Phylogenetic Relationships of the Western North American Phoxinins (Actinopterygii: Cyprinidae) As Inferred from Mitochondrial 1

Home , Algansea, Blue chub, Gila (fish), Lavinia (fish), Phenacobius, Pogonichthys

MOLECULAR PHYLOGENETICS AND EVOLUTION Vol. 9, No. 2, April, pp. 308–329, 1998 ARTICLE NO. FY970467

Phylogenetic Relationships of the Western North American Phoxinins (Actinopterygii: Cyprinidae) as Inferred from Mitochondrial 12S and 16S Ribosomal RNA Sequences Andrew M. Simons1 and Richard L. Mayden *Department of Biological Sciences, The University of Alabama, Box 870344, Tuscaloosa, Alabama 35487

Received June 6, 1997

able in appearance, biology, and distributions, ranging Parsimony analysis of the mitochondrial 12S and 16S from the small, herbivorous Eremichthys with a maxi- rRNA sequences of North American phoxinin taxa mum length of 65 mm to the large, piscivorous Ptycho- indicated the existence of three major clades, the cheilus that can reach lengths of up to 1.8 m. Members Western Clade, the Creek Chub Clade, and the Open of this fauna have previously been hypothesized to be Posterior Myodome Clade. The monophyletic Western related to both Asian (Howes, 1984; Miller, 1959, 1965) Clade identified in this paper contained fewer taxa and eastern North American taxa (Bailey, 1956; Uyeno, than that identified by previous authors. This clade 1961). However, few studies on relationships of the contained species restricted to drainages west of the Continental Divide, such as Gila, Acrocheilus, Relic- western fauna have included appropriate taxa neces- tus, Eremichthys, Siphateles, Ptychocheilus, Lavinia, sary to assess their interrelationships to other cy- and Orthodon, and the wide-ranging genus Phoxinus prinids. The traditional view of a uniquely derived found in eastern North America and Eurasia. Within this western fauna is complicated by the presence of wide- Western Clade Phoxinus was the basal sister group. Gila spread taxa such as Rhinichthys and the presumed was recognized as a monophyletic group exclusive of relationships between western genera, such as Richard- Siphateles and Snyderichthys, clades traditionally recog- sonius, and eastern genera, such as Clinostomus. The only nized as subgenera of Gila. The genus Gila was most explicitly phylogenetic study that clearly identified a clade closely related to the genus Acrocheilus and together of western cyprinids was that by Coburn and Cavender these formed the sister group to the genus Relictus. These (1992); however, based on their analysis, the western relationships were supported independent of weighting cyprinid fauna does not constitute a monophyletic group. schemes used in analyses. Identification of Phoxinus as Much of the western cyprinid fauna was described in the the basal sister taxon of the Western clade implies that late 1800s and early 1900s particularly by Baird, Girard, other major clades of North American Phoxinins likely Snyder, Jordan, and Evermann. Their work was an invalu- haveAsian or European relatives. ௠ 1998 Academic Press able first step in understanding this fauna. Unfortunately, in the process of describing this fauna they generated a confusing plethora of names, particularly with taxa related INTRODUCTION to Gila. Some taxa were even described a number of times, even within the same publication (Girard, 1856). In 1945, The cyprinids of western North America have long Miller revised the genera Gila, Siboma, Tigoma, Cheonda, presented an interesting taxonomic problem. The west- and Richardsonius. Of these genera, Miller recognized only ern cyprinid fauna has traditionally been identified Richardsonius and Gila. Cheonda, Siboma, and Tigoma largely on geographic criteria, containing taxa native to were subsumed under Gila. Within Gila, Miller recognized drainages west of the Continental Divide. The fauna is the subgenera Gila, Siboma, and Temeculina and erected viewed as distinctive due to the high number of en- the subgenus Klamathella to contain Gila bicolor (Girard). demic genera in the region, roughly 50% of which are The diagnoses of these subgenera were based on pharyn- monotypic (Miller, 1959). These taxa are highly vari- geal tooth counts, scale radii, and point of origin of the dorsal fin relative to that of the pelvic fins. Bailey (1956) 1 To whom correspondence and reprint requests should be ad- subsumed the genera Richardsonius and Clinostomus into dressed at current address: The University of Massachusetts, Depart- the genus Gila based on their general appearance, ecology, ment of Biology, Graduate Program in Organismic and Evolutionary and distribution. Bailey’s classification was not widely Biology, Morrill Science Center, Box 35810, Amherst, MA accepted, and Uyeno (1961), in his study of the osteology Sequence data from this article have been deposited with the EMBL/GenBank Data Libraries under Accession Nos. AF 023183, AF and phylogeny of Gila and relatives, resurrected Richard- 023186, AF 023188, AF 023192, AF 023194, AF 023195, AF 023197, sonius (including Clinostomus) from synonymy with Gila. AF 023198, AF 023201, AF038468–AF038498. Uyeno combined the genera Siphateles and Snyderichthys

308 1055-7903/98 $25.00 Copyright ௠ 1998 by Academic Press All rights of reproduction in any form reserved. PHYLOGENETIC RELATIONSHIPS OF PHOXININS 309

FIG. 2. Hypothesized phylogenetic relationships of the Western FIG. 1. Hypothesized relationships of the genera Gila, Richardso- Clade based on osteological and lepidological characters, from Co- nius, and Ptychocheilus from Uyeno (1961). burn and Cavender (1992). with Gila, an action that caused nomenclatural difficulties, large genetic difference between Notemigonus and the as the name bicolor was then used for the blue chub, Gila remainder of the included taxa, consistent with the bicolor, and the tui chub, Siphateles bicolor. Bailey and view that Notemigonus is a leuciscin, not closely related Uyeno (1964) resolved this by selecting Algansea bicolor as to the remainder of the North American fauna, which the name to retain in Gila for the tui chub and designating are phoxinins (Cavender and Coburn, 1992). Of the Gila coerulea as the oldest available name for the blue native California cyprinids included in Avise and Aya- chub. Uyeno (1961) recognized three subgenera within la’s (1976) analysis, two groups can be identified, one Gila, Gila, Siphateles, and Snyderichthys, and presented a containing Richardsonius and Pogonichthys and the other containing Ptychocheilus, Mylopharodon, Lavinia, phylogeny (Fig. 1) wherein Gila and Siphateles were 3 depicted as sister taxa, sister to Snyderichthys. The genus and Hesperoleucas. Within the second group, S. bicolor Gila was identified as the sister group to Richardsonius clusters with Orthodon. This cluster is sister to Hespero- (including Clinostomus), and this clade was sister to Ptycho- leucas plus Lavinia and Mylopharodon plus Ptychochei- cheilus. Although the subgenus Temeculina has been recog- lus. Indeed, the genetic distances between Mylophar- nized by some (Barbour and Miller, 1978) this taxonomy odon and Ptychocheilus and between Hesperoleucas and renders the subgenus Gila paraphyletic based on Uyeno’s Lavinia were so small that Avise and Ayala (1976) sug- (1961) tree. gested that they should be considered congeners. Avise and Ayala (1976) examined genetic variability Coburn and Cavender (1992) performed a phyloge- at 24 loci in eight species of California cyprinids and netic analysis on almost all North American cyprinid Notemigonus crysoleucas.2 Their analysis revealed a taxa using lepidological and osteological characters. They proposed a monophyletic western clade containing Gila and relatives (Fig. 2). This clade was largely 2 They utilized three algorithms to determine relationships among composed of taxa restricted to the Colorado system, the these taxa: (1) UPGMA, (2) WPGMA, and (3) Modified Wagner. Of these methods, the modified Wagner analysis requires fewer assump- tions regarding evolutionary rates and therefore will be the only 3 We refer G. bicolor to the genus Siphateles as our results indicate analysis discussed herein. that this taxon is not closely related to Gila. 310 SIMONS AND MAYDEN

Great Basin, and Pacific drainages. Their analysis Specimens Examined indicated that some taxa previously considered rela- The specimens examined included the following: tives of Gila, such as Clinostomus and Richardsonius Abramis bjoerkna (Brunnsviken Bay, Stockholm, Upp- (Bailey, 1956; Uyeno, 1961), were more closely related land, Sweden, UAIC 85533.03); Acrocheilus alutaceus to other major clades of North American cyprinids. (Kettle River, British Columbia, Canada, UAIC Coburn and Cavender were unable to resolve relation- 11365.01); Couesius plumbeus Mill Creek, 100 miles ships among many western taxa; only their Gila clade north of Ft. Nelson, British Columbia, Canada, UAIC is resolved (Fig. 2). Based on their analysis the genus 11366.01); Eremichthys acros (Soldier Meadows, Hum- Gila, as recognized by Uyeno (1961), is polyphyletic. Of boldt Co., NV); Gila atraria (Sevier River, Piute Co., the taxa that Coburn and Cavender included in the UT); Gila coerulea (Upper Klamath Lake, Klamath Co., western clade, we consider the following to be members OR, OS015083 (two specimens)); Gila conspersa (Rio of the open posterior myodome (OPM) clade: Mylocheilus, San Isabel at General Trias, Chihuahua, Mexico, UAIC Pogonichthys, Rhinichthys, Tiaroga, and Agosia (Simons, 7910.01); Gila cypha (Colorado River, Coconino Co., 1996). We refer to the Plagopterins plus the Creek Chubs AZ); Gila ditaenia (Sycamore Canyon, Santa Cruz Co., as the Creek Chub Clade (Simons and Mayden, 1997). AZ); Gila nigrescens (Mimbres River, NM); Gila orcutti In a study based on sequences of 12S rRNA, Simons (CA, UAIC 11562.01); Gila pandora (Rio Chama, NM); and Mayden (1997) removed the Plagopterins and Gila robusta (Aravaipa Creek, Graham Co., AZ); Lavinia Snyderichthys copei from the Western Clade, resolving exilicauda (Putah Creek, Yolo Co., CA, OS015074 (two them as sister to the Creek Chubs, a clade containing specimens)); Leuciscus rutilis (Brunnsviken Bay, Stock- Semotilus and its relatives. Furthermore, our data and holm, Uppland, Sweden, UAIC 8533.04); Margariscus phylogenetic analyses align the genus Rhinichthys as a margarita (Plover River, Marathon Co., WI, UAIC member of the OPM Clade. The analyses described 10241.07); Notropis atherinoides (Waupaca River, herein resolved a clade of western cyprinids referred to Waupaca Co., WI, UAIC 10485.06); Notemigonus cryso- as the Western Clade. This group included the genera leucas (Mill Creek, Portage Co., WI, UAIC 10527.02); Acrocheilus, Eremichthys, Gila, Lavinia, Orthodon, Orthodon microlepidotus (Putah Creek, Yolo Co., CA, Phoxinus, Ptychocheilus, Relictus, and Siphateles. Other OS015075 (two specimens)); Phoxinus erythrogaster taxa that may be part of this clade but have not yet (Spring River, Lawrence Co., MO, UAIC 11560.01); been examined by us include Yuriria, Algansea, Iotich- Phoxinus neogaeus (Orange River, ME); Phenacobius thys, Moapa, and Mylopharodon. Our analyses exam- mirabilis (Cottonwood River, Chase Co., KS, UAIC ined the monophyly and relationships of the Western 11564.01); Platygobio gracilis (Mississippi River, Jack- Clade with respect to other major groups of North son Co., IL, UAIC 11558.01); Ptychocheilus oregonensis American cyprinids utilizing the complete sequence of (Errock Lake, Errock, British Columbia, Canada, UAIC the 12S and 16S ribosomal RNA genes plus the interven- 11561.01); Relictus solitarius (Phalan Spring, Butte Valley, ing Valine tRNA gene. Elko Co., NV); Rhinichthys atratulus (Russell Creek, Clair- borne Co., TN, UAIC 9850.01); Siphateles bicolor (Pyramid MATERIALS AND METHODS Lake Indian Reservation, Washoe Co., NV); and Sny- derichthys copei (Sevier River, Garfield Co., UT). We have sequenced the entire 12S and 16S mitochondrial rRNA plus intervening Valine tRNA in 31 speci- DNA Amplifications and Sequencing mens of cyprinids representing the three major clades Standard phenol/chloroform procedures (Hillis et al., of North American phoxinins plus outgroup taxa. Taxa 1990) and the QIAGEN QIamp tissue kit (Catalogue were included in the analysis to represent the three major No. 29304) were used to extract genomic DNA. Seven clades proposed by Coburn and Cavender (1992) and the taxa, G. atraria, G. nigrescens, G. pandora, G. ditaenia, groups recognized in our previous study of North American G. cypha, G. robusta, and S. copei, were obtained as cyprinid relationships (Simons and Mayden, 1997). pure mtDNA preps from Dr. T. Dowling. Genes were Abramis, Leuciscus, and Notemigonus were included amplified in three overlapping fragments with Tfl DNA as outgroup taxa. The Western Clade was represented polymerase (Epicentre Products Catalogue No. F10250) by Acrocheilus, Eremichthys, Gila, Lavinia, Orthodon, under conditions recommended by the manufacturer. Phoxinus, Ptychocheilus, and Relictus. The Creek Chub Amplification primers used were PHEa and 12Sb for clade was represented by Couesius, Margariscus, and the first fragment, 12Sa and 16Sd for the second Snyderichthys. The OPM clade was represented by fragment, and 16Si and LEU25 for the third fragment Notropis, Phenacobius, Platygobio, and Rhinichthys. (Table 1). Initial amplification was from genomic DNA Coburn and Cavender’s (1992) concept of the western in a volume of 25 µl. The amplification profile included clade was represented by Acrocheilus, Eremichthys, denaturation at 94°C for 1 min, annealing at 50°C for 1 Gila, Lavinia, Orthodon, Ptychocheilus, Relictus, and Rhin- min, and extension at 72°C for 1 min 30 s for 30 cycles. ichthys. Their chub clade was represented by Couesius, Amplification products were gel purified on 2.5% Margariscus, Phoxinus, Phenacobius, and Platygobio. Their NuSieve agarose. Single-stranded DNA amplification shiner clade was represented by Notropis. (Gyllensten and Erlich, 1988) was performed in a PHYLOGENETIC RELATIONSHIPS OF PHOXININS 311

TABLE 1

Primers Used in the Ampliﬁcation and Sequencing of the 12S and 16S Genes for North American Cyprinids

Name Sequence (5Ј = 3Ј) Strand Reference

12Sa AAACTGGGATTAGATACCCCACTA L Kocher et al. (1989) 12Sb AGGAGGGTGACGGGCGGTGTGT H modiﬁed from Kocher et al. (1989) 12Sc GGAAAGAAATGGGCTACA L Titus (pers. comm.) 12Sd GGGTTGGTAAATCTCGTGC L Titus and Larson (1995) 16Sb TTACCCTAGGGATAACAGCG L This study 16Sc-L GACGAGAAGACCCTTTGGAGC L This study 16Sd CAAGAGGCGATGTTTTT H Titus (pers. comm.) 16Sd-L AAAAACATCGCCTCCTG L Simons and Mayden (1997) 16Se GGATCAGGACATCCTAATGGTGC L Titus (pers. comm.) 16Sh GCATAATAATCTAGCCAG L Titus and Larson (1996) 16Si GAAGTGGATAGAAGTTCAGCCT L This study 16S700 ACAGAAGGAAGTTTATTAT L This study LEU25 AGGGCTTAGGCCTTTCGCAA H This study PHEa AAAGCACAGCACTGAAGATG L Titus and Frost (1996) VALa TTAAGCATCTCACTTACACC L Modiﬁed from Titus (1992)

volume of 50 µl. The amplification profile was identical Phylogenetic Structure to double-stranded DNA amplification but was run for The g-statistic (Hillis, 1991) was calculated based on 35 cycles. Primer concentrations were generally used at 10,000 randomly generated trees using sequence data a ratio of 10:1, although concentrations were varied for for the 12S, Valine, 16S, and ALL data. The g-statistic some templates. Asymmetric amplifications were puri- was generated for all taxa and for a reduced data set fied in Millipore filter units (Millipore Ultrafree MC containing only one species from each genus to exclude polysulfone 30,000 MNWL, Catalogue No. UFC3TTK00) the possibility that phylogenetic structure was due to and resuspended in 30 µl dH2O. Sequenase 2.0 (United congeneric taxa (Titus and Larson, 1995). States Biochemical) was used in sequencing reactions; manufacturer’s instructions were followed. Sequencing Phylogenetic Analysis primers used were PHEa, 12Sb, 12Sa, 12Sc, 12Sd, Phylogenetic trees were generated using PAUP 3.1.1 VALa, 16Sh, 16Si, 16S700, 16Sd, 16Sc, 16Sb, and 16Se (Swofford, 1993). The data were analyzed in four differ- (Table 1). Termination products were electrophoresed ent sets: 12S, 12SϩVal, 16S, and ALL data. Within each on 8% Long Ranger gels (AT Biochem). set, gaps were coded as missing or as binary characters. Analyses Transversions were weighted with respect to transitions: 1:1, 2:1, or 3:1. Bootstrap analyses with 1000 Multiple alignment of sequences was performed us- replicates were performed on ALL data using each of ing CLUSTAL W (Thompson et al., 1994) with a gap the three weighting schemes. Analyses are referred to opening penalty of 10 and a gap extension penalty of 5. by data set and weighting scheme. For example, an The resulting alignment was then adjusted based on analysis of the complete data set (ALL) with transver- secondary structure models for Cyprinus carpio (Van de sions weighted three times with respect to transitions Peer et al., 1994; De Rijk et al., 1994). Insertion or is referred to as ALL, 3:1. deletion events that were not autapomorphic were coded as binary characters. This binary data matrix RESULTS was appended to the sequence matrix and included in some analyses. The numbers of variable and poten- Sequence data were complete for all taxa with two tially phylogenetically informative sites were calcu- exceptions. P. erythrogaster lacked positions 1–50 for lated for the sequence data matrix and the sequence data matrix plus the binary gap matrix (Table 2). TABLE 2

Pairwise Comparisons Variable and Potentially Phylogenetically Informa- Pairwise comparisons of all taxa were generated tive Sites for the Sequence Data with and without using PAUP* 4.0d51 (provided by D. L. Swofford). Coded Gaps Comparisons of absolute numbers of transitions and 12S ϩ 16S ϩ ALL ϩ transversions (Fig. 3) and the transition:transversion 12S gaps Valine 16S gaps ALL gaps ratio (Fig. 4) were calculated and plotted versus genetic distance for the 12SϩValine genes, the 16S gene, and Variable 200 203 11 476 483 687 697 all genes combined (ALL data). Informative 110 114 7 333 342 450 463 312 SIMONS AND MAYDEN the 12S rRNA gene and Eremichthys lacked positions 860–926 of the 16S rRNA gene. These data were missing due to technical problems in sequencing reactions. Pairwise comparisons of absolute number of transitions and transversions versus genetic distance did not indicate saturation of transitions or transversions in any of the three data sets (Fig. 3). However, examination of the transition: transversion ratio in all three data sets indicated a decrease in this ratio with increas- ing genetic distance. This ratio reached an asymptotic value of 2.5 among all data sets at genetic distances of 8% or more (Fig. 4), suggesting that saturation was occurring at some sites (Mindell and Honeycutt, 1990) and that weighting of transversions with respect to transitions was appropriate. Estimated g1 values for analyzed sequences indicated that the data were more structured than random data (P ϭ 0.1) using critical values presented in Hillis and Huelsenbeck (1992). This was also true of the 12S

FIG. 4. Ratio of transitions/transversions versus percent divergence for all pairwise comparisons. (A) 12SϩVal. (B) 16S. (C) ALL data.

and 16S when examined separately (P ϭ 0.1). Estima- tion of the g-statistic using only the Valine tRNA sequence indicated that these data were not more structured than random data. This was not surprising as there are only seven phylogenetically informative sites in the Valine tRNA sequence. The g1 values for 12S, 16S, and ALL data with congeneric taxa removed also indicated that the data were more structured than random data. Phylogenetic analysis of the combinations of character weightings and character sets produced varied results. Rather than present all the resolutions obtained, resolutions common to all analyses and alternative resolutions among major groups of taxa are discussed. The basal taxon was used when naming clades: for example, the Relictus clade refers to Relictus plus FIG. 3. Absolute number of transitions and transversions versus the taxa that comprise its putative sister group. Inclu- percent divergence for all pairwise comparisons. (A) 12SϩVal. (B) sion of gaps as binary characters made little difference 16S. (C) ALL data. (W) Transitions. (N) Transversions. in analyses of most character sets. In cases where inclusion PHYLOGENETIC RELATIONSHIPS OF PHOXININS 313

of bootstrap replicates in analyses of All, 1:1 (Fig. 5). In all other analyses where sister group relationships were resolved, the Creek Chub Clade was sister to the OPM Clade and together this clade was sister to the Orthodon Clade (12S 2:1 and 12SϩVal 1:1, 2:1, 3:1) or the Western clade (16S 2:1, 3:1, and ALL 2:1, 3:1). Bootstrap analyses of ALL 2:1 did not resolve a sister group relationship of the Creek Chub Clade in more than 50% of replicates (Fig. 6). A sister group relationship with the OPM Clade was resolved in 55% of bootstrap replicates of ALL 3:1 (Fig. 7). The OPM Clade and the Western Clade were never resolved as sister taxa. Within the Western Clade, all analyses supported a monophyletic group, containing, G. atraria, G. cypha, G. conspersa, G. pandora, G. ditaenia, G. robusta, G. nigrescens, and G. orcutti. Bootstrap analysis of ALL data indicated that this clade was supported in 97–99% of bootstrap replicates (Figs. 5–7). Although relationships among members of this clade varied with the analyses, analysis of ALL data produced the same sets of relationships regardless of weighting scheme employed. Four pairs of sister taxa were always resolved. G. atraria was sister to G. cypha, G. conspersa was sister to G. pandora, G. nigrescens was sister to G. orcutti, and G. ditaenia was sister to G. robusta. G. conspersa, G. pandora, G. nigrescens, and G. orcutti always formed a monophyletic group.

FIG. 5. Strict consensus of four trees produced by analysis of ALL data, TV:TS 1:1. of gaps did affect resolution, the affected nodes were ones found to be unstable in other analyses. Hence, inclusion of gaps did little to clarify relationships of taxa. In all analyses, four clades were always present: the Phoxinus Clade, containing the two representatives of the genus Phoxinus; the OPM Clade, containing Rhin- ichthys, Phenacobius, Platygobio, and Notropis; the Creek Chub Clade, containing Couesius, Margariscus, and Snyderichthys; and the Orthodon Clade, containing the genera Acrocheilus, Eremichthys, Gila, Lavinia, Orth- odon, Ptychocheilus, Relictus, and Siphateles (Figs. 5–7). In bootstrap analyses of the ALL data set, these clades were supported in 98–100% of all replicates (Figs. 5–7). In topologies produced in 12S and 12SϩVal analyses, the Phoxinus Clade was either sister to a clade composed of the OPM Clade, the Creek Chub Clade, and the Orthodon Clade (i.e., Phoxinus was the basal taxon of the ingroup) or in an unresolved polytomy with these clades. In all analyses of 16S and ALL data, the Phoxinus Clade formed the basal sister group in the Western Clade, sister to the Orthodon Clade. The monophyly of the Western Clade was supported by 100% of all bootstrap replicates (Figs. 5–7). The Creek Chub Clade was resolved as the sister group to the Western Clade in two analyses (16S, 1:1, FIG. 6. Strict consensus of six trees produced by analysis of ALL and ALL, 1:1). This relationship was supported in 54% data, TV:TS 2:1. 314 SIMONS AND MAYDEN

DISCUSSION

Phylogenetic analysis of 12S, 16S, and ALL data produced alternative trees among cyprinid taxa examined. These differences may be due, in part, to noise in the disparate data sets or to independence of the 12S and the 16S genes such that the history of the genes does not reflect the phylogenetic history of the organ- isms. In the case of the former, where differences in the various topologies are the product of noise in the data set, resolution may be achieved by analyzing the data in total, sensu Kluge (1989). This approach may not be justified, however, if the different topologies actually reflect different histories of the genes (Swofford, 1991). The genes examined herein are adjacent mitochondrial genes separated by the Valine tRNA. Thus, it is un- likely that these genes possess independent histories that may be exhibited by some nuclear genes or between nuclear genes and those of the mitochondrial genome. Therefore, we view the different topologies produced in the analyses as the product of either noise or insufficient phylogenetic signal. This view is supported by examination of the number of phylogenetically informative sites contained in the sequence data for each gene (Table 2). The 16S gene contains three times as many phylogenetically informative sites as the 12S. Hence, in this instance, we consider a total evidence approach (Kluge, 1989) as the best approach FIG. 7. Strict consensus of two trees produced by analysis of ALL to take when dealing with the different topologies data, TV:TS 3:1. produced by the separate analyses. Our results agreed with the phylogenetic conclusions of Coburn and Cavender’s (1992) analysis based on In weighted analyses of the 16S data, Acrocheilus morphology to the extent that the cyprinid fauna of west- plus G. coerulea formed the sister to the rest of the ern North America is not a monophyletic group. However, genus Gila. In weighted and unweighted analyses of phylogenetic analysis of mitochondrial ribosomal RNA ALL data, G. coerulea was the basal sister group to the gene variation supports a more restrictive concept of the rest of Gila; Acrocheilus was sister to all members of Western Clade than that proposed by previous authors. the genus Gila (Figs. 5–7). Our analyses support recognition of three clades within the The genus Relictus was resolved as sister to Acrochei- North American phoxinin fauna: the Western Clade, lus plus Gila in analyses of ALL data except when gaps the Creek Chub Clade, and the OPM Clade. The OPM Clade is congruent with Mayden’s (1989) were coded binary and nucleotide transformation was OPM clade except that Rhinichthys, usually considered unordered. In the latter case, Relictus was unresolved related to the western fauna (Coburn and Cavender, with respect to Acrocheilus plus Gila and all other basal 1992), is a member of this clade. Coburn and Cavender taxa in the Orthodon Clade. (1992) discussed alternative analyses of their data that Relationships among Eremichthys, Lavinia, Siphat- suggested that Rhinichthys may not be a member of eles, and Ptychocheilus varied depending upon the data their western clade. When scale characters were re- and weighting scheme used. In analyses of 12S se- moved from their analysis, Rhinichthys and Agosia quences these taxa formed a monophyletic group, while were resolved as members of their chub clade. Simi- in analyses of 16S and ALL data, they formed a larly, when the Asian taxa Tribolodon and Rhinichthys paraphyletic grade, basal to the Relictus clade. We were removed from their analysis, Rhinichthys and consider the relationships of these taxa to be unre- Agosia could not be resolved as members of either their solved within the Western Clade. western clade or their chub clade. In an analysis of the Orthodon is basal to all taxa in the Western Clade interrelationships of the Creek Chubs, based on the with the exception of Phoxinus in all analyses of the nucleotide sequences of the 12S rRNA, we found that 16S and ALL data sets. In bootstrap analyses of ALL the OPM Clade was the sister to all other phoxinin taxa data, Orthodon was resolved as sister to the above taxa included in the analysis (Simons and Mayden, 1997). In in 98–100% of replicates. analyses of only 12S rRNA sequences in the current PHYLOGENETIC RELATIONSHIPS OF PHOXININS 315 study, the OPM Clade and the Creek Chub Clade either consider them unresolved with respect to each other were sister taxa or were unresolved with respect to the and to the Relictus clade. Additional data are necessary Western Clade. The addition of the Valine tRNA se- to resolve relationships among these taxa. quence data resulted in a sister taxon relationship The recognition of three major clades of North Ameri- between the OPM Clade and the Creek Chub Clade. can phoxinins and the resolution of their relationships Analysis of the 16S rRNA or ALL data in an un- as described herein lead to the consideration of the weighted analysis resulted in resolution of the Creek relationships of Eurasian Phoxinins to the North Ameri- Chub Clade as the sister taxon to the Western Clade. can taxa. The Orthodon Clade as recognized herein is Weighting of transversions with respect to transitions, composed of taxa whose collective ranges are restricted however, resulted in the monophyly of the Creek Chub to drainages west of the Continental Divide. Phoxinus, plus OPM Clade. Given the evidence for saturation of the sister to the Orthodon Clade has an holarctic transitions with respect to transversions at some sites distribution, with representatives in eastern North we suggest that, based on all the evidence, the Creek America, Europe, and Asia. It is, however, absent from Chubs form the sister group to the OPM Clade. western North America. The genus Phoxinus is com- Within the Western Clade, the genus Gila was re- posed of nine species, the relationships of which were solved as a monophyletic group, exclusive of Siphateles discussed by Chen (1994). Chen (1994) hypothesized bicolor and S. copei. Gila coerulea, classified in the that Phoxinus was monophyletic and sister to a clade of subgenus Klamathella, formed the sister group to the Asian phoxinins, Eupallasella and Lagowskiella plus remaining members of the genus, the Gila Clade. Rhynchochypris. Chen (1994) further hypothesized that Among taxa included in this study, the relationships Phoxinus and the Eupallasella clade were sister to a within this clade are not consistent with the subgeneric clade containing Margariscus, Couesius, and Semoti- groupings proposed by Miller (1945) or with the phylog- lus. Basal to all the aforementioned taxa was the genus eny of Uyeno (1961). The only consistency between our Hemitremia. Based on our previous analysis of the hypothesis of relationships and that of Uyeno is a relationships of the Creek Chubs (Simons and Mayden, in relationship between G. pandora and G. conspersa. press) and Coburn and Cavender’s (1992) analysis, Hemi- Contrary to our analysis, Coburn and Cavender (1992) tremia is sister to Semotilus. Coburn and Cavender (1992) found no support for a monophyletic Gila, even after included Rhynchocypris and Tribolodon in their analysis. the exclusion of Siphateles. Our analysis did not in- The phylogenetic position of the Phoxinus clade with clude Iotichthys, Moapa, or Algansea, taxa that (based respect to the remainder of the North American taxa on Coburn and Cavender’s results) may render Gila has interesting implications with regard to the biogeo- paraphyletic. graphic hypotheses that may be proposed. If Phoxinus In analyses of ALL data, regardless of weighting is sister to the remainder of the Western Clade one scheme, A. alutaceus was resolved as the sister taxon to Gila, and the genus Relictus was sister to Acrocheilus would predict that the Creek Chub Clade and the OPM plus Gila. Hubbs et al. (1974) suggested a close relation- Clade would themselves have Asian representatives or ship of Relictus with Siphateles or Rhinichthys and together would form the sister group to an Asian taxon. noted that G. atraria is probably a near relative. Hubbs If Phoxinus is basal to the North American phoxinins, it et al. (1974) concluded that the combination of primi- is possible that they may form a monophyletic group tive and derived traits made the relationships of Relic- exclusive of any Asian or European forms. Analyses of tus difficult to determine. Coburn and Cavender (1992) the complete sequence data set, and the 16S rRNA data hypothesized that Relictus and Eremichthys, another separately, resolved a monophyletic Western Clade spring dwelling Great Basin endemic, were sister taxa inclusive of Phoxinus. This contrasts with analysis of and that together they formed the sister taxon to the 12S rRNA sequence data. In this case Phoxinus was Siphateles. Relictus and Eremichthys were not resolved sister to all other North American phoxinin taxa or was as sister taxa in any of our analyses. Similarly, Relictus unresolved with respect to the major clades found in all and Siphateles were never resolved as sister taxa. analyses. Addition of the Valine tRNA sequence did not Eremichthys and Siphateles were resolved as sister resolve the phylogenetic position of Phoxinus. Examina- taxa in 1:1 and 2:1 analyses of ALL data (Figs. 5 and 6). tion of the characters supporting a basal position of These resolutions were not strongly supported by boot- Phoxinus with respect to remaining North American strap analyses and were not observed in 3:1 analyses of phoxinins indicated that removal of a single character ALL data (Figs. 2–7). Similarly, we cannot make strong from the 12S data resulted in the resolution of Phoxi- statements regarding the relationships of Ptychochei- nus as sister to the remainder of the Western Clade lus and Lavinia. Although relationships among Er- rather than the basal taxon with respect to the North emichthys, Siphateles, Lavinia, and Ptychocheilus were American phoxinins. The combination of the small obtained in various analyses, these resolutions were amount of support for a basal Phoxinus in the 12S data not strongly supported by bootstrap analyses and and the results produced by analysis of the complete changed depending upon the weighting scheme ap- 12S and 16S data lead us to conclude that Phoxinus is plied. Given the instability among these taxa, we the sister taxon to the remainder of the Western Clade. 316 SIMONS AND MAYDEN

APPENDIX PHYLOGENETIC RELATIONSHIPS OF PHOXININS 317 318 SIMONS AND MAYDEN PHYLOGENETIC RELATIONSHIPS OF PHOXININS 319 320 SIMONS AND MAYDEN PHYLOGENETIC RELATIONSHIPS OF PHOXININS 321 322 SIMONS AND MAYDEN PHYLOGENETIC RELATIONSHIPS OF PHOXININS 323 324 SIMONS AND MAYDEN PHYLOGENETIC RELATIONSHIPS OF PHOXININS 325 326 SIMONS AND MAYDEN PHYLOGENETIC RELATIONSHIPS OF PHOXININS 327 328 SIMONS AND MAYDEN

We also hypothesize that there exist Asian taxa that are Chen, X.-Y. (1994). ‘‘Morphology, Phylogeny, Biogeography and Sys- closely related to the Creek Chub and the OPM clades. tematics of Phoxinus (Pisces: Cyprinidae),’’ unpublished Ph.D. dissertation, Univ. Kansas, Lawrence, Kansas. ACKNOWLEDGMENTS Coburn, M. M., and Cavender. T. M. (1992). Interrelationships of North American cyprinid fishes. In ‘‘Systematics, Historical Ecol- We are grateful for help with this project from the following people ogy, and North American Freshwater Fishes’’ (R. L. Mayden, Ed.), and institutions. T. Dowling provided mtDNA preparations of Gila pp. 328–373, Stanford Univ. Press, Stanford. atraria, G. ditaenia, G. nigrescens, G. pandora, G. robusta, Phoxinus De Rijk, P., Van de Peer, Y., Chapelle, S., and De Wachter, R. (1994). neogaeus, and Snyderichthys copei. P. Harris and D. Markle proved Database on the structure of large ribosomal subunit RNA. Nucleic tissues of Gila coerulea, Lavinia exilicauda, and Orthodon microlepi- Acids Res. 22: 3495–3501. dotus. M. Andersen provided specimens of Eremichthys acros, Relic- Girard, C. (1856). Researches upon the cyprinoid fishes inhabiting tus solitarius, and Siphateles bicolor. G. R. Haas and J. D. McPhail the freshwaters of the United States, west of the Mississippi valley, provided specimens of Acrocheilus alutaceus and Couesius plumbeus. from specimens in the museum of the Smithsonian Institution. W. Dimmick provided specimens of Platygobio gracilis. B. R. Ku- Proc. Acad. Nat. Sci. Phila. 8: 165–213. hajda. R. Matson, E. B. Jones, R. M. Wood, and S. R. Layman provided assistance in the field. T. A. Titus and E. O. Wiley provided Gyllensten, U. B., and Erlich, H. A. (1988). Generation of single- many of the primers used in this study. B. R. Kuhajda, J. C. Keiser, C. stranded DNA by the polymerase reaction and its application to Lydeard, and K. Roe provided valuable comments on the manuscript. direct sequencing of the HLA-DQA locus. Proc. Natl. Acad. Sci. We are especially grateful for help provided by K. E. Knott and B. R. USA 85: 7652–7656. Kuhajda. This research was supported by the Department of Biologi- Hillis, D. M. (1991). Discriminating between phylogenetic signal and cal Sciences at the University of Alabama and by National Science random noise in DNA sequences. In ‘‘Phylogenetic Analysis of DNA Foundation Grant 5-31943 to R.L.M. Sequences’’ (M. M. Miyamoto and J. Cracraft, Eds.), pp. 278–294, Oxford Univ. Press, New York. REFERENCES Hillis, D. M., and Huelsenbeck, J. P. (1992). Signal, noise and reliability in molecular phylogenetic analyses. J. Hered. 83: 189– Avise, J. C., and Ayala, F. J. (1976). Genetic differentiation in speciose 195. versus depauperate phylads: evidence from the California min- Hillis, D. M., Larson, A., Davis, S. K., and Zimmer, E. A. (1990). nows. Evolution 30: 46–58. Nucleic acids III: Sequencing. In ‘‘Molecular Systematics (D. M. Bailey, R. M. (1956). A revised list of the fishes of Iowa, with keys for Hillis and C. Moritz, Eds.), pp. 318–370. Sinauer, Sunderland, identification. In ‘‘Iowa Fish and Fishing’’ (James R. Harlan and Massachusetts. E. B. Speaker, (Eds.), 3rd ed., State Conservation Committee, Iowa. Howes, G. (1984). Phyletics and biogeography of the aspinine cypri- Bailey, R. M., and Uyeno, T. (1964). Nomenclature of the blue chub nid fishes. Bull. Br. Mus. Nat. Hist. (Zool.) 47: 283–303. and the tui chub, cyprinid fishes from the western United States. Copeia 1964: 238–239. Hubbs, C. L., Miller, R. R., and Hubbs, L. C. (1974). Hydrographic history and relict fishes of the north central Great Basin. Mem. Barbour, C. D., and Miller, R. R. (1978). ‘‘A Revision of the Mexican Calif. Acad. Sci. 7: 1–259. Cyprinid Fish Genus Algansea,’’ No. 155, Misc. Publ. Mus. Zool. Univ. Mich. Kluge, A. G. (1989). A concern for evidence and a phylogenetic Cavender, T. M., and Coburn, M. M. (1992). Phylogenetic relation- hypothesis of relationships among Epicrates (Boidae, Serpentes). ships of North American cyprinids. In ‘‘Systematics, Historical Syst. Zool. 38: 7–25. Ecology, and North American Freshwater Fishes’’ (R. L. Mayden, Kocher, T. D., Thomas, W. K., Meyer, A., Edwards, S. V., Paä¨bo, S., Ed.), pp. 293–327, Stanford Univ. Press, Stanford, California. Villablanca, F. X., and Wilson, A. C. (1989). Dynamics of mitochon- PHYLOGENETIC RELATIONSHIPS OF PHOXININS 329

drial DNA evolution in animals: Amplification and sequencing with Swofford, D. L. (1991). When are phylogeny estimates from molecular conserved primers. Proc. Natl. Acad. Sci. USA 86: 6196–6200. and morphological data incongruent? In ‘‘Phylogenetic Analysis of Mayden, R. L. (1989). Phylogenetic studies of North American DNA Sequences’’ (M. M. Miyamoto and J. Cracraft, Eds.), pp. minnows, with emphasis on the genus Cyprinella (Teleostei: Cypri- 295–333, Oxford Univ. Press, New York. niformes). Misc. Publ. Univ. Kansas Mus. Nat. Hist. 80: 1–189. Thompson, J. D., Higgins, D. G., and Gibson, T. J. (1994). CLUSTAL Miller, R. R. (1945). A new cyprinid fish from southern Arizona, and W: Improving the sensitivity of progressive multiple sequence Sonora, Mexico, with the description of a new subgenus of Gila and alignment through sequence weighting, position specific gap penal- a review of related species. Copeia 1945: 104–110. ties and weight matrix choice. Nucleic Acids Res. 22: 4673–4680. Miller, R. R. (1959). Origin and affinities of the freshwater fish fauna Titus, T. A. (1992). ‘‘A Phylogenetic Analysis of the Desmognathinae of western North America. In ‘‘Zoogeography’’ (C. L. Hubbs, Ed.), (Caudata: Plethodontidae): Evolutionary Patterns Inferred from pp. 187–222, Vol. 51, Am. Assoc. Adv. Sci., Washington, DC. Mitochondrial DNA Sequences,’’ unpublished Ph.D. dissertation, Miller, R. R. (1965). Quaternary freshwater fishes of North America. Univ. Kansas, Lawrence. In ‘‘Quaternary of the United States’’ (H. E. Wright and D. G. Frey, Titus, T. A., and Frost, D. R. (1996). Molecular homology assessment Eds.), pp. 569–581, Princeton Univ. Press, Princeton. and phylogeny in the lizard family Opluridae (Squamata: Iguania). Mindell, D. P., and Honeycutt, R. L. (1990). Ribosomal RNA in Mol. Phylogenet. Evol.. 6: 49–62. vertebrates: Evolution and phylogenetic applications. Annu. Rev. Titus, T. A., and Larson, A. (1996). Molecular phylogenetics of Syst. 21: 541–566. desmognathine salamanders (Caudata: Plethodontidae): A reevalu- Simons, A. M. (1996). ‘‘Phylogenetic Relationships of the North ation of evolution in ecology, life history, and morphology. Syst. American Phoxinins (Actinopterygii: Cyprinidae) Inferred from Biol. 45: 451–472. Mitochondrial Ribosomal DNA Sequences,’’ unpublished Ph.D. Titus, T. A., and Larson, A. (1995). A molecular perspective on the dissertation, Univ. Alabama, Tuscaloosa. evolutionary radiation of the salamander family Salamandridae. Simons, A. M., and Mayden, R. L. (1997). Phylogenetic relationships Syst. Biol. 44: 125–151. of the Creek Chubs and the Spine-Fins: An enigmatic group of Uyeno, T. (1961). ‘‘Osteology and Phylogeny of the American Cyprinid North American cyprinid fishes (Actinopterygii: Cyprinidae). Clad- Fishes Allied to the Genus Gila,’’ unpublished Ph.D. dissertation, istics. 13: 187–205. Univ. Michigan, Ann Arbor. Swofford, D. L. (1993). ‘‘PAUP: Phylogenetic Analysis Using Parsi- Van de Peer, Y., Van den Brock, I., De Rijk, P., and de Wachter, R. mony, Version 3.1,’’ computer program distributed by the Illinois (1994). Database on the structure of small ribosomal subunit RNA. Natural History Survey, Champaign. Nucleic Acids Res. 22: 3488–3494.