SystematicBotany (2004),29(1): pp. 177– 187 q Copyright 2004by the AmericanSociety of PlantTaxonomists

AChloroplast DNAPhylogeny of Solanum Section Lasiocarpa

LYNN BOHS DepartmentofBiology ,University ofUtah, Salt Lake City,Utah 84112 ([email protected])

CommunicatingEditor: J amesF .Smith

ABSTRACT. Solanum section Lasiocarpa includes about adozen species witha center ofdiversity in the New Worldtropics. Solanumla siocarpum and S. repandum (sometimesconsidered tobe conspeciŽc as S. ferox)have an OldW orlddistribution in Asia and theP aciŽc Islands.Several species inthis section produce edible fruits,and two, the lulo ornaranjilla ( S.quitoense ) and the cocona ( S.sessiliorum )are cultivated commercially.Phylogene tic relationships in Solanum section Lasiocarpa were investigated usingsequence data fromthe chloroplast trnT-trnL spacer, the trnL-trnF spacer, and the trnL gene, including the trnL intron.Sampling included 24 accessions fromsection Lasiocarpa and 14 accesssionsofother Solanum species as outgroups.Allspecies considered tobelong to section Lasiocarpa bypreviou sauthorsw ere examined withthe exception of therecently described S. atheniae.Solanumrobustum and S. stagnale,sometimesconsidered tobelong to section Lasiocarpa, are excluded fromthe group on thebasis ofthe trn data. The remainingspecies inthe section forma monophyletic group,with threew ell-supportedclades withinit: S. hirtum, S.pectinatum -sessiliorum -stramonifolium ,and theremainder of the species inthe section. Sequencesof S.lasiocarpum and S. repandum are extremelysimilar, and thesetw oAsian taxa cluster withthe New World S. candidum and S.pseudolulo on the trn trees.

Solanum section Lasiocarpa (Dunal)D’ Arcy compris- group ofthe genus Solanum,usually recognized as So- es approximately adozen species ofperennial shrubs lanum subgenus Leptostemonum .Previous authors such or small trees with acenter ofdistribution in north- asDunal (1852),Seithe (1962),Danert (1970),D’ Arcy western SouthAmerica. Morphologicalcharacters that (1972),and Whalen (1984)have regarded subgenus deŽne the section include difoliate sympodial units, Leptostemonum asa naturalgroup based on the shared large repand leaves, unbranchedinorescences ,stel- presence in most species ofspines, stellate hairs, and latecorollas, an dfruits covered with stellate hairs with tapered anthers. Molecularph ylogeneticstudies based reduced lateral rays (Whalen et al.1981). The section on chloroplast DNArestriction sites (Olmstead and wasmonograph ed byWhalen et al.(1981), who rec- Palmer, 1997)and nuclear and chloroplast sequence ognized 13species. Eleven are native tothe northern data(Bohs and Olmstead 1997,1999, 2001; Bohs, in Andes ofV enezuela,Colombia, Ecuador, and Peru, and press) indicate that Solanum species thatbear spines as three haveranges thatextend into Central America ( S. well asstellate hairs comprise amonophyletic group, candidum, S. hirtum)or northeastern SouthAm erica termed the Leptostemonum cladeby Bohs (in press). through the Guianasinto northern Brazil ( S.stramoni- Solanum wendlandii ,arepresentative of Solanum section folium).Several species in the section produceedible Aculeigerum Seithe,fallsoutside the cladecom prised of fruits and two, S.quitoense (the luloor naranjilla) and the other spiny Solanum taxa(Bohs and Olmstead 1997, S.sessiliorum (the cocona)are economicallyimportant 1999,2001; Bohs, in press). Solanum section Aculeigerum fruitcrops in Latin America (Heiser 1969,1985a). So- includes six species thatbear spines butlack stellate lanum quitoense has been introduced toPanama, Costa hairs. In this paper, Solanum subgenus Leptostemonum Rica,and Guatemalaand is now naturalized in Central is used in the traditional sense torefer toall taxa of America. Solanum lasiocarpum and S. repandum are the genus thatbear spines. The term Leptostemonum found in Asia and the PaciŽc Islands. Although treated cladeis used in accordancewith Bohs (in press) to asseparate taxa by Whalen et al.(1981), H eiser (1996) refer tothe monophyletic group ofspin y Solanum taxa considered them conspeciŽc under the name S. ferox. exclusive of Solanum section Aculeigerum . Dunal (1852),M orton (1976),and Hunziker (2001) Molecularstudies based on chloroplast DNArestric- included the SouthAm erican S. robustum in section La- tion sites and chloroplast ndhF sequence datausing a siocarpa,butWhalen et al.(1981) exclude dit from the broadrange ofsampling from Solanum indicate that section due todifferences in branching pattern, leaf section Lasiocarpa may bea relativelybasallineage shape,and fruittrichom es. Subsequent toWhalen et within the Leptostemonum cladeand thatit may be al.’s (1981)treatment, S. stagnale wasremoved from the sister to Solanum section Acanthophora Dunal (Olmstead section and placedin the S.polytrichum group within and Palmer 1997;Bohs, in press). Whereas these broad Solanum subgenus Leptostemonum (Dunal)Bitter scalestudies sampled only one totw ospecies from the (Whalen 1984;Child 1998;N ee 1999).Symon (1985) section, species-level relationships in section Lasiocarpa described S. atheniae from NewGuineaand postulated have been the subjectof n umerous investigations using thatit belonged tosection Lasiocarpa. morphologicaldata, crossing studies, isozyme electro- Solanum section Lasiocarpa belongs tothe spinysub- phoresis, karyotype analyses, and cpDNArestriction

177 178 SYSTEMATIC BOTANY [Volume 29

TABLE 1. Sources of Solanum DNAaccessions used inthis study. Seeds, leaves, orD NAextracts provided by 1 L. Bohs, University ofUtah, Salt Lake City,UT. 2 R. G.Olmstead,U niversityof W ashington, Seattle,WA. 3 A. Bruneau, McGill University,Montreal, Canada. 4 C. B. Heiser, Jr., Indiana University,Bloomington,IN. 5 J.Miller, Amherst College ,Amherst,MA. a Forfurther collection and voucher data see Appendix inWhalen et al. (1981). BIRMsamples bear theseed accession numberof the Univ ersityof Birmingham Solanacea e collection. Nijmegenaccession numbersrefer to the Solanacea ecollection attheU niversityof Nijmegen, The Netherlands.

Solanum section Lasiocarpa:S. candidum Lindl.3—Stoutamires.n. (IND) from Heiser S249a,Mexico: Veracruz (AY266250). S. candidum Lindl.1—Bohs 2898 (UT),Costa Rica: La Cangreja (AY266237). S. felinum Whalen4—Benitez de Rojas 8915 (IND), Venezuela: Colonia Tovar (AY266252). S. hirtum Vahl3—Whalen 730 (QCA),Ecuador (AY266254). S. hirtum Vahl3—Jones s.n. (IND) from Heiser S404a,Costa Rica: Guanacaste (AY266253). S. hyporhodium A.Braun &Bouche´3—Whalen 717 (BH), Vene- zuela: Sucre (AY266238). S. hyporhodium A.Braun &Bouche´4—Carreno Espinosa 8214 (IND),V enezuela: Sucre (AY266255). S. lasiocarpum Dunal4—Ansyar 9605 (IND),Indonesia: Pandang (AY266256). S. pectinatum Dunal4—Peeke 8512 (IND), Ec- uador: Limoncocha (AY266227). S. pectinatum Dunal1—Bohs 2899 (UT),Bolivia: Santa Cruz (AY266230). S. pseudolulo Hei- ser3—Plowman etal. 4276 (GH)a,Colombia: Meta, Sierrade la Macarena (AY266258). S. pseudolulo Heiser5—BohsDNA ex- tract 995, Nijmegen#824750021 (AY266242). S. quitoense Lam.1—Bohs 2873 (UT),Costa Rica (AY266228). S. quitoense Lam.4—Heiser s.n. BohsDNA extract 996, Ecuador: Quitomarket (A Y266243). S. repandum G. Forst.3—Heiser 8215 (IND), Fiji(A Y266229). S. repandum G. Forst.4—Ashley 8627 (IND),Solomon Islands: Malaita (AY266234). S. sessiliorum Dunal3— Dickson 458 (BH) from Whalen 859 (HUT),Peru (AY266261). S. sessiliorum Dunal var. sessiliorum 4—Heiser 8255 (IND), Ecuador: Yanzatza (AY266260). S. stramonifolium Jacq. var. inerme (Dunal) Whalen 4—Pickersgill 154 (IND),Peru: Iquitos (AY266244). S. stramonifolium Jacq. var. inerme (Dunal) Whalen 3—Whalen &Salick 860 (BH),Peru: Pasco, Iscozacin (AY266263). S. vestissimum Dunal3—Dickson 456 (BH) from Plowman 13431 (F),V enezuela (AY266264). S. vestissimum Dun- al4—Movilla s.n. (IND) from Heiser S432a,Colombia: Santa Marta (AY266247). Outgroups: S. abutiloides (Griseb.)Bitter & Lillo 2—RGO S-73 (WTU),BIRM S.0655 (AY266236). S. acerifolium Dunal1—Bohs 2714 (UT),Costa Rica (AY266249). S. capsicoides All.1—Bohs 2451 (UT),Peru (AY266251). S. dulcamara L.2—no voucher, USA(AY266231). S. jamaicense Mill.2—RGO S-85 (WTU),BIRM S.1209 (AY266239). S. luteoalbum Pers.1—Bohs 2337 (UT), BIRMS.0042 (AY266257). S. mammosum L.2—RGO S-89 (WTU),BIRM S.0983 (AY266232). S. melongena L.2—RGO S-91 (WTU),BIRM S.0657 (AY266240). S. palinacanthum Dunal1—Bohs 3151 (UT),Bolivia (AY266233). S. pseudocapsicum L.2— no voucher, BIRMS.0870 (AY266241). S. robustum Wendl.4—Bohs 3084 (UT),Argentina: Corrientes,Pericho ´n(AY266259). S. sisymbriifolium Lam.1—Bohs 2533 (UT),Argentina (A Y266235). S. stagnale Moric.4—Carvalho 3213 (IND),Brazil: Bahia, Val- enc¸a(AY266262). S. tenuispinum Rusby1—Bohs 2475 (UT),Bolivia (AY266245). S. torvum Sw.1—RGO S-101 (WTU), BIRM S.0839 (AY266246). S. wendlandii Hook. f.1—no voucher, BIRMS.0488 (AY266248).

site data(H eiser 1972,1985b, 1987, 1989; Whalen et al. MATERIALSAND METHODS 1981;Whalen and Caruso1983; Bernarde llo et al.1994; Allspecies placed insection Lasiocarpa byWhalen et al. (1981) Bruneau et al.1995). Many ofthese studies were aimed were sampled, including S. stagnale and S. robustum. Solanuma th- atexamining the evolutionary history ofthe Asian dis- eniae isknown only from the type (Symon1985) and no material juncts and the origin and evolution of S.quitoense . De- was available forsampling. Inmost cases, twoaccessions were spite the accumulationof an impressiv eamountof sampled fromeach species ofsection Lasiocarpa.Outgroup taxa data,a consensus has not been reached regarding the included ten species from Solanum subgenus Leptostemonum and fourspecies representingtaxa fromv arious non-spiny Solanum phylogeneticrelationships ofthe taxaof this group due clades. Outgroupsw ere chosen torepresent a variety ofdiv erse tocon icting topologies from different datasets and Solanum clades based on previous molecular studies. Inaddition, tolow resolution in somepartsof the trees. Evidence samplingincluded Žve representatives from Solanum section Acan- suggests thatthe Asian species S. repandum and S. la- thophora,whichwas identiŽed as thesister group tosection Lasio- carpa inprevious analyses based on chloroplastD NAdata (Olm- siocarpum are sister taxa(H eiser 1986,1987; Bernar- stead and Palmer1997; Bohs, inpress). Collection ,voucher, and dello et al.1994; Bruneau et al.1995) or even conspe- GenBank informationis giv en inT able 1. ciŽc (as S. ferox;Heiser 1996),but the closest relatives DNAwas extracted fromfresh or silica driedleaf material using ofthis cladeare debated. The inclusion of S. stagnale, protocols described inBohs and Olmstead(1997, 2001) and Bohs (inpress). AmpliŽ ca tionof the entire trnT (UGU)—trnF (GAA) S. robustum, and S. atheniae in section Lasiocarpa has not regionused primersa and fofT aberlet et al. (1991) in25 ml re- been critically examined and the datahave been in- actions as described inBohs and Olmstead(2001) witha PCR conclusive with respect tothe wild relatives ofthe pu- programof 92 8 Cfor7 minfollo wed by30 cycles of92 8 C for 1 tativedomestica tes S.quitoense and S.sessiliorum . min, 458 Cfor1 min,72 8 Cfor5 min,and asinglecycle of72 8 C The present study examines species-level phyloge- for7 min.PCR products were cleaned usingQiaQuick spin col- umns(Qiagen, Inc., Valencia, CA)and sequencedon an ABIa u- netic relationships in Solanum section Lasiocarpa using tomated sequencerusingprimers a throughf ofT aberlet et al. chloroplast trn sequence data.These datashed light on (1991). the circumscription ofthe section, the relationships of Sequence data were edited and contigsconstructed usingSe- the Asian taxa,and the wild relatives ofthe luloand quencher(Gene Codes Corp.)and sequences were aligned byeye usingSe-Al (Ramba ut 1996). Indelalignments took into account cocona,and demonstrate the utility of trn sequence themechanisms and patternsof evoluti on innon-coding sequenc- datafor examining species-level phylogeny within So- es outlined inK elchner (2000). Allsequences were submittedto lanum. GenBank (Table 1) and the data setsand representativetreesare 2004] BOHS:PHYLOGENY OF SOLANUMSECTION LASIOCARPA 179 deposited inTreeB ASE[accession numbersS907 (study)and sequence datamin us indels with the 32presence/ ab- M1490 (matrix)]. sence indel charactersadded. Ofthe 1575total char- The trn regionsampled here includes twocoding regions( trnL 59 and 39 exons),tw ointergenicspacers ( trnT–trnL and trnL–trnF actersin this dataset, 199w ere variablewith 89of spacers), and the trnL intron(for diagrams and sequences ofthis these parsimony-informative.This analysis resulted in regionin tobacco, see Yamada et al. 1986). Toexplore the infor- 3205trees of270 steps with aCIof0.667 and RIof mativenessof each ofthese regions in the context of Lasiocarpa 0.882.The Žnal parsimony analysis used the complete phylogeny,eachofthe non-codingregions was analyzed separate- lyusing parsimony and theresults were compared withthose aligned sequence datawith the coded indels, resulting fromthe complete data sets. in atotalof 2366 characters. Of these ,244w ere variable Toexplore theeffects of indels and indel coding on theph ylo- and 107were parsimony-informative.The analysis genetic results, several analyses were performedon the aligned found 3194trees of323 steps with aCIof0.688 and data set. The Žrstused thecomplete aligned nucleotidesequence RI of 0.882. data set, withgaps treated as missingda ta. The second excluded indelsfrom the sequence data matrix.F orsubsequen tanalyses, 32 All fourparsim onyanalyses described abovere- phylogenetically informativegap characters (i.e., thoseshared by solved the following clades, which were present in all twoormore taxa) whose homologycould be conŽdently assessed the strict consensus trees: 1)All the spinytaxaof So- were coded as separate presence/absence characters according to lanum (i.e., Solanum subgenus Leptostemonum ) with the thesimple indel coding schemeof Simmons and Ochoterena (2000). The thirdanalysis used thenucleotid esequence data with exception of S.wendlandii formed amonophyletic indelsexcluded and the32 presence/absence gap charactersadd- group with 100%bootstrap support in allanalyses. ed. The fourthanalysis used thecomplete aligned sequence data, Solanum wendlandii ,ananomalous spin ytaxonsom e- including indel regions,with the addition ofthe 32 presence/ab- times placedin subgenus Leptostemonum ,fell outside sence gap characters. the spiny clade. Solanum wendlandii wasalso excluded Parsimonyanalyses were conducted withP AUP* 4.0b10 (Swof- ford2002) usingthe heuristic search algorithmwith the TBR, from the Leptostemonum cladein previous molecular MulTrees, and Steepest Descent options, equal weightsfor all analyses (Bohs and Olmstead 1997,1999, 2001; Bohs, characters and character state changes, and 500 random-orderen- in press). 2)All species of Solanum section Lasiocarpa tryreplica tes. Bootstrap analyses were performedwith 500 repli- formed amonophyletic group,with 87–93% bootstrap cates usingthe heuristic search option, TBR and MulTrees, Max- treesset to 1,000, and rearrangementslimited to 1,000,000 per supportdependin gon the analysis. Solanum robustum replicate. and S. stagnale,sometimes putinto section Lasiocarpa, Sequence data fromthe ITSregion w ere obtained froma subset did not group with the traditional members ofthe sec- of the Lasiocarpa species used inthe trn studyusing protocols de- tion, butinstead emerged assister taxawithin the Lep- scribedin Bohs and Olmstead(2001). ITSsequence divergencewas extremelylow among Lasiocarpa taxa and provided littlephylo- tostemonum clade.3)The Žve species of Solanum sec- genetic information. These ITSsequence swere deposited in tion Acanthophora included in this study ( S.acerifolium , GenBank (numbersA Y263455—AY263467) ,but are notanalyzed S.capsicoides , S. mammosum, S.palinacanthum , and S. ten- furtherhere. uispinum)alsoformed amonophyletic group with 100%bootstrap support. 4) All analyses identiŽed a RESULTS cladewithin section Lasiocarpa consisting of S. sessili- The totallength ofthe trn aligned sequence dataset orum, S.stramonifolium , and S.pectinatum . Bootstrap was2334 n ucleotides, ofwhich 791represente dindels. supportfor this group ranged from 93–97% dependin g The totalunaligned length of trn sequencesranged on the analysis. 5)W ithin this latter clade,the twoac- from 1759to 2052 bp in species ofsection Lasiocarpa cessions eachof S.sessiliorum and S.stramonifolium and from 1673to 1955 bp in the outgroups(T able2). grouped together with 94–98% and 63–98% bootstrap Lengths ofthe trnT-trnL and trnL-trnF intergenicspac- support,respectivel y. ers and the components ofthe trnL gene are given in Other clades were resolved in one or more ofthe Table2 foreach accession sequenced. Eachof these re- analyses, buteither do not appearon the strict con- gions provided different numbers ofcharacters for sensus trees from the individualanalyses or were not phylogenetic analyses (Table3). resolved in allfour analyses. The 50%majority rule Ofthe 2334characters in the complete aligned se- consensus trees from Analyses 1and 2resolved iden- quence dataset, 212w ere variableand 75of these were ticalclades within theingroup (Fig. 2).In addition to parsimony-informative.Parsimonyanalysis ofthis data the groups described above,these analyses identiŽed set found 1907most parsimonious trees of263 steps, the following relationships: 1)The twoaccessions of S. with aconsistency index (CI;excludin guninformative hirtum grouped together and formed the basalbranch characters)of 0.761 and aretention index (RI)of 0.909 in the Lasiocarpaclade .2)A large cladew asidentiŽ ed (Fig. 1).In the second analysis, regions with indels consisting ofall accessions of S.vestissimum , S. hyporho- were excludedfrom the aligned sequence dataset. Of dium, S. felinum, S.quitoense , S.lasiocarpum , S. repandum, 1543total characters, 167 w ere variableand 57of these S. candidum, and S.pseudolulo .This wassister tothe S. were parsimony-informative.PAUP * found 796most pectinatum-S.stramonifolium -S.sessiliorum clade de- parsimonious trees of210 steps, with aCIof 0.747 and scribed above.Within this clade, S.vestissimum S432 anRI of 0.913. The third analysis used the aligned formed the basallineage ,which wassister tothe rest 180 SYSTEMATIC BOTANY [Volume 29

TABLE 2. Length of trnT to trnF regionin studied taxa. Values are rawsequence lengthin base pairs, notincluding indelsin the Žnal aligned version. a NA 5 Nota vailable.Firstca. 15 to19 bp ofsequence notreadable .

Totallength (trnT–L to trnT–trnL trnL trnL trnL trnL–trnF trnL–F Taxon spacer 59 exon intron 39 exon spacer spacers) Solanum section Lasiocarpa S. candidum S249 734 35 497 50 524 1840 S. candidum 2898 734 35 497 50 524 1840 S. felinum 8915 734 35 497 50 677 1993 S. hirtum S404 738 36 497 50 518 1839 S. hirtum 730 717 36 497 50 459 1759 S. hyporhodium 717 734 35 497 50 677 1993 S. hyporhodium 8214 734 35 497 50 622 1938 S. lasiocarpum 9605 752 35 497 50 524 1858 S. pectinatum 8512 749 36 497 50 676 2008 S. pectinatum 2899 749 36 497 50 676 2008 S. pseudolulo 4276 734 35 497 50 677 1993 S. pseudolulo 995 734 35 497 50 524 1840 S. quitoense 2873 734 35 497 50 524 1840 S. quitoense 996 734 35 497 50 524 1840 S. repandum 8215 734 35 497 50 524 1840 S. repandum 8627 734 35 497 50 519 1835 S. sessiliorum 458 749 36 497 50 523 1855 S. sessiliorum 8255 749 36 497 50 523 1855 S. stramonifolium 860 716 36 497 50 753 2052 S. stramonifolium 154 716 36 497 50 700 1999 S. vestissimum 456 734 35 497 50 524 1840 S. vestissimum S432 734 35 497 50 523 1839 Solanum subgenus Leptostemonum S. acerifolium 761 36 497 50 639 1983 S. capsicoides 752 36 497 50 525 1860 S. jamaicense NAa 36 497 50 631 NAa S. mammosum 731 36 497 50 519 1833 S. melongena NAa 36 497 50 462 NAa S. palinacanthum 716 36 497 50 526 1825 S. robustum 710 36 497 50 567 1860 S. sisymbriifolium 717 35 497 50 623 1922 S. stagnale 710 36 497 50 462 1755 S. tenuispinum 717 36 497 50 541 1841 S. torvum 806 35 497 50 567 1955 S. wendlandii 723 36 497 50 404 1710 Outer outgroups S. abutiloides 712 36 497 50 409 1704 S. dulcamara 677 36 497 50 413 1673 S. luteoalbum 714 36 501 50 400 1701 S. pseudocapsicum NAa 35 497 50 401 NAa

ofthe clade.3)W ithin the cladedescribed in #2abov e, did not cluster asa monophyletic group,butinstead the twoaccessions of S.quitoense grouped together formed agrade atthe baseof the Lasiocarpaclade .2) with 62–63% bootstrap support. 4) Also within this The S.stramonifolium cladeemerged assister toa larger clade, S. repandum 8627and bothaccessions of group consisting of S.sessiliorum plus S.pectinatum . S.pseudolulo formed alineage.This grouping received 3)The twoaccessions of S.hyporhodium plus S. felinum 60–63% bootstrap support. formed amonophyletic group within the large clade The 50%majority rule consensus trees from anal- described in #2above . yses 3and 4(i.e.,those thatincluded indels ascoded In general, adding the coded indel charactersin- presence/absencecharacters) differed only in the pat- creased resolutionon the majority rule consensus trees tern ofrelationsh ips among members ofsection Acan- butdecreased it slightly with respect tothe strict con- thophora in the Leptostemonum clade;ingroup relation- sensus trees (i.e.,in comparisonsbetween Analyses 1 ships were identical (Fig. 3).These analyses resolved vs. 4and 2vs. 3).This is because the indels exhibit a the sameclades asthose described abovewith the fol- fairam ountof homoplas y,aresult alsoseen bymap- lowing exceptions: 1)The twoaccessions of S. hirtum ping the coded indel charactersonto the trees from 2004] BOHS:PHYLOGENY OF SOLANUMSECTION LASIOCARPA 181

TABLE 3. Characteristicsof the trn sequence data matrixand parsimonyresults when each regionis analyzed separately. Search parameters described intext except: a 500 randomaddition replicates, withno morethan 150 trees $ 92 steps saved per replicate. b 500 randomaddition replicates,withno morethan 200 trees $ 55 steps saved per replicate.

Totallength (trnT–L to trnT–trnL trnL trnL trnL trnL–trnF trn L–F Coded spacer 59 exon intron 39 exon spacer spacers) indels Alignedlength including gaps (bp) 910 36 505 50 833 2334 32 #variable characters 77 0 41 0 94 212 32 #parsimony-informativecharacters 25 0 11 0 39 75 32 #mostparsimonious trees(length) .73352 — 67 — 28402 1907 .59042 (92)a (44) (123) (263) (55)b #nodes resolved instrict consensus 9 — 7 — 5 12 5 tree(# ingroup nodes) (4) (2) (1) (5) (1) RI(CIexcluding autapomorphies) 0.886 — 0.952 — 0.929 0.909 0.862 (0.778) (0.857) (0.778) (0.761) (0.593)

Alignedlength excluding gaps (bp) 578 35 494 50 386 1543 #variable characters 61 0 40 0 66 167 #parsimony-informativecharacters 20 0 11 0 26 57 #mostparsimonious trees(length) 170 — 64 — 364 796 (75) (43) (89) (210) #nodes resolved instrict consensus 5 — 6 — 7 12 tree(# ingroup nodes) (2) (1) (2) (5) RI(CIexcluding autapomorphies) 0.892 — 0.952 — 0.930 0.913 (0.767) (0.857) (0.750) (0.747)

sequence dataalone (datanot shown). About25 to 50% totw onodes in the ingroup,supporting the mono- ofthe coded indel charactersw ere homoplasticwhen phyly ofthe twelve Lasiocarpa species and placing the mapped onto the varioustrees. This is alsore ected twoaccessions of S.sessiliorum assister taxa.Thus, in the lower CIand RIv alues forthe coded indel data although the trnL-trnF spaceris apopularchoice for set ascompared tothe other separately-analyzedre- phylogenetic reconstruction in many plantgroups gions of the trn dataset (Table3). (e.g.,T aberlet et al.1991; Gielly and Taberlet 1994;v an The variousnon-coding regions ofthe trn data set Hamet al.1994; Kim et al.1996), the greatest resolu- provided different numbers ofph ylogeneticallyinfor- tion in the Lasiocarpa study wasprovided bydata from mative charactersand different levels ofph ylogenetic the entire trn array.It is difŽcult to predict from char- resolution (Table3). When analyzed separately,the acterv ariability or sequence divergencev alues alone trnL-trnF spacerregion including gaps provided the what or how muchseq uence datawill bedesirable in largest number ofph ylogeneticallyinformativ echar- examining aphylogeneticproblem. acters,y et this region resolved just one totw onodes Circumscriptionand Monophyly of Section within section Lasiocarpa.Ofthe individual regions of Lasiocarpa. The twelve species traditionallyconsid- the trn array, the trnT-trnL spacerprovided the great- ered tobelong to Solanum section Lasiocarpa emerge as est resolving power forboth ingroup and outgroup amonophyletic group in allanalyses. The cpDNAdata nodes (Table3). H owever,even greater resolution was show that S. stagnale and S. robustum are not closely achieved byincluding datafrom the twospacers plus related toother members ofsection Lasiocarpa.Solan um the intron (Table3), regardless ofwhether gaps were stagnale wasoriginally included in section Lasiocarpa in included or excluded from the datamatrix. the monograph ofWhalen et al.(1981), but at that time it wasonly known from several nineteenth century DISCUSSION herbarium collections. Itslarge repand leaves and stel- Utility of trn SequenceDa ta. Ofthe non-coding late-pubescent fruits were thought tounite S. stagnale regions sampled here,the trnL-trnF spacerwas the with therest ofthe species in section Lasiocarpa, al- most variableand provided the largest number ofpo- though Whalen et al.(1981) regarded it as‘ ‘phyloge- tentially phylogenetically informativecharacters.H ow- netically isolated’’ and morphologicallyanomalous ever, the trnL-trnF spaceralone did not provide re- within thesection. Whalen (1984)later removed S. stag- solving power over the entire phylogeny. Use ofthe nale from section Lasiocarpa and surmised thatit was trnL-trnF spacersequence dataalone resolved only Žve more closely related totaxa of his S.polytrichum group, toseven nodes in the strict consensus trees, versus 10 although it is anomalouswithin thatgroup due toits to12 nodes forthe complete sequence datasets. Fur- pubescentfruits and unarmed, weakly accrescentca- thermore, trnL-trnF spacerdata alone resolved just one lyces. The trn datashow that S. stagnale is not closely 182 SYSTEMATIC BOTANY [Volume 29

FIG.1. Oneof 1907 mostparsimonious treesof 263 steps fromthe complete aligned sequence data set(Analysis 1). Number ofnucleotide changes isindicated above the branches. related tomembers ofsection Lasiocarpa,butrather is (2001).Whalen (1984)tentative ly considered S. robus- sister to S. robustum,aspecies not included in the orig- tum tobelong tothe S.erythrotrichum species group, inal Lasiocarpa monograph butplaced within the sec- which alsohas pubescentberries. Although the trn tion byDunal (1852),M orton (1976),and Hunziker dataresolve S. stagnale and S. robustum assister taxa 2004] BOHS:PHYLOGENY OF SOLANUMSECTION LASIOCARPA 183

FIG.2. 50% majorityrule consensus treefrom the complete aligned data set(Analysis 1). Dashed linesare branches that collapse inthe strictconsensus tree. Bootstrap values (500 replicates) included on the branches. Arrowsmark Asian species of section Lasiocarpa;all othermembers of the section are New Worldtaxa. and thus suggest thatthey are more closely related tion Lasiocarpa.One consists solely ofthe twoacces- than Whalen (1984)believ ed, more extensive sampling sions of S. hirtum,which form either abasalgrade or within Solanum subgenus Leptostemonum is needed to cladein the section. Solanum hirtum is the most wide- conŽrm this conclusion.Likewise,further sampling is spread and variablespecies in section Lasiocarpa necessary toidentify the closest relatives tosection La- (Whalen et al.1981) and canbe distinguish ed from siocarpa within the Leptostemonum clade. other members ofthe section byits relativelydimin- Relationships Within Solanum Section Lasiocarpa. utive leaves and fruits and byits reexed calyxlobes. Three groups ofspecies canbe discerned within sec- Cladistic analyses ofm orphologicaland allozyme 184 SYSTEMATIC BOTANY [Volume 29

FIG.3. 50% majorityrule consensus treefrom the aligned data setwith indels excluded and coded indels included (Analysis 3). Dashed linesare branches that collapse inthe strictconsensus tree. Bootstrap values (500 replicates) included on the branches. Arrowsmark Asian species ofsection Lasiocarpa;all othermembers of the section are New Worldtaxa. charactersby Whalen et al.(1981) and Whalen and moderate success),and with S.pseudolulo (with difŽ- Caruso(1983) show ed S. hirtum tobelong toa clade culty)in greenhouse crossing trials (Heiser 1972,1989), including S.lasiocarpum , S. candidum, S.quitoense , and butno successfulintraspeciŽ c crosses were obtained S.pseudolulo ,and this relationship wasrecovered in a between accessions of S. hirtum from Trinidad and subset ofthe analyses ofBruneau et al.(1995) based CostaRica (H eiser 1972).Results from the trn data in- on morphologicaland isozymecharactersand chloro- dicatethat the sequencesofthe twoaccessions of S. plastD NArestriction sites. Solanum hirtum hybridizes hirtum (fromCosta Rica and Ecuador)are very similar with S.quitoense (easily), with S.stramonifolium (with and that S. hirtum forms anisolated basalbranch in 2004] BOHS:PHYLOGENY OF SOLANUMSECTION LASIOCARPA 185 section Lasiocarpa.These Žndings are atodds with pre- (1995)reanalyzed these datausing more characters vious datafrom morphological,isozyme ,and crossing and better plantmaterial and found that S. repandum studies, although there is somesuggestion ofthis phy- wassister to S.lasiocarpum , not S.sessiliorum . The trn logenetic position from the analysis ofcpD NArestric- dataagree with the conclusionsofH eiser (1987)and tion site datain Bruneau et al.(1995). Bernardel lo et Bruneau et al.(1995) that S.sessiliorum and S. repan- al.(1994) found thatchrom osomes of S. hirtum were dum are not sister taxaand supportthe sister relation- most similar in morphology to S.pectinatum , but the ship between S.sessiliorum and S.stramonifolium . trn data place S. hirtum and S.pectinatum on distinct Solanum pectinatum is the third member ofthis well- clades. supported clade.The relationships ofthis species have The second claderesolved in the trn analyses con- been enigmatic becauseit is the only member ofthe sists of S.pectinatum , S.stramonifolium , and S.sessilio- section with consistentlyunbranchedhairs and it is rum.These taxahave been consideredtobe ph yloge- reproductively isolated from other species ofsection netically isolated from eachother and from other Lasiocarpa (Heiser 1972,1989). Because details oftri- members ofthe section based on morphologicaland chomemorphology have been important in phyloge- karyotypicanalyses asw ell ascrossing studies (Heiser netic studies based on morphologicalcharacters, S. pec- 1972,1989; Whalen et al.1981; Whalen and Caruso tinatum wasexcluded from consideration in the mor- 1983;Bernardel lo et al.1994). A cladecontaining S. phological analyses ofWhalen and Caruso(1983). pectinatum, S.stramonifolium , and S.sessiliorum was However, the morphologicalanalyses ofBruneau et al. alsorecovered in the analysis ofcpD NArestriction site (1995)included individuals of S.pectinatum reported charactersin Bruneauet al.(1995; their Fig. 1).Anal- tobear stellate trichomes. In these trees, S.pectinatum yses ofm orphologicaland allozyme charactersalone emerged on acladealong with S. hirtum, S. candidum, or in combination(Bruneau et al.1995) as well askar- S.quitoense , S.pseudolulo , S. felinum, and S.vestissimum . yotype analyses (Bernardello et al.1994) failed tosup- Analyses ofisozym edatagav easimilar result (Whal- portthis relationship.However, when the morpholog- en and Caruso1983; Brunea uet al.1995). Th e trn data icaland allozyme charactersw ere combined with the conict with this placement and are instead consistent cpDNArestriction sites the three species formed abas- with the cpDNArestriction site datain identifyinga algrade sister tothe remaining species ofsection La- cladeconsisting of S.pectinatum , S.sessiliorum , and S. siocarpa (Bruneau et al.1995). Solanum pectinatum , S. stramonifolium . stramonifolium , and S.sessiliorum , along with S. hirtum, The S.pectinatum 2899accession from SantaCruz, are low elevationspecies found generally below1000 Bolivia used in the trn study is morphologically simi- m. Thus, the trn datasupport the hypothesis ofan lar totypical S.pectinatum butbears short- tomedium- early lowland radiation in section Lasiocarpa followed stalked stellate hairs on the leaves and stem. The cau- bydiversiŽ cation at middle and high elevations (Whal- line hairs often beargland-tipped midpoints thatare en et al.1981; Whalen 1983;Whalen and Caruso1983; longer than the lateral rays, and the rays themselves Bruneau et al.1995). are divergently spreading toascending .Exclusively Solanum sessiliorum ,commonly known asthe co- unbranched trichomesare ahallmark of S.pectinatum cona,is cultivatedfor its large edible fruits.There is and the 2899accession camefrom alocalityfar to the much variability in size,form,and avorof the fruits southeast ofother S.pectinatum collections (Whalen et and manylocally named cultivars exist in SouthAm er- al.1981). H owever, S.pectinatum is occasionallyculti- ica(Schultes and Romero-Castan˜eda 1962),but these vated forits edible fruits andthus couldhav ebeen are allconsidere dconspeciŽc with S.sessiliorum introduced toBolivia byhumans in recent times. Like- (Whalen et al.1981). Solanum sessiliorum var. georgi- wise,Bruneauet al.(1995) report thatsom especimens cum (R.E. Schult.)Whalen differs from the typicalv a- of S.pectinatum were found with sessile toshort- riety in having spiny stems and leaves and small glo- stalked stellate stem hairs with spreading or ascending boseberries and is thought toperhaps represent the rays and midpoints aslong asor longer than the lateral progenitor ofthe cocona.Although not nearly asim- branches. The trn sequences from both S.pectinatum portantor widely used as S.sessiliorum , S.stramoni- accessions were very similar.The taxonomicconceptof folium alsoproduces edible fruits and has bothspin y S.pectinatum probablyshould beexpanded toinclude and non-spinyforms, the latter formally recognized as variants with stellate trichomes. S.stramonifolium var. inerme (Dunal)Whalen. Whalen The third claderesolved bythe trn dataincludes S. et al.(1981) proposed that S.sessiliorum and S. stra- candidum, S. felinum, S.hyporhodium , S.lasiocarpum , S. monifolium may bedistantly related, butfavored aclos- pseudolulo, S.quitoense , S. repandum, and S.vestissimum. er relationship between S.sessiliorum and S. repandum Solanum lasiocarpum and S. repandum are native tothe on the basisof phenetic and cladisticanalyses ofmor- Old World; the remaining species are mainly montane phological data(Whalen et al.1981; Whalen and Ca- taxawith acenter ofdiversity in northwestern South ruso 1983).H owever, Heiser (1987)and Bruneau et al. America. This cladew asalso recov ered byBruneau et 186 SYSTEMATIC BOTANY [Volume 29 al.(1995) in their analyses ofcpDN Arestriction sites bridize only with difŽculty .Although S.quitoense and and combined cpDNA,morphol ogical,and isozyme S. candidum are notsister taxain the trn trees, there is data.H owever, their analyses ofmorpholo gicaland little charactersupport and resolution in this areaof isozyme data,alone and in combination, placed S. hir- the tree and aclose relationship between the twotaxa tum within this clade,whereas this species forms an cannotbe ruled out.H owever, the trn datarefute hy- isolated basalbranch in the trn trees. Datafrom cross- potheses ofclose associationsbetween S.quitoense and ing studies and karyotype analyses are equivocalwith S. hirtum, S.pectinatum , S.stramonifolium , and S. sessi- respect tosupport for this large group (Heiser 1972, liorum. 1989;Bernardel lo et al.1994). Likewise,the relationships ofthe twoAsian dis- Within this large clade,coded indel dataprovide juncts, S. repandum and S.lasiocarpum ,have been amat- somesupportfor the associationof S.hyporhodium with ter ofdebate .Whalen et al.(1981) and Whalen and S. felinum.In analyses without the coded indel data,all Caruso(1983) suggested that S. repandum and S. lasio- accessions ofthese twospecies along with S.vestissi- carpum were not sister taxa,but instead that S. repan- mum form abasalgrade in the large cladedescribed dum wasallied toand perhaps conspeciŽc with S. ses- above, with S.vestissimum S432comprising the basal siliorum, whereas S.lasiocarpum wasm ost closely re- branchin the entire large clade.All three ofthese taxa lated to S. candidum.Conversely,Heiser considered S. are high-elevationcloud forest species native toV ene- repandum and S.lasiocarpum tobe closely related and zuelaan dnorthern Colombia.Whalen et al.(1981) con- perhaps conspeciŽc (as S. ferox) and that S. candidum sidered the three species tobe closely related on mor- wassister tothe Asian taxa(H eiser 1986,1987, 1996). phological grounds. Solanum felinum and S.vestissimum The trn data,as w ell aspre vious datafrom crossing are extremely similar morphologically,with S. hyporho- and karytotype studies and analyses ofcpD NAand dium less so (Bruneau et al.1995). Solanum hyporhodium morphologicalcharacters (H eiser 1986,1987, 1996; Ber- and S.vestissimum clustered together in phenetic and nardello et al.1994; Bruneau et al.1995) supports the cladisticanalyses ofisozyme data(Whalen and Caruso close relationship between S. repandum and S.lasiocar- 1983;Brunea uet al.1995); S. felinum wasnot included pum and thus Heiser’s hypothesis.Furthermore, S. can- in these studies. Crossing and karyotypic studies did didum emerges asa member ofthe S. repandum/S. la- not supporta relationship among the three taxa(H ei- siocarpum clade,conforming toH eiser’s ideas ofrela- ser 1972,1989; Bernardel lo et al.1994), although S. hy- tionships. However, S. repandum and S.lasiocarpum did porhodium and S. felinum had similar chromosomechar- not form amonophyletic group in the trn analyses; acteristics (Bernardello et al.1994). Although the three rather, one accession of S. repandum formed aclade taxaare closely associatedin most ofthe trn trees, they with the two S.pseudolulo accesssions.This result do not form amonophyletic group.In addition, the should not beover-inte rpreted, however, sincethere is S432accession of S.vestissimum from Colombiais di- little charactersupport for the identiŽcation of lineages vergent from the other fourrepresenta tives ofthe within the large cladethat includes S. repandum, S. la- group,all of which are from Venezuela.H eiser (2001) siocarpum, S.pseudolulo , S. candidum, S.quitoense , S. hy- noted thataccessions identiŽed as S.vestissimum from porhodium, S.vestissimum , and S. felinum. Colombiaand Venezuela would not cross with each In general, the trn trees are quite similar tothose other and differed in their crossing behavior with S. obtained from analyses ofcpD NArestriction site data quitoense.Further taxonomicwork on species limits in (cf.Fig. 1in Bruneauet al.1995). This is not surprising, this complex and more intensive sampling with more given thatthe chloroplastgenome is asingle linked variablegenes is warranted toascertain the position of non-recombining genetic entity (Doyle 1992).Further these high altitude Colombianand Venezuelan taxa. molecular studies are underwayusing more variable Twoquestions thathav ebeen intensively studied nucleargenes in order toachiev ebetter resolutionof with respect tothis group ofspecies concern the wild phylogeneticrelationships among the species ofsec- relatives of S.quitoense and the origin and relationships tion Lasiocarpa,toincrease supportfor previousl yiden- ofthe twoOld World taxaof section Lasiocarpa.Sola- tiŽed clades, and tocom pareph ylogenies derived num quitoense ,the luloor naranjilla, is acommonly from maternally inherited chloroplast genes with those cultivated fruitcrop in Andean SouthAmerica. Its based on biparentally inherited nuclearmark ers. range has recently spread toinclude Central America, ACKNOWLEDGMENTS .Ithank C. Heiser, A. Bruneau, J.Miller, where it is naturalized in Panamaand CostaRica. So- P.Diggle,R. Olmstead,L. D.Go´mez, and the Botanic Garden at lanum quitoense has been consideredbysom etobe the Universityof Nijmegen, The Netherlands, forproviding seed known only from cultivation,although spiny and feral and DNAsamples; B. Hammel and R. Aguilar forŽ eld assistance; forms exist in northwestern SouthAmerica. Heiser R. Dirigfor informa tionregarding vouche rsa tBH; C. Heiser, A. (1972)proposed on morphologicalgrounds that S. qui- Bruneau,E. E. Dickson, J.Miller, P .Diggle,S. Knapp ,and D.Spoo- nerfor comments on the manuscriptand help withsample and toense is most closely related to S. candidum, but the voucherinformatio n; M.J ohnson, A. Freeman, S.King-Jones,A. twospecies have different habitatpreferenc es and hy- Egan, A. Moore, and K. Klatt forlabora toryassistance; and C. 2004] BOHS:PHYLOGENY OF SOLANUMSECTION LASIOCARPA 187

Heiser, M.N ee, R. Olmstead,and S.Knapp forinsight into all tematics, ed. W.G.D’Arcy.N ew York:Columbia University thingssolanaceo us. Fundingfor this w orkw as provided byN SF Press. grantsD EB-9996199and DEB-0235339 and bythe U niversityof ———.1987. Originsof Solanumlasiocarp um and S. repandum. Utah. American Journal of Botany 74: 1045–1048. ———.1989. ArtiŽcial hybridsin Solanum sect. Lasiocarpa. System- atic Botany 14: 3–6. LITERATURE CITED ———.1996. Reappraisalof Solanumferox , S.lasiocarpum , and S. repandum. SolanaceaeNewsletter 4(2): 44–50. BERNARDELLO,L. M., C. B. H EISER, and M. PIAZZANO. 1994. Kar- ———.2001. InterspeciŽc hybridizationand the improvementof yotypic studies in Solanum section Lasiocarpa (Solanaceae). thenaranjilla ( Solanumquitoense ).Pages 307–310 in Solanaceae American Journalof Botany 81: 95–103. V:ad vances in taxonomy and utilization, eds. R. G.van den Berg, BOHS,L. Inpress. Majorclades in Solanum based on ndhF sequence G.W.M.Barendse ,G.M.v an derW eerden, and C. Mariani. analyses. InSolanacea e,William D’ArcyM emorial, eds., V. Nijmegen:Nijmegen U niversityPress. Holowell, R. Keating, W.Lewis, and T.Croat. St.Louis: Mis- HUNZIKER, A. T. 2001. The genera of Solanaceae .Ruggell: A. R. G. souri Botanical Garden Press. Gantner Verlag. ———and R. G.O LMSTEAD.1997. Phylogenetic relationshipsin KELCHNER,S.A. 2000. The evolution ofnon-coding chloropla st Solanum (Solanaceae)based on ndhFsequences. Systematic DNAand itsapplicat ion inplant systematics. Annalsof the Botany 22: 5–17. Missouri Botanical Garden 87: 482–498. ———and ———.1999. Solanum phylogenyinferredfrom chlo- KIM, J. H., H. ’T HART,and T.H. M.M ES.1996. The phylogenetic position ofEast Asian Sedum species (Crassulaceae) based on roplast DNAsequence data. Pages 97–110 in SolanaceaeIV: chloroplast trnL (UAA)—trnF (GAA)intergenic spacer se- advances in biology and utilization, eds. M.N ee,D.E. Symon, quence variation. ActaBotanica Neerlandica 45: 309–321. R. N.Lester, and J.P .Jessop.Kew: Royal Botanic Gardens. MORTON, C. V. 1976. Arevision of theArgentinespecies of Solanum . ———and ———.2001. Areassessmentof Normania and Triguera Co´rdoba, Argentina: Academia Nacional de Ciencias. (Solanaceae). Plant Systematics and Evolution 228: 33–48. NEE,M.1999. Synopsisof Solanum inthe N ew World. Pages 285– BRUNEAU, A., E. E. DICKSON, and S. KNAPP.1995. Congruenceof 333 in SolanaceaeIV:ad vances in biology and utilization, eds. M. chloroplastD NArestrictionsite characters withmorpholog - Nee,D.E. Symon,R. N.Lester, and J.P .Jessop.Kew: Royal ical and isozymedata in Solanum sect. Lasiocarpa. Canadian Botanic Gardens. Journalof Botany 73: 1151–1167. OLMSTEAD,R. G.and J.D .P ALMER.1997. Implicationsfor the phy- CHILD,A. 1998. Studiesin Solanum and related genera (6). New logeny,classiŽcation, and biogeographyof Solanum from infragenerictaxa forthe genus Solanum L. (Solanaceae). Fed- cpDNArestrictionsite v ariation. Systematic Botany 22: 19–29. des Repertorium 109: 407–427. RAMBAUT,A. 1996. Sequence alignmenteditor .Version1.0 alpha DANERT,S.1970. Infragenerische Taxa derGattung Solanum L. Kul- 1. Distributedon the web athttp://evolve.zoo.ox.ac.uk/Se- turpanze 18: 253–297. Al/Se-Al.html. D’ARCY,W.G.1972. Solanaceaestudies II:typiŽ cation ofsubdi- SCHULTES,R. E. and R. R OMERO-CASTAN˜ EDA.1962. Edible fruits visions of Solanum. Annalsof the Missouri Botanical Garden 59: of Solanum inColombia. Botanical MuseumLea ets of Harvard 262–278. University 19: 235–286. DOYLE,J.J. 1992. Gene treesand species trees: molecular system- SEITHE,A. 1962. Die Haararten derGattung Solanum L. und ihre atics as one-character taxonomy. Systematic Botany 17: 144– taxonomische Verwertung. BotanischeJahrbu¨cher fu¨rSystematik, 163. Panzengesch ichte und Panzengeogr aphie 81: 261–335. DUNAL., M-F.1852. Solanaceae. Pages 1–690 in Prodromussyste- SIMMONS,M.P .and H. O CHOTERENA.2000. Gaps as characters in matis naturalis regni vegetabilis, ed. A. P.de Candolle.Vol. sequence-based phylogenetic analyses. Systematic Biology 49: 13(1). Paris: VictorisMasson. 369–381. GIELLY, L. and P. TABERLET.1994. The use ofchloropla stD NAto SWOFFORD,D.2002. PAUP*: phylogenetic analysis usingparsi- resolveplant phylogenies: noncodingv ersus rbcL sequences. mony.V ersion4.0b10. Sunderland:Sinauer Associates. Molecular Biology and Evolution 11: 769–777. SYMON,D.E. 1985. The SolanaceaeofN ew Guinea. Journalof the HAM, R. C. H. J. VAN, H. ’T HART, T. H. M. MES, and J. M. SAND- Adelaide Botanic Gardens 8: 1–171. TABERLET, P., L. GIELLY, G. PAUTOU, and J. BOUVET.1991. Universal BRINK.1994. Molecular evolution ofnoncoding regions of the chloroplastgenome in the Crassulaceae and related species. primersfor ampliŽ cation ofthree non-coding regions of c hlo- roplast DNA. Plant Molecular Biology 17: 1105–1109. CurrentGenetics 25: 558–566. WHALEN,M.D .1984. Conspectus ofspecies groups in Solanum HEISER, C. B., JR. 1969. Nightshades: theparadoxical plants. San Fran- subgenus Leptostemonum . GentesHerbarum 12: 179–282. cisco: W.H. Freemanand Co. ———and E. E. C ARUSO.1983. Phylogenyin Solanum sect. Lasio- ———.1972. The relationshipsof the naranjilla, Solanumquitoense . carpa (Solanaceae): congruenceofmorpholog ical and molec- Biotropica 4: 77–84. ular data. Systematic Botany 8: 369–380. ———. 1985a. Of plants and people. Norman: Universityof ———, D. E. COSTICH, and C. B. HEISER.1981. Taxonomyof So- OklahomaPress. lanum section Lasiocarpa. GentesH erbarum 12: 41–129. ———.1985b. Ethnobotanyofthe naranjilla ( Solanumquitoense ) YAMADA, K., K. SHINOZAKI, and M. SUGIURA.1986. DNAsequenc- and itsrelativ es. Economic Botany 39: 4–11. es oftobacco chloroplastgenes for tRN A Ser (GGA), tRNAThr ———.1986. Anew domesticated variety and relationships of So- (UGU), rRNALeu (UAA), tRNAPhe (GAA):the tRNA Leu gene lanum lasiocarpum .Pages 412–415 in Solanaceae: biologyand sys- contains a503 bp intron. Plant Molecular Biology 6: 193–199.