I,EANDRA VI-VII DEZI]MBRO DE 1977 Ne7

I}YRSONI},{OIDTAE, A NI]W SL]BF'AT{ILY OF THI] I{ALPIGHIACEAI

\flilliam R. Anderson,r

'I'he Malpigliiaceaecan easiiybe dividcd into In'o groui)sby their fruits, which are usually either winged or unwinged. Specieswith winged fruits are generally but not always vines, while speciesri'ith unwinged fruits are usually shrubs or trees. On tl.rebasis of thesecharacteristics, plus the shapeof the reccptacle,the family N'as divided two subfamilies by Franz Niedenzu, rvho published his rnonographof the family in 1928. Niedenzu used thc nanrc Pyramidotoraefor the group with rvinged fruits and Planitoraefor the groulr rvrtir unwinged fruits. Morton(1968) pointed out that the Code of Nomenclaturerequires that tl.renamcs of subfamiiiesbe basedon namesof included genera. lior Nieclenzu'sPyramido- toraehe proposedthe nane Gaudichauclioideae;for tire Planitoractirc naruervould hnve to be Malpighioideaebecausc the subfamrlyas circumscribcdby NiedenzLr irrcludedthc Malpighia, thc type o1' the family. The subfamily Malpighioideaesensu Morton is rnostly a n:rturai assemblage of genera linked together by nutnerous ciraracters. Horvcver, ccrt,rin generr included by Niedenzu do nclt scem to me to belong n'itlr thc othcrs, ancl I would excludethen.r. Unfortunatcly, orre of the excludedgcrlera is h{alpighia,and wherr it is removed the nanreof the subfamily goes rviih it. Therefore it is necessar)' to describe a ne\\r subfarnily i:or the rcmaining genera. In this paper I shall describeand circumscribethat subfamilt,lncl tlrcn discussnly reasonsfor excluding It{alpigliia ancl scveralotlrcr genera.

' Tliq ljnies.sity of Michigon Herbarium, Ann Arbor, Michigon 48109 USA. Reseorcb supported in port by Ncrlionol Science Foundation Grcrnt GB-373i4. Byrsonimoideaew R Anderson, Ir{alpighiacearutnsubfaru no\r.

Arbores ct f ruiices, rafo suffruticcs. Stipulae bene evolutae, plerurnque intrapetiolares,liberac vel connltac. Inflorescentiathyrsus ex cincinnis2-1O-floris constans,vel pseudoracemusvel spica vcl fasciculus(re vera ex cincinnisunifloris -+- constans).Gemrna floris saepecircinata. Antherae inter se similares.Pollen plerumcluetricolporatrrnt, raro tetracolporatum (in Galphimia Sect. Galphimia parasyncolporatum),10-30 p in diametro. Styii plerumque3, subulati,stigmate minuto (in Spachea2-3, crassi,stigmate truncato vel subpeltato). Fructus caroosus vel siccgs,nunquam alatus nec setifer, ex 3 carpellis liberis vcI z'3 lnericarpiis constansvel compositusloculis 1-1

-j (-hronrosotnatttm ntllncrtls: t) 6 vel tl'

Typus. Byrsonima1.. c. Richarclex Flun'rboldt,Bonpland .\ Kttnth, Nov Gen. Sp. (.1red) . ):I47 1s21(1822)' of subfamilyByrsonirnoideae are all nativesof the tropics anclsubtropics of the New .J(orld. There are no vines among thetn, irnd I believethat the vining habit 6as never evolvedin this group, This is in rnarkedcoutrast to many other genera(e. g. Banisteriopsis,HeteroPterys, Tetrapterys), in'n'hich the vining habit is probably ancestraland extant shrubs are probably derived. The best single characterfor recognizingthis subfamily is the presenceof tlrrce subulate styles sith minute stigmas; the only excePtionis Spachea,which cannot be excluded b:causeit seemsto be closelyrelatecl to Lophanthera. Smali tricolporatepollen is anoiherimportant charactcrthart ur.rites tiris subfamily,and anotheris a chrotnosome nurnberof n - 6 or n : 12. Almost all chromosomenuntbers knorvn for other New World lr{alpighiaceaearc n : 10 or a multiplecf 10,or eneuploidderivatives Of such rnultiples.However, very few chromosomecounts have been pubiishedfor more lire lvfalpighiaceae,ancl this gencmlizationnray or miy not be supportedrvllen data are available. It would be especiallyinteresting to have a chrotttosomecount for the troublesomegenus Spachea. any No member of this subfamilyhas the fruit rvingedor setose,nor is there evidenceof reductionfrom a *,inged fruit comparableto the u'inglessspecies found rn Banisteriopsis,Heteropterys, Stigmaphyllon, I\{ascagnia' etc. In tnost generathe fruit is probably dispers.,iby rvater or blown about with other detritus-likeseeds and fruits. Byrsonimahas acirieveda rnaior distinction by evolving a fleshy, bird- dispersedfruit, and it is probably no coincidencethat Byrsonimais the only geuus in ihe subfamilyn'ith more than 1i .\flith the adventof small birds thete has probably been strong selectionin various lines for fleshy fruits. and it is my opinion that suchfruits have evolvedseparately and repeatedlyin the . It happenedonly once in this subfamily,in the ancestorof Byrsonima,which pro- bably had a friut like that of Blepharandra. Somewhatsimilar fruits also evolved in the ancestorsof Malpighia and Bunchosia,but other characteristicsof those generado not suPpoft a probable origin in this subfamily,and they are therefore r:xcluded.

KtsY TO THI] TRIBES OT BYRSONIIVIOIDEAI]

I (-;rrpelsconnate in flower, bearing apical styles. 2 . F'ruit quite indehiscent,neither opening nor breakingapart into mericarps I . Byrsonimeae 2. F-ruit breaking apart into dry 1-seededrnericarps, the nrericarpsoften dorsally loculicidal z. Galphimieae I . Carpels free from each other, borne on a flat or slightly pyramidal torus, bearing ventriflxed sub apical to almost basal styles f. i.,,,"nrt "..u. i ByrsonimeaeW. R. Anderson,trib. nov.

Gemma floris saepecircinata. Glandescalycis l0 vel nullae. Staminat0 (vel rn Diacidia spp. 9-6), aniherisglabris vel piliferis, inter se -+ similaribus. Carpella 3 omnino (praeter stylos) connata,ovario 3- vel abortu 2- vel 1-loculato,stylis i apicalibussubulatisque, stigmate nrinuto apicali vel parum interno. Fructusinde, hiscens,drupaceus vel siccus,1-3-spermus. Typus. ByrsonimaL. C. Richardcx Humboldt,Bonpland, & Kunth, Nov, Gen. Sp (4r ed.) 1:1471821 (1822).

KEY TO THE GENERA OIi BYRSONIMEAE

Leaves,bracts, and bracteoleseglandular (except for gland-tipped marginal teeth or cilia on the bractsand bracteolesof some species). 2. Fruit drupaceous;hairs on the anthers,if. any,medifixed or sub-medifixed, with 2 branches;New World tropics and subtropics . . . BytsonimaRich. ex H. B. K 2. l"ruit dry; hairson thc anthersbasifixed Petalsall yellow; anthersbcaring 2(-4) stout, apical,ax'n-like hairs, these strongly differentiated from any other hairs on the stamen; AmazonianBrazil, southernVenezuela, and easternColombia Diacidia Grisebach Petals white an

All of thesegenera are well representedin the GuayanaHighland and I shall describeand discussthem further in my fotthcoming treatmentof the Malpighiaceae frorn that region. I am also preparing a monographof the whole genusByrsonima,

). GalphimieaeNiedenzu, Ber. Deutsch. Bot. Ges. 8:192 1890

Floriferous peduncie well developed,cven rvhen cinciuuus is reduced to 1 flower. Calyxglands 10 or fewer through fusionand loss,to none. Petalssubequal or the posteriorpetal moderatelydifferentiatcd. Stnmcns10, tlrc filamentsopposite thc sepalsusually longer than those opposite the petals, the anthers +- alike, glabrous. Carpels 2-3, connate in flower for their whole length (except styles), both or all 3 loculesuniovulatc; styles 2-3, aprcal. Fruit breaking apart into 2-3 (or fewer by abortion) 1-seededmericarps, the mericarpsoften dorsally loculicidal but not opening sufficiently to rcleasethe seed; baseof the mericarp often with an ernpty or aerenchyma-filledcavity.

Type. GalphimiaCavanilles, Icon. et descr. pl . 5:61. 1799

*- 8 ,,- KEY TO THE GENERA OF GALPHIMIEAE

Stylesslender and subulate,tapered distally to a minute stigma. 2. Anthers unwinged, at most apically verruculate;bracteoles egiandr,lar. 3 . Anthers without apicalornamentation; calyx glands 5 or fewer, often none; leavesusually bearingglands on baseof lamina, sometimeson petiole;from Texasto Argentina Galphimia Cav, ) . Anthers ornamented at apex rvith 2 wart-like outgrowths (<>);calyx glands 10, with some neighboring pairs partially fused; leaveseglandular; Bahia and Amazonas,Brazil . . . Verucularia Adr. Juss. 2. Anthers longitudinally winged; alternatebracteoles bearing rarge glands, Amazonian Brazil, Venezuela,and probabiy Colonbia Lophanthera Adr. Juss, I . Styies stout, truncatc or subpeltatc at tl.reapex; northern Soutir America, Panam6,and the W'est Indics SpacheaAdr. Juss. As circumscribedhe rc this tribe is drastically changed from the group recognizedby Niedcnzu in 1928, Sevcralof his genera have been excludedfrom the subfamilf, and others have been assignedto other tribcs. The four genera left form two natural pairs (Gatphimia and Verrucularia vs. Lophanthera antl Spachea)that have in common similar fruits and a tendencytoward reducliol ol' tire calycineglands. tr'Iostof thesegenera were recentlystudiecl by a4acBryde(1970). I-Iegave little considerationto thc relationshipsbetween g€nera, but he did describe tltcm thoroughlyand his paper is r usefulsoilrcc of clescriirti'edata. It is not clearto what extent the rncricarirsof this tribe arc naturallyloctrli- cidal. In Gaiphimia and Verrttculariathcy are probably usr-rallyso, althou*h I have scenno eviclcncetl'rat the seedis releasecl. They appearto be at leasttardily ioculi- cidal in Lophanthera latifolia and L. penclula,but not in L. sprr.rceanaaneciesin Panam{lncl tlre West Indies),and Pterandra(onc speciesin Panami). I Acmandrereae\w . R. Andcrson. trib. nov

Folia, brecteae,et bracteolaeeglandulosae. Flores I in quaque bractea (i. e. cincinnusuniflorus). Pedicellussessilis. Glandes calycis l0 vel nullae. Stamina -r 10, filamentis liberis, antheris glabris, inter se -r similaribus. Carpella iibera, in toro plano vcl parum pyranridaliportata, omnia 3 uniovulata,stylis ventrifixis 'r'el subapicalibus fere basalibus,gracillimis subulatisque,stigmate minuto apicali vcl parurn interno. Fructus ex 3 (vel aboriu 2-t) coccissiccis indehiscentibus constans. Typus. Acmanthera (Adr de Jussieu) Grisebach in Martius, Fl . bras. t2(l):29. 1818; Pterandrasect. Acmanthera Adr. de Jussieu,Malp, Synopsis, Ann . Sci Nat . (ZeSdr. Bot.) l1:j28.1840.

KEY TO THE GENNRA OF AC}TANTIIEREAIT

Antherswinged. 2. Inflorescencean elongatedpseucloracerne terrninating an axillary shoot; stipules1.t-11 cm long, deciduous;northcrn AmazonianBrazil ...... Acmanthera(Adr. Juss.) Griseb. 2 Inflorescencea tight umbellatc fascicle,sessile or subsessile,axillary to leavesor bractsor leaf scarson older stems;stipules up to 0.6 cm long, persistent;Brazil, western Guyana, southern Venezueia,Colombia, and Panamd, Pterandra Adr. Juss

Anthers rrn*'ingcd; Pari and F'renchGuiana ColeostachysAdr. Juss

'fhe generainciuded in this tribe q'ere treatedby MacBryde(1970), and I recently published a brief revision of Acmanthera (Anderson, 197)). Pterandra is interestingfor the reductionof its inflorescence,and Coleostachysfor the rarity of its only species,C. genipifolia Adr. Juss., and for its reportedly poricidal ar:thers(IvlacBryde, L97O).

GENERA EXCLUDED FROM THE SUBFA}TILY BYRSONIMOIDEAE

It is norv appropriateto make some comrnentsabout thc generoincluded in thc subfarnilyPlanitorae by Niedenzu (Malpighioideaesensu Morton) and ex- ciuded from the subfamily Byrsonimoidcae.I do not pretend to know where the strongestaffinities of all thesegenera lie. ThereforeI shali emphasizemy rcasons

- 11 * for cxclu4ine them frorn thc llyrsonin.roidelc, whosc dcfinition is tirc purpose of pllcctlrent of somc problenlatical r-lcnera for (his 1ra1rcr,aucl Icirvc thc prolrer futurc siudies.

I\{alpighia L.

plants of Malpighia are shrubs,and their fruits are fleshy. Those trvo charac- teristicsare tire only basesfor placing the genus in proximity to Byrsonimaand its relatives. Against such placement are the following features of Malpighia: Stipulesminute and interpetiolar. Flowers strongly zygomorphic,with anthersof different srzesand (in some species)the stylessigmoid, as in Hiraea. Pollen large, polyporatc, the 4-8 poles not all equatorial,and with furrows not equivalent to the colpi in thc Byrsonimoideae.Styles with large oblique-internalstigmas and dorsallytruncate or hooked at the apex. Chromosomenumber: n: 10 (2n -)6 in IVLurens fide Foudt, 1966). The fruits of Malpighia descrvespecial attcntion, since their superficialsirni' larity to the clrul:esof Bytsonima has been the causeof much confusion. When one rernovesthc flesh from a fruit of Byrsonimaone finds a single spheroidstonc, smooth ol fugose.This stone,which is very hard and bony, is the result of com- plete fusion of the endocarl>sof the threc carpelsthat form the ovary' Is is quite ir.rdehiscer-rt,but if broken open rvill be found to contain tirree one-seededlocuies (reduced to two in some species). The cntire f ruit of Blephamndrais very iike t5e stole of Byrsonima,ancl I believe tJrata fruit like that of Blepharandrapro' bably gavc rise to the drupe of Byrsonirnathrough gradual increasein the fles- hincss of the exocarpin responseto selectionby srnall birds. A very different situatiol is for-rr.rdrvhen the flesh is rernovedfrom a fruit of tr{alpighia.Here one finds three seParatestones wiricl.r, once tire exocarPis gone, are joined only along a narrow cer-rtralaxis, where tlrey are bornc on a conttnoltpyramidal torus. There is r.rocontinuity of the endocarps. Each stone, or Pyfene' is derived f rom one carpel,ar.rd contains one secd. Its wall is tough and fibrous; it bearstwo lon- gituclinallateral ridges or winglcts, a dorsal ridge or winglet, and usually smallcl parallel or transversericlges in betwecn. This arrangementis very suggestiveof a mascagnioidmericarP in which thc wings did not develop. Pleasesee Plates I and II for a comparison of tire florvers and fruits of Byrsonima and Malpighia. Taking all of the abovefacts into account,I can only concludethat Malpighia is probably derived from one of the wing-fruited genera' most likely a nascag- nioid genus (c. g. Ir{ascagnia,Hiraea, Jubelina,N{ezia, Cabi, etc' ) ' Therc

-. 12 -- arc cxtant nrascagnioids,e.g. Cabi, in wirich thc fruit is reducedto three wingless mericarps. If such a fruit were modfied by the addition of a fleshy exocarp,it would resemblethat of Malpighia. The superficial resemblanceof the fruits of Byrsonima and Malpighia is surely a case of parallel evolution. It is interesting that Malpighia has also abandonedthe vining habit of its ancestors,but as I have noted elsewirerein this paper, that particular shift in habit has happenedso often in this family that its occurrencehere rvould be a slim basisindeed for postulating a relationshipbetween Malpighia and Byrsonima.

liunchosiaRich. ex H. U,K.

This is a genusof shrubsand trecs. Collectorsoccasionally describe Bunchosia as a climber, but this probably meansa lax or leaning shrub rather tiran a twining vine. Tire stipules are small and intcrpetiolar or epipetiolar. The leaves bear large glands enibeddedin the lamina and alternatebracteoles bear dorsal glands. 'I-he glands of the calyx are 8-10, the number apparentlyreduced through lateral fusion in some specics and through loss in others. The pollen is poiyporate,with )-8 non-equatorialpores in the speciesexamined, and rvithout furrows or colpi 'I'here are two or tlrree carpels,quite connate in the ovary. The two or threc styles are stout, frcc to connatc, with large terminal stigmas that are cepitatc, peltate,or rcniform. f'he fruit consistsof trvo or three one-seededpyrenes in a common fleshy exocarp. \When tire flesh is removed the pyrenesare found to bc clongated,round or elliptical in cross-section,free from eachother, rvitir a smooth, brittie, cartilaginousrvlll . The only published chromosomenumber is that of l.oudt (1966), 2n -- 72 f.or Bunchosiamontana. Fouet'shigher counts are susPect becausethey were n-radefrom sectionsrather than squashes,and I consider this report an inadequatebasis for any phylogenelicspeculation. The abovedescription should makc it abundantlyclear tirat Bunchosiais ntlt at all closely related to Byrsonima. Moreover, there is little basisfor postulating a rnuch closer relationshipto Malpighia. Its fleshy fruit was surely derived from sorneother ancestor,and representsanother case of parallel evolutionin this family. \7hcn I come to considerwhat may be the closestextant relativesof Bunchosia, I find the genus an enigrna. The genus that seernsclosest to Bunchosiais Hc- ladena , rvl.richhas similar infloresccnces,pollen, and stigmas. The cocci of He- ladena,if envelopedin a common fleshy exocarp,rvould constitutea fruit similar lo that of Bunchosia.Btrnchosia is also somewirat similar to Srracheain thcrr' laminarand bracteolarglancls ancl thcir stylcsand stigmas.

lf Dicella Grisebach 'fhe Dicella is a genus of woody vines. stipulcs are tiny and interpetiolar. The calyx glands are eight on the four lateral sepals,none on the anterior sepal (this is the characteristiccondition for vining genera). The pollen is large and polyporate with non-equatorialpores, and has broad fissuresthat are probably not equivaient to the colpi of tl.re Byrsonimoideae.The ovary comprisesthree cornpletelyconnate carpels,trr,o of them rvell devcloped and one usually rudi- mentary. The trvo fertilc locules each bear onc ovule, but one often fails tcr produce a seed. The stylesale stout, with truncate,oblique-internal stigmas, and are dorsally angled at tlre apex; when only two carpelsare fertile, the third style is rudintentaryor absent.The fruit is a spheroid,indehiscent, nut-like structure sith r t.lrick,fibrous wall. At maturity it is dry and hard, not flesl.ry,and the only apparentadaptation for dispersalis the enlarged,rving-like sepals, whicl-r are mr:cl'rlarger than the fruit in so,nespecies but smaller in others. The one or two large seedscontain a small embryo and abundant perisperrn,an interesting ano- maly in this family (I have observedperisperm formed during grorvth of the seed in Gaudichaudia,but by maturity it iras been completell'resorbed). Thc chromo- sone numberis unknown. There is not a single feature of Dicella that suggestsderivation fronr or close relationship to the Byrsonimoideae.The fruit is not evcn fleshy, although a rvorker rvho had never seenit alive could be forgiven for thinking it so. Dicella certair.rlybelongs with other vining genera,but beyond that I cannot speculateat tlris time. \(hen sufficient data are availableabout cl.rromosomenumbers, pollen morphology,and ernbryologyin many genera,perhaps its correctphyletic position s ili bcconreafparent

Thryallis Martius 'Ihis is a peculiar genus of shrubby vines, almost uniquc in the family in their stellatcpubescence. The tiny stipulesare ePiPetiolar.The calyx is eglan- dular. The pollen is polyporate and without colpi or furrows. The three styles bear oblique-internalstigmas. The fruit is small and pyramidal, comprising three dry indehiscentcocci, each with a hard, rugosewall . The cocci can be separated, but they do not seemto fall apart in nature. The calyx is dry and Persistent,and ptobably aids in dispersalby the wind of the whole fruit as a single unit. Each locule contains a tiny, hardly-developedembryo; the rest of the spaceis filled by an aerenchyma-liketissue, perhaps modified perisperrn The chromosome number .isunknox.n. r4 I do not knos' v,here this genus belongs,but its habit, pollen, and stigmas stggest derivation from one of the wing-fruited vines. Ther" j, no basis what, s()everfor placing it in the Byrsonirnoideae.

Clonodia Grisebach

This is a sn.rallge'us of rvoody vincs, strongly si'riiar to I:leteropterysspp. but with the mericarp dorsally verrucoseor cristateinstearl of rvinged. Th. fuct that Niedenzu u'as u'illing to irlacethis gentrsso far from the u,ing-fruitedgenera demonstrates that he \\ras morc interested in n]aintaining intact his arbitrarily defined hierarchy than in devisingthe most naturai taxonomypossible. For further discussionof Clonodia scc Cuatrecasas,l9(r0.

HeladenaAdr. Juss.

Heladenais a genusof shrubsand rvoodyvines rvith snraileprpetiolar stipulcs. The calyx bears eight stalked gla'ds or the four lateral ,.prir. The pollen is Iarge, without furrows, and polyporatc, tlrc pores not ali eqiratorial. The styles have large obliquc-internalstigmas. The f ruit comprisesthrcc unrvinged i'de- hiscent cocci that fall frorn a pyramitrartorus. The chromosomcnun.rber is not known. The [o.r or five speciesof Heladena are all rare and poorly known. They are similar to Bunchosiain their inflorescence,pollen, and stigmas,and the two generamay actually be fairly closelyrelated. Niedenzu probably placed Heladena near Spacheabecause their stigmasand fruits are similar. It is possiblethat their fruits do indicate a close relationship betq,eenthesc two g.n.ir, in whicir casc Heladenamight constituean intriguing link between the Byrsonimoideaeand thc oiher i{alpighiaceae. However, the similarity could also be coincidental,anJ Heladena may be derived from some wing-fruitecl vining genus with a reducecl fruit. At prese't, a possible relationship to spachea,itself an anomaly in the Byrsonimoidcae, is hardly e'ough to fustify including Heladena in the new sub family.

SUMMARY

The exclusion of Malpighia from Niedenzu's subfamiiy <>forces the recogrritionof a new sr.tbfamilyof Malpighiaceae.It is narnedByrsonimoideae, described,and divided int.r the tribes Byrsonimeae,Galphimieae, and Acmanthereae. Each tribe is describedand a key to its generais pro'ided. The reasonsfor exclu- ding six genera included by Niedenzu in his <>are discussed. _ r5 _ LITERATURE CITED

ANDERSON, WLLIAM R. 19?5 The of aunonthers (Ma'- p.gh-aceae).C,ontr. Univ. Mich. Herb. 11(2):41-50. GuATRECASAS, JOS10. 1960 - Prima Flora colombiana. 2A, Malpighiaceae, Apdndice 1. Webbia 15(2):393'398. FOUtT, MICHELLE. 1966 - Contributron a 1'6tude cyt+taxinomique des Mal- prghiac6es. Adansonia 6:457-505. IOBREAU, DANIELLE. 196? - Contribution a l'6tude du pollen des Mai' pighiaceae d'Afr-que. Poilen et Spores 9|o)i24L-277. nIACBIIYDE, BRUCE. 19?0 - A revision of t'rle Galphirnihwe Ndz., MaI- pighiaceae. Unpublished Ph. D. thes:s, Washington University' MORTON. C. V. 1968 - A typification of some subfamily, sectional, and sub- sectional names in the family Malpighiaceae. Taxon 77i314-324. NIEDENZU, FRANZ. 1928 -. Malpighiacess in A. Engler' Das Pflanzenreich IV. Vol. 141,PP. 1-8?0.

RESUMO

A ercluslo de Malpighia da subfamilia <>de Niedenzu iorqa a nccessidadede se instituir uma nova subfamilia de Malpighiaceae.E nomeada Byrsonimoideae,descrita, e dividida nas tribos Byrsonimeae,Galphimieae, e Acmanthereae.Cacla tribo 6 descritac uma chave para os gdneros6 incluida' As raz6espara excluirscis gdneros incluidos por Niedcnzuna subfamilia. sio discutidasneste trabalho.

LEGEND FOR THE PLATES

Malpigbia Platc I (Figules 1 - 11): Comparison of florvers of Byrsonima and

coccigera' x ? 4 Fig. 1 - Androecrum and gynoecium of lv{alpighia viminilolia' x 5-3 fii. Z - Androecium and gynoecium.of- B-orsonima 3 ni!. f - Gynoecium of Byrsonima viminifolia' x 5 FG. 4 - Gynoecitrin of Matpighia-eoccigera' x 8'5 x 1250 niE, s - pollen of Byrsonima basiloba' polar view' fii. O - Polien of Dyrsonima basiloba' side view, x 1250 F'i;. ? - Pollen of Malpighia glabra, x 1250 Fi;. B - Stigma oi Malpighia coccigera'.x 31 fig. I - Stigma of Byrsonima viminilolia' x 120 Fi;. 10- Oviry of Malpighia coccige-ra, cross-sectioD, x 21 Fi;. 11 - Ovary of Byrsonima basiloba' cross-section' x 14 )Dl,-.-

,t6 H t'1 F

Fl

t ;flHeJ$fr * '199.';8fi 'f, EHx *ElE^ i :g;Cs;$n iIfiIiiiiE E A:EESgF: I i giiffilir s sini;igtf I -I$IIgEi-Iiiili$i;

,