Quick viewing(Text Mode)

Preliminary Observations on Floral Biology in Mangrove Rhizophoraceae Author(S): P

Preliminary Observations on Floral Biology in Mangrove Rhizophoraceae Author(S): P

Preliminary Observations on Floral Biology in Author(s): P. B. Tomlinson, R. B. Primack, J. S. Bunt Source: Biotropica, Vol. 11, No. 4 (Dec., 1979), pp. 256-277 Published by: The Association for Tropical Biology and Conservation Stable URL: http://www.jstor.org/stable/2387918 . Accessed: 30/08/2011 15:41

Your use of the JSTOR archive indicates your acceptance of the Terms & Conditions of Use, available at . http://www.jstor.org/page/info/about/policies/terms.jsp

JSTOR is a not-for-profit service that helps scholars, researchers, and students discover, use, and build upon a wide range of content in a trusted digital archive. We use information technology and tools to increase productivity and facilitate new forms of scholarship. For more information about JSTOR, please contact [email protected].

The Association for Tropical Biology and Conservation is collaborating with JSTOR to digitize, preserve and extend access to Biotropica.

http://www.jstor.org PreliminaryObservations on Floral Biology in Mangrove Rhizophoraceae

P. B. Tomlinson, Harvard University,Harvard Forest, Petersham, Massachusetts 01366, U.S.A. R. B. Primackt Department of Botany, Universityof Canterbury, Christchurch, New Zealand and J. S. Bunt Australian Instituteof Marine Science, Townsville, ,

ABSTRACT The tribe Rhizophoreae (Rhizophoraceae) includes the largesttaxonomic assemblage exclusivelyof ,with 4 genera and about 18 . is pan-tropical;, , and have an Indo-Malayan dis- tribution.All species have the same basic floral structure,but field observationsdemonstrate a wide varietyof pollina- tion mechanisms.Variation in such featuresas size and orientationof ,number of flowersper inflorescence,num- ber of ,time of stamendehiscence, and method of dischargecan be shown to have direct relevance to pol- lination biology. Pollen vectorsmay be predominantlywind (Rhizophora, as deduced from circumstantialevidence) or animals (remaining genera, as observed directly). An explosive method of pollen release occurs in Bruguiera and Ceri- ops tagal, but is modifiedto suit a range of flowervisitors such as birds in the large-floweredBruguiera species (e.g., B. gymnorrhiza)or butterfliesand other insectsin the small-floweredBruguiera species (e.g., B. parviflora) and moths in . lacks the explosive mechanismof pollen release,as does Kandelia. Since these taxa often formmixed communitiesa high degree of partitioningof pollinator resource is achieved.

INTEREST IN T HE REPRODUCTIVE BIOLOGY of man- mentalstudies of fluid-preservedmaterial, which will groves has largelycentered on seedling development form the basis for a separate comparative account. and dispersal since embryodevelopment on the par- All taxa in the section Rhizophoreae of Rhizophor- ent to a seedling stage (vivipary) is more fre- aceae have been studied, and the comparative data quently shown in this group of than in any accumulated permit some speculation about floral other. Studies on floral ecology are very few, as van evolution,although this must remain tentativeuntil Steenis (1958) points out, but knowledge of repro- comparable informationfor non-mangrove Rhizo- ductive capacity and interspecificsterility barriers phoraceae has been assembled. must begin with a considerationof function. That the Rhizophoreae include distinctivefloral The most extensive studies are those of Guppy mechanismsis evident from the early summaryand (1906) on Rhizophora. The present account is a reportsof Gehrmann (1911) and Porsch (1924, p. summaryof informationobtained during field work 611) who recorded the explosive method of pollen in mangrove communities,mainly in Papua New release in Bruguiera sexanzgula,associated with bird Guinea and Queensland, and surveys floral biology visitors,an observationmore recentlyconfirmed by of Rhizophoraceae in a provisional way. Field work Davey (1975) in southeastAfrica. We can now re- provided the opportunityto observe flower visitors port that all Bruguiera species have explosive pollen and floral mechanisms in Bruguiera, Ceriops, and release,but that two main agents triggerthe process, Rhizophora. Additional informationhas come from in the small-floweredspecies, birds in the other parts of the South Pacific (mainly Fiji and large-floweredspecies. Ceriops tagal is similarlyex- ) and in South Florida. No field ob- plosive, but triggeredmainly by moths. Ceriops de- servationson Kandelia have been made. Although candra lacks explosive dehiscence,and pollen vectors informationis incomplete it is sufficientto dem- are as yet unknown. onstrate a wide range of pollination mechanisms Despite the abundance and wide distributionof which occur in plants with a fundamentallysimilar Rhizophora there is no extended treatmentof its floral construction,suggesting a fairlycomplete di- floral biology. Guppy (1906) provided information vision of pollinator resources. about floral mechanism in Fijian Rhizophora spe- The field observationshave been supplemented cies, but did not commenton vectors.Gill and Tom- by detailed morphological,anatomical, and develop- linson (1969) have suggested, but without much supportingevidence, that R. mangle is wind-pollin- lPresent address: Department of Biology, Boston Univer- sity,Boston, Massachusetts02215, U.S.A. ated; Kress (1975) observed visitors to Rhi-

256 BIOTROPICA 11(4): 256-277 1979 zophora in South Florida; and there is one recordof unopened flowers.The antherswere maceratedcom- bees visiting the flowersof R. stylosain Java (Doc- pletely in several drops of filteredsea water. Addi- ters van Leeuwen, 1927). Our comparativeobserva- tional filtered sea water was added to bring the tions lead us to the conclusion that the flowersof suspensionof particulatesup to a volume of approxi- most Rhizophora species are basically wind-pollin- mately 10 ml. The suspensionwas violentlyagitated ated, although they may be a pollen source for in- for 30 seconds to freeany pollen grains still enclosed sects and may even retain vestiges of insect attrac- by the anther tissue, and then filteredthrough a 50 tion in odor and nectar. appears micronscreen mesh to remove tissue debris.Virtually to be pollinated by small bees and wind. no pollen grains were trapped in this screen,nor did Since the taxa involved are wide-ranging (pan- theyclump or settle,at least within the few minutes tropical in Rhizophora) and our field work is ne- that the suspension was utilized. A Coulter Counter cessarilygeographically restricted, it is clear that our (model TAII, Coulter Electronics,Hialeah, Florida), descriptionsmust serve only as an introductionand normallyused for countingcell numbersas in blood our conclusions may not be universallyapplicable. samples, was then used to count the numberof pol- Further field studies in differentgeographic areas len grains in a subsample of this suspension,or if are needed. At the same time our observationsare the pollen densitywas still too high to count, in a sufficientlycomplete to indicate an example of adap- more dilute suspension. Successive subsamples from tive biological radiation in a widespread and eco- the same pollen suspension gave virtuallyidentical nomically importanttropical group which is prob- counts.The numberof pollen grains in each of two ably monophyletic. anthers of ,Ceriops tagal, and were counted completelyafter the antherswere maceratedin a drop of iodine solution. MATERIALS AND METHODS In order to confirm the results of the Coulter Most detailed observationswere made at field sites Counter,two antherseach of Bruguieragymnorrhiza, in Queensland (3 Mile Creek and Cape Ferguson, Rhizophorazmucronata, R. X lamarckii,and R. apicu- Townsville and Missionary Bay, Hinchinbrook Is- lata were macerated,agitated and suspended in 4 ml land), but these have been supplemented by more of a solution of 1 part aqueous iodine to 2 parts gly- general observationsin Australasia.Unless otherwise cerine. The pollen grains suspended in one drop stated,observations refer to the Queensland localities. (0.025 ml) of this solutionwere counted.This man- ual countingmethod gave similarresults to the Coul- ter Counter figures,supporting the validity of both TABLE 1. The numberof pollen grainsper flower(P), approaches.The resultsare presentedin table 1. ovulesper flower(0), and pollengrains pro- ducedper ovule (P/O) in speciesof the Rhi- zophoraceae.The numberof pollengrains per STIGMA RECEPTIVITY.-The onset of stigma recep- flowerand per ovule are given in thousands; ovule numbersare the most commonlyob- tivityin Bruguiera exaristataand Rhizophora stylosa servedvalues. was determinedby floating stigmas in a naphthyl- acetate. This produced a color reaction, i.e., meta- Pollen and species P x 103 0 P/O x 103 vector bolically active surface cells indicative of esterase activity,by coupling with hexazotised pararosanilin, Rhizophoraapiculata 5256 4 1314 R. mucronata 2096 4 524 Wind which in turn is indicative of stigma receptivity R. X lamarckiia 4000 4 1000 (Heslop-Harrison et al. 1975). Bruguiera (Type 1) B. gymnorrhiza 1800 6 300 B. exaristata 1080 6 180 Birds Bruguiera (Type 2) B. cylindrica 176 6 29 The section Rhizophoreae of Rhizophoraceae in- B. parviflora 4.8 6 0.8 Insects Ceriops decandra 84 4 21 (Insects?) cludes four genera distinguishedby their exclusively Ceriops tagal 9.2 6 1.5 Moths mangrovehabitat and markedvivipary such that the Kandelia candel 659 6 170 (Insects?) hypocotylof the embryo protrudes from the fruit apollen frequentlyaborted. while still on the parent ,the seedling being the unit of dispersal (e.g., figs. 4C F; 6B; 7C). They DETERMINATION OF POLLEN-OVULE RATIOS.-TO are furtherrecognized by a combination of wood determinethe numberof pollen grains produced per anatomical charactersas distinctiveas in many entire flowerand the numberof pollen grains produced per families (van Vliet 1976). Of the four genera Rhi- ovule (P/O ratio), two antherswere dissected from zophora (figs. 8 and 9) is pantropicalwith three or

Floral Biology in Rhizophoraceae 257 FIGURE 1. Longitudinalsections of flowersin Rhizophoraceae,drawn with a camera lucida, all with the same orientation which is not necessarilyas in nature (see text). Flowers drawn were at a comparable stage of development,i.e., at or shortlybefore anthesis. Hatching in anthers indicates approximate extent of sporogenous tissue. Bracteoles occur below the flowersin D-I. Bruguiera: large-floweredspecies representedby Briuguieragymnorrhiza (A), and B. exaristata(B); small-floweredspecies representedby B. parviflora (C), and B. cylindrica (D); Rhizophora, representedby Rhizophora stylosa (E), and R. apiculata (F); Kandelia candel (Chai s.n., Sarawak) (G), and Ceriops, representedby C. tagal (H), and C. decandra (I). All material collected by P. B. Tomlinson except for Kandelia candel.

FIGURES 2 and 3 Bruguiera (large-floweredspecies). Figure 2. Bruguiera gymnorrhiza(from a specimen cultivatedat "The Kampong" Douglas Road, Miami, Florida). A, branched shoot with apposition growth (x 1/4); B. terminal bud enclosed by stipules (x ?2), open removed,the detailed figurewith the stipules removed (x 3); C, detail of in- side of detached stipule (x 3) showing numerous colleters; D, from inside partlyunfolded (x 3), to show main diagnostic featuredistinguishing this species from B. sexangula. Figure 3. (from fluid-preserved material,P. B. Tomlinson, 30 October 1974, Bootless Bay, East of Port Moresby,). A floral diagram; B, flowerin 1. s. (x 3/2) with the enveloping the stamens; inset detail of stigma (x 3); C, flower from the side (x 1); D, flower from above (x 3/2), all petals and stamens "sprung"; E, diagram of pair with dotted outline of enclosing petal; F, open petal, abaxial (outside) view (x 3); G, open petal, adaxial (inside) view (x 3).

258 Tomlinson, Primack, and Bunt ru PF~~I

5utcrx ocxsmzut*

Floral Biology in Rhizophoraceae 259 z EruW - rb p?xrvt,lora,

A.oM

G

VA~~~~~~~~

260 Tomlinson, Primack, and Bunt A~~~~~~~~~

PF MEE

FIGURE 5. Ceriopsdecandra (from fluid-preserved material, P. B. Tomlinson,20 June1977, Deluge Inlet,Hinchinbrook Island,Queensland). A, shootwith erect seedling (x ?6); B, congestedinflorescence, apparently dichotomizing (x 3); C, floraldiagram with bracteoles; D, flowerbud in 1. s. (x 8); E, singlepetal (x 10); F, singlestamen (x 10). fourspecies formingan "Atlantic"group, four or linson 1978), which occursin Fiji (Guppy 1906, fivespecies forming an "Indo-Pacific"group with an Richmondand Ackerman1975), has recentlybeen overlapbetween the two groupsin the Fiji-Tonga shownto occurin New Caledonia.This breakdown area.There is good evidencethat R. X lamarckiiin of sterilitybarriers between species which are closely theSouth Pacific (Tomlinson and Womersley1976) relatedstill requires an adequategenetic explanation. and R. X harrisoniiin West Africaand SouthAmer- The othergenera are restrictedto the Indo-Pa- ica (Breteler1969) are sterilehybrids rarely pro- cific regionand theirtaxonomy and nomenclature ducingseedlings. A sterilehybrid R. X selala (Tom- seemhighly stable, although one quitewidely distrib-

FIGURE 4. Bruguiera(small-flowered species). A-C,Bruguiera p!rviflora (from fluid-preserved material, P. B. Tom- linson,23 October1974, Labu Lagoon,Lae, Papua New Guinea); A, insertionof twoinflorescences on shootwith leaves removed,terminal bud not showncompletely (x ?2); B, singleflower, side view (x 3); petalsand stamens"sprung"; C, maturedetached seedling (x 1/2).D-K, Bruguieracylindrica (from fluid-preserved material, P. B. Tomlinson,5 No- vember1974, Taurama, Link Road ca. 2 km northeastof Barracks,Port Moresby, Papua New Guinea); D, distalend of floweringshoot (x ?2); E, three-floweredinflorescence to showcymose construction (x 3); F, maturedetached seedling (x 1/2);G, 1. s. unopenedflower (x 4); H, floraldiagram; I, singlepetal after pollen release (x 9); J,1. s. openedflow- er (x 4) withpetals and stamens"sprung"; K, antesepalous(left) and antepetalous(right) stamen(x 9).

Floral Biology in Rhizophoraceae 261 criop" PF

PF

262 Tomlinson, Primack, and Bunt utedspecies of Bruguierawas describedonly recent- pressedin lateralbranches from the momentof in- ly (Ding Hou 1957). Bruguiera(figs. 2-4) includes ception (Halle et al. 1978:189). In Rhizophora six speciessome more widespread than others. Ceri- distal branchesmore or less graduallyadopt this ops (figs.5 and 6) includesat leasttwo species,one form,as illustratedin Halle et al. (1978:53), where widespread.Kandelia (fig. 7) is monotypicand has thecondition is referredto by themore general term themost restricted range. No evidenceof hybridiza- ' plagiotropyby appositiongrowth." The net result tion occursin thesetaxa. The ecology,morphology, is a seriesof sympodiawith each unit terminating in and nomenclatureof the familyin Malesia is sum- a leafyrosette (fig. 2A). In Ceriops(fig. 5A, 6A) marizedby Ding Hou (1958). Tables 2 and 3 sum- and the small-floweredspecies of Bruguiera(fig. marize relevantinformation on floralmorphology, 4D) branchesremain more erect, so thatthe overall discussedin detailbelow. crownshape tends to be morenarrowly conical. To- tal heightin all speciesdepends very much on eda- MORPHOLOGY phic conditions;Rhizophora, for example,varies ARCHITECTURE.-In theirvegetative habit most spe- fromscrubby plants 2 or 3 m tall,to majesticsingle- cies correspondto the definitionof Attims'smodel trunkedtrees 30 m tall,and supportedby massive (Halle and Oldeman 1970, Halle et al. 1978) be- aerialroot buttresses. cause theyhave continuousgrowth with usually in- All sequentiallyreproduced appendages in these termittent(i.e., non-episodic) branching,lateral taxa,including flowers, are lateral,i.e., sexual repro- flowers,and brancheswhich essentiallyrepeat the ductionhas no influenceon architecture,and all de- patternof growthof theparent axis. Bruguiera gym- velop withinterminal buds. Prolepsisoccurs rarely norrhizaand B. sexangulaconform strictly to Aubre- and onlyafter damage to a terminalbud (Halle et al. ville's model since "Terminalia-branching"is ex- 1978:42).

TABLE 2. Floral features in Rhizophoraceae. Kandelia Ceriops Bruguiera Rhizophora decandra tagal large fl. spp. small fl. spp. (fig. 7) (fig. 5) (figs. 6, llB) (figs. 2, 3, 10) (figs. 4, llA)(figs. 8, 9, 11, C, D) No. of species 1 1 1 3 3 5-9 Inflorescencetype 3 3 3 1 2 3 Flower orientation pendulous ? erect ? erect recurved erect ? pendulous Flowervisitors (most frequent) ? ? moths birds butterflies bees ? Pollen vector (main) insects (?) insects (?) moths birds insects wind No. lobes and petals 5 (-6) 5 (-6) 5(-6) 8-14 8-10 4 No. stamens (basic) numerous 10-12 10-12 16-28 16-20 8-16 Precocious anther dehiscence - +? + + Explosive pollen release - - + + + Filamentlength long short long long long + sessile Antherlocules 4 4 4 4 4 multi-locellate Petal color at anthesis white white white brown brown creamy-white Petal appendages filamentous club-shaped club-shaped bristles bristles hairs or absent (fig. 71) (fig. 5E) (fig. 61) (figs. 2D; 3F, G) (fig. 41) (figs. 8H, 91) Ovule number 6 6 6 (4-)6(-8) 4(-6) 4 Flowersper 2-10 2-10 (some- 2-10 1 2-10 2-numerous inflorescence times numerous) aDetailed informationfor Rhizophora is provided in table 3.

FIGURE 6. Ceriops tagal (from fluid-preservedmaterial, P. B. Tomlinson,21 March 1976, Barune, Papua New Guinea). A, branchedshoot with flowers(x ?2); B, maturedetached seedling (x 1/2); C, details of terminalbud with expanding inflorescences(x 3): inset, cupular bracteoles (x 4) with flower removed to show colleters; D, terminal bud (x 3) shown in figureC with stipules removed to show nodal series of colleters;E, detailed stipule fromwithin (x 3) to show extensivedevelopment of colleters (inset: single colleter (x 20); F, floral diagram; G, flowerfrom above (x 6), petals and stamens "sprung"; H, flower including bracteole pair in l.s. (x 6); I, single petal (x 10); J, Detail of stamensand presumed nectaries (x 10).

Floral Biology in Rhizophoraceae 263 Kzn&tiacb&1t XPF

264 Tomlinson,Primack, and Bunt BUD MORPHOLOGY.-A characteristicfeature of bud (figs. 2-11) and are initiated within the terminal morphologyin Rhizophoraceae is the pair of cadu- bud. They become visible when the enclosing stipule cous stipulesassociated with the youngestpair of ex- of the subtendingleaf abscises (e.g., figs. 4A; 6C). panded leaves (figs. 2B, 6D). Until they fall they Development is essentiallycontinuous and seemingly enclose all youngerorgans and form a cavity which synchronous with vegetative shoot extension, al- is filled with either fluid (as in Rhizophora) or though there is a longer or shorterperiod of inflor- semi-solid mucilaginous material (as in Ceriops). escence maturationbefore flowersopen. This period This exudate is apparentlyproduced by the series of is longest in most Rhizophora species because in- colleterson the inner surfaceof the stipule base and florescencebranching before final flower initiation at its attachment(figs. 2C, 6D, E; 7E, F). Similar is extensive (e.g., fig. 8A). In the extremeexample colletersoccur in the axils of bracts (fig. 6C, inset). of R. apiculata the flowersare not fullymature until There is evidence that the fluid exudate of Rhizo- the inflorescenceoccupies a position below the ter- phora contains sugar and is attractiveto birds (Pri- minal leafy rosette (figs. 9A, B), a feature shown mack and Tomlinson 1978), a device which possibly laterto have importantbiological consequences.Most reduces those insect predatorswhich are likely to be taxa have some periodicityof flowerproduction, de- eaten by birds. terminedendogenously or exogenously,but in many cases by a combinationof both mechanisms.Gill and INFLORESCENCE MORPHOLOGY.-Reproductiveaxes Tomlinson (1971) showed that Rhizophora in in all species occur singlyin each axil of a pair South Florida is essentiallyever-flowering but with

TABLE 3. Floral featuresin Rhizophora.

Species Distributiona Inflorescenceb Stamenc Petals Style Reference0 No. of branch 1st flower No. lengthd orders node number A. "Fertile" group apiculata Indo-Malayan 1 (-2) 2 2 (-4) 12 glabrous short 3, 6 mangle Atlantic 1-2(-3) 2, 3 2-5 8 hairy long 4 mucronata Indo-Malayan 2-4 2 4-16 8 hairy short 3 racemosa Atlantic 2-5 2 4-numerous 8 hairy long 1, 2 samoensis Pacific 1-2(-3) 2, 3 2-5 8 hairy long 5, 7 stylosa Indo-Malayan -Pacific 2-4 2 4-16(?) 8 hairy long 3, 6 B. "Sterile" group (+ putative parents) X harrisonii Atlantic 2-5 2 4-numerous c.8 hairy long 1, 2 (mangle X racemosa) X lamarckii Australasia 2 2 4 c.16 sl. hairy long 6 (apiculata X stylosa) X selala Fiji, New 1-2(-3) 2, 3 2-7 8 hairy long 5, 7 (samoensis X Caledonia stylosa) aDistributionis very generalized and distinguishesessentially between "Indo-Malayan" (i.e., to the western Pacific) from "Atlantic" (west Africa-Caribbean,but including in part the Pacific coast of South America). More re- stricteddistributions are specified. bInflorescencebranches refersto number of bifurcations(2) or trifurcations(3) below flowersand usual range of ac- tual flowernumbers. "Numerous" refersto numbersgreater than 32. cStamen number refersto usual number,where it is not 8, the range may be considerablee.g., 7-21 in R. X lamarckii. dStylelength merelydistinguishes short (sessile stigmas or style < 1 mm) fromlong (style > 1 mm). eReferencesto details of floral morphologyas follows: 1. Breteler (1969), 2. Breteler (1977), 3. Ding Hou (1958), 4. Gill and Tomlinson (1969), 5. Guppy (1906), 6. Tomlinson and Womersley (1976), and 7. Tomlinson (1978).

FIGURE 7. Kandelia candel (from fluid-preservedmaterial, Paul Chai, s.n., Sarawak). A, terminalleafy shoot with flow- er buds (x 1/3); B, entirefew-flowered inflorescence (x 1/2) C, detached matureseedling (x 1/3); D, floral diagram; E, Vegetativebud representedby enveloping stipules (x 3/2) (inset: diagram of bud in t.s.; stipules cross-hatched,leaves and branch axes solid black); F, details of vegetativebud in E with stipules removedto show colleters (x 3) (inset: de- tail of detached stipule to show colleters); G, flower fromthe side (x 3/2); H, flower in I.s. (x 3) (inset: detail of stigma); I, a single petal (x 4).

Floral Biology in Rhizophoraceae 265 seasonalfluctuation in therate of flowerproduction. cymesoccasionally are initiated. In Queenslandthere is evidentseasonality in flower- 2. FLOWERS IN REGULAR CYMES. These occur in ing, fruiting,and seeding,even in Rhizophoraspe- Bruguieracylindrica, B. hainesii,and B. parviflora cies. Here Bruguierais in flowerfrom May to Au- which may be referredto collectivelyas "small- gustand mustconstitute a majornectar source at a floweredBruguieras" (figs. 4, 1lA). The flowers time when otherterrestrial plants are generallynot are relativelysmall (1 cm longor less) and typically flowering.Three basic flowerarrangements exist: erect,have peduncles and pedicelsextended and bract- 1. FLOWERS SOLITARY. These occurin Bruguiera eoles vestigial.The basicunit is a regulardichasium. exaristata,B. gymnorrhiza,and B. sexangula,which Protectionof the maturefloral parts is again pro- may be referredto collectivelyas "large-floweredvided solely by the individualflower cups and sepal "(fig. 10). The flowersare large (2 cm lobes. long or more), pedicellate,without bracteoles, typic- 3. FLOWERS IN DICHOTOMIZING PANICLES. In Cer- ally recurvedso theypoint away fromthe terminal iops (figs.5B; 6C), Kandelia(figs. 7A,-B), and Rhi- vegetativebud, and are pendulous.Protection of the zophora (figs. 8A, 9B), the flowersare moderately maturefloral parts is hereprovided only by themas- largetO quite small,peduncles and pedicelsare var- sive floralcup and sepal lobes. The distinctionbe- iouslyextended, the flowersoften sessile, and the tweenthis, and the next typeis not absolute,since bractsand bracteoleswell-developed, enclosing in-

FIGURE 8. stylosa(from fluid-preserved material, P. B. Tomlinson, X 74C, BootlessBay, Port New Guinea), inflorescenceand floral A, leafyshoot with leaves cut offto showmuch-branched in- Papua PF morphology. 30 florescencewithinRhizophora the leafy rosette (x positionof oldestleaf pair shown by arrow; B, open flowerin l.s. (xMoresby, 3); C, open flowerfrom above (x 3); D, floral?2),diagram; E, undehiscedstamen (x 3); F, dehiscedstamen (x 3); G, style(x 3); H, petal (x 3). D zytoza 4tota FIGURE 9. Rhizophoraapiculata (from fluid-preserved material, P. B. Tomlinson,18 X 74C, Singaua,Lae, Papua New Guinea), inflorescenceand floralmorphology. A, leafyshoot with young flower buds (x ?4); B, floweringshoot with foliageleaves cut offto showmature flowers below leafy rosette (x ?2), positionof oldestleaf pair shownby arrows; C, open flowerwith bracteolar cup in side view (x 3); D, openflower in l.s. (x 3); E, open flowerfrom above (x 3); F, detailsof ? sessilestigma (x 3); G, floraldiagram; H, dehiscedstamen, adaxial view (x 3); I, Petal (x 3).

266 Tomlinson, Primack, and Bunt < P~~~~~~~~~~~~~~~~~F

tHt1

Floral Biology in Rhizophoraceae 267 i.~* 1- ~~~~~i sT r ..... | \,

268 Tomlinson, Primack, and Bunt ~~~~~~~~~~~~- A

FIGURE 11. Floral aspects in small-floweredBruguiera, Ceriops, and Rhizophora. A, Bruguiera parviflora,Missionary Bay,Hinchinbrook Island, North Queensland; leafy rosette from above to showerect flowers. B. Ceriopstagal, same lo- cality;opening and open flowers.C and D, RhizophoraX lamarckii,same locality. C, recentlyopened flower from below, some stamensalready lost; D, leafyshoot to showpendulous flowers.

FIGURE 10. Bruguiera,floral aspects in large-floweredspecies. A-D. B. gymnorrhiza,Missionary Bay, Hinchinbrook Island,North Queensland. A, leafyrosette with 2 axillaryflowers at the timeof floweropening; B, flowerfrom below withpetals in dosed position,before pollen dispersal; C, old flowerwith petals and stamenslost; D, flowerfrom below immediatelyafter explosive discharge of pollenby openingof petals.E-F. B. exaristata,Three Mile Creek,Townsville, NorthQueensland. E, flowerbuds and open flowersin characteristicpendulous position; F, two flowersfrom below, the rightone withunopened petals (cf. fig.B), the leftwith petals and stamenslost.

Floral Biology in Rhizophoraceae 269 florescencebranches and flowersto a late stage of sense, so that larger flowers are either solitary,or development.Protection of developingparts, includ- borne on little-branchedinflorescences. However, in ing flowers,to quite a late stageof developmentis the most elaborate inflorescenceof R. X. harrisonii, providedby the thickfleshy bracts. The bractsof the flowersare still largerthan those in Ceriops. The Rhizophoraapiculata are quite corky(fig. 9B). extensiverange of flowersizes is illustratedin figure This thirdtype of inflorescenceis unusual among 1 at a standard magnification.Flower size is corre- angiospermsbecause the axis bifurcatesby equal lated more closely with the size of the pollinator. dichotomyin planes mutuallyat rightangles, each forkbeing associatedwith a pair of thickbracts FLORAL PLAN.-All rhizophoraceousflowers have a whichsubtend the next higher branch order and en- well-developed,thick, fleshy floral cup (or hypan- close younger,developing parts. The same typeof thium) with free valvate which enclose the inflorescenceis known for other Rhizophoraceae, e.g., inner parts until anthesis and always persist in the Ca?rallia.The flowersthemselves are subtendedby a fruit (e.g., fig. 3H). The calyx cup is either flat pair of emptybracteoles. When equal dichotomyis (often with a distinctdisc), or ranges fromshallowr regular,a simple power law (base 2) determines to verydeep. Petals are free,equal in numberto and flowernumber according to sequence2, 4, 8, 16 . . . alternating with the sepals, and usually provided etc.,but for higherorders irregularities appear by with diagnosticallyuseful appendages (hairs or club- abortionof certainaxes so thatthe actual number of shaped structures). Stamens are usually twice the flowersis less than the potentialnumber as deter- number of petals, with one series antesepalous, the minedby numberof forks.Rarely is reduction(or other antepetalous,and all inserted on the rim of abortion)so completethat a singleflower is devel- the calyx cup. These common featuresare illustrated oped,as we have observedin R. mucronata.An un- in the floraldiagrams in figures3-9. Kandelia is ex- derlyingcymose construction is also suggestedby a ceptional in the numerous (30-40) stamens (fig. veryoccasional dichasium in thisspecies. Rhizophora 7D). Anthersdehisce within the flowerbud, except mangleand R. X selala are exceptionalin thataxes in Ceriops decandra and Kandelia, although this ac- frequentlytrifurcate at the first(rarely subsequent) tion does not lead to automatic self-pollination.The node so thatnumbers occur which are initiatedby ovary is inferior,with two or more locules. Ovules thistrifurcation and subsequentdichotomy (e.g., 3, are few,large and relativelyfixed in numberfor each 5, 7). For Ceriops thereis some indicationof a taxon (the numbers cited in table 1 refer to the divisioninto fouraxes at the firstnode (e.g., fig. most common values). There is a simple persistent 5B). stylewith two or more obscure stigmas.The number Flowernumber varies in this inflorescencetype of parts is relativelyfixed and thereforediagnostic but within narrowlimits for most species, two for each taxon. Informationis summarizedin tables (rarelyfour) in R. apiculata,four in R. X lamarckii, 2 and 3 and details are shown in figures2-9. up to over100 in vigorousspeciments of R. X harri- Figure 1 also shows that the shape of the floral soniiand R. racemosa;where branching, up to seven cup is very variable and to a large extent correlated orders,has occurred.For Ceriops,flower numbers are with flower size. Thus the large-floweredBruguiera usuallylow (2-11), but in Queenslandpopulations species (figs. 1A, B) have a deep calyx cup which of C. decandraexceptionally high numbers(30-40) becomes nectar-filledat anthesis.Small-flowered Bru- have been observed. guiera species (figs. IC, D) and Ceriops (figs. 1H, The essentialmechanical and biologicalfeatures I) have a well-developed,flat-topped, or lobed disc of thistype of inflorescenceprovide the continuous which encloses the base of the stamens.The disc of protectionof the youngestunits by a successiveser- Rhizophora is scarcelydeveloped, and the top of the ies of organs(bracts, bracteoles, and finallysepals). ovary is more or less conical (figs. 1E, F). Ceriops Flowersare normallyinverted, this being conspicu- tagal has a very shallow, half-inferiorovary but ap- ous in long-pedunculatespecies like R. racemosa parent nectar glands alternate with the stamens and R. stylosawhere the whole inflorescence becomes (fig. 6J). The calyx cup of Kandelia is shallow (fig. invertedby curvatureof its axis. This featureseems IG) . biologicallyadaptive in termsof pollendispersal, as Functionallyall flowers are comparable in that suggestedlater. Chai (pers. comm.) recordsthe the valvate calyx lobes separate at anthesis,diverging flowersof Kandeliato be invertedat maturity. to greateror lesser degree,to expose the petals which adopt various configurations.Petals and stamens fall FLOWER SIZE.-This feature is correlated with the or are blown fromthe open flowerafter they abscise elaborationof the inflorescencein a verygeneral within one (Rhizophora),or a few (other genera),

270 Tomlinson, Primack, and Bunt days. This leaves the central stigma alone in the After the petal-stamencatapult has "fired," the "empty" flower. At no time is there obvious stig- petals lie back against the calyx lobes with the empty matic secretionor divergenceof stigmas; stigma re- stamens twisted and disorganized (figs. 3D, 10D). ceptivityhas been established by chemical tests, as All abscise and usually fall within a day or two (fig. described earlier. Older stigmas become black, but 1OC). Untriggeredpetals may retaintheir tension for the style usually persists in the fruit. Unfertilized up to 10 days, and they may be lost without ever flowers abscise early. opening. Accidental triggeringof petals, at least in Despite this uniformfloral plan at least six, quite the large-floweredBruguiera species, seems rare so elaborate, pollination mechanisms are involved, to that a flowervisitor appears essential for pollen dis- some extent the categoriescutting across taxonomic persal. boundaries.Three of the categoriesinvolve explosive release of pollen by the same mechanismand may be BIRD-POLLINATION IN LARGE-FLOWERED BRU- described collectively,but the basic subdivision is GUIERAS (Type 1).-In Queensland the flowers of still establishedby the pollen vector. B. exairistata(figs. 10E, F) are visited by the yellow- breasted sunbird (Nectarinia jugularis) and honey eaters (Meliphagidae). Birds often forage in pairs, a DIRECT OBSERVATIONS ON male and a female,calling frequently,with the male POLLINATION MECHANISMS activelychasing other birds out of the territory.The EXPLOSIVEPOLLEN RELEASE.-In all Bruguiera spe- birds perch on the stout branches and turn their cies and in Ceriops tagal the stamens become en- heads upward to take nectar via their curved bills; closed in pairs by the petals, i.e., each petal functions theywere observed to trip the explosive mechanism, as a pouch enclosing both an antepetalous and an releasing clouds of pollen but without being dis- antesepalous stamen. The latter is initiated opposite turbed by the explosion. Birds also move frequently a sepal, but early in developmentit twists (either to between plants,and could act as effectivepollen dis- the rightor left,but always in the same directionin persers; they feed on a few flowers on each plant a single flower) so that as the petal base enlarges it and also eat insects. grows around the stamen. This antesepalous stamen Flowers are pendulous and open slowly over a is slightlylower in the flowerthan the antepetalous period of about 24 hours,the petals at firstexposed stamen because of the basal twist in the filament as brown pouches alternating with the spreading (e.g., figs.3E, 4K). Tension is developed because the sepals which are conspicuous against the foliage. stamenspress against the interlockedmargins of the Stigmas are not receptiveon the firstday the flower petal, whose dorsal attachmentis thickened.When is open, but become receptiveon the second or third the flowerexpands the petal tends to bend back, but day and remain so afterthe stamensand petals have is retainedin an erect position by the adherentven- fallen. tral margins. Marginal hairs seem to be important Individual flowers are visited for as much as in holding the petals in the folded position. Shortly eight days afteropening, and in one set of measure- before the flower opens the stamens dehisce. At ments stamens continued to be triggeredseparately anthesis, the sepals diverge presenting the closed, over a period of 7 days. Copious nectar accumulates erect petals in the "cocked" position. Suitably trig- in the deep calyx cup and is retained by the petal gered, the petal margins unzip instantaneouslyand base and its associated hairs. In unvisited flowersit fly apart, releasing the stamens which catapult the overflowsonto the petals where it becomes viscous loose pollen toward the center of theeflower, often by evaporation and so inhibits the explosive mech- as a visible cloud. Presumablymuch of the pollen anism. would be projected onto the head of an animal vec- Stamens persist in either the triggeredor tripped tor,but none were trappedto obtain more precise in- state for up to 10 days in measured examples, but formation.Flowers in Bruguiera gymnorrhizabefore they eventuallyfall with the petals as a single unit and after pollen release are shown in figures lOB or in smaller groups of stamens and petals whether and lOD, respectively. or not theyhave been fired. The mechanism for triggeringa flowervaries in After the petals and stamens have fallen nectar differentsized flowers and according to the flower continues to be secreted,the stigma remains recep- visitor, as described below. Each petal behaves in- tive, and birds continue to visit the "empty"flowers. dependentlyso that multiple visits are possible, and This action and the initiallydelayed stigma receptiv- individual flowerswith combinationsof closed and ity constitutea partial,but weak, outbreedingmech- open petals can be found. anism via partial protandry.

Floral Biology in Rhizophoraceae 271 Triggeringof petal explosion is effectedby beenobserved by us in B. parviflora.Here becauseof touchingsensitive basal hairs,especially those which the smallerflowers, the explosionis not violentand projectinto the center of theflower over the entrance no visiblecloud of pollenis observed;petals can be to thecalyx cup. Neitherthe tip of thepetal nor its triggeredby a delicate touch to the distal parts. apical hairs are sensitive.Appropriately, therefore, These flowersare not likelyto be robbedby birds triggeringcan onlybe stimulatedby fairlyvigorous becausethey secrete very little nectar. They have no probinginto the calyxcup. evidentodor. It is likelythat other insects, e.g., bees, The flowersof B. gymnorrhiza(figs. 10A-C) are visitthe flowersand mayalso functionas vectors. largerthan thoseof B. exaristata,and the calyx is Adaptationsto butterflypollination in this spe- red.We have observedflowers visited by the grace- cies maybe summarizedas follows,although it might ful hopeyeater (Meliphaga gracilis) and less fre- be said equallythat these are reasonswhy the flowers quentlyby the duskyhoney eater (Myzomelaob- are not bird-pollinated:Flowers are not recurved, scura). Birds again feed in pairs; theyfly between and are displayedto the outsideof the treecrown the treeswith the malescalling frequently in a way (figs.4A, 1lA); flowersare small (comparedwith whichsuggests territorial warning. Pollinator activity bird-pollinatedflowers); branchlets are thinand in- is high,and in one measuredsample five out of sufficientfor bird perching;petals are greenish-yel- sevenflowers possessed exploded petals only one day low; calyxcup is flat,nectar is producedin small afterthe flowershad opened.Bruguiera sexangula, quantities;and pollen is releasedby delicate,distal illustratedin detailin figure3, is distinguishedmain- stimulationof petals. lyby the petal appendages (figs. 2D; 3F; G), and has Bruguieracylindrica is verysimilar to B. parvi- the same explosivemechanism as otherspecies. We floraand is here illustratedin detail (figs. 4D-J). have not had the opportunityto watchpollination The floralcup is shorterthan in B. parviflora,with a ourselves,but birdvisitors are well known (Davey distinctcalyx cup. We have not observedflower 1975). visitors,but expect that these will be insects.The re- In summiarywe note the followingadaptations mainingspecies, B. hainesii,also has explosivepol- forbird pollination: 1. Recurvedpedicels of solitary len release,but furtherdetails are lacking. flowersto facilitateaccess by birdsperched below and perhapsalso to preventrain fromdiluting the MOTH-POLLINATION IN Ceriopstagal (Type 3). nectar;twigs are stout.2. Flowersare all attractive The flowerof thisspecies, although with the same byvirtue of thespreading stellate calyx lobes and in mechanismas theprevious two groups,is specialized two species (B. gymnorrhiza,B. sexangula) by the to a differentpollen vector in its function(fig. 6). reddishouter surface of the hypanthium.3. Copious In NorthQueensland we have observedflowers to nectarproduction, secreted into a deep calyx cup, open mainlyin the laterafternoon (fig. 1lB) and suitablefor an activepollinator. 4. Largeflowers with emit a faintbut characteristicfragrance during the a heavyconstruction suitable for a powerfulpollin- night.The whitepetals are not conspicuouswhen ator like a bird. foldedbut becomeso afterthey have unfolded.We We also note thatat MissionaryBay, Hinchin- have seen mothsvisiting the flowersin the early brookIsland, B. gymnorrhizawas the only nectar- evening,and theyare thepresumed agents of pollen producingplant to flowerin winter(July) and at transfer,but otherinsects could easilyperform the Townsville(3 Mile Creek),where B. exaristatawas functionsince the nectariferousdisc is exposed. observed,few otherplants were in flowerat the Adaptationsfor moth pollination are as follows: same time.Consequently, in winter,when theseob- smallflower and flowerparts suited to a weak pol- servationswere made, Bruguiera becomes a veryim- linator;delicate explosive mechanism, triggered by a portantnectar source for birds. Clearly also theman- delicatestimulation of distal parts which will not grove environmentis edaphicallysuited to winter injure the pollinator;petals white,and visible at flowering,since treesare not limitedby soil mois- night; flowersfragrant at night,characteristic of ture.Birds are the majorpollinators of manyother mothflowers; and a tinyamount of nectar,suggest- Australiantree species (Ford and Paton 1976,Paton ing a smallinsect vector. and Ford 1976) so theremay be a highdependence The illustrationof petals in C. tagal in Ding in birdson nectarsupply. Hou (1958) whichshows them fused to forma ring and theirdescription as "coherentat the base with BUTTERFLY-POLLINATION IN SMALL-FLOWERED uncinatehairs" refers to themafter explosion but in BRUGUIERAS (Type 2) .-Large butterfliesfeeding driedspecimens. The same artefactcan be observed on open flowersand so triggeringpollen release have in fluid-preservedmaterial. It is not a primaryfea-

272 Tomlinson, Primack, and Bunt ture of the pollinationprocess, since it refersto jux- The stage of pollen release is short,petals and taposition of the petals (fig. 61) after they have stamens usually falling within 24 hours of flower been triggred (fig. 6G). opening, as noted by Guppy (1906). This condition should be contrastedwith that of Bruguiera where NON-EXPLOSIVE POLLEN RELEASE BY Ceriopsde- flowers remain functional at the unopened petal candra (Type 4) .-Flowers (figs. 5C-F) in this stage for up to 10 days. species lack the petal-stamenconfiguration on which the explosive mechanism of C. tagal depends. Sta- The stigma of Rhizophora stylosabecomes recep- mens have relativelyshort thick filaments(fig. 5F) tive on the second day afterthe flower has opened and remain in their original antesepalous or ante- and hence normallyafter the stamensand petals have petalous position. The petals (fig. SE) become in- fallen. Pollen grains accumulate on a specific region terlockedmarginally by basal hairs, i.e., as illustrated of the stigma only when the stigma is receptive.As by Ding Hou (1958) for C. tagal, and this circum- a result,an individual floweris generallyincapable of stance produces a shortcorolla tube, crowned by the self-pollination.In the population under study,most series of clavate appendages. flowers fall with their petals still attached, before We have no details of flowervisitors in this spe- they become receptive to pollen. Some other pre- cies, but it is evident that the mechanismis less elab- sumred endogenous mechanism controls fruit set orate than for C. tagal with which it is much con- since pollination does not seem limiting. trasted.Chai (pers. comm.) refersto the flowersof (fig. 9E) shows the wind- this species as non-scented and being visited by pollination mechanism in its most complete form. Trigonid bees, especially during early morning. The calyx lobes diverge least, the petals neitherre- curve nor retain pollen since they are glabrous, and WIND-POLLINATION IN Rhizophora(Type 5).- a pepper-pot mechanism is sufficientto dis- Despite the variation in stamen number and petal simple the dust. Slow development of the hairiness, the floral mechanism is essentially the perse pollen such that they are mature only below the same in all Rhizophora species, and we interpretit flowers leafy rosetteon each branch (e.g., fig. 9B) can be as facilitatingwind-pollination, providing evidence interpretedas a simple device which facilitatespol- (below) frompollen-ovule ratios and other sources. len dispersal by minimizinginterference of the tree's Flowers open at any time of day, and the calyx lobes own foliage. Equally this may facilitatecapture of may separate quite widely,as in R. mangle, or only pollen by receptive flowers. slightlyas in R. apiculata and R. racemosa.The mul- tilocellate anthers (figs. 8E, F; 9H) dehisce before Insect visitorsto the flowersof most Rhizophora the flower opens via the adaxial flap which falls species are not common. We have recordedoccasion- against the base of the style.When the styleis short, al pollen-collectingbees visitingflowers of R. mangle as in R. apiculata (figs. 9D, F), the stigma may be in South Florida and of R. X lamarckiiin Queensland. occluded by the adaxial flap, although this does not Bees visit certain populations of R. stylosa in abun- seem to be part of the floralmechanics. For all spe- dance to collect pollen. These bees, as well as occa- cies of Rhizophora except R. apiculata (fig. 91) and sional butterfliesand birds, visit the flowers after R. X lamarckii,the petal margins are provided with the petals and stamens have abscised, seeminglybe- a dense weft of filamentoushairs (fig. 8H) which cause theirscars will exude fluid.In other large pop- grow into the space between the anthers in the ulations of R. stylosa at the same time of year, no closely packed flower bud. Because of precocious flowervisitors were observed. anther dehiscence the hairs become coated with the Evidence for the wind-pollinationmechanism in powdery pollen mass, whose release is facilitatedby Rhizophora may be summarized as follows: light the multilocellatecondition and the method of de- powderypollen is produced in great abundance (see hiscence by a flap. As the floweropens much of the pollen-ovule ratios below); anthesis in bud, pollen pollen is drawn out by the petals,which become more siftedout with no animal agent necessaryfor disper- or less recurved in all the species with hairy petals sal; pollinator attractantsmore or less absent, i.e., (fig. 8C). Dispersal of the pollen is easily facilitated little nectaror fragranceproduced, and the abundant when the floweris shaken by the wind and is further pollen is diffuselypresented; stamens are short-lived, promoted by hygroscopic movement of the inter- fallingthe same day as the floweropens; flowersare twined petal hairs, which are released in a flicking typicallypendulous at maturityand, for R. apiculata, motion. The marked tendencyfor flowersto be in- opening below the leafy rosettes; extensive pure vertedor pendulous (figs. 1 C. D) seems important stands of occur in a windy environment;and in this mechanism. greased slides suspended in stands of Rhizophora

Floral Biology in Rhizophoraceae 273 both in Queenslandand SouthFlorida retain abun- 1 Pollinationby animals1 dantwind-borne pollen. 3 Pollinationmechanism explosive (anthers Againstthis evidence, however, it mustbe noted dehiscentin the flowerbud). thatthe stigma is notelaborated in themanner usual 4 Pollinationby day-timevisitors. for wind-pollinatedspecies. 5 Pollinationby birds large-floweredBruguiera species INDIRECT OBSERVATIONS (e.g., B. exaristata,B. gymnorr- hiza,B. sexangula) 5 Pollinationby butterflies POLLINATOR PREDATORS.-Supportingevidence that small-floweredBruguiera species pollinatorsin pollinationTypes 2 and 3 above are B. cylindrica,B. hainesii,B. insectscomes from the observationthat spiders (e.g., parviflora) (family Thomisidae) position themselvesbelow 4 Pollinationby night-timevisitors. flowers.We have observedan unidentifieddark- 6 Pollinationby moths colored spider catchingmoths on Ceriops tagal. Ceriops tagal Green crab-spiders,which are muchthe same color 3 Pollinationmechanism not explosive as the inflorescenceof Bruguieraparviflora, position (anthersnot dehiscentin the flower themselvesbelow flowers,but we did not observe any insectcaptures. bud). short-tongued In contrast,no crab spiderswere observedon 7 Pollinationby small, -Ceriops decandra2 Rhizophoraflowers or the flowersof the bird-pol- animals linatedBruguiera species (Type 1). We have evi- 7 Pollinationby fairlylarge, longer- animals- Kandeliacandel2 dence that Rhizophorapromotes bird visitorsvia tongued sugar secretionsfrom its terminalbuds (Primack These biologicalfeatures can be relateddirectly and Tomlinson1978). In both examplesattracted to morphologicalfeatures which include differences birdswould eat spiders. in size of flowers,differences in detailsof inflor- escenceand floralconstruction, and differencesin POLLEN-OVULE RATIOS.-Comparative evidencethat floralmechanisms (summarized in table2 and illus- contraststhe different types of pollinationmechanism tratedin the severalfigures). The most elaborate is providedin the measurementsof pollen-ovulera- typeinvolves an explosiveprojection of pollentrig- tios in table 1. Rhizophorahas appreciablyhigher geredby flowervisitors. Early authors who investi- values than the bird-pollinatedBruguiera species gatedthe bird mechanism of triggeringin Bruguiera (whichhave usuallymore stamens). This largepro- gymnorrhiza(Gehrmann 1911, Porsch 1924) sug- ductionof pollen is typicalof wind-pollinatedspe- gestedthat the bird visitswere secondaryand that cies. The bird-pollinatedBruguiera species produce the explosive mechanismof dispersal was for over 100 timesas muchpollen per ovule as the in- anemophily.However, it seems unlikelythat so sect-pollinatedspecies of Bruguieraand Ceriops. elaboratea mechanisminvolving copious nectar Thesevalues and differencesare comparableto those production,a massivecalyx, and preciseflower or- obtainedby Cruden (1976, 1977). ientation,could be developedin relationto anemo- phily,especially as we now know that firingof a flowervisitor. Further- DISCUSSION themechanism depends on more, the differentiationof flowersaccording to BIOLOGICAL CONSIDERATIONS.-The informationsize, scent, number per inflorescence,and orientation presentedabove, although it is incompleteat many suchthat birds on the one hand and differentkinds points,is sufficientto show thatalthough the Rhi- of insectson the otherbecome specializedflower zophoreaehave a similarfloral construction, they visitorsmakes it unlikelythat the mechanismis di- exhibita wide rangeof pollinationmechanisms. The rected toward anemophily.Undoubtedly where a situationmay be summarizedin the followingdiag- cloud of lightpollen is releasedwind dispersalis nostickey: possible,especially in the large-floweredBruguiera 1 Pollinationpredominantly by wind (anthers species,but thisseems incidental to the real vector. dehiscentin the flowerbud). Floralenergetics clearly vary in thedifferent pol- 2 Matureflowers within the leafycluster ------Rhizophora (most species) lOnly the usual and presumed most effectivepollinator is 2 Matureflowers below the leafycluster included, flowersmay receive a diversityof visitors. --- R. apiculata 2No directobservations made.

274 Tomlinson, Primack, and Bunt linationtypes. A high nectarand pollen production its major flowervisitor (Nectarinia jugularis). How- by Bruguieraexaristata and by B. gymnorrhizaall ever, it is not usual for a high degree of interde- suggestthat bird pollinationis energeticallycostly pendence to develop between plant and bird in bird- and also relativelyinefficient at moving pollen pollinated species (Carpenter, 1976). grainsfrom the anthersto receptivestigmas. One Further extensive work on the rhizophoraceous presumedadvantage to theplants of birdpollination mangroves, especially in different geographical is thatthe birds forage actively in weatherconditions areas, is an obvious requirementto substantiatemany unsuitablefor insects.Furthermore, though birds of the claims in this paper. It is hoped that this out- maybe inefficientin theirdispersal of pollen,their line will prove useful in orienting future research tendencyto fly activelyfrom plant to plant may workers. compensatein providinga highdegree of cross-pol- A serious deficiencyin our approach is that we lination.In contrast,the low pollen/ovuleratios of physiology and genetic the insect-pollinatedspecies suggestsa more effi- remain uncertain of floral for isolation. That there is some mech- cient systemless costlyto the plant. Here flowers mechanisms of self-sterilityis suggestedby the relative in- adaptedfor insect pollination may place the pollen anism 0-11 percentof moreprecisely on the pollinator,in contrastto the frequencyof fruitset in Rhizophora, all flowers according to measurementsat different bird-pollinatedflowers in whichthe flowerappar- 1971, and Kress entlyprojects a cloudof pollenonto the head of the times of year (Gill and Tomlinson 1974) in Florida. bird.These Bruguiera species may be comparedwith Metrosideros(Myrtaceae) studied by Carpenter We have commentedupon the likelymechanismr (1976) in Hawaii. Here bird-pollinationseems par- for out-crossingprovided by the partial protandryof ticularlyefficient in relationto a presumedpartial large-floweredBruguiera species, and if stigma re- self-compatability,and large-floweredBruguiera spe- ceptivityis universallydelayed, dichogamy may be cies mnaybe comparable.However, a detailedinvesti- an effectivedevice. In contrast,self-compatibility is gation using the carefulquantification adopted by suggested for Rhizophora by Kress (1974) on the CarpenLerstill remains to be carriedout. basis of the ability of isolated greenhouse-grown We have also providedevidence on the basisof plants to set fruitand the ratio of albino to normal floralnmechanisms, orientation, and highpollen-ovule seedlings in individual trees which show this mark- ratiothat Rhizophora is predominantlywind-pollin- er. It is clear that futureresearch must concentrate ated; the greatestspecialization for wind-pollination on stigma receptivityand the genetic mechanisms shownby R. apiculatawith its pepper-potmechan- for preventingselfing. ism and essentiallyramiflorous condition; and the leastspecialization shown by R. stylosawhich is ac- EVOLUTIONARYCONSIDERATIONS.-The wide range tivelyvisited by bees. The abundanceand wide dis- of mechanismsdeveloped by the Rhizophoreaeallows persabilityof Rhizophorapollen is shown in the some speculationabout theirevolutionary divergence workof Muller(1959) on sedimentsin theOrinoco since it seems reasonable to assume a monophyletic delta,where there is no otherexisting genus of Rhi- origin for them in view of vivipary and the other zophoraceaeto confusethe picture.Rhizophora pol- common featuresthey share (van Vliet 1976). Kar- len predominatesin mostof his samples.His con- delia representsthe least specialized type of floral clusionthat Rhizophorais wind-pollinatedis fully mechanism which may approximate the ancestral supportedhere and the abundanceof wind-borne state. However, we still lack details about its floral pollen in existingRhizophora communities is easily biology. Ceriops decandra is comparativelyunspe- verified.More recentlyit has been shownthat pol- cialized and may also reflectan ancestral state, but len in differentRhizophora species is scarcelydis- again we lack details of flower functioning.The ex- tinguishable(Muller and Caratina1977). This find- plosive mechanismrepresents a considerablespecial- ing wouldsuggest a uniformpollination mechanism. ization and a monophyleticorigin seems likely. Cer- UndoubtedlyRhizophora is the most successful iops may be the key genus in this process since it in- mangrovein termsof wide distributionand abun- cludes both specialized and unspecialized taxa. Adap- danceof individuals.Wind-pollination may partially tive radiation leads to the Bruguiera condition. Rhi- accountfor both the successof Rhizophoraand the zophora representsa differentline of evolution,di- frequencyof putativehybrids. In contrastthe range rected toward anemophily,but probably having had restrictionassociated with specialized flower visitors an ancestral entomophilous condition; vestiges of is suggestedby Bruguieraexaristata, whose limited this may be seen in taxa like R. mangle and R. stylosa geographicaldistribution coincides quite closely with on the basis of reports of flower odor, sometimes

Floral Biology in Rhizophoraceae 275 nectar, and flower visitors (mainly for pollen) ACKNOWLEDGEMENTS (Kress 1974). Rhizophora apiculata in contrast Visits by P. B. Tomlinson to the westernPacific in 1973, seems most specialized for wind pollination,the pol- 1976, and 1977 were sponsored by grants from the Na- tional Geographic Society and Grant INT #76-24479 of len being shaken out of a half-open flower, borne the National Science Foundation Office of International below the leafy crown of the twig. Rhizophora may Programs. Additional support has come from the Atkins still be in an active state of evolution, since inter- Garden Fund and Cabot Foundation of Harvard Univer- sity. The following have provided direct logistic support: specificsterility barriers seem to break down readily, Director, Australian Institute of Marine Science, Towns- leading to hybridswarms of which R. X lamarckiiin ville, Queensland; Director General, O.R.S.T.O.M., and the the New Papuasian area (Tomlinson and Womers- Director of its Center in Noumea, New Caledonia; Chief, Division of Botany, Department of Forests,Lae and Con- ley 1976), R. X harrisonji in the Caribbean-West servatorof Forests,Port Moresby,Papua New Guinea; Su- African area (Breteler 1969, 1977), and R. X selala perintendent,Suva Herbarium, Fiji, and the Curator,Har- vard UniversityHerbaria. Individuals who have provided in Fiji and New Caledonia (Tomlinson 1978) rep- field assistanceinclude N. C. Duke, D. R. Frodin, M. Ga- resentexamples. The problem here is of understand- lore, E. E. Henty,Camilla Huxley, P. Morat, Thelma Rich- ing the genetic basis-forthe breeding mechanismof mond, J.-M. Veillon, Kevin White, and J. S. Womersley. We are gratefulto Priscilla Fawcett,Botanical Illustra- Rhizophora. tor at FairchildTropical Garden, for the black-and-whiteil- lustrations;to Paul Chai, ForestBotanist, Kuching, Sarawak, Our present understandingis thereforeprelim- , for fluid-preservedmaterial of Kandelia and in- inary and incomplete.There is much scope for fur- formation about its floral behavior, and to Mrs. E. C. Sweeney, "The Kampong" 4013 Douglas Road, Miami, ther extended work on the floral biology of man- Florida 33133, for access to cultivatedBruguiera specimens. groves. Permissionby the Editor-in-Chiefto reproduceillustra- tions of Rhizophora apiculata and R. stylosa which first appeared in "Contributionsfrom Herbarium Australiense" is gratefullyacknowledged.

LITERATURE CITED

BRETELER, F. J. 1969. The Atlanticspecies of Rhizophora. Acta. bot. neerl. 18: 434-441. . 1977. America's Pacific species of Rhizophora. Acta. bot. neerl. 26: 225-230. CARPENTER, F. L. 1976. Plant-pollinatorinteractions in Hawaii: Pollination energeticsof Metrosideroscollina (Myrta- ceae). Ecology 57: 1125-1144. CRUDEN, R. W. 1976. Fecundityas a functionof nectar production and pollen-ovule ratios. In, Burley, J. and B. T. Styles. (Eds.). Variation, Breeding and Conservationof Tropical ForestTrees. Pp. 171-178. Academic Press,Lon- don. . 1977. Pollen-ovuleratios: a conservativeindicator of breeding systemsin floweringplants. Evolution, Lancaster, Pa. 31: 32-46. DAVEY, J. E. 1975. Note on the mechanismof pollen release in Bruguiera gymnorrhiza.Jl. S. Afr. Bot. 41: 269-272. DING Hou. 1957. A conspectusof the genus Bruguiera (Rhizophoraceae). Nova Guinea (n.s.) 8(1): 163-171. . 1958. Rhizophoraceae in Flora Malesiana. Ser 1. 5: 429-493. DOCTORS VAN LEEUWEN, W. M. 1927. Blumen und Insektenauf einer kleinen Korallen-Insel.Annls. Jard.bot. Buitenz. 37: 1-31, & 4 plates. FORD, H. A., AND D. C. PATON. 1976. Resource partitioningand competitionin honeyeatersof the genus Meliphaga. Aust. J. Ecol. 1: 281-287. GEHRMANN, K. 1911. Zur Blutenbiologie der Rhizophoraceae. Ber. dt. bot. Ges. 29: 308-318. GILL, A. M., AND P. B. ToMLINSON. 1969. Studies on the growth of red mangrove ( L.) 1. Habit and general morphology.Biotropica 1: 1-9. - - . 1971. Studies on the growthof red mangrove (Rhizophora mangle L.) 3. Phenology of the shoot. Biotropica 3: 109-124. Guppy, H. 1906. Observationsof a naturalistin the Pacific between1896 and 1899. Vol. II. Plant dispersal.MacMillan, New York. HALLE, F., AND R. A. A. OLDEMAN. 1970. Essai sur l'architectureet la dynamique de croissance des arbres tropicaux. Masson et Cie. Paris. , AND P. B. ToMLINSON. 1978. Tropical trees and forests,an architecturalanalysis. Springer Verlag, Heidelberg and New York. HESLOP-HARRISON, J.,R. B. KNOX, Y. HESLOP-HARRISON, AND 0. MATTSON. 1975. Pollen-wall proteins: emissions and role in incompatabilityresponses. In, Duckett, J. G. and P. A. Racey. (Eds.). The Biology of the Male Gam- ete. Pp. 189-202. Academic Press, London. KRESS, W. J. 1974. The floral biology of Rhizophora mangle in South Florida. Undergraduatehonors thesis.Biology De- partment,Harvard University. MULLER, J. 1959. Palynologyof recentOrinoco delta and shelf sediments: reportsof the Orinoco shelf expedition,Vol. 5. Micropaleontology5: 1-32.

276 Tomlinson, Primack, and Bunt , AND C. CARATINI 1977. Pollen of Rhizophora(Rhizophoraceae) as a guide fossil.Pollen Spores 19: 361-389. PATON, D. C., AND H. A. FORD. 1976. Pollinationby birdsof nativeplants in Soutlh Australia.Emu 77: 73-85. PORSCH, 0. 1924. Vogel BlumenstudienI. Jb. wiss. Bot. 63: 553-706. PRIMACK, R. B., AND P. B. TOMLINSON. 1978. Sugar secretionsfrom the buds of Rhizophora.Biotropica 10: 74-75. RICHMOND, T. DE A., AND J. M. ACKERMAN. 1975. Flora and fauna of mangroveformations in Viti Levu and Vanua Levu. In, Walsh, G. E., S. C. Snedaker,and H. J. Teas. (Eds.). Biology and Managementof Mangroves. I.A.F..S Gainesville,Florida. VAN STEENIS, C. G. G. J. 1958. in Ding Hou (loc. cit.) VAN VLIET, G. J. C. M. 1976. Wood anatomyof Rhizophoraceae.Leiden Bot. Ser. 3: 20-75. TOMLINSON, P. B. 1978. Rhizophorain Australasia,some clarificationof taxonomyand distribution.J. Arnold Arbor. 59: 156-169. , AND J. G. WOMERSLEY. 1976. A Rhizophoranew to Queenslandand New Guinea, with notes relevantto the genus. Contrib.Herb. Austral.19: 1-10.

Floral Biology in Rhizophoraceae 277

Top View