096009* Sacramento North America Leuciscinae Leuciscinae Phoxinins AF036194 ** Asia Xenocyprinae Leuciscinae Xenocyprins stymphalicus AF090766 Stymphalia Europe Leuciscinae Leuciscinae Leuciscins Pseudorasbora parva Y10453 Rhoˆne Europe Gobioninae Rasborinae Rasborins Pseudorasbora parva AF051873 ** Asia Gobioninae Rasborinae Rasborins Raiamas guttatus AF051875 ** Asia Danioninae Rasborinae Rasborins atratulus AF452078* Lake Erie North America Leuciscinae Leuciscinae Phoxinins Rhodeus ocellatus AF051876 ** Asia Acheilognathinae Acheilognathinae Acheilognathins Rhodeus sericeus Y10454 Saoˆne Europe Acheilognathinae Acheilognathinae Acheilognathins balteatus AY096011* Bonneville North America Leuciscinae Leuciscinae Phoxinins caspicus AF095610 Samur Europe Leuciscinae Leuciscinae Leuciscins Rutilus rutilus Y10440 Saoˆne Europe Leuciscinae Leuciscinae Leuciscins acarnanicus AF090775 Trichonis Europe Leuciscinae Leuciscinae Leuciscins prenanti AF051880 ** Asia Barbinae Cyprininae Barbins Semilabeo prochilus AF051881 ** Asia Labeoninae Cyprininae Labeonins Sinocyclocheilus grahami AF051879 ** Asia Cyprininae Cyprininae Cyprinins bicolor AY096010* ** North America Leuciscinae Leuciscinae Phoxinins Spinibarbus denticulatus AF051878 ** Asia Barbinae Cyprininae Barbins Squaliobarbus curriculus AF051877 ** Asia Leuciscinae Cyprininae Xenocyprins cephalus Y10446 Rhoˆne Europe Leuciscinae Leuciscinae Leuciscins Z75927* Guadiana Europe Leuciscinae Leuciscinae Leuciscins T I. Continued

Cavender & Chen et al. (1984) Howes (1991) Taxa EMBL no. Drainage Origin Coburn (1992) subfamilies subfamilies lineages

Telestes souffia Y10439 Saoˆne Europe Leuciscinae Leuciscinae Leuciscins Tinca tinca Y10451 Saoˆne Europe Tincinae Incertae sedis Tincins hellenicus AF090776 Pinios Europe Leuciscinae Leuciscinae Leuciscins melanops AF090778 Strymon Europe Leuciscinae Leuciscinae Leuciscins Xenocyprioides carinatus AF036201 ** Asia Xenocyprinae Rasborinae or Leuciscinae Xenocyprins Xenocyprioides parvulus AF036207 ** Asia Xenocyprinae Rasborinae or Leuciscinae Xenocyprins Xenocypris argentea AF036196 ** Asia Xenocyprinae Leuciscinae Xenocyprins Xenocypris dividi AF036195 ** Asia Xenocyprinae Leuciscinae Xenocyprins Xenocypris fangi AF036205 ** Asia Xenocyprinae Leuciscinae Xenocyprins Xenocypris hupeinensis AF036165 ** Asia Xenocyprinae Leuciscinae Xenocyprins Xenocypris microlepis AF036206 ** Asia Xenocyprinae Leuciscinae Xenocyprins Xenocypris yunnanensis AF036208 ** Asia Xenocyprinae Leuciscinae Xenocyprins 936 .   . ambiguous alignments were found. Levels of saturation were examined by plotting the pairwise number of transitions (Ti) v. the pairwise number of transversions (Tv) for all nucleotides and each codon position separately. Phylogenetic reconstructions were obtained using the neighbour-joining (NJ) (Saitou & Nei, 1987) and maximum parsimony (MP) (Fitch, 1971) methods as implemented in PAUP* v.4.0d (Swofford, 1998). When the number of nucleotides is small relative to the number of taxa (as is the case in this study), simple search algorithms and nucleotide substitution models are as efficient as more exhaustive searches and complicated models at recovering the true topology (Nei & Kumar, 2000; Takahashi & Nei, 2000). Consequently NJ analysis was based on Jukes and Cantor distance and MP analysis was performed using a heuristic search (NNI branch swapping; MULPARS option) with characters unweighted and weighted [tranversions (Tv) given three times the weight of transitions (Ti)]. Support values for internal nodes of the trees were estimated using the bootstrap resampling procedure (Felsentein, 1985) with 1000 replicates. The monophyly of subfamilies was examined using the Kishino–Hasegawa (K–H) (Kishino & Hasegawa, 1989) and Templeton (Templeton, 1983) tests as implemented in PAUP. Topologies contrasted were obtained from unconstrained MP searches (desig- nated MP) and those constrained (designated CON) to fit the previous hypotheses (subfamily affiliations as defined previously by Chen et al., 1984; Howes, 1991; Cavender & Coburn, 1992; Fig. 1). Placement of Xenocyprioides was not clearly defined in Howes (1991); therefore, this genus was tested in both the Rasborinae and Leuciscinae as suggested by Banarescu & Coad (1991) and Rainboth (1991), respectively. Several taxa included here were not examined by Cavender & Coburn (1992). These were placed through personal communication (T. Cavender and M. Coburn), excluded from analysis [ (Kessler)], or placed in the leuciscins [ pictus (Heckel & Kner), Phoxinellus prespensis (Karaman), Pseudophoxinus stymphalicus (Valenciennes)] as in Zardoya & Doadrio (1999).

RESULTS Of the 1140 nucleotides of cyt b, 596 sites were variable with 516 of these phylogenetically informative using the parsimony criterion. The mean base composition was similar to numerous fishes (Cantatore et al., 1994) and also for cyprinids (Dowling & Naylor, 1997; Briolay et al., 1998; Tsigenopoulos & Berrebi, 2000): lower G content (mean=15·7%) and similar content of A (mean=27·6%), C (mean=28·3%), and T (mean=28·4%). The average pairwise ratio of transitions (Ti) v. transversions (Tv) was 2·48. Most of the variability among sequences was observed in the third codon position. First and second positions show very slight saturation, whereas substitutions at the third position are strongly saturated (unpubl. data). Even when sites appear to be saturated (through comparison of Ti : Tv ratios), they may include phylogenetic informa- tion (Broughton et al., 2000). The present examination of the data are consistent with these findings as separate analyses in which third positions and transitions were excluded yielded considerably less resolution. Therefore, all sites in all analyses were included. Neighbour-joining clustering of Jukes & Cantor (1969) distances yielded the topology given in Fig. 2. Topologies recovered from unweighted and weighted (Tv three times Ti) parsimony analyses were similar, recovering four and 20 most-parsimonious trees, of 7177 and 10 338 steps, respectively. The weighted analysis yielded a more highly resolved topology that was congruent to the one recovered by NJ analysis. Phylogenetic analyses using NJ and MP identified five major lineages supported by moderate to high bootstrap values (Fig. 2): (1) all European I II III IV

F. 2. Neighbour-joining tree of recovered from cyt b sequence data from European, American and Asian taxa using JC distance. Branch lengths are proportional to the expected mean number of substitutions per site. Percentage bootstrap supports obtained from neighbour-joining (JC distance) and the maximum parsimony analyses are presented above and below the branches, respectively. The first bar (I) corresponds to geographic distribution and the following bars to the morphological classifications of (II) Chen et al. (1984), (III) Howes (1991) and (IV) Cavender & Coburn (1992). I (Geographic Distribution): , Europe; , America; , Asia; , Holartic; , Paleartic. II [Chen et al. (1984) hypothesis]: , Acheilognathinae; , Gobioninae; , Xenocyprininae; , Cultrinae; , Leuciscinae; , Daneoninae; , Cyprininae , Barbinae; , Labeoninae; , Tincinae. III [Howes (1999) hypothesis]: , Acheilognathinae; , Leuciscinae; , Cultrinae; , Alburninae; , incertae sedis; , Rasborinae; , Rasborinae or Leuciscinae; , Gobioninae; , Cyprininae. IV [Cavender & Coburn (1992) hypothesis]: , leuciscins; , phoxins; , cultrins; , xenocyprins; , acheilognathins; , gobionins; , rasborins; , tincins; , cyprinins; , barbinins; , labeonins. 938 .   . leuciscins (including Notemigonus)+North American phoxinins [including Phoxinus phoxinus (L.)] (81% and 56% bootstrap in NJ and MP, respectively); (2) European gobionins+Pseudorasbora; (3) primarily Asian groups [cultrins+ acheilognathins+gobionins (excluding Abbotina)+xenocyprinins]; (4) Abbottina+ Sinocyclocheilus+Acrossocheilus and (5) cyprinins [apart from Sinocyclocheilus, and Acrossocheilus]+barbins+labeonins. It was not possible to determine the phylogenetic position of two taxa (European Rhodeus and Tinca) and relation- ships among these five lineages as topologies were not well-resolved at this level. Placement of some taxa in groups or subfamilies was similar among the three morphological studies, however, there are some differences (Fig. 1). Morpho- logical classifications proposed by Chen et al. (1984) and Cavender & Coburn (1992) are generally in agreement (Fig. 1). Excluding the Leuciscinae and placement of Tincinae, the groups considered by the latter correspond to the subfamilies of Chen et al. (1984).InHowes (1991), the subfamilies Cyprininae and Leuciscinae included the labeonins, barbins and cyprinins and xenocyprinins, respectively, of Cavender & Coburn (1992). Cavender & Coburn (1992) included the subfamily Alburninae of Howes (1991) in their leuciscinin group. When cyt b sequence data were constrained to fit the hypothesis of Chen et al. (1984) (Fig. 1), the MP search recovered 10 CON trees that were 402 steps longer and significantly different from the MP trees (Table II). When the analysis was constrained to maintain the two lineages proposed by Chen et al. (1984), Leuciscini and Barbini, the topologies recovered were significantly longer than those recovered in the MP search from the data. To assess the impact of individual groups on this result, monophyly of each of the subfamilies was examined. It was not possible to test Danioninae, Tincinae and Labeoninae because each group was only represented by one taxon. When the Acheilognathinae, Cultrinae and Leuciscinae were constrained, a heuristic search retained four, four, and 60 CON trees that were 47, 31 and 50 steps longer, respectively, than the MP trees; however, these CON trees were not significantly worse than the MP trees. Analysis of monophyly of the Barbinae, Cyprininae, Gobioninae and Xenocyprinae under imposed constraints yielded eight, four, eight and four CON trees that were 44, 80, 148 and 117 steps longer and significantly different from the MP tree (Table II). Results obtained with equally weighted characters were similar to those above, however, CON trees for the Barbinae were not significantly longer than the MP tree. Therefore, rejection of the Chen et al. (1984) hypothesis may result from misalignment of taxa to the Barbinae, Cyprininae, Gobioninae and Xenocyprinae. Howes (1991) did not provide a resolved hypothesis of subfamily relationships but did assign subfamilies to two major lineages. When subfamilies were constrained to fit this hypothesis, 10 CON trees were recovered that were 402 steps, significantly longer than the MP tree. When each of the subfamilies were constrained, topologies recovered for the Acheilognathinae and Cultrinae were 47 and 13 steps longer, respectively, than the MP trees; however, these differences were not significant. Contrasts examining monophyly of the Alburninae, Cyprininae, Gobioninae, Leuciscininae (with and without Xenocyprioides) and Rasborininae (with and without Xenocyprioides) subfamilies were significantly longer than the MP trees (Table II). In analyses where characters were weighted     939

T II. Results of Kishino & Hasegawa (1989) tests of monophyly. Results obtained by Templeton’s (1983) method (not shown) were the same as provided by Kishino–Hasegawa tests

No. Significantly Length .. best Length t worse than diff. (diff.) trees best tree?*

Chen et al. (1984) hypothesis Chen et al. hypothesis 10 10 740 402 48·84–51·79 8·23–7·76 Yes Two major lineages 8 10 481 80 25·79–28·38 3·10–2·82 Yes (leuciscini-Barbini) Acheilognthinae 4 10 385 47 24·49–29·31 1·92–1·79 No Barbininae 8 10 382 44 17·23–20·62 2·55–2·13 Yes Cultrinae 4 10 369 31 20·59–22·54 1·51–1·38 No Cyprininae 4 10 418 80 26·25–28·06 3·05–2·85 Yes Gobioninae 8 10 486 148 32·36–35·91 4·03–4·57 Yes Leuciscinae 60 10 388 50 30·61–32·94 1·63–1·52 No Xenocyprininae 4 10 455 117 23·61–26·72 4·96–4·38 Yes Howes (1991) hypothesis Howes’ hypothesis (including 15 1 156 84 83·66–84·27 13·82–13·72 Yes Xenocyprioides in Leuciscinae) Howes’ hypothesis (including 1 11 479 1330 90·77–90·87 14·65–14·63 Yes Xenocyprioides in Rasborinae) Two major lineages 8 10 277 128 26·53–27·27 4·83–4·69 Yes Acheilognathinae 4 10 385 47 24·49–26·31 1·92–1·79 No Alburninae 8 10 514 176 30·06–31·78 5·85–5·54 Yes Cultrinae 8 10 351 13 14·60–17·59 0·89–0·74 No Cyprininae 10 10 413 75 28·91–31·66 2·59–2·37 Yes Gobioninae Yes Leuciscinae (excluding 5 10 629 291 38·07–39·22 7·64–7·42 Yes Xenocyprioides) Leuciscinae (including 1 10 669 331 37·70–39·79 8·78–8·32 Yes Xencyprioides) Rasborinae (excluding 8 10 442 104 28·47–32·02 3·65–3·25 Yes Xenocyprioides) Rasborinae (including 2 10 506 168 41·48–43·09 4·05–3·90 Yes Xenocyprioides) Cavender & Coburn (1992) hypothesis Cavender & Coburn hypothesis 9 10 802 464 49·29–52·42 9·41–8·85 Yes The two subfamilies 8 10 391 53 27·48–30·91 1·93–1·71 No Cyprininae 4 10 390 52 28·43–30·27 1·83–1·72 No Leuciscinae 4 10 413 75 26·49–28·24 2·83–2·66 Yes Acheilognathins 4 10 385 47 24·49–26·31 1·92–1·79 No Barbins 8 10 354 16 21·17–24·30 0·76–0·66 No Cultrins 4 10 369 31 20·49–22·32 1·51–1·39 No Cyprinins 8 10 407 69 28·06–30·49 2·46–2·26 Yes Gobionins 8 10 503 165 27·49–30·95 6·00–5·33 Yes Leuciscins 20 10 338 0 12·50–0·00 0·00–0·00 No Phoxinins 4 10 357 19 29·60–27·93 0·64–0·68 No Rasborins 8 10 442 104 28·47–32·02 3·65–3·25 Yes Xenocyprins 4 10 433 95 22·55–26·92 4·21–3·53 Yes

*Significantly worse than best tree for P<0·05. 940 .   . equally, Rasborinae (excluding Xenocyprioides) and Cyprininae were not sig- nificantly worst than MP trees but topologies for the remaining subfamilies were still significantly longer. When data were constrained to fit the hypothesis of Cavender & Coburn (1992), the MP search recovered nine CON trees that were 464 steps longer and significantly different from the MP trees (Table II). In addition, constraining topologies to enforce monophyly of Leuciscinae and Cyprininae yielded eight CON trees that were 53 steps longer than, but not significantly different from the MP trees. When Leuciscinae and Cyprininae were independently constrained, topologies obtained were 75 and 52 steps longer, respectively than MP tree, with only the leuciscin CON trees significantly different from the MP tree (Table II). Groups considered by Cavender & Coburn (1992) were also evaluated, but the tincins and labeonins could not be examined as they were represented by only one taxon. When the acheilognathins, barbins, cultrins, leuciscins and phoxinins were constrained, topologies recovered were 47, 16, 31, zero and 19 steps longer (respectively) than the MP trees; however, these differences were not significant. When the cyprinins, gobionins, rasborins and xenocyprins were constrained, analysis yielded topologies that were significantly longer (69, 165, 104 and 95 steps, respectively) than those recovered in the MP search (Table II). When characters were weighted equally, constraint analysis identified only two groups, gobionins and xenocyprinins, which were significantly longer than those recovered in the MP search.

DISCUSSION One of the five lineages stabilized in the tree was the European and North American Leuciscinae (including Alburninae sensu Howes, 1991) which formed a monophyletic lineage in both analyses. North American cyprinids were not monophyletic, as the North American genus Notemigonus was allied with Old World Leuciscinae (Briolay et al., 1998). Other North American Leuciscinae (Siphateles, , Acrocheilus and Orthodon) also seemed to be more closely related to European Leuciscinae+Alburninae than to other American represent- atives. Phoxinus phoxinus, the only Eurasian and American cyprinid in the phoxinin lineage (Cavender, 1991), was placed as the sister group to a monophyletic North American+European Leuciscinae assemblage as reported by Briolay et al. (1998), Zardoya & Doadrio (1998) and Ha¨nfling & Brandl (2000). Low bootstrap support, however, makes paraphyly of phoxinins and placement of Phoxinus tenuous. Gobionins were found in three different lineages: (1) European Gobio clustered with the rasborin Pseudorasbora (66% and 63% in NJ and MP, respectively), (2) Asian Gobiobotia clustered with other representatives of Asian Xenocyprinae (Distoechodon, Xenocypris, Pseudobrama), Acheilognathinae (Acheilognathus, Rhodeus) and the rasborin Raimas (<50% and 53% bootstrap support in NJ and MP, respectively), and (3) Abbotina was associated with Sinocyclocheilus and Acrossocheilus (72% and <50% in NJ and MP bootstrap analyses, respectively). Xenocyprinae was not monophyletic when the genus Xenocyprioides was assigned to this subfamily, consistent with the results of Xiao et al. (2001) and their request for revaluation of this genus. The present results suggest that this     941 revaluation must be extended to other genera of this subfamily, such as Xenocypris. This was not in agreement with Xiao et al. (2001); however, they did not include the necessary taxa to reach that conclusion (e.g. Gobiobotia). The present study does not support the monophyly of Acheilognathinae. The European acheilognathan Rhodeus sericeus (Pallas) was not closely related to Asian representatives of the group but was affiliated with European Leuciscinae as in Gilles et al. (2001). These results contradict Okazaki et al. (2001), which supported monophyly of Rhodeus; however, bootstrap support for monophyly was low and they only examined a limited number of outgroups. To understand the relationships of the Acheilognathinae with the other subfamilies, more basal taxa of this subfamily should be included, such as Tanakia (Okazaki et al., 2001). Despite belonging to different subfamilies, most Asian taxa (except for Abbotina and members of the Cyprininae) formed a distinct lineage (84% and <50% in NJ and MP bootstrap analyses, respectively). Although the cultrin genera Megalobrama, Cultrichthys and Parabramis formed a single lineage, this sub- family was rendered paraphyletic by the close relationship of Xenocyprioides carinatus Chen & Huang (a member of Leuciscinae) with Megalobrama.In general, Asian specimens were more closely related to each other than to other specimens of their genera or subfamilies from Europe or North America, indicating the need for further study and taxonomic revision of many Asian cyprinid groups. The taxonomic position of Tinca tinca (L.) has been unclear, as it has been placed in the Cyprininae (Chen et al., 1984), Leuciscinae (Cavender & Coburn, 1992), or in its own subfamily Tincinae (Bogustkaya, 1986). It was not possible to clearly affiliate this group to any other; therefore, the present results are consistent with recommendations of Arai (1982), Howes (1991) and Ha¨nfling & Brandl (2000) that this species should be considered as incertae sedis. Gilles et al. (2001), however, recently suggested that Tinca is the sister group of the all non-rasborines and non-cyprinines, based on mitochondrial DNA control region and cyt b and 16S rRNA. Their study did not include as many taxa or the Asian rasborins, therefore, continued examination of this hypothesis is necessary. Topologies recovered from cyt b sequences were generally inconsistent with those based upon morphological criteria. The barbelled cyprinines (Cyprininae, Gobioninae and Rasborinae) and barbel-lacking leuciscines (Leuciscinae, Alburninae, Cultrinae and Acheilognathinae) sensu Howes (1991) did not form distinct groups. Therefore, presence or absence of barbels as suggested by Howes (1991) is not a useful character for defining cyprinid relationships. The two major lineages defined by Chen et al. (1984) and Howes (1991) (Fig. 1) were not supported by the present data. The morphological classification that was most consistent with cyt b sequences was provided by Cavender & Coburn (1992) as the present analysis supported monophyly of their subfamilies Leuciscinae and Cyprininae. While Cavender & Coburn’s (1992) major lineages are consistent with cyt b data, several of the groups proposed by them as well as Chen et al. (1984) and Howes (1991) (Fig. 1) were not supported (Table II). Failure to support monophyly for many of these groups seemed to be related to misalignment or absence of taxa, particularly involving gobionins and xenocyprins. Rasborin groupings were not supported as representative genera (e.g. Pseudorasbora, 942 .   .

Raiamas, perhaps Xenocyprioides) and were scattered throughout the tree. Failure to support monophyly for the gobionins was due to the distinctiveness of European Gobio relative to the Asian taxa Gobiobotia and Abbottina. Likewise, monophyly of the xenocyprinins was not supported as its members were scattered among several different lineages (e.g. Xenocypris fangi Tchang and Xenocyprioides parvulus Chen). Given the impact of these Asian fishes appeared to play in the conflict among molecular and morphological results, re-evaluation of these taxa is necessary. Resolution of groups in this paper provide a good start towards understanding cyprinid relationships; however, this study does not include any of the diversity present on the African subcontinent and ‘undersamples’ Asian taxa. Therefore, further understanding of relationships for the entire family will require further comparative morphological and molecular studies on Asian cyprinids and taxa representative of the diversity found in Africa. We are grateful to T. Cavender, M. Coburn and M. Kottelat for the help with the subfamilies and groups species assignment. This study was supported by Centro de Biologia Ambiental and by grants from Fundac¸a˜o da Cieˆncia e Tecnologia to M. M. Coelho (Praxis XXI 2/2.1/BIA/149/94) and to N. Mesquita (Praxis XXI BD/19751/99) and NSF support to TED (DEB-9220683).

References

Alves, M. J., Coelho, M. M., Collares-Pereira, M. J. & Dowling, T. E. (1997). Maternal ancestry of the Rutilus alburnoides complex (Teleostei,Cyprinidae) as determined by analysis of cytochrome b sequences. Evolution 51, 1584–1592. Arai, R. (1982). A chromosome study on two cyprinid fishes Acrossocheilus labiatus and pumila, with notes on Eurasian cyprinids and their karyotypes. Bulletin of the National Science Museum, Tokyo, Series A 8, 131–152. Banarescu, P. & Coad, B. W. (1991). Cyprinids of Eurasia. In Cyprinid . Systematics, Biology and Exploitation (Winfield, I. J. & Nelson, J. S., eds), pp. 127–155. London: Chapman & Hall. Bogustskaya, N. G. (1986). To the position of the Tinca tinca (L.) in the system of cyprinid fishes (Cyprinidae). Proceedings of the Zoological Institute of Leningrad 154, 49–65. Briolay, J., Galtier, N., Brito, R. M. & Bouvey, Y. (1998). Molecular phylogeny of Cyprinidae inferred from cytochrome b DNA sequences. Molecular Phylogenetics and Evolution 9, 100–1408. Brito, R. M., Briolay, J., Galtier, N., Bouvet, Y. & Coelho, M. M. (1997). Phylogenetic Relationships within Genus Leuciscus (Pisces, Cyprinidae) in Portuguese Fresh Waters, Based on Mitochondrial DNA Cytochrome b Sequences. Molecular Phylogenetics and Evolution 8, 435–442. Broughton, R. E., Stanley, S. E. & Durrett, R. T. (2000). Quantification of homoplasy for nucleotide transitions and transversions and a reexamination of assumptions of weighted parsimony analysis. Systematic Biology 49, 617–627. Cantatore, P., Roberti, M., Pesole, G., Ludovico, A., Milella, F., Gadaleta, M. N. & Saccone, C. (1994). Evolutionary analysis of cytochrome b sequences in some perciformes: Evidence for a shower rate of evolution than in mammals. Journal of Molecular Evolution 39, 589–597. Cavender, T. M. (1991). The fossil record of Cyprinidae. In Cyprinid Fishes. System- atics, Biology and Exploitation (Winfield, I. J. & Nelson, J. S., eds), pp. 127–155. London: Chapman & Hall. Cavender, T. & Coburn, M. (1992). Phylogenetic relationships of North American Cyprinidae. In Systematics, Historical Ecology, and North American Freshwater Fishes (Mayden, R., ed.), pp. 293–327. Stanford, CA: Stanford University Press.     943

Chen, X. L., Yue, P. Q. & Lin, R. D. (1984). Major groups within the family Cyprinidae and their phylogenetic relationships. Acta Zootaxonomica Sinica 9, 424–498. Dowling, T. E. & Naylor, G. J. P. (1997). Evolutionary relationships of in the genus Luxilus (Teleostei: Cyprinidae) as determined from cytochrome b sequences. Copeia 4, 758–765. Felsentein, J. (1985). Confidence limits on phylogenies: an approach using the bootstrap. Evolution 39, 783–791. Fitch, W. M. (1971). Toward defining the course of evolution: Minimum change for a specific tree topology. Systematic Zoology 20, 406–416. Gilles, A., Lecointre, G., Faure, E., Chappaz, R. & Brun, G. (1998). Mitochondrial phylogeny of the European Cyprinids: implications for their systematics, reticulate evolution, and colonization time. Molecular Phylogenetics and Evolution 10, 132–143. Gilles, A., Lecointre, G., Miquelis, A., Loerstcher, M., Chappaz, R. & Brun, G. (2001). Partial combination applied to phylogeny of European cyprinids using the mitochondrial control region. Molecular Phylogenetics Evolution 19, 22–33. Ha¨nfling, B. & Brandl, R. (2000). Phylogenetics of European cyprinids: insights from allozymes. Journal of Fish Biology 57, 265–276. doi:10.1006/jfbi.2000.1304. Howes, G. J. (1991). Systematics and biogeography: an overview. In Cyprinid Fishes. Systematics, Biology and Exploitation (Winfield, I. J. & Nelson, J. S., eds), pp. 1–33. London: Chapman & Hall. Jukes, T. H. & Cantor, C. R. (1969). Evolution of protein molecules. In Mammalian Protein Metabolism (Munro, H. N., eds), pp. 21–132. New York: Academic Press. Kishino, H. & Hasegawa, M. (1989). Evaluation of the maximum likelihood estimate of the evolutionary tree topologies from DNA sequence data, and the branching order in Hominoidea. Journal of Molecular Evolution 29, 170–179. Machordom, A. & Doadrio, I. (2001). Evidence of a Cenozoic Betic-Kabilian connection based on freshwater fish phylogeography (, Cyprinidae). Molecular Phylogenetics and Evolution 18, 252–263. Maddison, W. P. & Maddison, D. R. (1992). MacClade: Analysis of Phylogeny and Character Evolution. Sunderland, MA: Sinauer Associates. Nei, M. & Kumar, S. (2000). Molecular Evolution and Phylogenetics. New York: Oxford University Press. Okazaki, M., Naruse, K., Shima, A. & Arai, R. (2001). Phylogenetic relationships of bitterlings based on mitochondrial 12S ribosomal DNA sequences. Journal of Fish Biology 58, 89–106. doi:10.1006/jfbi.2000.1430. Rainboth, W. J. (1991). Cyprinids of South East Asia. In Cyprinid Fishes. Systematics, biology and exploitation (Winfield, I. J. & Nelson, J. S., eds), pp. 156–210. London: Chapman & Hall. Saiton, N. & Nei, M. (1987) The neighbor-joining method: a new method for reconstructing phylogenetic trees. Molecular Biology and Evolution 4, 406–425. Sambrook, J., Fritsch, E. F. & Maniatis, T. (1989). Molecular Cloning: a Laboratory Manual. New York: Cold Spring Harbor Laboratory Press. Schmidt, T. R. & Gold, J. R. (1993). Complete sequences of the mitochondrial b gene in the cherryfin shiner Lythrurus roseipinnis (Teleostei: Cyprinidae). Copeia 1993, 880–883. Swofford, D. L. (1998). PAUP*: Phylogenetic Analysis Using Parsimony (*and Other Methods), v. 4.0. Sunderland, MA: Sinauer Associates. Takahashi, K. & Nei, M. (2000). Efficiencies of fast algorithms of phylogenetic inference under the criteria of maximum parsimony, minimum evolution, and maximum likelihood when a large number of sequences are used. Molecular Biology and Evolution 17, 1251–1258. Templeton, A. R. (1983). Phylogenetic inference from restriction endonuclease cleavage site maps with particular reference to the evolution of humans and the apes. Evolution 37, 221–244. Tsigenopoulos, C. S. & Berrebi, P. (2000). Molecular phylogeny of North Mediterranean freshwater barbs (genus Barbus: Cyprinidae) inferred from cytochrome b 944 .   .

sequences: biogeographic and systematic implications. Molecular Phylogenetics and Evolution 14, 165–179. Xiao, W., Zhang, Y. & Liu, H. (2001). Molecular systematics of Xenocyprinae (Teleostei: Cyprinidae): , biogeography, and coevolution of a special group restricted in East Asia. Molecular Phylogenetics and Evolution 18, 163–173. Zardoya, R. & Doadrio, I. (1998). Phylogenetic relationships of Iberian Cyprinidae: Systematic and biogeographical implications. Proceedings of the Royal Society of London, Series B 265, 1365–1372. Zardoya, R. & Doadrio, I. (1999). Phylogenetic relationships of Greek Cyprinidae: molecular evidence for at least two independent origins of the Greek cyprinid fauna. Molecular Phylogenetics and Evolution 13, 122–131.