SystematicBotany (2004),29(1): pp. 77– 96 q Copyright 2004by the AmericanSociety of PlantTaxonomists

Molecular Phylogeny and Biogeography of Ribes (Grossulariaceae),with an Emphasis on Gooseberries (subg. Grossularia )

LISA M. SCHULTHEIS1,2,3 and MICHAEL J. DONOGHUE1 1Department ofEcology and Evolutionary Biology,Yale University,P.O.Box 208105, New Haven, Connecticut06520; 2The Arnold Arboretum ofHarvard University,125Arborw ay,JamaicaPlain, Massachuse tts 02130; 3Present address: Biological and Health Sciences Division, Foothill College,12345El Monte Road, Los Altos Hills, California94022

CommunicatingEditor: Matt Lavin

ABSTRACT. Gooseberries are oftendistinguishe dfromcurrants as adistinctgenus ( Grossularia )orsubgenus ( Ribes subg. Grossularia ),but recent molecular phylogenetic analyses ofc hloroplast and nuclear data disagree as tothe monophylyof this group.Wereportne wsequence data fromthe 18–26S nuclear rDNAITSand ETSregions and fromthe chloroplast psbA- trnH intergenicspacer that, incombina tionwith pre viously reported data, suggestsubg. Grossularia ismonophy letic and nested within Ribes.Two main lineages are evident withinsubg. Grossularia ,correspondingto the true gooseberries(subg. Grossularia sect. Grossularia )and aclade ofglabrous-st yledwestern N orthAmerican gooseberries(subg. Grossularia sect. Robsonia, subg. Hesperia, Lobbia).Biogeograp hic analyses based on DIVAoptimizationssuggest a westernN orthAmerican originfor subg. Grossularia ,withsubsequent dispersal toeast Asia givingrise to a well-supportedclade ofAsian gooseberry species insect. Grossularia .Thisexample contrastswith the w ell-documented pattern ofdispersal fromAsia toN orth America, and highlightsthe need toinvestiga te additional groups distributedwidely throughthe N orthernH emisphere.

Ribes L. comprises approximately 150species of placedalong aseparatelineage .Asecond study using shrubby plants with strikingly diverse oraland fruit ITSsequence dataalso supported the monophyly of features.The infragenericclassiŽcation for Ribes varies the true gooseberries (sect. Grossularia ),but did not among the primary treatments (e.g.,J anczewski 1907; sample glabrous-styled gooseberries or spinycurrants Coville and Britton 1908;Berger 1924;Rehder 1940; (Fenton et al.2000). Given the long tradition ofrec- Sinnott 1985),creating adifŽcult taxonomy (reviews in ognizing the gooseberries asa distinct genus (e.g.,Co- Spongberg 1972;Sinnott 1985).One ofthe more nota- ville and Britton 1908;Berger 1924)or subgenus (Jan- ble and consistent divisions within the genus is be- czewski 1907;Sinnott 1985),one goalof this study was tween the currants (subg. Ribes)and the gooseberries toclarify the monophyly ofsubg. Grossularia . (subg. Grossularia ).Currants are mostly spineless AŽrst step towards addressing the monophyly of shrubs bearing multi-ow ered racemes with jointed the gooseberries (subg. Grossularia )is tocombine the pedicels, whereas gooseberries havenodal spines and availablechloropla st (Messinger et al.1999) and ITS sometimes bristly stems, bearfew- ow ered racemes (Senters and Soltis 2003)datasets, and toobtain addi- with non-jointed pedicels, and have highly reexed se- tional datato resolve apparentcon icts evident in pals(Senters and Soltis 2003).The differencesbetween these previous studies. Additional datamay result in gooseberries and currants have led manyresearchers aconvergence onto similar topologies, indicating that torecognize gooseberries atthe subgeneric(subg. the apparentcon ict was w eak or due toinsufŽ cient Grossularia )(Janczewski 1907;Sinnott 1985)or generic data,or alternatively,may reinforce the apparentcon- (Grossularia )(e.g.,Coville and Britton 1908;Berger ictbetw een the chloroplast and nuclear genomes, in- 1924)le vel. Previous phylogenetic analyses, however, dicating possible hybridization and introgression are atodds regarding the statusof the gooseberries as events in the history ofthe group (Rieseberg and Wen- anaturalgroup (Fig. 1).Sequence datafrom the nu- del 1993;A vise 1994).A dditional datasets reported clear18S– 26S rD NAinternal transcribed spacer(ITS) here include sequence datafrom the nuclearencoded region indicated monophyly ofsubg. Grossularia (Sen- 18S–26S rD NAexternal transcribed spacerregion ters and Soltis 2003),while restriction site datafrom (ETS)and from the chloroplast encoded psbA-trnH in- the 18S–26S n uclearrD NAregion (Messinger et al. tergenic spacerregion. Previous studies reported levels 1993)and from twochloroplas tregions indicated poly- ofv ariation in the ETS similar toor greater than those or paraphyly (Fig. 1;M essinger et al.1999). The chlo- typicalof the ITSregions (Baldwin and Markos 1998; roplastdata strongly resolved the morphologicallyin- Bena et al.1998; Linder et al.2000). Studies have also termediate spiny currants (subg. Ribes sect. Grossular- reported sufŽcient variation within the psbA-trnH in- ioides)asa paraphyletic grade subtending only the true tergenic spacerregion forph ylogeneticreconstruction gooseberries (subg. Grossularia sect. Grossularia ), while within genera (e.g.,Sang et al.1997; Mast and Givnish acladeof glabrous-sty led gooseberries (subg. Hesperia, 2002;M ort et al.2002). Lobbia, and Grossularia sect. Robsonia)were weakly Asecond goalof this study wasto consider Ribes

77 78 SYSTEMATIC BOTANY [Volume 29

FIG.1. AsimpliŽed representation ofprior phylogenetic hypotheses for Ribes.Chloroplast restrictionsite data fromMes- singeret. al.(1999) suggestthat gooseberries(subg. Grossularia )are non-monophyletic.ITS data fromSenters and Soltis(2003) support the monophylyof gooseberries. in the contextof N orthern Hemispherebiogeography. ofSouth America, butis especially diverse in western The Northern Hemisphereproblem has recently at- North America, bothin terms ofth enumber ofspecies tracted attention from aphylogeneticperspective,by and the representationof major subclades.Lik ewise, bothzoologists (e.g.,Sanmart ´‡net al.2001) and bota- subg. Grossularia ,which is the focusof our analysis, nists (e.g.,Manos and Donoghue 2001).It has become occursaround the Northern Hemisphere,butis most clearthat  oristic similarities among the major areas diverse in western North America. SpeciŽcally ,nu- ofendemism ,such aseastern Asia and eastern North merous glabrous-styled gooseberries (subg. Hesperia, America(see Bouffordand Spongberg 1983;W en Lobbia, and sect. Robsonia)are distributed in western 1999),were established in several ways (e.g.,move- North America, while the true gooseberries (sect. Gros- ment through Beringia or the North Atlantic)and at sularia)are foundthroughoutN orth America and Asia different times in different groups (e.g.,Tiffn ey 1985a; (including Taiwanand Japan),with one species in Eu- Manchester1999;W en 1999;Xiang et al.1998, 2000; rope.Better knowledge ofph ylogeneticrelationships in Donoghue et al.2001; Fritsch et al.2001; Manos and Ribes,and especially in Grossularia ,would make it pos- Stanford2001; Tiffney and Manchester 2001;Xiang sible toassess geographic patterns ofdiversiŽ cation and Soltis 2001).Se veral recent studies have highlight- and directions ofmoveme nt in the group. ed plantgroups thatappear to have diversiŽed ini- tially in Asia and subsequently moved toN orth Amer- MATERIALSAND METHODS icavia the Bering Land Bridge,apparently atse veral Samples. New data reported here include 12 ITSsequences , different times (e.g.,Donoghu eet al.2001; Xiang and 73 ETSsequences (from57 species), and 53 psbA-trnH sequences. Soltis 2001).This iterative trans-Beringian movement Previous studies provided an additional 67 ITSsequences (Senters resembles the pattern described formammals earlier and Soltis2003) and restrictionsite data fromtw oampliŽed chlo- roplast regions( rbcL to accD and rpoC1 to rpoC2)for32 species in the Tertiary (Beard 1998),and generally suggests (Messingeret. al. 1999). Oursampling w as weightedtow ards thatAsia has been aprimary source areafor N orthern subg. Grossularia ,the focus ofthis study. Weincluded all ofthe Hemispherediversity.However,this may reect the true gooseberryspecies (sect. Grossularia )recognized bySinnott (1985), all ofthe Asian true gooseberryspecies recognized byBer- sample oftaxa that has been examined todate ,and ger(1924), representatives fromeach ofthe three glabrous-sty led other patterns, including movement from North Amer- gooseberrytaxa (sect. Robsonia, subg. Hesperia, Lobbia), and both icato Asia, havealsobeen described (see examples in spinycurrant species (sect. Grossularioides ).In total, 82 ofthe ap- ‡ proximately150 species of Ribes were included inthis study (T able Sanmart´net al.2001). It is noteworthythatN orthern 1), representingall ofthe infragenerictaxa recognized byBerger Hemisphereplantgroups thatare especially diverse (1924), but withnomencla ture followingSinnott (1985). Itea, rep- today in western North America haveseldom been the resented by I.virginica and I.ilicifolia ,was chosen as theoutgroup focusof ph ylogeneticbiogeographicanalyses. based on previouslyp ublished studies (Soltiset al. 1990, 1993; Morgan and Soltis1993; Soltisand Soltis1997), inwhich Ribes, Ribes is very broadly distributed around the North- Itea,and Saxifragaceae s.s. appeared tobe closely related members ern Hemisphere,and extends south in the mountains ofSaxifragales . 2004] SCHULTHEISAND DONOGHUE:PHYLOGENY AND BIOGEOGRAPHY OF RIBES 79

DNAIsolation. Total DNAwas isolated fromfresh, silica-des- ETSprimer (Baldwin and Markos1998) under the followingpa- iccated, and herbariummaterial. Allextractions were performed rameters:initial denaturation (94 8C, 2min),follow ed by35 cycles usingQiaGen DNeasy Plant Minikits, following manufact urer’s ofdena turation (94 8C, 30 sec), annealing (65 8C, 1min),extension instructions,except that plant material was groundin w armed (728C, 1.5 min),and concludingwitha Žnal extension(72 8C, 7 AP1 buffer(provided inQiaGen kit)ra therthan liquidnitrogen. min).PCR reactions contained 2.5 mL10X AmpliTaq buffer(P erkin-

SomeDN Aaliquots were provided bySenters and Soltis(U niver- Elmer), 2.5mL10mM dNTPs, 2.5 mL 25mM MgCl2, 0.5mL 10mM sityof Florida), as indicated inT able 1. ETS-Rib1 primer,0.5 mL 10mM18S-ETS primer,0.1 mL AmpliTaq AmpliŽcation and Sequencin g. PCR and sequencingreactions DNApolymerase(Perkin-Elmer),and –10ng DNA, toa total vol- were performedusing a Perkin-ElmerCorpora tionGeneAmp 9600 ume of 25mL. Cleaned ampliŽca tionproducts were sequencedin orMJ Research Inc.DN AEngine y ThermalCycler .PCR products bothdirections with primers ETS-Rib1 and 18S-ETS. Iteailicifolia were cleaned usingQiaGen QiaQuickPCR PuriŽcation kits.Se- isa partial sequence,missingthe Ž rst107 bp. quencingreactions were performedwith ABI Prism BigDye T er- PrimerspsbAF and trnHR (Sanget al. 1997) were used forboth minatorCycle SequencingReady Reaction Kits, usinghalf orfull ampliŽca tionand sequencingofthe psbA-trnH intergenicspacer reactions, and were cleaned usingeither EtO H/NaOAcprecipita- region.PCR reaction parameters were: initial denaturation(94 8C, tion, followingrecommend ationsin BigDye kits, or Edge Biosys- 2min),followed by35 cycles ofdena turation (94 8C, 30 sec), an- temsP erformaDTR y Gel Filtrationsystems. In some cases, se- nealing (618C, 30 sec), extension(72 8C, 1min),and concluding quences were improved withthe addition ofDMSO to sequencin g witha Žnal extension(72 8C, 7min).PCR reactions contained reactions (approximately6% Žnal volume). Sequencingreactions 2.5mL10X AmpliTaq buffer(P erkin-Elmer),2.5 mL10mM dNTPs, were resolved on 5% polyacrylamide gelswith an Automated Bio- 2.5mL 25mM MgCl2, 1.25mL 10mMpsbAFprimer, 1.25 mL 10mM systems(ABI) 377 sequencer, loaded withstandard orRapid Load trnHR primer,0.1 mLAmpliTaq DNApolymerase (Perkin-Elmer), membranecombs (The Gel Company).Sequench erv ersion3.1.1 and1–10ng DNA, toa total volume of25 mL. Forsome samples, was used toexamine and edit sequence chromatograms. ampliŽca tionw as aided bythe addition of2.5 mLBSAto the PCR ITSampliŽ cation products forthe 12 new sequencesreported reaction. Ribes viburnifolium was notsuccessfully sequenced,pos- inthis study were generated usingprimers ITS4 (White et al. 1990) siblydue tohomopoly merstrings close toboth the 5 9 and 39 prim- and ITS-I (59-GTCCACTGAACCTTATCATTTAG-3 9;designedbyL. ersites. A central portion ofapproxima tely159 bp could notbe E. Urbatsch, Louisiana State University)with the followingparam- recovered from Iteavirginica and was coded as N’s inthe matrix. eters: initial denaturation(97 8C, 1min),follow ed by35 cycles of Sequence identitiesw ere veriŽed byperforming BLAST searches denaturation(97 8C, 10 sec), annealing (48 8C, 30 sec), extension (National Centerfor Biotechnol ogyInformation ,National Insti- (728C, 20 sec increasing 4sec witheach cycle), and concluding tutes ofH ealth) usingsequences from Ribes burejense . witha Žnal extension(72 8C, 7min).PCR reactions contained Alignment. Sequenceswere initiallyaligned inClustal XMul- 2.5mL10X AmpliTaq buffer(P erkin-Elmer),2.5 mL10mM dNTPs, tiple Sequence AlignmentProgram v ersion1.81 under thedefa ult

2.5mL 25mM MgCl2, 1.25mL BSA, 1.25mL 10mMITS4primer, settings,then adjusted byeye usingMacClade 4sequencingeditor 1.25mL 10mMITS-Iprimer, 0.1 mLAmpliTaq DNApolymerase features(Maddison and Maddison 2000). All Ribes sequenceswere (Perkin-Elmer),and 1–10ng DNA, toa total volume of25 mL. Am- readily alignable byeye ,as were Itea sequences,but alignment pliŽca tionproducts were sequenced withITS4, ITS2(White et al. between Ribes and Itea was oftendifŽ cult. Ambiguousregions 1990), ITS-Iand ITS3B(the re verse complementof ITS3; Whiteet were aligned sotha tthenumber of informativ esiteswas mini- al. 1990). Mostthe lengthof ITS 2 was sequenced inboth direc- mized. Base pairs thatcould notbe assigned withconŽ dence due tions.Sequenci ngprimers for ITS 1 oftenproduced shortsequenc- tow eak ornoisy signal were coded with‘ ‘N’’ orwith IUP AC-IUB es withminimal overlap ,such thatsequences were largelybased ambiguitysymbols. Gaps insertedfor alignment in regions of in- on asinglestrand. Aportion ofthe 5.8S regioncould notbe re- ferredindels were coded as ’’–’’and treated as missingda ta. Seven covered from R. cynosbati (41 bp) or R. echinellum (77 bp), and was large indelscoded as present (1)or absent (0)were added tothe thuscoded as N’s inthe ma trix.Of the 67 ITSsequence sobtained psbA-trnH matrix.The seven coded indelsincluded threeinser- fromSenters and Soltis(2003), Žve were missinga large portion tions in Itea (positions79– 88, 199–207, 418–434), twodeletions in ofITS 1 (approx. 100bp for R. giraldii and R. montigenum ; approx. Itea (positions110– 114, 359–372), an insertionconsisting primarily 240 for R. missouriense ;approx. 130 for R. watsonianum ; approx. ofT Arepeats inmembers of sect. Grossularia (positions279– 295), 230 for R. rubrum). and adeletion in R. himalense , R. manshuricum , and R. rubrum (po- Sequencesfromthe external transcribedspacer of18– 26S nucle- sitions111– 116). Datasets are available at TreeBASE(study acces- ar rDNA(ETS)w ere obtained followingthe strategyof Baldwin sionnumber s1001). and Markos(1998). Ribes odoratum (subg. Ribes sect. Symphocalyx ) DatasetCombinabilit y. Overlap amongthe fourda tasets (ITS, and R. hirtellum (subg. Grossularia sect. Grossularia )were chosen ETS, psbA-trnH, restrictionsites) w as notcomplete .Ofthe 84 spe- forinitial long-distancePCR ofthe inter-genic spacer region(IGS) cies included inthis study ,50 were represented inat least three inorder to design an internal primerconserv ed across Ribes. ofthe four da tasets (Table 1). Aseriesof partition homogeneity Long-distance PCR ofthe IGSw as conducted withthe following tests(F arriset al. 1995) were performedin P AUP* ver.4.0 (Swof- parameters:initial denaturation (94 8C, 1min),follow ed by35 cy- ford2001) toassess dataset combinability.Each testconsisted of cles ofdena turation(94 8C, 30 sec), and combinedannealing and 100 replicatesemployin gheuristic searches withsimple taxon ad- extension(72 8C, 6min).PCR reactions contained thefollowing dition, TBR branch swapping, maxtreesset to 1000, and invariant components:2.5 mL10X KlenTaq LAP olymeraseMix (Clontech characters excluded. Laboratories,Inc.), 2.5 mL10mM dNTPs, 2.5 mL 10mM 18S-IGS Initialpartition homogeneitytests indicated thatthe psbA-trnH primer(Baldwin and Markos1998), 2.5 m L10mM26S-IGSprimer and restrictionsite datasets were combinable (p 50.66), while all (Baldwin and Markos1998), 0.5 mLKlenTaq LAD NApolymerase, otherda taset combinationsw ere incongruent(p # 0.01). Toexplore and 1–10 ngD NA, toa total volume of25 mL. AmpliŽca tionprod- potential sources ofincongruen ce,weconducted additional ho- ucts were sequencedinone direction usingprimer 18S-E (Baldwin mogeneitytests with individua ltaxa orgro ups oftaxa excluded and Markos1998), yieldingapproxima tely480bp ofreadable se- based on differentialand well-supported(bootstrap .70%) place- quence. PrimerETS-Rib1 (5 9 GAACTGTTGTCGCGTGCGTCGT3 9) mentin trees, oron differentialplacemen tregardlessof support was designed5 9 ofthe 18S-E primingsite, froma conserved region levels(deQueiroz et al. 1995). Forthe ETS and ITScomparison , withinETS. Attemptsw ere made tosequence furtherinto the ETS taxa were excludedone atatimeand ingro ups ifthey represented regionusing primer ETS-Rib2 (5 9 ACGACGCACGCGACAA- composites ofmultiple accessions (Table 1) and exhibiteddiffer- CAGTTC 39),thereverse complementof ETS-Rib1, but readable ential placementsin ETS versus ITStrees. Samples of R. oxyacan- sequenceswere onlyobtained from R. hirtellum ,hinderingefforts thoides subsp. oxyacanthoides were also excluded inETS versus ITS toidentify a conserved sitefurther 5 9 ofETS-Rib1. comparisonsbecause therewas divergencebetween thetwo in- Short-distancePCR ofthe 3 9 portion ofthe ETSregion was per- cluded ETSsequences (sequenced forthis study), thus itw as not formedusing the ETS-Rib1 primerin conjunction withthe 18S- clear which should be combinedwith the ITS sequence (obtained 80 SYSTEMATIC BOTANY [Volume 29

TABLE 1. Taxa included inthis study ,withGenBank accession numbers.F ollowingMessinger et al. (1999), taxonomycorresponds toSinnott (1985) with Grossularia subg. Hesperia and subg. Lobbia treated under Ribes. ITSsequences noted inboldface are fromSenters and Soltis(2003). Sentersand Soltisprovided the DNAforitalicized sequences.Restriction sitedata are fromM essingeret al. (1999). Voucher informationiscited as per Sentersand Soltis(2003) orM essingeret al. (1999) when thedata orDN Acame fromthose sources. Herbariumcodes followthe Index H erbariorum,EighthEdition. NPGR refersto the USD A-ARSN ational Plant Germplasm Repository. UCBG refersto the U niversityof California Botanical Garden. RBGE refersto the Ro yal Botanic Garden, Edinburgh.

Subg. Ribes L. Sect. Berisia Spach (Alpine Currants) R.acuminatum (Hook. F.&Thomson)Jancz. Chase 3585 (K): ITS AF426376. R. alpinum L.ArnoldArboretum 678– 80E: ETSA Y138021; psbA-trnH AY138094. Chase 3587 (K): ITS AF426378. NPGR 6640: restrictionsites. R.diacanthum Pallas ArnoldArboretum 1852– 81B: ETSAY138022; psbA-trnH AY138095; ITSAY138047. NPGR 34 ( Messinger 315;OSC):restriction sites. R. giraldii Janczewski ArnoldArboretum 609– 74B: ETSA Y138023; psbA-trnH AY138096. M. Mort s.n. (WS): ITS AF426381. R. glaciale Wall. UCBG 91.0285: ETSAY138024; psbA-trnH AY138097; ITSAY138048. R.komarovii Pojark. UCBG 91.078: ETS(1): AY138025. ArnoldArboretum 1094– 82A: ETS(2): AY138026; psbA-trnH AY138098. R.maximowiczii Batalin T.S. Elias 10940 (RSA): ITS AF426380. NPGR 267: restrictionsites. R.orientale Desf. Boufford etal. 28317 (A):ETSA Y138027; psbA-trnH AY138099. R. tenue Janczewski Boufford etal. 27604 (A):ETS A Y138028; psbA-trnH AY138100. Chase 3610 (K): ITS AF426377. R.vilmorini Janczewski Chase 3612 (K): ITS AF426379. Sect. Calobotrya (Spach) Jancz. (OrnamentalCurrants) R. afŽne H.B.K. M.Medina 2517 (NY): ITS AF426326. R.brandegeei Eastwood s.c. 350568 (WS):ITS AF426331. R. cereum Douglas UCBG 93.1213: ETSA Y138013; psbA-trnH AY138087. NPGR 237.001: restrictionsites. Ross 3425 (NY): ITS AF426328. R.ceriferum Coville and Rose s.c. 05017194 (MO):ITS AF426333. R. ciliatum H. and B. Diggs 2625 (NY): ITS AF426327. NPGR 670.001 ( Messinger 311 ;OSC):restriction sites. R. dugesii Greenman Siplivinsky 3939 (WS): ITS AF426329. R.glutinosum Benth. Chase 3594 (K): ITS AF426340. R.indecorum Eastwood UCBG 86.0903: ETSA Y138014; psbA-trnH AY138088. Mort 1370 (WS): ITS AF426336. R.malvaceum SmithUCBG 91.1481: ETSA Y138015; psbA-trnH AY138089. Johnson s.n. (WS): ITS AF426338. R.mogollonicum Greene NPGR 294.001: ITS AF426332;restrictionsites. R.neglectum Rose Villarrea 4940 (NY): ITS AF426330. R.nevadense Kellogg UCBG 89.1635: ETSAY138016; psbA-trnH AY138090. Mort 1373 (WS): ITS AF426339. R.sanguineum Pursh UCBG 90.0193: ETSA Y138017; psbA-trnH AY138091. Mort 1372 (WS): ITS AF426335. NPGR 46: restriction sites. R.tortuosum Benth Breedlove 62230 (MO): ITS AF426325. R.viscosissimum Pursh Grimes 1878 (NY): ITS AF426334. NPGR 281.001 ( Fredricks 394 ;OSC):restriction sites. R. wolŽi Rothrock Siplivinsky 4587 (NY): ITS AF426341. Sect. Coreosma (Spach) Jancz. (Black Currants) R.americanum Mill.NPGR 93: restrictionsites. Nee 24196 (NY): ITS AF426375. R.bracteosum Douglas UCBG 89.1645: ETSA Y138033; psbA-trnH AY138103; ITSAY138049. R. fragrans Pallas s.c. 4378976 (MO):ITS AF426373. R.hudsonianum Richardson B.Ertter3807 (NY): ITS AF426372. var. petiolare (Douglas)Jancz. NPGR 278 ( Fredricks 390 ;OSC):restriction sites. R.janczewskii Pojark. Chase 3597 (K): ITS AF426370. R. nigrum L.NPGR 215.001 (OSC):ITS AF426374. R.viburnifolium A.Gray UCBG 65.1431: ETSAY138034. NPGR 762.001: ITS AF426371;restrictionsites. Sect. Grossularioides (Jancz.) Rehd. (Spiny,orGooseberry-stemmed Currants) R. lacustre (Pers.)Poiret Lesica 4710 (NY): ETS(1): AY138018; ITS AF426366. ArnoldArboretum 777– 93A: ETS(2): AY138019; psbA-trnH AY138092. NPGR 45: restrictionsites. R.montigenum McClatchie Bugham &Miller s.n. (WS): ETS AY138020; psbA-trnHA Y138093 ; ITS AF426367. NPGR 864.001 (Messinger 254 ;OSC):restriction sites. Sect. Heritiera Jancz. (DwarfCurrants) R.erythrocarpum Coville and LeibergNPGR 860.001 ( Messinger 249 ; OSC): ITS AF426342;restrictionsites. R. howellii Greene NPGR 449.001 ( Messinger 333 ; OSC): ITS AF426343;restrictionsites. R.glandulosum Grauer UCBG 89.0750: ETS(1): AY138035. Schultheis 589–00 (YU):ETS (2): AY138036; psbA-trnH AY138105. NPGR 231: restrictionsites. Chase 3605 (K) ( 5 R.prostratum L’Her .): ITS AF426345. R.laxiorum Pursh NPGR 439: restrictionsites. Goodrich 19052 (WS): ITS AF426344. Sect. Parilla Jancz. (AndineCurrants) R. andicola Jancz. Friere-Fierro2577 (NY): ITS AF426368. R.fasciculatum Sieb.& Zucc. UCBG 88.0615: ETS(1): AY138043; psbA-trnH AY138104. ArnoldArboretum 1879: ETS(2): AY138044. Chase 3592 (K): ITS AF426346. R.ovalifolium Jancz. Gentry s.n. (MO): ITS AF426369. R.valdivianum Phil. Messinger 314 (OSC):restriction sites. Sect. Ribes L.(Red Currants) 2004] SCHULTHEISAND DONOGHUE:PHYLOGENY AND BIOGEOGRAPHY OF RIBES 81

TABLE 1. Continued.

R.himalense Decaisne Boufford etal. 28903 (A):ETS A Y138037; psbA-trnH AY138106. s.c. 04714258 (MO):ITS AF426369. R.manshuricum Komarow ArnoldArboretum 67– 7: ETSAY138038; psbA-trnH AY138107. Chase 3599 (K): ITS AF426320. R. petraeum Wulfen Chase 3604 (K): ITS AF426318. R. rubrum L.ArnoldArboretum 1119– 78A: ETS(1): AY138039; psbA-trnH AY138108. ArnoldArboretum 214– 96: ETS(2): AY138040. Schuhwerk 7039 (NY): ITS AF426321. R. sativum (Rehbeh)Syme Missouri Botanical Garden: ITS AF426323. cv.’Diploma’NPGR 747 (Thompson46; OSC):restriction sites. R. spicatum Robs. Chase 3609 (K): ITS AF426322. R. triste Pallas UCBG 94.1114: ETS(1): AY138041. ArnoldArboretum 407– 94: ETS(2): AY138042; psbA-trnH AY138109. Mes- singer 314 (OSC):restriction sites. Baldwin 2269 (WS): ITS AF426319. Sect. Symphocalyx Berland. (Golden Currants) R. aureum Pursh UCBG 56.0948: ETS(1): AY138030. ArnoldArboretum 460– 81A: ETS(2): AY138029; psbA-trnH AY138101. NPGR 769: restrictionsites. Clement 26 (WS): ITS AF426382. R. odoratum Wendl. UCBG 86.0126: ETS(1): AY138032. ArnoldArboretum 1192– 74A: ETS(2): AY138031; psbA-trnH AY138102. NPGR 691: restrictionsites. Subg. Grossularia (Mill.)Pers. Sect. Grossularia (Mill.)Nutt. R.aciculare Sm. A.K. Skvortsov sn (A):ETS A Y137976; psbA-trnH AY138060; ITSA Y138050. R. alpestre Wall. ex Decne. Boufford etal. 28437 (A):ETS A Y137975; psbA-trnH AY138059. Chase 3586 (K): ITS AF426349. R.burejense F.SchmidtNPGR 259.001: ETSA Y137977; psbA-trnH AY138061; restrictionsites. Messinger 260 (OSC): ITS AF426355. R. curvatum J.K. Small R.Kral 53094 (GH): ETSA Y137979. R.cynosbati L.ArnoldArboretum 1108– 78A: ETS(1): AY137981; psbA-trnH AY138063; ITSAY138051. Schultheis 590–00 (YU): ETS(2): AY137980. R.divaricatum Douglas UCBG 85.1596: ETS(1): AY137982; psbA-trnH AY138064. Chase 3591 (K): ETS(2): AY137983; ITS AF426347. R.echinellum (Coville) Rehd. ArnoldArboretum 234– 96: ETSA Y137984; psbA-trnH AY138065; ITSAY138052. R.formosanum Hayata Wang 1096 (MO):ETS (1): AY137985; ITS AF426354. RBGE 19934301: ETS(2): AY137986; psbA-trnH AY138066. R.grossularioides Maxim. RBGE 19592035: ETSA Y137987; psbA-trnH AY138067; ITSA Y138053. R.hirtellum Michx. UCBG 86.0125: ETS(1): AY137988. Schultheis 588–00 (YU):ETS (2): AY137989; psbA-trnH AY138068. Chase 3595 (K): ITS AF426353. R. inerme Rydb. Layser 1770 (UW):ETS (1): AY137990; ITS AF426356. ArnoldArboretum 225– 80A: ETS(2): AY137991; psbA- trnH AY138069. R.missouriense Nutt. W.R. Smith 11396 (GH): ETSAY137993; psbA-trnH AY138071. Solomon 859 (WS): ITS AF426348. R. niveum Lindl. Messinger 226 (OSC): ETS AY137994; psbA-trnHA Y138072 ; ITS AF426351. NPGR 777.001 ( Messinger 226 ; OSC):restriction sites. R.oxyacanthoides L. subsp. oxyacanthoides NPGR 139.001: ETS(1): AY137996. Alverson 1645 (WIS):ETS (2): AY137995; psbA-trnH AY138073. Peterson& Annable4440 (WS): ITS AF426383. subsp. cognatum (Greene)Sinnott Woodland s.n. (WS): ETS AY137978; psbA-trnHA Y138062 ; ITS AF426357. subsp. irriguum (Douglas)Sinnott RBGE 19794064: ETSA Y137992; psbA-trnH AY138070; ITSAY138054. NPGR 773.001 (Messinger 221 ;OSC):restriction sites. subsp. setosum (Lindl.)Sinnott Arnold Arboretum 1395– 83A: ETSA Y137998; psbA-trnH AY138075. Lackschewitz 7845 (UW): ITS AF426352. R.rotundifolium Michx. Mitchell &Barbour10,112 (NYS):ETS AY137997; psbA-trnH AY138074; ITSAY138055. R.stenocarpum Maxim. RBGE 19698970: ETSA Y137999; psbA-trnH AY138076; ITSAY138056. R.uva-crispa L.UCBG 87.0719: ETS(1): AY138001. ArnoldArboretum 1404– 80B: ETS(2): AY138000; psbA-trnH AY138077. Hill 20953 (NY) ( 5 R.grossularia L.): ITS AF426350. Subg. Hesperia A. Berger R. amarum McClatchie UCBG 89.1081: ETSA Y138002; psbA-trnH AY138078. R.californicum Hook. and Arn.UCBG 82.1692: ETSAY138004; psbA-trnH AY138080; ITSA Y138057. R.menziesii Pursh Messinger 233 (OSC): ETS AY138006; psbA-trnHA Y138082 ; ITS AF426364. NPGR 769.001 ( Messinger 233 ; OSC):restriction sites. R. roezli Regel Nelson s.n. (WS): ETS AY138007; psbA-trnHA Y138083 ; ITS AF426365. var. cruentum (Greene)Regel NPGR 772.001 ( Messinger 217 ;OSC):restriction sites. Subg. Lobbia A. Berger R.binominatum A.Heller Messinger 260 (OSC): ETS AY138003; psbA-trnHA Y138079 ; ITS AF426359. NPGR 867.001 ( Messinger 260;OSC):restriction sites. R. lobbii A.Gray UCBG 85.1496: ETSAY138005; psbA-trnH AY138081. Smith Jr.7443 (WS): ITS AF426361. R.velutinum Greene Annable2503 (NY): ETS AY138011; psbA-trnHA Y138085 ; ITS AF426358. NPGR 865 (Messinger 255.1 ; OSC):restriction sites. R.watsonianum Koehne Patrick s.n. (UW): ETS AY138012; psbA-trnHA Y138086 ; ITS AF426360. Sect. Robsonia Berland. R.speciosum Pursh UCBG 84.0004: ETS(1): AY138008; psbA-trnH AY138084. Mort 1371:WS:ETS (2): AY138009; ITS AF426362. NPGR 901.001: restrictionsites. R.thacherianum Johnson s.n. (WS): ETS AY138010; ITS AF426363. 82 SYSTEMATIC BOTANY [Volume 29

TABLE 1. Continued.

Outgroups Iteavirginica L.New YorkBotanic Garden 76/98C: ETSA Y138045; psbA-trnH AY138110. Messinger 337 (OSC):restriction sites. Ware 94 (WS): ITS AY231368. I.ilicifolia OliverUCBG 86.0414: ETSA Y138046; psbA-trnH AY138111; ITSAY138058.

fromSenters and Soltis2003). Itis possible thatincongruence be- sularia,but relationshipswithin this clade were leftunresolved ,as tween ETSand ITSdatasets was due tointraspeciŽ c variation, an were relationships amongall remainingspecies. Comparisons as yetunexplore dissue within Ribes that maylimit future use of were between treesfrom constrained branch and bound searches composite accessions. The presence ofETS and ITSregions on the forthe chloroplastda ta, and fromconstrained heuristic searches same transcript, similarGC contents(ETS 5 mean 0.49; ITS 5mean forthe rDN Adata. 0.58), similarproportion sofinv ariant characters (ETS 5 297/462; AncestralArea Reconstru ction. Ancestral areas were inferred ITS5438/700), and similarlevels ofra te heterogeneityacross sites usingD IVA, which assignsareas tointernal nodes such that dis- (ETS: gamma 5 0.65; ITS:gamma 5 0.57) argued against different persal and extinction eventsare minimized(Ronquist 1996, 1997). evolutionary processes ofsequence evolution as an explanation for Default costswere used (vicariance 5 0, dispersal 5 1, extinction incongruence. 51), and the numberof inferred areas per node was leftuncon- Inorder to maximize taxon samplingand data, testswere also strained. These analyses used treesfrom the combined psbA-trnH used toassess combinabilityof taxa represented inthree of the and rDNAdatasets (Fig.8), withspecies assigned toone ormore foura vailable datasets. While thisrequired the inclusion oftaxa offour areas (easternN orthAmerica, westernN orthAmerica, forwhich one dataset was coded as missing,the decrease inphy- eastern Eurasia, and westernEurasia). Geographic distributions logenetic accuracy was expected tobe insigniŽcant (Weinsand followed Berger(1924), Janczewski(1907), and Sinnott(1985). Sect. Reeder 1995). Calobotrya, R. aureum, R. bracteosum , R. watsonianum ,and the clade Phylogenetic Analysisof Sequence Data. Analysesof single containing theremaining glabrous-sty led gooseberries( Hesperia, (ITS,ETS, and psbA-trnH)and combinedda tasets were performed Lobbia, Robsonia)were all coded withwestern N orthAmerican dis- usinga parsimonycriterion in P AUP* 4.0. Allsites and all changes tributions.Sect. Grossularioides was coded forboth N orthAmerica were equally weighted, incontrast tothe 1.3 gains:1.0 lossw eight- and eastern Eurasia. Subg. Berisia was coded forboth eastern and ingscheme used byM essingeret al. (1999) intheir analyses of westernEurasia. Ribes triste was coded forall areas except western restrictionsite data. psbA-trnH data were analyzed bothwith in- Eurasia. Itea was coded forboth eastern NorthAmerica and east- dels treated as missingdata, and withsev en large indelsincluded ernEurasia. as binarycharacters .Positions309– 327 and 448–465 ofthe aligned Because DIVAanalyses do notpermit polytomies ,we treated R. psbA-trnH dataset were excluded fromall analyses due toambig- aureum and R. bracteosum as sistertaxa. Withinsect. Grossularia , uous alignments. membersof the Asian gooseberryclade were all coded as eastern Analysesof the ITS,ETS, and psbA-trnH datasets consisted of Eurasia. The remainingspecies ofthe large polytomyall live in heuristic searches with100 replicatesof random addition, andTBR westernor eastern NorthAmerica. Weresolved thispolytomy in branch swapping. Nomorethan 400 treesgrea terthan aspeciŽed several differentw ays, soas toexplore someof the possibilities. length(234 forETS; 556 forITS; 137 for psbA-trnH) were saved Forexample ,insome analyses wecreated one eastern North per replicate,orin the case of psbA-trnH withindels, no morethan American and one westernN orthAmerican clade. Inothers, the 250 treesgreater than length144. The limitingtree length w as NorthAmerican species were arranged pectinatelyinrelation to determinedbased on theshortest trees found inprior, incomplete theAsian clade. searches, and was imposed toprev ent computational overload searchingsub-optimal trees. Analysesof all datasets combinedor ofcombined c hloroplast datasets were conducted withthe branch- RESULTS and-bound algorithm,with taxa excludedbased on partition ho- mogeneitytest results. Fordatasets combiningETS 1ITS, SequenceChara cteristicsandAnalyses. The aligned ETS1psbA-trnH,psbA-trnH 1rDNA, orall data forthose taxa rep- resentedin three of the four datasets, analyses consisted ofheu- 18–26S rD NAITSd ataset,in cluding 5.8S(positions 283– risticsearches with100 replicates ofrandom addition, TBR branch 452),w as700 bp (282bp 5ITS1; 248bp5ITS2)with 2.8% swapping, and maxtreesset to 40,000. ofth edatam atrixcod ed asmissing data.F or the region Clade support was assessed usingbootstrap analyses (Felsen- asa whole,sequences ranged from 438–662bp (v aria- stein1985) as implemented inP AUP* 4.0 with300 replicatesof heuristic searches, each with5 replicatesofrandom taxon addition tion due in partto missing terminal sequence data), and Nearest NeighborInterchange (NNI) branch swapping. For uncorrected pairwise distances within Ribes ranged the combinationofall fourda tasets, and forthe combinedchlo- from 0.0–0.08, distances with Itea ranged from 0.27 roplast datasets, clade support was assessed using100 bootstrap replicateseach withbranch and bound searches. (with R. missouriense )to0.32 (with R. malvaceum ), and Kishino-HasegawaTests. The parsimonyimplementationof the mean GCcontent was0.58. Of the 700aligned base- Kishino-Hasegawatest(Kishino and Hasegawa1989) inP AUP* 4.0 pairs,528 were constant within Ribes,and 183w ere was used tocompare thehypothe sisof subg. Grossularia monophyly totha tofsubg. Grossularia non-monophyly,inw hichsect. Grossu- parsimonyinformative.ITS1display ed higher levels of larioides and sect. Grossularia forma clade.The parsimonyversionof divergenceand ahigher proportion ofinformativ e the testw as used because comparisonsinclu ded restriction siteda ta. charactersthan did ITS2(ITS1 5 0.0–0.15, 65/ 282in- Asrecomme nded byGoldma net al. (2000), comparisonsw ere be- formative; ITS2 5 0.0–0.11, 25/ 248informativ e).In- tween apriorih ypothesesbased on previouslyp ublished phyloge- nies(M essingeret al. 1999; Sentersand Soltis2003). cluded ITS1sequences ranged from 34–258bp, with the Analysesw ere run enforcingeach oftw oconstraint trees. The lower limit due totaxa for which muchof ITS1 se- Žrstconstraint treeresolved subg. Grossularia as monophyletic, but quence datawas missing. ITS2sequence sranged from leftrela tionshipswithin subg. Grossularia and amongall remain- ingspecies unresolved. The second constraint treeenforced res- 216–241bp .5.8Sranged from 161–164bp ,butw asonly olution ofa clade containing sect. Grossularioides plus sect. Gros- 142bp in R. sanguineum due toa large deletion. 2004] SCHULTHEISAND DONOGHUE:PHYLOGENY AND BIOGEOGRAPHY OF RIBES 83

Lengths and GCcontents fell within the ranges re- conict with amonophyletic subg. Grossularia (Fig. 3; ported fora surveyofangiosperm taxa(Baldwin et al. trees from analyses including indels are not shown). 1995).Analyses ofthe ITSand 5.8Sregions generated Twomain lineages may beidentiŽ ed within subg. 36,400trees (length 5556; CI50.65,0.58 excludin gun- Grossularia ,the true gooseberries (sect. Grossularia ) and informativecharacters;RI 50.84). the glabrous-styled gooseberries ( Hesperia, Lobbia, Rob- AmpliŽed productsfrom the 18–26S rDN Ainter- sonia).The glabrous-styled gooseberries are resolved as genic spacerregion were approximately 2kbin R. odor- acladein ITS,restriction site,and psbA-trnH trees atum (sect. Symphocalyx ) and 3kb in R. hirtellum (sect. (Figs. 1,3; M essinger et al.1999). ETS trees alsoresolve Grossularia ),the twospecies used todesign internal the glabrous-styled gooseberries ( Hesperia, Lobbia, Rob- primers foram plifying 3 9 ETS. Only 406bpof the 3 9 sonia)asa cladewith the exception of R. watsonianum end ofthe ETS region and 56bpof the 5 9 end of 18S (subg. Lobbia),placed with strong supportin the true were sequencedin this study. Included sequences gooseberryclade(sect. Grossularia )(Fig. 2).The integ- ranged from 299–462bp in size,with some variation rity ofthe individual sections comprising the glabrous- due tomissing terminal sequence data,and had a styled gooseberries ( Hesperia, Lobbia, Robsonia) is un- mean GCcontent of0.49. U ncorrected pairwise dis- clear.The true gooseberries (sect. Grossularia ) are re- tances among ETS sequenceswithin Ribes ranged from solved asa cladein restriction site,ITSand ETS trees 0.0to 0.07. P airwise distances between Ribes and Itea (including R. watsonianum in the latter),and in psbA- ranged from 0.26(with R. oxyacanthoides ,accession 2) trnH trees when indels are included (Figs. 1,2; M es- to0.32 (with R. triste,accession 1).Of the 462aligned singer et al.1999; Senters and Soltis 2003).If indels are basepairs,388 w ere constantwithin Ribes,60w ere par- ignored, psbA-trnH trees resolvethe majority oftrue simony informative,and 2%w ere coded asmissing gooseberries asa clade,and are consistent with a data.ETS analyses generated34,005trees (length 5234; monophyletic sect. Grossularia (Fig. 3).W ithin sect. CI50.80,0.74 excluding uninformative characters; Grossularia ,the ETS datasetprovided strong support RI50.93;Fig. 2). foran Asian clade,including the European R. uvacris- The aligned psbA-trnH intergenicspacerdataset was pa.ITStrees were consistent with anAsian clade,but 502bp,including a3 9 portion of psbA (positions 1–37) excluded R. uvacrispa . psbA-trnH trees were alsocon- and a 59 portion of trnH (position 484–502). One per- sistent with anAsian clade,with the weakly support- cent ofthe matrix wascoded asmissing data.Included ed inclusion ofthe North American R. oxyacanthoides sequencesranged from 355–445bp in length, with subsp. irriguum.Restriction site trees had limited sam- some variation due tomissing terminal sequence data, pling in sect. Grossularia ,butw ere consistent with an and had amean GCcontent of0.29. The spacerregion Asian versus North American divergence(M essinger alone ranged from 352–378bp within Ribes, and was et al.1999). 369bp in Itea,based on complete sequences. Uncor- All datasets exceptthe restriction sites resolved a rected pairwise distances (including psbA and trnH cladecomprising sect. Calobotrya, sect. Grossularioides , terminals) within Ribes ranged from 0.0to 0.04. Dis- and subg. Grossularia (sect. Grossularia , sect. Robsonia, tances between Ribes and Itea ranged from 0.15(with subg. Lobbia, Hesperia).M embers ofthis cladeare large- R. odoratum,accession 2)to 0.21 (with R. bracteosum ). ly from western North America, with sect. Grossularia Within Ribes,405/465included charactersw ere con- having awider distribution. ETS trees included the stant,and 34/465were parsimonyinformative.Ex- blackcurrant Ribesbracteosum (sect. Coreosma) within cluded charactersrepresen ted apolyA region ranging this clade,in anunresolv ed position (Fig. 2).ITS trees from 4–13bp, and a16–18bp insertion in Itea (relative included the SouthAmerican sect. Parilla in this clade, to Ribes)with which 5–6bp segments of Ribes could be in anunresolve dposition (Senters and Soltis 2003).In aligned in variousw ays. Analyses ofsequence data alltrees, sect. Calobotrya wasresolv ed asm onophyletic, from the psbA-trnH intergenicspacerproduced 34,000 including some members ofsect. Heritiera in restriction trees with little homoplasy(length 5137; CI50.92, 0.88 site and ITStrees (only one species ofsect. Heritiera excludinguninformative characters;RI 50.94;Fig. 3). wasincluded in ETS and psbA-trnH analyses, and falls Analyses including indels coded asbinary characters elsewhere). Section Grossularioides wasresolved with produced 18,001trees (length 5144; CI50.92,0.90 ex- strong supportin ETS and ITStrees (Figs. 1,2; Senters cluding uninformative characters;RI 50.95). and Soltis 2003),but is unresolved or strongly para- TreesGen eratedby Analysesof Individual phyletic in the psbA-trnH and restriction site trees re- Datasets. Results from the expanded ITSdataset spectively (Figs. 1,3; M essinger et al.1999). concurwith those presented bySenters and Soltis The cladecontaining sects. Calobotrya, Grossularioides (2003)and are not illustrated here.Subg. Grossularia and subg. Grossularia wasclearly nested within Ribes wasresolved asmonoph yletic in ITSand ETS trees, in ITS,ETS and psbA-trnH trees (Figs. 2,3; Senters and with weak tomoderate bootstrapsupport (Figs. 1,2). Soltis 2003).Among the remaining sections of Ribes, psbA-trnH trees showed little resolution,butdid not alldatasets provided strong supportfor sect. Berisia, 84 SYSTEMATIC BOTANY [Volume 29

FIG.2. The majority-ruleconsensus of34,005 treesresulting from analyses ofETS sequence data (CI 50.80, 0.74 excluding uninformative characters; RI 50.93). Numbersabove the branches indicate the percentage ofbootstrap replicates inwhich the clade appeared. Numbersin parentheses below the branches indicate the percentage oftrees in which the clade appeared, when lessthan 100%. 2004] SCHULTHEISAND DONOGHUE:PHYLOGENY AND BIOGEOGRAPHY OF RIBES 85

FIG.3. The majority-ruleconsensus of34,000 treesresulting from analyses of psbA-trnH sequence data (CI 50.92, 0.88 excluding uninformative characters; RI 50.94). Numbersabove the branches indicate the percentage ofbootstrap replicates in which the clade appeared. Numbersin parentheses below the branches indicate the percentage oftrees in which the clade appeared, when lessthan 100%. 86 SYSTEMATIC BOTANY [Volume 29 with the inclusion ofsome species ofsect. Ribes in ITS trees (Senters and Soltis 2003).Section Symphocalyx was strongly resolved in ETS and restriction site trees (Fig. 2;M essinger et al.1999), but w asunresolve din psbA- trnH trees (Fig. 3)(ITS trees included only one species). Section Heritiera appearedto be polyph yletic in ITS and restriction site trees (only one species ofsect. Her- itiera is included in ETS or psbA-trnH analyses) (Mes- singer et al.1999; Senters and Soltis 2003).The mono- phyly ofsects. Ribes and Coreosma remain unclear.For those species sampled, sect. Ribes wasresolved as monophyletic in restriction site trees (Messinger et al. 1999),as m onophyletic excluding R. triste in ETS and psbA-trnH trees (Figs. 2,3), and aspolyphyl etic in ITS trees, with some species clearly related tosect. Berisia (Senters and Soltis 2003).Section Coreosma was re- solved aspolyph yletic in ETS and restriction site trees (Fig. 2,M essinger et al.1999), and aspara- or poly- phyletic in ITStrees (Senters and Soltis 2003)(only one species wasincluded in the psbA-trnH dataset).Pre vi- FIG.4. The strictconsensus ofsix trees resulting from ousrestriction site analyses of18– 26S rDN Asuggested analyses of psbA-trnH sequence data combinedwith chloro- thatmembers ofsect. Coreosma were divergentfrom plast restrictionsite data (Messingeret al. 1999) (CI 50.90, 0.79 one another and from other Ribes species (Messinger excluding uninformative characters; RI 50.87). Data combina- et al.1993). Sampling from sect. Parilla wastoo limited bilitydetermined by partition homogeneitytests (p 50.66). Numbersabove the branches indicate the percentage ofboot- todraw strong conclusions, butthe ITStrees support- strap replicates inwhich the clade appeared. ed its monophyly tothe exclusion of R. fasciculatum , the only non-SouthAm erican member ofthe section (Senters and Soltis 2003).Relationsh ips among the sec- Supportfor sect. Grossularia , sect. Calobotrya, and the tions outside ofthe predominantly western North glabrous-styled gooseberries ( Hesperia, Lobbia, Robsonia) American clade(sects. Calobotrya, Grossularioides , subg. increased relative toanalyses ofeither datasetalone Grossularia )varied markedly among datasets,with lit- (Fig. 3;M essinger et al.1999). Support for a paraphy- tle supportfor any one resolution(Figs. 2,3). letic sect. Grossularioides decreased slightly relative to TreesGenerat ed fromAnalysesof Combined the restriction site analyses (Fig. 4;M essinger et al. Datasets. Datasetcombinabilit yis typically based on 1999). acriticalv alueof p 5 0.05(i.e .,datasets are incongru- Analyses ofthe combined ETS and ITSrD NAdata ent when p , 0.05).H owever,studies suggest thata (p50.04;excluded R. fasciculatum , R. glandulosum , R. criticalvalue of p 5 0.05may betoo strict such that himalense, R. manshuricum , R. nevadense , R. rubrum, R. the null hypothesis ofdataset congruence will befalse- oxyacanthoides subsp. oxycanthoides , R uvacripsa) gener- ly rejected (Huelsenbeck et al.1996; Sulliv an1996; ated 150trees (length 614;CI 50.77,0.70 excludin gun- Cunningham 1997),and thata valuesuch asp 5 0.01 informativecharacters;RI 50.85)(Fig. 5).Subg. Gros- may bem ore appropriate(Cunningham 1997).Many sularia wasm onophyletic,as w ere thetrue gooseber- ofthe datasets reported here were not combinable ries (sect. Grossularia )(Fig. 5).The glabrous-styled based on intitial partition homogeneity tests, butthe gooseberries ( Hesperia, Lobbia, Robsonia)were strongly incongruence wasgenerally eliminated when problem- resolved asm onophyletic with the exception of R. wat- atictaxa were excluded. sonianum (sect. Lobbia),placedstrongly assister tothe Analyses ofthe readily combinablechloroplas t psbA- true gooseberries. Within sect. Grossularia , only the trnH and restriction site data(p 50.66)generated six Asian cladew asresolved with strong support.The trees (length 144;CI 50.90,0.79 excludin guninforma- cladecontaining subg. Grossularia , sect. Grossularioi- tive characters;RI 50.87).In agreement with analyses dies, and sect. Calobotrya wasagain evident, with sect. ofrestriction site dataalone ,true gooseberries (sect. Grossularioides resolved assister tosubg. Grossularia . Grossularia )were not placedas sister tothe remaining Supportfor the glabrous-styled gooseberries ( Hesperia, glabrous-styled gooseberries ( Hesperia, Lobbia, Robson- Lobbia, Robsonia), sect. Calobotrya, sect. Grossularioides , ia),suggesting thatsubg. Grossularia is not monophy- and the Asian cladewithin sect. Grossularia increased letic (Figs. 1,4). The true gooseberries were nested in relative tosupport from ETS or ITSdatasets analyzed aparaphyleticsect. Grossularioides ,asin the restriction individually (Figs. 2,5; Senters and Soltis 2003). site trees, with sect. Calobotrya weakly placedas sister . The most evident difference between chloroplast and 2004] SCHULTHEISAND DONOGHUE:PHYLOGENY AND BIOGEOGRAPHY OF RIBES 87

FIG.5. The strictconsensus of150 treesresulting from analyses ofcombined ETS and ITSsequence data (CI 50.77, 0.70 excluding uninformative characters; RI 50.85). Data combinabilitydetermined by partition homogeneitytests (p 50.04). Num- bersabove the branches indicate the percentage ofbootstrap replicates inwhich the clade appeared. 88 SYSTEMATIC BOTANY [Volume 29

FIG.6. The strictconsensus offour trees resulting from analyses ofthe fourdatasets combined(ETS, ITS, psbA-trnH, chloroplast restrictionsites). Data combinabilitywas determinedby partition homogeneitytests (p 50.75). Numbersabove the branches indicate the percentage ofbootstrap replicates inwhich the clade appeared. rDNAtrees wasthe position ofsect. Grossularioides . Robsonia).These groups again formed awell-support- Subgenus Grossularia is monophyletic in rDNAtrees, ed clade,nested within Ribes (Fig. 6).The strongly sup- with sect. Grossularioides placedw eakly assister (Fig. ported sister-grouprelationsh ip between the true and 5).In contrast,subg. Grossularia is polyphyletic in chlo- glabrous-styled gooseberries (Fig. 6)is consistent with roplasttrees, with sect. Grossularia nested within sect. the monophylyofsubg. Grossularia ,butcould change Grossulariodes , and sect. Calobotrya sister tothat group- with theinclusion ofsect. Grossularioides . ing (Fig. 4).Analyses ofthe combined chloroplast and ETS and psbA-trnH combined analyses (p 50.04; ex- rDNAdatasets (p 50.75;excluded R. glandulosum , R. cluded R. oxyacanthoides subsp. irriguum)generated329 lacustre, R. montigenum )generatedfourtrees (length trees (length 372;CI 50.84,0.77 excludin guninforma- 556; CI50.88,0.69 excludin guninformative characters; tive characters;RI 50.90)(Fig. 7).Subgenus Grossularia RI50.76)(Fig. 6).T axonsampling waslimited, due to wasresolved asmonophy letic, within which bothth e the limited overlap among datasets,but trees showed glabrous-styled and thetrue gooseberries were strong- strong resolutionofsect. Calobotrya, sect. Grossularia , ly supported, the latter including R. watsonianum and the glabrous-styled gooseberries ( Hesperia, Lobbia, (subg. Lobbia).The Asian cladewas w ell supported 2004] SCHULTHEISAND DONOGHUE:PHYLOGENY AND BIOGEOGRAPHY OF RIBES 89

FIG.7. The strictconsensus of329 treesresulting from analyses ofcombined ETS and psbA-trnH data (CI50.84, 0.77 excluding uninformative characters; RI 50.91). Dataset combinabilitywas determinedby partition-homogeneity tests (p 5 0.04). Numbersabove the branches indicate the percentage ofbootstrap replicates inwhich the clade appeared. within sect. Grossularia ,including the European R. uva- Combined analyses of psbA-trnH and rDNAdatasets crispa. Both sect. Calobotrya and sect. Grossularioides (p50.02;exclud ed R. oxyacanthoides subsp. irriguum and were strongly supported, butneither wasresolved as taxaex cluded from ETS 1ITSanalysis )generated 3080 sister tosubg. Grossularia .The cladecomprising subg. trees (length 5 719; CI50.80,0.74 exclud ing uninfor- Grossularia , sect. Grossularioides , and sect. Calobotrya mative characters; RI 50.85).Sects. Grossularioides , Calo- wasagain evident, nested within Ribes. botrya, Grossularia ,and the glabrous-styled gooseberries 90 SYSTEMATIC BOTANY [Volume 29

(Hesperia, Lobbia, Robsonia)were eachstrongl ysupport- result hinges on resolutionatthe base,which remains ed,andagain formed astronglysupported claden ested highly uncertain and requires more intensivesampling in Ribes (Fig. 8).Support for the monophyly ofsubg. of Ribes lineages outside ofsubg. Grossularia . Grossularia wasincrease drelativetosupport seen with anyindividual dataset(Figs. 2,3, 8). DISCUSSION Combinability (p 50.12)of all datasets forthose taxa GooseberryM onophyly. Results from previous represented in three ofthe fourdatasets required ex- phylogeneticanalyses based on nuclear18S– 26S rD NA clusion ofcomposite taxaproblema ticwithin the ITSdata (Senters and Soltis 2003)and chloroplast re- rDNAdataset,as w ell asexclusion oftaxa problema tic striction site data(M essinger et al.1999) differed re- forthe chloroplast versus rDNAdatasets.Combined garding gooseberry monophyly. Adding the additional analyses generated 4,802trees (length 5768; CI50.80, nuclear18S– 26S rD NAETS and chloroplast psbA-trnH 0.74excludin guninformative characters;RI 50.85).Sec- datasets couldincrease supportand resolutionin re- tion Calobotrya, sect. Grossularia ,and the glabrous- sulting topologies,or couldreinforce the differences styled gooseberries ( Hesperia, Lobbia, Robsonia) exclud- between thenuclearand chloroplast topologies,with ing R. watsonianum were eachw ell supported, and to- eachre ecting accuratebut separate histories. gether formed astrongly supported clade(Fig. 9).Sub- Initial examinationofthe trees produced with the genus Grossularia wasresolved asmonophyletic, additional datasets reinforced the apparentcon ict be- bearing in mind thatsect. Grossularioides wasexclude d tween nuclearand chloroplast genomes. ITS,ETS, and from the analyses. combined rDNAdatasets supported the monophyly of While allcombined datasets resolved the cladecon- subg. Grossularia (Figs. 1,2, 5). In contrast,the com- taining sects. Calobotrya, Grossularioides , and subg. bined chloroplastdatasets (restriction sites plus psbA- Grossularia asnested within Ribes (Figs. 4–9), relation- trnH)suggested acloser relationship between the true ships among other lineages were unclear,asin the and glabrous-styled gooseberries than did the restric- analyses ofindividual datasets (Figs. 2,3). Basal rela- tion site dataalone (Fig. 1;M essinger et al.1999), but tionships were unresolvedin the analyses ofcombined still resolved subg. Grossularia asnon-mon ophyletic, chloroplast data(Fig. 4).The combined ETS and psbA- with sect. Grossularioidies forming agrade atthe base trnH analyses suggested basalpositions forgolden of sect. Grossularia (Fig. 4).The increase in tree length currants (sect. Symphocalyx )(Fig. 7).All remaining when combined chloroplast datawas constrained to analyses suggested basalpositions forsects. Berisia resolve amonophyletic subg. Grossularia (p50.03) or and Ribes (Figs. 5,6, 8, 9). Basal relationsh ips were not when combined rDNAdatawas constrained toresolve well-supported in any analysis. sects. Grossularioides and Grossularia as a clade Kishino-HasegawaTests. Trees generated from (p50.05)also indicated datasetincongrue nce. analyses constraining the monophyly ofsubg. Gros- The discrepancy between nuclearand chloroplastto- sularia versus trees constrained toresolv easister re- pologies primarily involved the restriction sites dataset lationship between sect. Grossularioides and sect. Gros- and not the psbA-trnH dataset.The psbA-trnH dataset sularia were marginally,butsigniŽ cantly different in provided little resolution (Fig. 3),con icting with nei- boththe chloroplast (p 50.03)and the rDNA(p 50.05) ther the rDNAnor the restriction sites topologies. datasets. Analyses combining the psbA-trnH datawith either the AncestralArea R econstructions. DIVAreconstruc- ETS data (p50.04;Fig. 7)or the rDNAdata(p 50.02; tions ofancestral areassupport a western North Amer- Fig. 8)produced trees resolving subg. Grossularia as icanorigin fora large segment of Ribes, beginning monophyletic with increased supportrelative toETS with node 2in Fig. 8.W ithin this cladeit is alsothe or rDNAdataalone (Figs. 2,5). Section Grossularioides casethat subg. Grossularia is inferred tohave diversi- waseither sister tosubg. Grossularia (psbA-trnH 1 Žed Žrst within western North America(Fig. 8,node rDNA;Fig. 8)or wasunresolve din apolytomy with 6).Subsequen tmovement toeastern Asia is inferred to subg. Grossularia and sect. Calobotrya (psbA-trnH 1 have occurred somewhere within sect Grossularia , but ETS; Fig. 7).The monophyly ofsubg. Grossularia was the ancestral areafor the true gooseberries is equivo- alsosupported byanalyses ofall datasets combined, cal,and caninclude virtually anycombinationofareas whether including taxarepresent ed in allfour datasets dependingon theexactresolutio nofthe large poly- (p50.75;Fig. 6)or in three ofthe fourdatasets tomy atthe baseof this clade(Fig. 8,node 8).Im por- (p50.12;Fig. 9).Analyses combining alldatasets re- tantly,ourresults indicate thatthe Asian species with- quired exclusion ofsect. Grossularioides ,thus leaving in sect. Grossularia form aclade,implying asingle mi- subg. Grossularia monophylyuncertain. However, in- gration from North America toAsia. The ancestral cluding sect. Grossulariodes in spite ofdataset incon- area for Ribes is reconstructedtobe w estern North gruence (p50.01)produced trees resolving amono- Americaand western Eurasia,or western North Amer- phyletic subg. Grossularia sister toa monophyletic sect. icaand allof Eurasia (Fig. 8,node 1).H owever, this Grossularioides (notshown). 2004] SCHULTHEISAND DONOGHUE:PHYLOGENY AND BIOGEOGRAPHY OF RIBES 91

FIG.8. The strictconsensus of3,080 treesresulting from analyses ofcombined psbA-trnH and rDNAdata (CI 5 0.80, 0.74 excluding uninformative characters; RI 5 0.85). Dataset combinabilitywas determinedby partition homogeneitytests (p 50.02). Numbersabove the branches indicate the percentage ofbootstrap replicates inwhich the clade appeared. DIVA(Ronquist1996) was used toinfer ancestral areas at the numberednodes. The ancestral areas at nodes two throughseven were reconstructed as westernNorth America. Node one was reconstructed as westernNorth America plus eitherwestern Eurasia orall ofEurasia. There were multiple possible reconstructionsat node 8, including (1)western North America and eastern Eurasia, (2)North America and eastern Eurasia, (3)North America, or(4) western North America. 92 SYSTEMATIC BOTANY [Volume 29

FIG.9. The strictconsensus of4,802 treesresulting from analyses ofall fourdatasets (ITS,ETS, psbA-trnH, chloroplast restrictionsites), including taxa represented inat least threedatasets (CI 50.80, 0.74 excluding uninformative characters; RI50.85). Dataset combinabilitywas determinedusing partition homogeneitytests (p 50.12). Numbersabove the branches indicate the percentage ofbootstrap replicates inwhich the clade appeared. 2004] SCHULTHEISAND DONOGHUE:PHYLOGENY AND BIOGEOGRAPHY OF RIBES 93

The general picturethat em erges from these analy- posses multiple rDNArepeat types, with different re- ses suggests thatsubg. Grossularia is monophyletic peattypes sampled in ourITS v ersus ETS datasets. (Figs. 2,5– 9). rD NAdatasupported gooseberrymono- Eachof these possibilities has been reported in other phyly (Figs. 2,5). psbA-trnH data,while lacking sufŽ- planttaxa (W endel et al.1995a; Cam pbell et al.1997 cient variation toresolve gooseberry monophylyor forpotential recombinantrDNArepeat types: Suhet non-monophyly,supported gooseberry monophyly in al.1993; W endel et al.1995b for incom plete homoge- combinationwith rDNAdata(Figs. 7,8). Finally ,anal- nization ofrD NArepeat types). yses ofall datasets combined supported gooseberry The true gooseberries (sect. Grossularia )have abroad monophyly,although these analyses excluded the range throughoutthe northern hemisphere, with nine spinycurrants (sect. Grossularioides )(Figs. 6,9). What species in North America (Sinnott 1985)and seven spe- requires further investigationis the role hybridization cies in Eurasia (Berger 1924).In his treatment ofN orth may have played in the history ofthe spinycurrants American gooseberries, Sinnott (1985)divided his nine (sect. Grossularioides )and the true gooseberries (sect. species into Žvegroups based on phenetic analyses. Grossularia ).M essinger et al.(1999) en couraged further The groups consisted of(1) R. cynosbati , with spiny exploration ofthis possibility in Ribes,in which they fruits, (2) R. niveum, R. curvatum, and R. missouriense noted potential chloroplast captureof th esect. Gros- with highly exserted stamens and long sepals, (3) R. sularia chloroplast type bysect. Grossularioides . Crosses rotundifolium and R. divaricatum ,with medium length have not been successfulbetw een species ofsect. Gros- stamens and purple sepals, (4) R. inerme and R. hirtel- sularioides and other sections within Ribes,and are usu- lum,with intermediate length stamens and small oral ally unsuccessfulbetw een sections or subgenera (Keep features,and (5)the Žve subspecies of R. oxyacanthoi- 1962).ConŽ rmation ofthe pattern seen in trees gen- des,with short stamens. Sinnott (1985)recommen ded erated from the chloroplast restriction site data(M es- that R. echinellum beexclude dfrom the true gooseber- singer et al.1999) is needed from additional chloro- ries and aligned with the glabrous-styled gooseberries plastdatasets thatprovide greater resolutionthan did ofw estern North America( Hesperia, Lobbia, Robsonia). the psbA-trnH dataset. The analyses presented here showed strong support Relationships Withinth eGooseberries. Subgenus forthe inclusion of R. echinellum asw ell asth eEurasian Grossularia comprises twomain lineages (Figs. 2,5– 9), species within the true gooseberry lineage (Figs. 2–3, the true gooseberries (sect. Grossularia )and acladeof 5,7– 9). Resolutio namong the North American species glabrous-styled gooseberries ( Hesperia, Lobbia, Robson- waslacking or not well-supported (Figs. 2–9) and thus ia).The fourgooseberry sections are traditionallydis- neither supported nor conicted with Sinnott’s hy- tinguished from eachother bybasally pubescent styles pothesized relationships. in sect. Grossularia (with the exception ofsome Asian The seven Eurasian species ofsect. Grossularia (Ber- species), byfour v ersus Žveparted owers in sect. Rob- ger 1924)include R. uvacrispa in Europe, R. alpestre and sonia,byanthers broader atthe basein subg. Hesperia, R. aciculare in the Himalayanregion, R. stenocarpum and bythe absenceof the aboveset offeatures in subg. and R. formosanum in southern and eastern Asia, and Lobbia.The glabrous-styled gooseberry lineage,com- R. burejense and R. grossularioides in northern and east- prising 27species distributed through western North ern Asia. Acladeof Asian species is well-supported America into Mexico(Berger 1924),w asw ell-support- (Figs. 2,5, 7– 9), with possible inclusion of R. uvacrispa ed given oursampling, butsections within this lineage (Figs. 2,7). The psbA-trnH datasetincludes R. oxyacan- were not resolved. Ribesw atsonianum (subg. Lobbia) thoides subsp. irriguum within the Asian clade(Fig. 3), wassurprisingl yplacedoutside ofthe glabrous-styled butthis is weakly supported (Fig. 3)and requires fur- gooseberrylineage,assister tothe true gooseberries ther substantiation, particularlysince R. oxyacanthoides (Figs. 2,5, 7– 9). ITS data resolv ed R. watsonianum with- subsp. irriguum is restricted tonorthw estern North in the glabrous-styled gooseberry lineage (Senters and America and may itself beof h ybrid origin from cross- Soltis 2003),but ETS datastrongly resolved the species es between R. oxyacanthoides subsp. setosum and R. in- within the true gooseberry lineage (sect. Grossularia ) erme (Sinnott 1985),both distributed in western North (Fig. 2).This conict within the rDNAdatamay reect America. Resolutionwithin the cladeof Asian species hybridizationbetween R. watsonianum and a member suggests acorrespondanceto geographic and oristic of sect. Grossularia followed byŽ xationwithin the R. regions within Asia. Whether the Asian andNorth watsonianum genome ofa recombinant rDNArepeat American true gooseberries represent divergentline- type.Accountsof Ribes hybrids in the Želd are rela- ages or amonophyletic Asian cladenested within a tively rare (Messler et al.1991) and do not implicate North American grade is unclear given the available R. watsonianum ,butfertile hybrids canbe obtained data. from artiŽcial crosses between glabrous-styled goose- OtherGroups within Ribes. Since this study fo- berries and true gooseberries (Keep 1962,citing R. lob- cused on the gooseberries, sampling within other sec- bii 3 R. divaricatum ).Alternatively, R. watsonianum may tions of Ribes wassometim es limited. Nevertheless, the 94 SYSTEMATIC BOTANY [Volume 29 datadid supportthe monophyly ofsects. Calobotrya, Grossularia ),and in eastern North America bythree Parilla, Symphocalyx , and Berisia. Section Calobotrya is a (somespecies of Heritiera, Ribes, and Grossularia ). One group of21 western North American species (Berger approachfor sorting through acomplexbiogeograph ic 1924)known asthe ornamental currants. The mono- history,asmight beexpected in Ribes,is toexamine phyly ofsect. Calobotrya wasw ell-supported,with the phylogeneticpatterns in componentclades and, ide- inclusion ofsome dwarf currants (sect. Heritieria) (Figs. ally,todate divergences. 2–9; M essinger et al.1999; Senters and Soltis 2003). Ouranalyses indicate thatsubg. Grossularia is nested Sampling waslimited in sect. Parilla,agroup of41 within apredominantly western North American dioecious SouthAm erican species (Janczewski 1907), clade(Figs. 2,5– 9), and thatit diversiŽed initially in butthe group wassupported in ITStrees, excluding thatregion (Fig. 8,node 6).Subsequent ly,within sect. the east Asian R. fasciculatum ,the section’sonly non- Grossularia ,it appearsthat m ovementoccurred from SouthAmerican species (Senters and Soltis 2003).A western North Americato eastern Asia, presumably possible sister relationship between sect. Parilla and through Beringia, followed byvicariance and the ori- sect. Calobotrya (unpubl.data cited in Weigend and gin and diversiŽcation of the well-supported Asian Binder 2001)is consistent with the position ofsect. Pa- gooseberry clade.Donoghue et al.(2001) and Xiang rilla seen with ITSdata (Senters and Soltis 2003).Sec- and Soltis (2001)highlight ed casesof moveme nt outof tion Symphocalyx ,the golden currants, includes Žve Asia into North America through the Bering Land species distributed through northern Mexico,and cen- Bridge. Ribes appearsto provide acaseof m ovement tral and western North America (Berger 1924).The in the oppositedirection. Movementin bothdirections golden currants were well-supported given the avail- is, ofcourse ,tobe expected. Whatremains tobe de- ablesam pling (Fig. 2,4, 9). The alpine currants, sect. termined in futurestudies ofdisjunct taxais exactly Berisia,are agroup of17 dioecious species distributed which groups moved in which directions, and whether through Eurasia (Janczewski 1907).The monophyly of there are anysigniŽcant generaliza tions thatcan be sect. Berisia wasw ell-supported (Figs. 2–9; M essinger made aboutthese different patterns. et al.1999), but with the inclusion ofsome red currants Elsewherein Ribes there are other possible casesof (sect. Ribes)in ITStrees (Senters and Soltis 2003). movement between North America and Asia. One case The monophyly ofsects. Ribes and Coreosma was involves R. lacustre of sect. Grossularioides ,which is dis- questionable (Figs. 2,3, 5). The red currants (sect. Ri- tributed in eastern Asia asw ell asin North America. bes)are agroup of15 Eurasian species, with one spe- Other possible casesrelate toresolutio natthe baseof cies, R. triste,alsodistributed in North America. Sec- Ribes,where relationships remain uncertain. The basal tion Coreosma,the blackcurrants, includes six species positions ofsect. Symphocalyx in trees resulting from in North America and six in Eurasia (Berger 1924).Sec- psbA-trnH dataand from the combinationof psbA-trnH tion Heritiera,the dwarfcurrants, appearedto be poly- and ETS datasuggest awestern North American ori- phyletic, asseen in Messinger et.al. (1999) and Senters gin forthe entire clade(Figs. 2,7). This would imply and Soltis (2003).This group ofsix species (Berger early dispersion toAsia, giving rise tosects. Berisia 1924)is deŽned primarily bya prostrate habit,and, as and Ribes.However,allother datasets and datasetcom- suggested byM essinger et al.(1999), is likely acaseof binations suggested abroader Eurasian plus western convergence. North American distribution atthe base(Fig. 8,node Biogeography. The spread oftaxa around the 1),owing tothe basalpositions ofsects. Berisia and Northern Hemispherehas been facilitatedby the avail- Ribes.This would imply early vicariancein volving ability atv arioustim es oftw omajor migration Asia and North America,and movement backto Asia routes—the Beringian and the North Atlanticland within the western North American clade.Future stud- bridges (Tiffney 1985a,1985b; Donoghue et al.2001; ies should include more sampling ofsects. Berisia, Ri- Sanmart´‡net al.2001). Especially in large and relative- bes, Coreosma and Symphocalyx tohelp resolve basalre- ly old clades, which have becomewidesprea daround lationships and biogeographicpatterns in Ribes. the Northern Hemisphere,it is possible (perhaps even The fossil record for Ribes consists largely ofleaves, likely) thatboth pathways were used. Informationon with few reports ofseeds (Kremenetski 1998),fruits the timing ofk ey divergences and on the direction of (Cevallos-Ferriz 1995),and owers (Gandolfoet al. movement within subcladeswill becritical in sorting 1998).The leafrecord for Ribes in North America may outthe possibilities in particularcases. Ribes provides extend from approximately 2mya(Hannibal1911; anexcellent example ofsuch agroup,being represent- Dorf 1930;Axelrod 1966)to at least 34.5mya (Mac- ed in Eurasia bysix sections ( Berisia, Ribes, and some Ginitie 1953;Mancheste r2001),and possibly to45 or species from Heritiera, Coreosma, Grossularioides , and 50mya (Axelrod 1998;W ehr and Hopkins 1994).Al- Grossularia ),in western North America bynine ( Sym- though leaffeatures are seldom used todistinguish phocalyx, Calobotrya, Robsonia, Hesperia, Lobbia, and extant taxaof Ribes,Wolfe(1964) used leafserration somespecies of Coreosma, Heritiera, Grossularioides , and and venation features todistinguish between fossil 2004] SCHULTHEISAND DONOGHUE:PHYLOGENY AND BIOGEOGRAPHY OF RIBES 95 leaves ofsubg. Ribes and subg. Grossularia from Ne- DONOGHUE,M.J., C. D.B ELL, and J. LI.2001. Phylogenetic patterns vada.If these features are reliable,this dates subg. innorthern hemisphere plant geography. International Journal of Plant Sciences 162: S41–S52. Grossularia toat least 14mya (Fig. 9,node 6).A critical DORF, E. 1930. Pliocene orasof California. Carnegie Institutionof next step in understanding Ribes biogeographywill be WashingtonPublicati on 412. toaccurately assign Ribes fossils toparticular sub- FARRIS, J. S., M. KA¨ LLERSJO¨ , A. G. KLUGE, and C. BULT.1995. Test- clades so asto infer the timing ofintercontin ental di- ingsigniŽ cance ofincongruen ce. Cladistics 10: 315–319. FELSENSTEIN,J.1985. ConŽdence limitson phylogenies: an ap- vergences. proach usingthe bootstrap . Evolution 39: 783–791. FENTON,B., A. N.E. B IRCH, G. MALLOCH, P. G. LANHAM, and R. ACKNOWLEDGEMENTS .Wethank the followingindividua lsand M. BRENNAN.2000. Gall mitemolecular phylogenyand its institutionsfor providing plant material: B. Alverson, D.Boufford, relationship tothe ev olution ofplant hostspeciŽ city. Applied R. Mitchell, ArnoldArboretum of Harv ard University,University Acarology 24: 831–861. ofCalifornia Botanical Garden, Royal Botanic Garden, Edinburgh, FRITSCH,P.W., C. M.M ORTON, T. CHEN, and C. MELDRUM. 2001. New YorkBotanical Garden, and theUSD A-ARSN ational Plant Phylogenyand biogeographyofthe Styracaceae. International GermplasmRepository.A. Sentersand D.Soltisgraciousl ypro- Journalof Plant Sciences 162: S95–S116. vided numerousDN Aaliquots, DNAsequences,and access toear- GANDOLFO, M. A., P. HOC, S. N. SANTILLANA, and S. A. MARENSSI. lymanuscript s. H. Schornand D.Erwinprovided assistance with 1998. Una orfo ´silmorfolo ´gicamente af ´‡nalas Grossulari- paleobotanical literature.M.La vin, P.Herendeen, and twoanon- aceae (ordenRosales) de la Formacio´nla Meseta (Eoceno me- ymousreviewe rsprovided numeroushelpful comments.This proj- dio)Isla Marambio, Anta´rtida. AsociacionPaleontologica Argen- ect was supported inpart bya Putnam Fellowship throughthe tina PublicacionEspecial 5, Paleo´geno deAme´rica del Sury de la ArnoldArboretum of Harv ard University. Pen´‡nsula Anta´rtica : 147–153. GOLDMAN, N., J. P. ANDERSON, and A. G. RODRIGO.2000. Likeli- hood-based testsof topologies inphylogen etics. Systematic LITERATURE CITED Biology 49: 652–670. HANNIBAL,H. 1911. APliocene ora fromthe Coast Ranges of AVISE, J. C. 1994. Molecular markers, naturalhistory and evolution. California. Bulletin of the Torrey Botanical Club 38: 329–342. New York:Chapman &Hall. HUELSENBECK,J.P ., J.J. B ULL, and C. W. CUNNINGHAM. 1996. Com- AXELROD, D. I. 1966. The PleistoceneSoboba ora of southernCalifor- binineddata inphylogene tic analysis. Trends in Ecology and nia. Universityof California Publicationsin Geologica lSci- Evolution 11: 152–158. ences 60. JANCZEWSKI,M.E. 1907. Monographiesdes groseilliers, Ribes L. ———. 1998. ThunderM ountain, Idaho. Universityof California Me´moires Societe de Physique etD’ Histoire Naturelle de Gene`ve Publicationsin Geologica lSciences 142. 35: 199–517. BALDWIN,B. G.and S.M ARKOS.1998. Phylogenetic utilityof the external transcribedspacer (ETS)of 18S– 26S rDNA: Congru- KEEP,E. 1962. InterspeciŽc hybridizationin Ribes. Genetica 33: 1– ence ofETS and ITStrees of Calycadenia (Compositae). Mo- 23. lecular Phylogenetics and Evolution 10: 449–463 KISHINO, H. and M. HASEGAWA.1989. Evaluationof the maximum likelihoodestima te ofthe evolutionarytreetopologies from ———, M. J. SANDERSON, J. M. PORTER, M. F. WOJCIECHOWSKI , C. DNAsequence data, and the branchingorder in H ominoidea. S. CAMPBELL, and M. J. DONOGHUE.1995. The ITSregion of nuclear ribosomalDN A: avaluable source ofevidence on Journalof Molecular Evolution 29: 170–179. angiospermphylogen y. Annalsof the Missouri Botanical Garden KREMENETSKI,C. V.1998. Holocene historyof the northern range 82: 247–277. limitsof some trees and shrubsin Russia. Arctic and Alpine BEARD,K. H. 1998. East ofEden: Asia as an important center of Research 30: 317–333. taxonomicoriginationin mammalian evolution. Bulletin of the LINDER,C. R., L. R. G OERTZEN, B. V. HEUVEL, J. FRANCISCO-OR- Carnegie Museumof NaturalHistory 34: 5–39. TEGA, and R. K. JANSEN.2000. The complete external tran- BENA, G., M. F. JUBIER, I. OLIVIERI, and B. LEJEUNE.1998. Ribosom- scribedspacer of18S– 26S rDNA: ampliŽca tionand phylo- al external and internal transcribedspacers: combineduse in genetic utilityat lowtaxonomiclevelsin Asteraceae and thephylogene tic analysis of Medicago (Leguminosae ). Journal of closely allied families. Molecular Phylogenetics and Evolution 14: Molecular Evolution 46: 299–306. 285–303. BERGER,A. 1924. Ataxonomicreview ofcurrants and gooseberries. MACGINITIE, D. 1953. Fossil plants of the Florissant beds, Colorado. NewY ork Agricultural Experimental StationT echnical Bulletin Carnegie Institutionof W ashingtonPublicatio n599. 109: 1–118. MADDISON,D.R. and W.P.M ADDISON.2000. MacClade 4. Sun- BOUFFORD,D.E. and S.A. S PONGBERG.1983. Eastern Asia-Eastern derland: Sinauer Associates. NorthAmerican phytogeographical relationships-a history MANCHESTER,S.R. 1999. Biogeographical relationshipsof N orth fromthe timeof Linnaeus tothe twentiethcentury. Annals American Tertiary oras. Annalsof the Missouri Botanical Gar- of theMissouri Botanical Garden 70: 423–439. den 86: 472–522. CAMPBELL,C. S., M.F .W OJCIECHOWSKI , B. G. BALDWIN, L. A. AL- ———.2001. Update on the megafossil ora ofFlorissant, Colo- ICE, and M. J. DONOGHUE.1997. Persistentn uclear ribosomal rado. Proceedingsof the DenverM useumof Nature& Science, DNAsequence polymorphismin the Amelanchier agamic Series4, No.1: 137. complex (Rosaceae). Molecular Biology and Evolution 14: 81–90. MANOS,P.S.and M.J. D ONOGHUE.2001. Progressin northern CEVALLOS-FERRIZ,S.R. S.1995. Fruitsof Ribes fromthe Princeton hemispherephytogeogr aphy: an introduction. International chert, BritishColumbia ,Canada. American Journalof Botany Journalof Plant Sciences 162: S1–S2. 82(6, supplement):84. ———and A. M.S TANFORD.2001. The historical biogeographyof COVILLE,F.V.and N.L. B RITTON. 1908. Grossulariaceae . North Amer- Fagaceae: trackingthe tertiary history of temperate and sub- ican Flora 22: 193–225. tropical forestsof the northern hemisphere . International Jour- CUNNINGHAM,C. W.1997 Can threeincongruen ce testspredict nal of Plant Sciences 162: S77–S93. when data sholdbe combined? Molecular Biology and Evolution MAST,A. R. and T.J.G IVNISH.2002. Historical biogeographyand 14: 733–740. the originof stoma tal distributionsin Banksia and Dryandra DEQUEIROZ, A., M. J. DONOGHUE, and J. KIM.1995. Separate versus (Proteaceae)based on theircpD NAphylogeny. American Jour- combinedanalysis ofphylogene tic evidence. AnnualR eview of nal of Botany 89: 1311–1323. Ecology and Systematics 26: 657–681. MESSINGER, W., K. HUMMER, and A. LISTON. 1999. Ribes (Grossu- 96 SYSTEMATIC BOTANY [Volume 29

lariaceae)phylogenyas indicated byrestriction-si te polymor- ular evolution and phylogenetic implicationsof internal tran- phismsof PCR-ampliŽ ed chloroplastDN A. Plant Systematics scribedspacer regionsof ribosomal D NAinW interaceae. and Evolution 217: 185–195. American Journalof Botany 80: 1042–1055. ———, A. LISTON, and K. HUMMER 1993. Restriction sitemapping SULLIVAN,J.1996. Combiningdata withdifferent distribution sof of Ribes nuclear ribosomalD NA. Acta Horticulturae 352: 175– among-siterate variation. Systematic Biology 45: 375–380. 184. SWOFFORD,D.2001. PAUP*. Phylogenetic analysis usingparsi- MESSLER,M.R., R. J.C OLD, and P. WILSON.1991. Natural hybrid- mony.V ersion4. Sunderland: Sinauer Associates. izationin w esterngooseberries ( Ribes subgenus Grossularia : TIFFNEY,B. H. 1985a. Perspectives on theorigin of the  oristic Grossulariaceae). Madron˜o 38: 115–129. similaritybetween eastern Asia and eastern NorthAmerica. MORGAN,D.R. and D.E. S OLTIS.1993. Phylogenetic relationships Journalof the Arnold Arboretum 66: 73–94. amongmembers of the Saxifragace ae sensu lato based on ———.1985b. The Eocene NorthA tlantic land bridge:its impor- rbcL sequence data. Annalsof the Missouri Botanical Garden 80: tance intertiary and modernphytogeogra phy ofthe northern 631–660. hemisphere. Journalof the Arnold Arboretum 66: 243–273. MORT,M.E., D.E. S OLTIS, P. S. SOLTIS, J. FRANCISCO-ORTEGA, and ———and S.R. M ANCHESTER.2001. The use ofgeological and A. SANTOS-GUERRA.2002. Phylogeneticsand evolution ofthe paleontological evidence inev aluatingplant phylogeographic Macaronesian clade ofCrassulace ae inferredfrom nuclear hypothesesin the northern hemisphere tertiary. International and chloroplastsequence data. Systematic Botany 27: 271–288. Journalof Plant Sciences 162: S3–S17. REHDER, A. 1940. Manual of cultivated treesand shrubs. New York: WEHR,W.C. and D.Q.H OPKINS.1994. The Eocene orchards and The MacMillan Company. gardensof Republic, Washington. Washington Geology 22: 27– RIESEBERG,L. R. and J.F .W ENDEL.1993. Introgressionand itscon- 34. sequences inplants. Pp.70–109 in Hybrid zones and the evo- WEIGEND, M. and M. BINDER.2001. Threenew species of Ribes L. lutionary process, ed. R. G.Harrison. New York:Oxford U ni- (Grossulariaceae) fromCentral and SouthAmerica. System- versityPress. atic Botany 26: 727–732. RONQUIST,F.1996. DIVA, version1.1. Computerprogramand WEINS,J.J. and T.W.R EEDER.1995. Combiningda ta setswith manual available byanonymo us FTP fromUppsala Univer- differentnumbers of taxa forphylogene tic analysis. System- sity(h ttp://ftp.uu.se). atic Biology 44: 548–558. ———.1997. Dispersal-vicariance analysis: anew approach tothe WEN,J.1999. Evolution ofthe eastern Asian and eastern North quantiŽcation ofhistorical biogeography. Systematic Biology American disjunct distributionsin oweringplants. Annual 46: 195–203. Review of Ecology and Systematics 30: 421–455. SANG, T., D. J. CRAWFORD, and T. F. STUESSY.1997. Chloroplast WENDEL, J. F., A. SCHNABEL, and T. SEELANAN.1995a. Anunusual DNAphylogeny,reticulate evolution, and biogeographyof ribosomalDN Asequence from Gossypium gossypioides reveals Paeonia (Paeoniaceae). American Journal of Botany 84: 1120– ancient, cryptic intergenomicintrogression. Molecular Phylo- 1136. genetics and Evolution 4: 298–313. SANMART´‡N, I., H. ENGHOFF, and F. RONQUIST.2001. Patternsof ———,———,and ———.1995b. Bidirectional interlocus con- animal dispersal, vicariance and diversiŽca tionin the Hol- certed evolution followingallopolyp loid speciation incotton arctic. BiologicalJournalof theLinneanSociety 73: 345–390. (Gossypium). Proceeding of the National Academy of Sciences USA SENTERS,A. E. and D.E. S OLTIS.2003. Phylogenetic relationships 92: 280–284. in Ribes (Grossulariaceae) inferredfrom ITS sequence data. WHITE, T. J., T. BRUNS, S. LEE, and J. TAYLOR.1990. AmpliŽca tion Taxon 52: 51–66. and direct sequencingoffungal ribosomalRN Agenesfor SINNOTT,Q.P .1985. Arevision of Ribes L. subg. Grossularia (Mill.) phylogenetics. Pp.315–322 in PCR protocols: aguide tomethods Pers. Sect. Grossularia (Mill.)Nutt. ( Grossulariaceae ) in North and applications, eds. M.Innis, D .Gelfand, J.Sninsky ,and T. America. Rhodora 87: 189–286. White.San Diego: Academic Press. SOLTIS,D.E. and P.S.S OLTIS.1997. Phylogenetic relationshipsin WOLFE, J. A. 1964. Miocene oras from Fingerrock Wash, Southwestern Saxifragaceae sensu lato: acomparison oftopologies based Nevada. GeologicalSurvey Professionalpaper 454-N. on 18S rDNAand rbcL sequences. American Journalof Botany XIANG,Q.Y .and D.E. S OLTIS.2001. Dispersal-vicariance analyses 84: 504–522. ofintercontine ntal disjuncts: historical biogeographical im- SOLTIS,D.E., M.T .C LEGG, and M. DURBIN. 1990. rbcL sequence plicationsforangiosperms in the northern hemisphere . Inter- divergence and phylogenetic relationships inSaxifragace ae national Journalof Plant Science 162: S29–S39. sensu lato. Proceedingsof the National Academy of Sciences, USA ———,and P.S.S OLTIS.1998. The eastern Asian and eastern and 87: 4640–4644. westernN orthAmerica oristicdisjunction :congruent phy- ———, E. R. MORGAN, A. GRABLE, P. S. SOLTIS, and R. KUZOFF. logenetic patterns inseven diverse genera. Molecular Phylo- 1993. Molecular systematicsof Saxifragace ae sensu lato. genetics and Evolution 10: 178–190. American Journalof Botany 80: 1056–1081. ———,———,———,S.R. M ANCHESTER, and D. J. CRAWFORD. SPONBERG,S.A. 1972. The genera of Saxifragaceae inthe south- 2000. Timingthe eastern Asian-easternNorthAmerican o- eastern United States. Journalof theArnold Arboretum 53: 109– risticdisjunction: molecular clock corroborates paleontologi- 498. cal estimates. Molecular Phylogenetics and Evolution 15: 462– SUH, Y., L. B. THIEN, H. E. REEVE, and E. A. ZIMMER.1993. Molec- 472.