FLORA VITIENSIS NOVAANEWFLORAOFFIJI(SPERMATOPHYTESONLY)ALBERT C. SMITHVOLUME2Angiospermae: Dicotyledones, Families44-116Lawai, Kauai, Hawaii1981CONTENTSOF VOLUME2Introduction. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . ISupplementaryReferences. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3Division Angiospermae (Magnoliophyta)(continued) . . . . . . . . . . . . . . . . . . . . . . . 4ClassDicotyledones(Magnoliatae). . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4SubclassMagnoliidae. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6Order Magnoliales . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7Family 44. Degeneriaceae. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7OrderAnnonales.. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. 13Family 45. Annonaceae 13Family 46. Myristicaceae. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. 40OrderAristolochiales 52Family 47. Aristolochiaceae. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. 52OrderPiperales 56Family 48. Piperaceae. .. . . .. . .. . . .. . .. . .. . . .. .. . . . . .. . . . . . .. 56Family 49. Peperomiaceae. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. 75OrderChloranthales 97Family50. Chloranthaceae 97OrderLaurales 101Family51. Trimeniaceae 102Family52. Monimiaceae 104Family53. Hernandiaceae 108Family54. Lauraceae 113Family55. Cassythaceae 141Family56. Gyrocarpaceae 143OrderNymphaeales 145Family57. Nymphaeaceae 146Family58. Ceratophyllaceae 147SubclassRanunculidae 148OrderRanunculales 149Family59. Menispermaceae 149Family60. Ranunculaceae 150OrderPapaverales 153Family61. Papaveraceae 154SubclassHamamelididae 155OrderUrticales 155Family 62. Ulmaceae 156Family 63. Cannabaceae 166Family64. Moraceae 167Family65. Urticaceae " 209OrderCasuarinales 251Family 66. Casuarinaceae 251OrderBalanopales 255Family 67. Balanopaceae 255SubclassCaryophyllidae 259Order Caryophyllales 260Family68. Phytolaccaceae 261Family69. Nyctaginaceae 262Family70. Aizoaceae 269Family71. Cactaceae 270Family72. Molluginaceae 272Family73. Caryophyllaceae 273Family74. Portulacaceae 275Family75. Basellaceae 280Family76. Amaranthaceae 281Family 77. Chenopodiaceae 295OrderPolygonales 296Family 78. Polygonaceae 296OrderPlumbaginales 307Family 79. Plumbaginaceae 307SubclassDilleniidae 310OrderDilleniales . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 311Family80. Dilleniaceae 311Order Theales . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 317Family 81. Ochnaceae . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. 318Family 82. Theaceae 321Family83. Saurauiaceae 326Family84. Clusiaceae 329Family85. Elatinaceae 348OrderMalvales 349Family86. Elaeocarpaceae 349Family 87. Tiliaceae 365Family88. Sterculiaceae 382Family89. Bombacaceae 411Family 90. Malvaceae 414Order Euphorbiales 439Family 91. Euphorbiaceae 439Order Thymelaeales 576Family92. Gonystylaceae 576Family 93. Thymelaeaceae 580OrderLecythidales 592Family 94. Lecythidaceae 593Family 95. Barringtoniaceae 594OrderRhizophorales 601Family 96. Rhizophoraceae 602OrderViolales 617Family 97. Flacourtiaceae 618Family 98. Violaceae 655VIFamily 99. Turneraceae 662Family100. Passifloraceae 663Family101. Bixaceae 669Family102. Cochlospermaceae 670Family103. Caricaceae 671Order Cucurbitales 673Family104. Cucurbitaceae 673OrderBegoniales 688Family105. Begoniaceae 688OrderCapparales 692Family106. Capparaceae 692Family107. Cleomaceae 698Family108. Brassicaceae 703Family109. Moringaceae 7I IOrderSaIicaIes 7I3Family110. Salicaceae 713OrderEricaIes 714FamilyI I I. Ericaceae 714Family112. Epacridaceae , .. 720OrderEbenales 724Family113. Symplocaceae 724Family114. Ebenaceae 729Family115. Sapotaceae 744OrderPrimulaIes 782Family116. Myrsinaceae 782vuINTRODUCTIONIn theIntroduction toVolume1 of thisFlora (p. 3) I stated that publication wasexpectedtobeinthree volumes. The first volume included the anticipatedmaterial,except that thefamily Orchidaceae was perforceomitted. As workcontinuedondicotyledons, however, it wasrealizedthat twoadditionalvolumeswouldnot ade-quately cover them and the orchids, assuming a degree of coverage similar to that ofthe gymnosperms and monocotyledons in Volume 1. It is now planned to extend FloraVitiensis Novatofour volumes. Volume2, herewith, includesthedicotyledonousfamilies assigned to the subclasses Magnoliidae, Ranunculidae, Hamamelididae,Caryophyllidae, andDilleniidae, in the approximate sequence of Takhtajan (1969)1.As nowplanned, Volume3will includethefamilies of thesubclass Rosidae, andVolume 4 will include those of the subclass Asteridae. Volume 4 will presumably alsocontain indices to the entire work as well as pertinent concludingmaterial; theOrchidaceae will be incorporated into either Volume 3 or Volume 4, unfortunately outof sequence.Ithad been hoped and expected that no new names wouldrequire publication inFloraVitiensisNova. However, it hasproved impracticable to avoid the use of a fewnewnames, asvariousprecursorpapershavenot coveredall families and genera insuch detail that all novelties or required combinations were adequately treated.Because it is inconvenient for the compilers of indices to search floristic works for newnames, I herelistthosepresentedin thisvolume:Peperomiaceae (fam. nov.; apparentlynotpreviously validly published)Euphorbiaceae: Glochidion atrovirens, G. brunnescens. G. multilobum. G. inus-itatum. G. collinum (spp. nov.)Flacourtiaceae: Casearia fissistipula. C. crassipes (spp. nov.)Ebenaceae: Diospyros gillespiei var. nandarivatensis (comb. nov.),D. phlebodes(comb. etstat. nov.)Sapotaceae: Planchonella brevipes (sp. nov.), P. umbonata (comb. et stat. nov.),P. smithii (comb. etstat. nov.)In this connection it has been noted with interest that L. K. Weresub and J. McNeill (inTaxon 29: 474. 1980) "... question the desirability of publishing taxonomic novelties inFloras..."Pertinent as such questioning may be, it can affect only future publication,since thousands of botanical names have been first used in regional Floras by botanistsof thelasttwocenturies.Itwouldindeedbemore convenientto compilers of plantindices if Floras presented merely compendia of already published knowledge, but somany exceptions are justifiable that it seems unwise to consider any legislated restric-tionsinthisrespect.Since the publication of Volume 1of the present Flora (or more or less coincidentwith it), many books and journal articles pertinent to the subject matter, or to relatedtaxonomic or nomenclatural topics, have been produced. A few are here discussed andlisted.The appearance of a new edition of the International Code ofBotanical Nomencla-, References indicated by parenthetical dates, if not otherwise modified, are listed in Volume I, pp. 84-88,andinthepresent volume, p. 3.2 FLORAVITIENSISNOVA Vol. 2ture(lCBN) is always noteworthy. The"Leningradedition"(Stafleuet aI., 1978)differs primarily in a fewmatters of detail from the preceding "Seattle edition," but itintroducesthe use of a decimal system for indicating each paragraph of Articles andRecommendations; thisdecimal system when pertinent will be used in future referen-cestotheICBNinthepresent Flora.Thesecondvolume ofthesecondeditionof Taxonomic Literature(Stafleu&Cowan, 1979), covering authors whose family names begin with H-Le, is now avail-able; togetherwiththefirst volume(cLVolume Iof this Flora, p. Il) it includesinvaluable information about botanical publications. Botanists await the completionof thismajor compilation withkeenanticipation.For 25 years the card index so valuable to botanists, Index Nominum Genericorum(lNG), has been supplemented by the efforts of many collaborators working on behalfoftheInternational Associationfor Plant Taxonomy. Publicationof theIndexinbookform(Farr, Leussink, &Stafleu, 1979)was an event of major significance (cLTaxon 29: 266. 1980). The compilers of the book edition are well aware that errors oromissions will require future correction. Indications of type species for genera (whennot obvious) are primarily documented suggestions, and specialists are free to proposeandtodocumentalternative choices.Studentsof flowering plants continue attempts to improve a system of classifica-tionthatwillmorenaturally group families, orders, andhigher taxa in accordwithadvances in knowledge in pertinent fields. A recent classification with many merits isthat of Dahlgren (1980), who recognizes only two subclasses of angiosperms (Magno-liidae andLiliidae, i. e. dicotyledons andmonocotyledons),utilizing superorders forthenext stepinhisdivision. InmanyrespectshisclassificationresemblesthoseofCronquist (1968), Takhtajan(1969), andThorne (1976), butseveral interesting newconcepts areintroduced.A. L. Takhtajan (1980)has recently summarized his concepts of the classificationof flowering plants in a comprehensive review inEnglish. This valuable contributionwas not perused in time to be considered in thepreparation of the present volume ofthis Flora, and at any rate it does not significantly alter Takhtajan's earlier (1967, 1969)familysequences, but it incorporatesdatafromrecent pertinentliterature andwillprovideuseful documentationtophylogenists.A useful analysis of plant communities on two islands of the Lau Group of Fiji hasbeen publishedby P. J.Garnock-Jones (1978), mentioned as a collector in VolumeI,p. 83, of this Flora.Comments of considerable interest have recently been made by P. S. Green (1979)inrespect tothe occurrence of presumedFijian endemics in the NewHebrides. As aresult ofthe 1971 Royal SocietyExpeditiontotheNewHebrides, several speciespreviouslythought to beendemictoFiji arenowknownfromthemorewesterlyarchipelago. Amongthese species is Collospermummontanum(Seem.) Skottsb.(Liliaceae), which is no longer to be considered a Fijian endemic as stated in Volume Iof the present Flora, p. 143. Other cases, occurringamongdicotyledons, will bementionedinthisandremaining volumes. The extension of some Fijian taxa to theNewHebridesis of course tobe anticipated, and those mentioned by Green illustrateagain that no Flora may be consideredcomplete indetail at the moment ofitspublication.A noteworthy publication is that by Briggs and Johnson (1979) on evolution in the1981 INTRODUCTION 3Myrtaceae; the main thrust of this treatment bears on the orders Myrtales andLythrales (subclass Rosidae, to be discussed in Volume 3 of this Flora), but importantdiscussionsareincludedonthefamilies Lecythidaceae andRhizophoraceae, whichhave often been included in the subclass Rosidae as relatives ofthe Myrtales but which,inthelight of recentevidence, appear betterplacedinthesubclassDilleniidaeandhenceareincludedinthepresentvolumeof thisFlora.In recent yearsW. R. Sykes andG. P. Buelow have made extensive collections ofplantsin Tonga, andbothhave kindly calledmy attention to occurrences that affecttheknowndistributionofFijianplants. AshasbeenthecaseintheNewHebrides,theserecent Tongancollectionshavedisclosedrange extensions of several taxa thathad previouslybeenconsideredendemictoFiji, orof taxapreviouslybelievedtoterminate the eastward extent of generic distributions in Fiji. Similarly, W. A.Whistler, whohasrecentlypublishedaninteresting discussionof thevegetationofEasternSamoa (1980), hasmadelarge collectionsthroughout Samoa, andina fewcases his findings have extended the known distribution of species or genera occurringinFiji.I take this opportunity to acknowledge the valuable assistance of G. L. Webster inmy preparation of manuscript for the family Euphorbiaceae; the key to genera of thatfamilyrecordedfromFiji waskindlycontributedbyhim.SUPPLEMENTARYREFERENCESBRIGGS, B. G., & L. A. S. JOHNSON. 1979. Evolution in the Myrtaceae-evidence frominflorescencestructure. Proc. Linn. Soc. NewSouthWales 102: 157-256.DAHLGREN, R. M. T. 1980. Arevisedsystem of classificationof theangiosperms. Bot. J. Linn. Soc. 80:91-124.FARR. E. R., J. A. LEUSSINK. &F. A. STAFLEU. 1979. IndexNominum Genericorum. Vo!. 1,2,3. RegnumVeg. 100, 101, 102. Utrecht.GARNOCK-JONES, P. J. 1978. Plant communitiesonLakeba andsouthernVanua Balavu, Lau Group, Fiji.Bull. Roy. Soc. NewZealand17: 95-117.GREEN, P. S. 1979. Observationsonthephytogeographyof theNewHebrides, LordHoweIslandandNorfolkIsland. InBramwell, D. Plants andIslands, 41-53. AcademicPress.STAFLEU, F. A., &R. S. COWAN. 1979. Taxonomic Literature. Ed. 2, Vo!. 2: H-Le. Regum Veg. 98: I-XVIII,1-991. Utrecht.__, et a!. 1978. International Code of Botanical Nomenclature (Twelfth International BotanicalCongress, Leningrad, 1975). Utrecht.TAKHTAJAN, A. L. 1980. Outlineof theclassification of floweringplants(Magnoliophyta). Bot. Rev. 46:225-359.WHISTLER, W. A. 1980. Thevegetationof EasternSamoa. Allertonia2: 45-190.4FLORAVITIENSISNOVA Vo!. 2DIVISIONANGIOSPERMAE (MAGNOLIOPHYTA)(continued)CLASSDICOTYLEDONES(MAGNOLIATAE)Users of thisFlora may recall my comments in Volume 1 (p. 6) as to the inherentlyunsatisfactorynatureof keystofamilies(andlarger taxa) thatpresume toindicateevolutionary relationships. It will be obvious that keys to taxa, in any systematic work,become less and less useful as the rank of the taxon becomes more inclusive. In dividinga species into subspecies or varieties one seldom has difficulty in producing a satisfac-torykeywithmutuallyexclusivealternatives. Inkeyingthespeciesof agenusthedifficultiesare somewhatgreater, but even here most keys are quite satisfactory. Toseparate the genera of a family by keying in a reasonable manner is still quite feasible,and even to prepare a key to the families of an order is possible, particularly if the orderis restricted in inclusiveness. Above the level of order the difficulties are compounded,as is evident to the user of any work in which keys of this nature are attempted. At thisstagethepreparerisassailedbydoubts of thenaturalnessof thehigher grouping,whether subclassor superorder. Indeed, modern studiesof plants fromincreasinglymultitudinousanglesmakeitobvious that noconsensushasbeenreachedas to thecomponents of any given superorder or subclass. Perhaps for this reason most angio-spermphylogenistspresent theirconclusions intheformof summaries anddiscus-sions, carefully avoiding keys to their major taxa. Such students as Cronquist (1968)aretobe complimentedontheirwillingnesstoattempt the impossibleby preparingkeys to subclasses and orders; such keys, however guarded and however surrounded bywarnings, cannot yet beverysatisfactory.Nevertheless, inthe present volumea keytothesubclasses of dicotyledons ispresented, in large part adapted from Cronquist's (1968) work, but the reader shouldnot expect it to be very informative. In such a subclass as the Dilleniidae, the key to itscomponent orders(occurring inFiji) here presented is definitely unsatisfactory, andthewriteris awareof hisrashnessin evenattempting it. Perhaps the difficulties arecompoundedbythenecessarily vague circumscriptions of the subclasses Dilleniidaeand Rosidae, which for their cohesiveness depend upon esoteric characters that are stillnotentirely analyzedandcertainly not very useful to fieldand herbarium botanists.The sequence of component orders of the Dilleniidae as outlined by Takhtajan (1969)is in general here followed, but certain transfers of taxa to and from the Rosidae haveseemed desirable. Quite possibly it is unrealistic to divide dicotyledonous plant fami-lies into seven cohesive groupings (subclasses), and future phylogenists may well agreewithThorne(1976) that a greater number ofbasiclargetaxa(superordersinhisterminology) isrequiredinordertoprovideanything likerealisticgroupings.Withthesecommentsthewriterwarnsnonbotanicalusersof thisFlora that thefollowingkeytosubclasses will beof verylittleusetothem, andthat thekeystocomponent orders, especiallythosereferredtothesubclass Dilleniidae, shouldbegivenlimitedcredence.1981 DICOTYLEDONES 5USEFULTREATMENTSOFCLASS(inadditiontomany works citedinVolume 1 of this Flora. pp. 84-88):Bailey. I. W.. & B. G. L. Swamy. The conduplicate carpel of dicotyledons and its initial trends ofspecialization. Amer. J. Bot. 38: 373-379. 1951(a highly significantpaper. for its date. indicating probableevolutionarytrends inthedicotyledonous carpel). Brewbaker. J. L. Thedistributionand phylogeneticsignificanceof binucleateandtrinucleatepollengrainsinthe angiosperms. Amer. J. Bot. 54: 1069-1083.1967 (an important contribution to our understanding of phylogenetic trends in the angiosperms as a whole.referring to monocotyledons as well as dicotyledons; the characters described will not be of use to the field orherbariumbotanist. but nevertheless. because of their evolutionary implications. they are mentioned in mykey tosubclasses of dicotyledons).Smith. A. C. An appraisal of the orders and families of primitive extantangiosperms. J. Indian Bot. Soc. 50A: 215- 226. 1972 (an attempted justification of the use of comparativelysmall orders and families of putatively"primitive" dicotyledons). Walker. J. W. Comparative pollenmorphology and phylogeny of the ranaleancomplex. In Beck. C. B. Originand Early EvolutionofAngiosperms. 241-299. 1976 (althoughprimarilydirectedtowardtheevolutionof polleninthesubclassMagnoliidae. this paper suggestsevolutionarytrends pertinent todicotyledons asawholeanddefinesterminologyhereutilized).KEYTOSllBCLASSESPlants putatively suggestive of the ancestral dicotyledonous condition; flowers usually apocarpous and witha well-developed perianth (except in Piperales and Chloranthales in our area). this always polypetalousor apetalous; stamens usually numerous and developing in centripetal sequence (but reduced in numberinAristolochiales. Piperales.Chloranthales. and some Laurales); pollen always binucleate when shed;carpels primitively conduplicate(this sometimes apparentonly at tip of style). withmodified laminarplacentation. unsealed (at least in distal portion) in earlystages. the pollen tube sometimes notpenetrating solidcarpellarytissues; ovulesalwaysbitegmicandcrassinucellate.Spherical idioblasts (ethereal oil cells) present in parenchymatous tissues (except in Nymphaeales); xylemcomparatively unspecialized;pollen grains anasulcateor uniaperturate or derivedfrom such types.MAGNOLIIDAESpherical idioblasts lacking; plantsherbaceous or secondarily woody; pollen grains never anasulcate oruniaperturate-derived. . RANUNCULIDAEPlants putatively more "advanced" in one or mere respects than those of the Magnoliidae or Ranunculidae;pollen grains never anasulcate or uniaperturate-derived; spherical idioblasts lackingin parenchy-matoustissues.Flowers more or less strongly reduced. often unisexual and with the perianth poorly developed or lacking.oftenborne in catkins but never grouped into bisexual pseudanthia. never with numerous ovules onparietal placentas; woody plants (except some Urticales); pollen usually binucleate when shed. rarelytrinucleate. . HAMAMELIDIDAEFlowers usually comparatively well developed andwith an evident perianth. or otherwise usually eithergroupedintobisexual pseudanthia orwithnumerousovulesonparietal placentas.Flowersusuallypolypetalous. lessfrequentlyapetalousor sympetalous(if sympetalous usually withstamens more numerous thancorolla lobes. or with stamens opposite corolla lobes. or withbitegmicor crassinucellateovules);carpels I-many. freeorunited.Placentationof varioustypes. oftenparietalorfree-central orbasal. sometimesaxile; stamens. ifnumerous. developinginacentrifugal sequence; species with the combination of few stamensandaxile placentationusuallywithseveral or manyovules per loculeor withsympetalouscorollasorboth.Pollen always trinucleate when shed; ovules bitegmic. crassinucellate. usually campylotropous oramphitropous; seeds often withperisperm; plants either withbetalainsinsteadof anthocya-ninsor withfree-central tobasalplacentation or both; none of the speciesoccurring inFijitreelike(except someNyctaginaceae) CARYOPHYLLIDAEPollen usually binucleate whenshed.infrequently trinucleate; ovules variousbut seldom campy-lotropous or amphitropous; seeds seldom with perisperm; plants never with betalains; placen-tation rarely free-central or basal (except in Primulales); species occurring in Fiji oftep. trees orshrubsbut sometimesherbsorvines. . DILLENIIDAEPlacentation seldom parietal;flowers seldom with free-central or basal placentation in a unilocular.compound ovary. but often with 2-several locules with only I or 2ovules each. usuallypolypetalous. lessoftenapetalous. rarelysympeta!ous; stamens. if numerous. developing in acentripetal sequence; pollenpredominantlybinucleatewhenshed. but oftentrinucleate.ROSIDAEFlowerssympetalous (with few exceptions). the stamens usually the samenumber as corollalobesorfewer. never opposite corolla lobes; carpels frequently 2. sometimes 3-5 or more; ovules unitegmicandtenuinucellate; pollenmost frequentlytrinucleatewhenshed. but oftenbinucleate.ASTERIDAE6FLORAVITIENSISNOVA Vol. 2SUBCLASSMAGNOLIIDAEThere has been great diversity in the grouping of the orders and families generallyconstrued ascomposing the "ranalean complex"(e. g. Cronquist(1968), Takhtajan(1969), andThorne(1976) aslistedinmyVol. I, pp. 84-88, andSmith (1972)andWalker(1976)aslistedaboveunder theclassDicotyledones). In the present work Iaccept Takhtajan'ssuggestion(1969: 207) of separatingthesubclassRanunculidaefromtheMagnoliidae, withminormodifications(Smith, 1972:218). Theproblemsinvolvedbear onlymarginallyonthefloraofFiji, but conclusionsimpliedinthearrangement of ordersandfamiliesin thepresent workmayimpelsomereaders toexamine the larger questions. Although there is no single set of characters that rigidlydefines the subclass Magnoliidae, there exist in its included orders some of the featuresthat probably characterized the earliest dicotyledons. Recognition of these features as"primitive"innosensesuggeststhedirect origins of other subclasses inthe extantMagnoliidae; the implication is merely that in this subclass certain features suggestingdicotyledonousforbearsseem tohavepersisted, whileinothersubclassesthese fea-tures havebeenobscuredbyacceleratedevolutionarychange. Furthermore, manybotanists are now inclined to question the origin of monocotyledons from any putativemagnoliideanancestry; it seems morereasonable toassumethatthetwoclasses ofangiosperms had independent origins in a common ancestral taxon (cf. Moore, H. E.,Jr.,&N. W. Uhl. The monocotyledons: their evolution and comparative biology. VI.Palms and the origin and evolution of monocotyledons. Quart. Rev. BioI. 48:414-436.1973). Thishypothesisdoes not carryimplicationsof abiphyletic origin of angio-sperms; theparallelpointedout byMoore andUhlis tobe foundin thebirdsandmammals, which presumably evolved fromrelated reptilian lines but may, inareasonablesense, be consideredmonophyletic.USEFULTREATMENTSOF SUBCLASS (in addition to works cited inVol. I of thisFlora or above under theclass Dicotyledones): Ehrendorfer, F., F.Krendl, E. Habeler, &W. Sauer. Chromosome numbers and evo-lution in primitive angiosperms. Taxon 17: 337-353. 1968. Walker, J. W. Evolution of exine structure in thepollen of primitive angiosperms. Amer. J. Bot. 61: 891-902.1974. Walker, J. W. Aperture evolution in thepollen of primitive angiosperms. Amer. J. Bot. 61:1112-1137. 1974. Walker, J. W., & J. J. Skvarla. Primi-tive columellaless pollen: a new concept in the evolutionary morphology of angiosperms. Science 187: 445-447. 1975. Okada,H. Karyomorphological studies of woody Polycarpicae. J. Sci. Hiroshima Univ., Ser. B,Div. 2, Bot. 15: 115-200. 1975. Walker, J. W. Evolutionary significance of the exinein the pollen of primitiveangiosperms. In Ferguson, I. K., &J. Muller. Theevolutionarysignificance oftheexine. Linn. Soc.SymposiumSer. 1: 251-308. 1976. The listed papersprovide valuable supplementary data supporting thehypothesis that the "ranalean complex" retains palynological and caryological features suggesting a "primi-tive"position among dicotyledons.KEYTOORDERSOCCURRINGINFIJIPlants mostly woody and terrestrial, with comparatively unspecialized xylem (but with true vessels in all ourrepresentatives) and with cambium;spherical idioblasts (ethereal oil cells) present in parenchymatoustissues.Flowers hypogynous (except epigynous in most Aristolochiales); ovules often several or many per carpelbutsometimes solitary;nodestrilacunar ormultilacunar.Perianthevident, sometimesdividedintosepalsandpetals; seedswithwell-developedendosperm;stipules (in ourrepresentatives)lacking.Flowers actinomorphic; carpels superior andusuallyfree, lessoftenunited.Perianth partsbasically spiralled(butinDegeneriaceae, the only familyof the order inFiji,thecalyx 3-lobed); stamens laminar, usually with3 or more traces; pollen grains (inDegeneria-ceae) anasulcate, boat-shaped, atectateand primitivelycolumellaless, psi late; carpels (inDegeneriaceae) solitaryorrarely2, withmanyovules. . ..... MAGNOLIALES (FAMILY44)Perianthparts basically in threes; stamens not laminar, constantlyI-traced; pollen grains diverse,rarely (but in no genera in our area) columellaless; carpelsI-many, each with I-many ovules.ANNONALES (FAMILIES45,46)1981 DEGENERIACEAE 7Flowers(intheonlygenus inour area)zygomorphicandwithout petals. thecalyxenlargedandpetaloid. theanthersunitedwiththestyleintoagynostemium. thepollengrainsanasulcate.inaperturate. globose or globose-oblate. tectate or semitectate.the ovary inferior. with numer-ousovulesin eachlocule. thefruit asepticidal capsule.....ARISTOLOCHIALES (FAMILY47)Perianthnone. the flowers crowdedinto a spadix; pollen grains anasulcate. inaperturate. subglobose.tectate; ovary I-locular. the ovulesolitary. erect (infamilies inour area); seedswithcopiousperisperm andscanty endosperm; stipulespresentorabsent. . ... PIPERALES (FAMILlES 48. 49)Flowersperigynous toepigynous (inour area hypogynous onlyinTrimeniaceae ofLaurales); pollengrains (in genera in our area) inaperturate or forate. subglobose, tectate or semitectate; ovules (in ourfamilies)solitary, pendulous; seedssometimeslacking endosperm; nodesunilacunar.Perianthlacking; plants(inAscarina. theonlygenus inour area) usuallyappearing dioeciousbutprobably basically monoecious. thecJ'flowers (in Ascarina in our area) with a single stamen; ovaryinferior, unilocular, theovuleorthotropous; stipulespresent. ...CHLORANTHALES (FAMILY50)Perianth evident but often small and lacking petals, rarely lacking; plants with bisexual. polygamous. orunisexual flowers; stamens few to numerous; carpelsI-many. each carpel or ovary locule with ananatropousovule; stipules (inourrepresentatives)absent LAURALES (FAMILIES51-56)Aquaticherbs, lacking vessels and without cambium; spherical idioblasts lacking; leaves floating. emersed.or submersed; pollengrains(inourrepresentatives)zonasulcate orinaperturate. tectate.NYMPHAEALES (FAMILlES57. 58)ORDERMAGNOLIALESTheorder Magnoliales is sometimes interpretedin anextraordinarilyextendedsense (e. g. Cronquist, 1968:135-144), but more logically it is construed as comprisingtwosuborders (Magnoliineae andAnnonineae)andseven families (cf. Smith (1972)and Walker (1976) cited above under the class). Since 1972, however, continuing studyhas convinced me that differences between the two suborders are of such a nature thatthe recognition oftwo orders is advisable. As thus defined, the Magnoliales form a verycoherent order composed of four families: the widespread and predominantly North-ern Hemisphere Magnoliaceae, the EupomatiaceaeandHimantandraceaeofNewGuinea and easternAustralia, and the Degeneriaceae, endemicto Fiji. Withthepossible exception of the order Winterales (composed of the sole family Winteraceae),theMagnolialesprobablypossessagreater accumulation of plesiomorphic ("primi-tive") characters than any other extant dicotyledons. Their secondary xylem, althoughnot vesselless, retains features indicating early stages of advancement. The stamens arebroad, dorsiventral, 3-veined microsporophylls with elongated, immersed sporangia.Thepollengrainsof theMagnoliaceae andDegeneriaceae are anasulcate andboat-shaped, and those of Degeneria and Eupomatia have atectate and primitively columel-laless exine. The carpel of Degeneria is best described as an adaxially folded, 3-veinedmegasporophyll with laminar placentation and flaring stigmatic surfaces that are notcoherent at the time of pollination. Such carpels are very similar to those of TasmanniaandBubbiaintheWinteraceae.FAMILY44. DEGENERIACEAEDEGENERIACEAE I. W. Bailey &A. C. Srn. in J. ArnoldArb. 23: 357. 1942.Trees; stipulesnone, the leaves alternate, petiolate, simple, pinnatinerved; flowerssolitary (butbearing 2 or 3 bracts near middle of pedicel), supra-axillary, ~ , hypogy-nous, the receptacle subglobose or convex, depressed in center; perianthclearlydifferentiated into calyx and corolla; calyx deeply lobed, the sepals 3 (rarely 4), muchsmaller than petals; petals numerous, 2-4-seriate, imbricate, carnose; stamens numer-ous, spiralled in2or 3series, carnose, laminar, roundedorsubtruncateat apex.3-veined, with 4 slender, parallel, elongated, extrorse,immersed sporangia dehiscing8FLORAVITIENSISNOVA Vol. 2by 2longitudinal clefts; pollen grains anasulcate, bilateral, boat-shaped, psilate,atectate and primitivelycolumellaless; staminodes withinthe stamens andfewer,similarintexturebut conspicuouslyintrorselycucullate;carpel solitary(or carpelsrarely2 andattachedat slightlydifferent levelsonthereceptacle), inaequilaterallyellipsoid, conduplicate, openin earlystages, the ventral marginsexternally flaring,with numerous, loosely interlocking, short, glandular hairs, the stigmatic areas extend-inginwardalongadaxial surfaces of carpel; ovules numerous, anatropous, biseriatebutsometimesvascularizedbybranches of bothventral and median traces; fruit anasymmetrical, oblong-ellipsoid, tardily dehiscent follicle with a long-persistent vascu-lar skeleton composed of branches of median, ventral, and supernumerary traces, thepericarp coriaceous and smooth, the endocarp intrusively lobed with spongyingrowths; seeds large, with copious, irregularly grooved and cleft, subruminateendosperm, the outer integument of mature seeds with a thick cuticle, a succulent outercoat bearing oil cells, andaninner stony coat; cotyledons3 or 4,veryrarely2.DISTRIBUTION: Endemic toFiji, withasingle species.USEFUL TREATMENTS OF FAMILY (in addition to many papers already listed in this Flora):Bailey, I. W., &A. C. Smith. Degeneriaceae, a new family of flowering plants from Fiji. J. Arnold Arb. 23: 356-365. 1942.Smith, A. C. Additional notes onDegeneriavitiensis. Op. cit. 30: 1-9. 1949. Swamy, B. G. L. Furthercontributions to the morphology of the Degeneriaceae. Op. cit. 30: 10-38. 1949. Lemesle, R., & A.Duchaigne. Contribution ill'etude histologique et phylogenetique du Degeneria vitiensis I. W. Bailey et A.C. Srn. Rev. Gen. Bot. 62:699-719. 1955. Dahl, A. 0., &J. R. Rowley. Pollen of Degeneria vitiensis. J.ArnoldArb. 46: 308-323. 1965.Although I. W. Bailey had long been concerned with the morphology and anatomyofmembersof the"ranaleancomplex,"thediscoveryof anewspecies, genus, andfamilyof thiscomplex andtheir descriptionin1942 gaveapronouncedimpetus toresearch in this area by him andhis associates and students. It is no overstatement toFIGUREI. Degeneria vitiensis, from DA15292; flower, showing petals and extrorse surfaces of stamensandstaminodes, x2.1981 DEGENERIACEAE 9suggest that the first paper on theDegeneriaceae led to arecrudescence of interest inthe "Ranales," as the putatively most primitive dicotyledons were then known. Since1942 several hundred studies of diverse aspects of "ranalean" plant taxa by botanists ofmany countries and disciplineshave contributed to our present understanding of theevolutionaryhistory of dicotyledons. Thisunderstandingisstill inadequate, but thenewknowledgesopromulgated has led, inless than40years, toavastlykeenerappreciation of some of the principles suggested by such perceptive pioneers asJussieu, deCandolle, andBessey.I. DEGENERIA I. W. Bailey & A. C. Srn. inJ. Arnold Arb. 23: 357.1942; A. C. Srn. in op.cit. 36:277. 1955.Characters anddistributionof thefamily.TYPE SPECIES: Degeneria vitiensis I. W. Bailey &A. C. Srn., the only known species.The original formaldescription of Degeneria vitiensis having beenbased on onlytwo collections, substantial amplification isnow possible. It has always surprised methat aplant soabundant inFiji escapedthenoticeof suchdiscerning collectorsasSeemann, Horne, Gibbs, Greenwood, and Gillespie, among others. However, exhaus-tive search of herbaria whereFijian collections might be deposited has convinced methat the species remained uncollected until May 7, 1934, when specimens wereobtained by me (no. 1754) in the lower Wainunu River Valley, Mbua Province, VanuaLevu; these specimens, in youngfruit and withample wood material, remainedunidentifiedtofamily. The secondcollectionnowknowntome isDA287,a sterilespecimen collected in Naitasiri Province, Viti Levu, in 1936 but without further localityor collector's name. A third collection was made by B. E.V. Parham (as DA1488) onMay 11, 1939, inthe vicinityofNanduna, near WaindrandraCreek, LomaivunaTikina, NaitasiriProvince,VitiLevu; this was also accompanied by immature fruits,butneither of the Department of Agriculture collections was available toBailey andme in1942. The fourthknown collection, and the first in flower, was Degener 14537,collectedFebruary 24, 1941, near Nauwangga, south of Nandarivatu, Mba Province,Viti Levu; thiscollectionweindicatedasthetype.Itisperhapsunderstandablethat earliercollectorsfailedtoobtainthespecies,which is a tree of the upper storey of the rain forest, mingling its branches and foliagewiththoseof many other species. Evenwhenthetreeisinfull floweror fruit thoseorgansarenotdiscerniblefromtheground; theflowersshattersoonafteropening.Once the plant is detected from a fallen fruit or the decaying vascular skeleton of one, itcan be located in the forest canopy. As a result I have now observed at least 100 individ-uals andhave collected material from many of them in such quantity that specimensareavailableinmost major worldherbaria andmanywoodsamplesareat hand.I. Degeneria vitiensis I. W. Bailey &A. C. Srn. in J. Arnold Arb. 23: 357.pl. 1-5. 1942;A. C. Srn. in Bull. Torrey Bot. Club 70: 537. 1943, inJ. Arnold Arb. 30: l.fig.I;pl.1. 1949; Swamy in op. cit. 30:lO.fig. 1-106. 1949; Lemesle &Duchaigne in Rev.Gen. Bot. 62: 699.fig. 1-8. 1955; J. W.Parham, PI. Fiji Isl. 46.fig. 23.1964, ed. 2.76. fig. 23. 1972. FIGURES1-3, 77.Atall forest treeattaininga height of30m. at maturity, withacomparativelycompact crownandastraight, slender trunkto70cm. indiameter, andwith3-7obvious rounded buttresses inthe lowest 1-2 m. Thedarkgraybarkhas regularfissures and presents no outstanding characteristics, although with experience it seemsreadilyrecognizedbyFijiansandotherforesters. Juvenileplantshavelargerleaves10 FLORAVITIENSISNOVA Vol. 2thanmaturetrees, thebladesbeing asmuchas45x 14 cm., long-decurrent onthepetiole, and deltoid-cuspidate at apex, with asmany as30 pairs of secondary nerves.Matureleaf blades may measure only 5-27x2.5-14 cm. and are rounded or slightlyemarginate at apex, with 8-18 pairs of secondary nerves. Specimens may be spectacu-larly loaded with flowers, even though these are seldom discernible from the ground.When fully open the flowers may attain a diameter of 6.2 cm., then emitting a deliciousfragrance suggestive of the flowers of Cananga odorata and some other Annonaceae.The 3 (rarely 4) green sepals may be somewhat more than 5 mm. in length and breadth,obscurely yellow-glandular, entire andinconspicuously ciliate. The petals are12-14,white or cream-whiteor the outer ones greenishwithout, spiralled in2-4series,carnose andbroadly imbricate, the exterior ones elliptic, 18-25 mm. long and10-13mm. broad, obscurelyyellow-glandular, the inner ones narrower and somewhatoblong. The stamens are 20-31, spiralled in 2 or 3 series, dull pink in bud, becomingyellowish to white or cream-colored at anthesis and obscurelyyellow-glandular,rounded or truncate at apex and there inconspicuously ciliate. The outer stamens maymeasure as much as 7x3 mm.,the inner ones being slightly smaller. The staminodesare 9-11 in number, 1- or 2-seriate, at first rich pink dorsally, nearly white ventrally butyellow distally, becoming cream-white at anthesis, with 3 parallel veins and conspicu-ously hooded, rarely bearing rudimentary sporangia; they are obovate, conspicuouslynarrowed at base andlarger than stamens, sometimes attaining a size of 12x5 mm.Thepurewhitecarpel bears22-30ovuleson 2 parallel placentas borne between themediantraceandthe2 ventral traces, somesessileandsomefunicled but not (asoriginally described) limited in this respect to one orthe other placenta. At anthesis thecarpel maybeas largeas 7x 3 mm. andwithitswall copiously immersed-yellow-glandular. The fruits at first are green, then pink to purple, and at maturity black, up toII x5 cm. The seeds (most ovules maturing) are embedded in pale green pulp; theirsucculent outercoat issalmon-pinktobright orange. Swamyhaspointedout theextraordinary factthat he did not discern dicotyledony in Degeneria (actually, how-ever, it does very rarely occur).Of the more than 300 seeds he examined, about 87%had3 cotyledons and about 13% 4 cotyledons. In the related Magnoliaceae tricotyle-donous embryos are only occasionally encountered. The distantly related Queenslandfamily Idiospermaceae (Laurales) has 3 or4 massive, peltately attached cotyledons of averydifferenttype thanthoseof Degeneria.TYPIFICATION: The holotype, as mentioned above, is Degener 14537 (A), collectedFebruary 24, 1941, near Nauwangga, valley of Nandala Creek south of Nandarivatu,MbaProvince, Viti Levu. There aremanyisotypes.DISTRIBUTION: Endemic toFiji andthusfarknownonly fromVitiLevu, VanuaLevu, and Taveuni, occurring at elevations between 30 and1,150 m. in dense or openforest or in second-growth forest. Because of the interest of the species I cite below allthecollectionsIhave examined. Certainlymanyother collectionsexist, sincemostbotanists who visit Fiji wish to see the species and have prepared herbarium material.A few"show trees" in southern NaitasiriProvince,readily accessible fromSuva, arewell knowntomembers oftheDepartmentsofAgricultureandForestryandareprotectedfortheconvenienceof interestedvisitors.LOCALNAMESANDUSES: Masiratu is the name best known in southern Viti Levu,while in the more northern uplands the name vavaloa (black shoe)is widely used.Infirst collecting the speciesonVanuaLevuIrecordedthename yaranggele, but thisname has not otherwise been noted.The timber of Degeneria vitiensis is occasionallymilled (although the species is too scattered in occurrence tobe deliberately sought);FIGURE 2. Degeneria viliensis, fromDA15292; A, flower with many petals removed, showing calyx, afewinner petals, andextrorsesurfacesof stamensandstaminodes, x4; B, extrorse surface of anouterstamen, x 16; C, extrorse surface of an inner stamen, x 16; D, four staminodes, showing introrse and lateralsurfaces, x 8.1981 DEGENERIACEAE II12FLORAVITIENSISNOVA Vol. 2FIGURE 3. Degeneria vitiensis. from DA/5292; A. carpel and bases of staminodes. x16; B, longitudinalsectionof carpel. showing thecopiouslyglandular wall. one stigmatic crest, and one row of ovules. x 16.the wood is considered potentiallyuseful as a buildingtimber, a casewood, forfurniture, and for peeled and sliced veneer. Many attempts have been made togerminate seeds and to establish the plant outside of Fiji for its potential ornamentaland scientific value; most such attempts have failed, but a few young plants may existinU. S. orEuropeangreenhouses.AVAILABLECOLLECTIONS: VITILEVU: MBA: Hills between Nandala and Nukunuku Creeks. along trailfromNandarivatutoward Lewa, Smith6/70. 6/90: westernslopesofMt. Nanggaranambuluta. east ofNandarivatu. Smith630/. 63/8; hillseast ofNandalaCreek. southofNandarivatu. Smith5923; Nau-wangga, valley of Nandala Creek. DA 3642; hills between Nggaliwana and Tumbeindreketi Creeks. east ofthesawmill at Navai. Smith 5875. 5880. 60/8;western andsouthern slopes of Mt. Tomanivi. Smith 5744.DA/2726 (Me/ville et at. 7I/5). /304/. MBA or NAITASIRI: Waimongge Creek. south of Mt. Tomanivi. DF/079. NANDRONGA &NAVOSA: Northern portion of RairaimatukuPlateau. betweenNandrau andNanga.Smith 5555; Nausori Highlands, Bola NH-12. DA 13892. DF I/44 (SI56/ /6). I/45 (S156 I /5); Nandronga &Navosa without further locality. DA14297. SERUA: Nambukelevu East. Berry 95; inland from Namboutini.Damanu 105. 106. DF456.457. I/05, 1126 (S1561 / 2), 1129 (SI56/ / I); hills north of Ngaloa. in drainage ofWaininggere Creek. Smith 9/89; Tumbarua, inland fromNgaloa.DF878. / /35 (S1561 /4). 1/36 (S156 I /3).NAMOSI: Hillsbordering Wainavindrau Creek. in vicinity of Wainimakutu. Smith 8600; northern slopes ofKorombasambasanga Range. in drainage of Wainavindrau Creek. Smith 870I; hills east of WainikoroiluvaRiver. near Namuamua. Smith8939; LombauRiver. Bola79; NambukavesiCreek. DF 230. BolaN/-I7.NAITASIRI: Vicinity of Nanduna. near Waindrandra Creek. DA1488 (coli. B. E. V. Parham. May11. 1939).1981 ANNONACEAE 133008.364/,3772.584/. /0/32, /0/46. /5223, /5292; opposite Nawanggambena District School, Stautfer &Kuruvoli 5852; Nawanggambena, DA1/854; Naivuthini, DA/533; WaimanuRiver, DAL./3244 (Berry54);Adi CakobauSchool water supply road, Sawani, Webster&Hildreth/4/0/;Naitasiri without furtherlocality, DA287 (colI. in1936), DA, June22 or 27, 1947. REwA: Mt. Korombamba, DA/6538. VANUALEVU: MOUA: Lower Wainunu River Valley, Smith /754 (May 7, 1934); north ofThongea, Wainunu River.DA/5773. TAVEUNI: Slopes of Mt. Manuka, east of Wairiki, Smith 8200; Nggathavulo Estate, DA/6937.ORDERANNONALES ManyphylogenistshavetakentheorderAnnonalesinaverybroadsense, sub-merging it in their concept of Magnoliales or taking it as the appropriate ordinal nameforan extended complex of magnoliidean families, At another extreme, Hutchinson(1973) limits the Annonales to the two families Annonaceae and Eupomatiaceae. It ismore generally considered, however, that the latter family is more strictly related to theMagnoliaceae than to the Annonaceae. Most recent students of "ranalean" taxa agreethat threefamilies, Annonaceae, Myristicaceae, andCanellaceae(not inFiji), areclosely related and group them into a suborder (of either Magnoliales or Annonales).In the present treatment this coherent cluster of families is taken to compose the orderAnnonales.KEYTOFAMILIESOCCURRINGINFIJIPlants with hermaphrodite flowers (our species) or rarely monoecious; perianth basically 3-whorled, I whorlcalycine andtheother2petaloid; stamens free, hypogynous; carpelsnumerous or few, free or rarelyunited; fruit composed of free carpels or these connate into a syncarp, the seeds I-many, sometimes (butnotconspicuously)arillate. . 45. ANNONACEAEPlants dioecious; flowers apetalous, the perianth composed of a 3(rarely 2-5)-lobed calyx; stamens usuallywithfilaments(andsometimes anthers) united, the