Plicate Staminal Filaments in Tillandsia Subgenus Anoplophytum

Amer. J. Bot. 76(10): 1478-1485. 1989.

PLICATE STAMINAL FHLAMENTS IN TILLANDSIA SUBGENUS ANOPLOPHYTUM (BROMELIACEAE)1

TIMOTHY M. EVANS AND GREGORY K. BROWN2 Departmentof Botany, University of Wyoming, Laramie,Wyoming 82071-3165

ABSTRACT Plication of staminal filamentsis an importantdiagnostic character for Tillandsiasubgenus Anoplophytum(ca. 45 species). The monophyletic integrity of subgenusAnoplophytum has recently been questioned, and we conducted an anatomical investigation of plicate staminal filamentsto bettercharacterize this putative synapomorphy.Developmental studies show that the filament plications, or folds, become visible duringor soon after anthesis. Serial sections of preplicationfilaments and filaments in sequential stages of plication were preparedand observedwith light microscopy.A uniform sequenceof parenchymacell collapse begins three to four cell layers out from the vascular bundle and proceeds centrifugallyto the epidermis. Eventuallythe epidermalcells collapse,leaving only the vascularbundle and a few surrounding parenchymacells intact. Above and below the zone of plication,all parenchymaand epidermal cells in the filament remain intact. Species traditionallyplaced in subgeneraTillandsia and Allardtiahave been found with plicate staminal filamentsthat are anatomicallyand developmentally identical to those studied from subgenusAnoplophytum. Alone, staminal filament plication does not appear to be a good diagnostic characterfor subgenusAnoplophytum, and doubtsconcerning the monophylesisof this subgenusare reinforced. The functionalsignificance of stamen filamentplication remains unknown.

ANOPLOPHYTUMwas originally proposed by This is clearly the case when comparing sub- Beer (1854) as a segregategenus of Tillandsia. genus Anoplophytumwith subgenusAllardtia, Tillandsiastricta, a commonly cultivated tax- where the diagnostic featuresare identical ex- on is the type species. Anoplophytumwas later cept for 1) relative stamen length and 2) the relegatedto sectional status within Tillandsia occurrenceof plicate staminal filaments. The by Grisebach(1 864). Baker(1887) maintained monophyletic nature of subgenusAnoplophy- Anoplophytumwithin Tillandsiabut upgraded tum has been in question (Gilmartin, 1972; it to the rank of subgenus. Mez (1934) recog- Brown and Gilmartin, unpublisheddata), and nized the essence of subgenusAnoplophytum, a morpho-anatomicalstudy of the plicated fil- as well as the six other subgeneraof Tillandsia aments was conducted to better characterize sensu Smith and Downs (1977). It was Smith the plication processassociated with taxa from (1935) who emended the typification of Ano- subgenus Anoplophytum.The gross morpho- plophytum and made the transverselyplicate logical similaritiesbetween plicate staminalfil- filaments diagnostic for the subgenus. aments and contractileroots were noted prior An examination of diagnostic charactersfor to the anatomical investigations. Consequent- the seven Tillandsia subgenera (Smith and ly, mechanistic parallels between root con- Downs, 1977) shows that these taxa are usually traction and filament plication processes re- distinguished by only one or two characters. ceived serious consideration. A secondary motive of this research was to evaluate the I possible functional significance of plicate fil- Received for publication3 November 1988; revision aments. accepted4 April 1989. We sincerelythank the collectorsand institutions(Table 1) for providingthe plant materialutilized in this study. MATERIALSAND METHODS-All flowers and We also thank Dr. Matt Lavin for direction in the use of buds used in this study were fixed in Camoy's the rotarymicrotome and the techniquesused in the prep- solution, as the primary use of this material aration of plant material for thin sectioning. W. Dennis Clark, Amy Jean Gilmartin, Harry Luther, and John F. was for chromosomal study. After a minimum Utley provided many useful suggestions on the manu- of 24 hr, they were transferredto 70%ethanol script. This researchwas sponsored by National Science and stored at or below 0 C. The collections Foundationgrants BSR-8607 187, BSR-8708267, an NSF acquiredfor use in this study (i.e., those with "ResearchExperiences for Undergraduates"award, and a researchgrant from the BromeliadSociety, Inc., to GKB, plicate filaments) were done so without prior andby the Departmentof Botany,University of Wyoming. knowledge as to their identification. The cri- 2 All correspondence. terion for study was based solely on the oc- 1478 October 1989] EVANS AND BROWN-STAMINAL FILAMENTS IN TILLANDSIA 1479

TABLE 1. Tillandsiaspecies examinedfor plicate staminalfilaments

Subgenus (SensuSmith & Species Downs, 1977) Voucher T. aequatorialis Allardtia Cultivated,Denver BotanicalGarden (RM) Lyman B. Smith T. butzii Mez Tillandsia Mexico, Edo. Chiapas Cultivated, Selby Botanical Garden (SEL 80-1598) T. caput-medusa Tillandsia Cultivated,Selby BotanicalGarden (SEL s.n.) E. Morren T. gardneriLindley Anoplophytum Brazil,Edo. Rio de JaneiroMartinelli 11575 (WS) T. sceptriformis Allardtia Ecuador,Prov. PichinchaBravo 870 (WS) Mez & Sodiro ex Mez T. stricta Anoplophytum BrazilCultivated, Selby BotanicalGarden (SEL 75-86-68) Solander in Sims T. xiphioides Anoplophytum Argentina,Prov. Salta Palaci 207 (WS) Ker-Gawler

currenceof plicate staminalfilaments. One col- parenchymaand epidermal cells collapse, we lection for each of the seven species (Table 1) speculate that structuralsupport provided by was studied anatomically. those cells declines, causing the filament to The filaments were dissected from anthesis bend. The possibility that plication causing stage flowers and buds of varying age, dehy- compressionalforces are generatedby filament drated in a graded ethanol series to 100%eth- elongation below the region of plication was anol, and stained with eosin. They were then investigated. However, no evidence for this infiltratedwith xylene, then paraplast,and fi- type of plication mechanism was found. nally embedded in paraplast. Parenchymacell collapse is firstvisible three Serial sections were cut at 7 ,umon a rotary to four cell layers from the vascular bundle microtome. The sections were afixed to slides (Fig. 5). The collapsing region expands cen- with Haupt's solution, stained with safranin, trifugally,leaving large intercellular spaces (Fig. then with fast green as a counter stain, and 6). Eventually even the epidermal cells col- mounted in Permount. Photographic docu- lapse, leaving only the vascular bundle, a few mentation was with Kodak Tech Pan 2415 parenchymacells nearthe vascularbundle, and film. the cuticle still intact (Fig. 7). The remaining The filaments for scanning electron micros- intact parenchymaare more intensely stained copy (SEM)were firsthydrated, using a graded with safranin than are preplication paren- ethanol series, and then put through an os- chyma cells. mium tetroxide-thiocarbohydrazide-osmium Collapse of transverse cell walls is the first tetroxide (OTO) series (Brown and Gilmartin, visible sign of cell collapse (Fig. 8). As the 1989). They were dehydratedin a graded eth- transversewalls collapse inward,the radialand anol series, critical-point dried, mounted on tangentialwalls fold along the longitudinalaxis SEM stubs, and sputter-coatedwith gold. in an accordionlikemanner. This results in an The scanningelectron microscopy was done overall decrease in cross-sectionalarea of the on a JEOLJSM-35C at the University of Wy- cells, but there is no initial decrease in cell oming. Photographicdocumentation was with length.It is not until the vascularbundle bends, Polaroid Type 665 film. crushing some of the cells, that there is a decrease in the length of any of the cells. This REsULTs-Plicate filaments were found to be transversecollapse of parenchymacells not only present in species traditionally classified in causes longitudinalwrinkling in the radialand subgenera Tillandsia and Allardtia as well as tangentialcell walls, but also on the surfaceof Anoplophytum(Table 1). There was no ob- the fully plicated filament (Fig. 9). servable variation in the anatomical devel- As the parenchymacells collapse, the vas- opment of the plications within the different cular bundle begins to bend within the inter- subgenera. cellular spaces created by the adjacent col- Plication, or folding, of the filament occurs lapsing cells. The initial collapse of the cells during or soon after anthesis. Plication for- causestransverse wrinkles in the epidermis(Fig. mation appears to be a direct result of cell 10), but the bends in the vascular bundle are collapse within the filament (Fig. 1-4). As the not internallyvisible until laterin the plication 1480 AMERICANJOURNAL OF BOTANY [Vol. 76

~~~~

Fig.1-4. Bar = 1 mm. 1. Cross section of T sceptrWormisbefore plication 2 Cross section of T butzizafter plication.3. Longitudinals ection of T. sceptriformisbefore plication. 4. Longitudinalsectiono_!f T . da

pcain

1. nS~~~~~~~~1 October1989] EVANSAND BROWN STAMINALFILAMENTS IN TILLANDSIA 1481

' A:~~~~~

Fig. 5-. Ba =i 1 m elclas einn 3t ellyr rmvsuarbnl nT uz.6 ag

T butzii. ( arrowsini 5Ce llasn trlansverseginnwalls).aer rmvaclr ude nT uti.6. ig 1482 AM CAN JOURNALOF BOTANY [Vol. 76

~~~~. .-S

AC4 A

Fig. 9-12. Bar = 1 mm. 9. Plicated T. aequatorialisfilament with longtudal wrinkleson surface.10. Earlyplication in T. aequatorialis.Plications first become visible as transversewrinkles in the epidermis.11. PlicatedT. strictafilament. Plicatedarea is 3 to 4 mm below anther. 12. Plicated T. gardnerifilament. Phicated area is 1 to 2 mm below anther. process. As more cells collapse, the transverse same region of the filament. In T. stricta the wrinklesgive way to longitudinalwrinkles, and plicated area is 3 to 4 mm below the anther, the vascular bundle begins to bend, until pli- while in T. gardneriplication occurs 1 to 2 mm cation is complete. The plication events de- below the anther (Fig. 11, 12). scribed here appear to be the same for each of those taxa listed in Table 1. DISCUSSION-There are striking external Although the plicated area of the filamentis similaritiesbetween plicate staminal filaments shortened due to the bending of the vascular and contractileroots in general.Thus we com- bundle, the filament as a whole continues to paredcontractile root growthmechanisms with elongate (Table 2). After plication, elongation the suspected contractile process in plicate continues in the filament below the plicated staminal filaments. Several mechanisms have area. The length of the filament above the pli- been proposed for contractile roots. Wilson cated region, if any, as well as the plicated area and Honey (1966), for example, propose that itself(once plicationis complete)remains fairly root contraction in Hyacinthus orientalis is constant (Table 2). Notice that for different caused by the radial expansion and longitu- species the plicated area is not always on the dinal contraction of inner parenchyma cells. October 1989] EVANS AND BROWN-STAMINAL FILAMENTS IN TILLANDSIA 1483

Ruzin (1979) describes root contraction in TABLE 2. Effect of plication on filament length Freesia (Iridaceae)to be the result of radially Base of plication expanding outer parenchyma cells, with no Total filament to base of longitudinal contraction taking place. Still Species length(mm) filament(mm) another mechanism of root contraction was T. aequatorialis proposed by Thoday (1926, 1931) for Oxalis Preplication 13.5 7.5 incarnata.He states that contractionis caused Postplication 36.0 30.0 by the collapse of transverse zones of cells. T. sceptriformis According to this model, transverse zones of Preplication 5.5 3.0 cells collapse as a result of a water deficit. As Postplication 10.0 6.0 water is lost through the transpirationstream T. stricta or absorbedby the bulb, certainlayers of paren- Preplication 8.0 4.0 chyma cells collapse. The unequaldistribution Postplication 13.0 8.0 of turgid and collapsed layers causes the stele to bend. Smith (1930) suggesteda similarmod- el for the contractionof Brodiaea lactea roots. In that case, transverselayers of cells collapse, by elongation in the proximal, nonplicate por- but prior to collapse all the parenchymacells tion of the filament (Table 2). expandradially and shortenlongitudinally. The Other observations that may serve to help actual cause of cell collapse was not deter- elucidate the role of plication in pollination mined. include 1) plicate regions are notably elastic. One requirement of any proposed mecha- Tension on the stamen with forceps will draw nism for plication is that it accountsfor enough out the plicate region, and upon release the force to cause the vascular bundle to bend. anther will "snap" back to its previous posi- Thoday (1931) observed that the bending of tion, the plications being restored. 2) Collec- the vascular cylinder in contractile roots will tively, the pleatlike folding of the filaments not occur if the parenchymacells directly ad- may form a pluglike barrierwithin the floral jacent to it remain turgid. Only if these cells tube. This may retard evaporation of nectar. collapsewill the stele be able to bend. In plicate The central most important difference be- filaments, however, the vascular bundle starts tween Tillandsia subgeneraAllardtia and An- bending even though most of the cells next to oplophytum(sensu Smith and Downs, 1977) it remainturgid. We do not know whethercom- is the length of the stamen relative to petal pressional forces within the filament vascular claw length. In subg. Allardtia the stamen ex- bundle or loss of ground tissue support, or a ceeds its respective petal claw and is more or combination of the two are ultimately respon- less equal to the total petal (claw + blade) sible for plication. length. In subg. Anoplophytumthe stamen is With respect to the plicated region on the shorterrelative to its petal, being more or less filament,plication is a contractileprocess. We equal to the petal claw in length. Of the 46 arenot certainifthe staminalfilament plication species in the subgenus(Table 3), 27 areknown described here is unique. To our knowledge, to have some degreeof filamentplication. Three however, contractile processes that are anal- species (T. ixioides, T. meridionalis, T. xiph- ogous to those of contractile roots have not ioides) are describedas having non-plicate fil- been previously known in floral structure. aments, though Luther (personal communi- The functional role of plicate staminal fila- cation) has occasionally observed plicate ments in pollination biology is not clear. Prior filamentsin T. meridionalis.Descriptive mor- to this investigation, it was hypothesized phology of staminal filamentsis lackingfor the (Brown,unpublished data) that plicationserved remaining 11 species of subg. Anoplophytum as a mechanismto drawdehiscent anthers back (Table 3). to the level of the stigma as a self pollination We were surprisedto discover that over half mechanism. The manifestationof filamentpli- of the species with plicate filaments examined cation at mid- to late-anthesis is consistent in this study (Table 1) are not members of with this initial hypothesis. However, obser- subgenus Anoplophytum. This dramatically vations and measurements of the positional underscores the fact that most bromeliad relationship between the anthers and stigma speciesare woefully lacking in descriptivefloral throughout anthesis do not support this idea morphology(see Brown and Gilmartin, 1984, (i.e., pre- and postanthesis anthers and stigma for more discussion of this issue). For example, maintain more or less fixed positions relative there is no descriptive account for the androe- to one another).Any decreasein filamentlength cium in T. aequatorialis;T. butziiand T. caput- due to plication is more than compensated for medusa are simply described as having an- 1484 AMERICAN JOURNAL OF BOTANY [Vol. 76

TABLE 3. Filament plication in Tillandsia subgenusAn- droecia of exserted stamens while T. sceptri- oplophytuma formis has stamens included. Since Smith (1935) assigned plicate staminal filaments as Taxon Filament condition diagnostic for subgenus Anoplophytum, this T. aeranthos (Loiseleur) plicate characterstate appearsto have been associated Lyman B. Smith only with taxa from Anoplophytum.Our find- T. albertiana F. Vervoorst unknown ing of plicate filaments in species traditionally T. araujei Mez in Martius plicate T. arequitae (Andre) Andre unknown placedin other Tillandsiasubgenera erodes the T. argentina C. H. Wright plicate diagnostic value of this characterand reflects T. bergeri Mez faintly plicate the more profoundproblems of subgenericcir- hT. bermejoensis Hromadnik straight cumscriptionswithin Tillandsia. T. brachvphylla Baker unknown Developmentally, filament plication is bT. buchlohii Rauh plicate viewed as a terminalcharacter (L0vtrup, 1978). T. caulescens Brongn. ex Baker plicate bT. chiltensis Rauh straight In light of hypotheseselaborated by Takhtajan T. diaguitensis Castell. unknown (1954, 1972) and more recently presented by T. didisticha (E. Morren) Baker plicate Tucker(1984), terminalfloral characters would T. esseriana Rauh & Lyman B. unknown most likely have diagnostic utility at the spe- Smith cific or intraspecificcategories of the taxonom- T. friesii Mez unknown ic hierarchy.This interpretationwould T. gardneri Lindley plicate again T. geminiflora Brongn. in plicate suggestan overemphasis on filament plication Duperrey as diagnostic at the subgenericlevel. T. geissei R. A. Philippi unknown Filament plication appearsto be a complex T. globosa Wawra slightly if at growth/degenerativeprocess. However, there all plicate are no direct data to address the genetic foun- bT. grazielae Ruby Braga & Sucre straight dation for the character.New T. heteromorpha Mez unknown observations of bT. horstii Rauh unknown variability in the occurrence of plicate fila- T. ixioides Griseb. straight ments suggestsa rathersimple genetic basis for T. jucunda Castell. plicate the character.Read (1988) makes a brief report T. linearis Vell. Conc. unknown on morphologicalvariation in T. bergeri,and T. lorentziana Griseb. plicate mentionsthat plications were not foundin some T. macbrideana Lyman B. Smith straight flowers,while indications of T. meridionalis Baker plicate filamentplication T. montana Reitz linear, were observed in others. Smith and Downs contorted (1977) describe T. bergerias having"filaments near apex faintlyplicate." In T. xiphioidesthe occurrence bT. muhrae W. Weber unknown of plicate filaments is variable. Smith and T. nana Baker unknown Downs (1977) describe T. xiphioidesas having T. neglecta E. Pereira straight "filamentsfiliform, straight" (Table T. nuptialis Ruby Braga & Sucre plicate 3). We find T. pohliana Mez in Martius plicate full manifestationof plication in the Palaci 207 bT. pseudomacbrideanaRauh plicate collection (Table 1) as well as in commercially T. reclinata Pereira & Martinelli unknown available cultivated material. FT.rosea Lindley plicate Muddling the interpretation of the plicate T. seideliana E. Pereira plicate filamentas a diagnosticcharacter are data from T. sprengeliana Klotzsch ex Mez unknown in Martius a recent survey of bromeliad stigma mor- T. stricta Sol. in Sims plicate phology that reveals a stigma type that cor- T. stricta Sol. var. albifolia plicate relates strongly with the established Anoplo- T. sucrei E. Pereira plicate phytum diagnostic characters. Brown and F. tenuifolia L. plicate Gilmartin (1989) have reportedon the stigma bT. toropiensisRauh plicate morphologytype for 13 species from subg.An- T. vernicosa Baker plicate and T. xiphioides Ker-Gawler filiform, oplophytum, have found all but two of the straight species (i.e., T. linearis, T. xiphioides) to have the simple erect stigma type (see Brown and a Sensu Smith and Downs (1977) and taxa subsequently Gilmartin, for definitions of described. 1984, stigma type b Author did not explicitly place into subgenus Anoplo- categories).The T. linearis collection and one phytum in original citation. of the two T. xiphioides collections examined had conduplicate-spiralstigmas. The othercol- lection of T. xiphioides had the simple-erect stigma form. Mez (1934) also illustrates the simple-erectform for T. xiphioides. Again, T. October 1989] EVANS AND BROWN-STAMINAL FILAMENTS IN TILLANDSIA 1485 xiphioides with plicate and nonplicate fila- der Bromeliaceennach ihrem Bliithenstande.Flora ments, stamens either included or exserted, 37: 346-349. style exserted,and broad petal blades (all traits BROWN,G. K., AND A. J. GILMARTIN.1984. Stigmastruc- ture and variation in Bromeliaceae-neglected taxo- of other subgenera)appears to be problematic nomic characters.Brittonia 36: 364-374. in subg. Anoplophytum (sensu Smith and , AND . 1989. Stigma types in Bromeli- Downs, 1977). aceae-a systematicsurvey. Syst. Bot. 14: 110-132. Alone, the simple-erect stigma type is not GILMARTIN,A. J. 1972. The Bromeliaceaeof Ecuador. diagnostic for subg. Anoplophytum,since this Phaner. Monogr. 4: 1-255. J. Cramer, Lehre, Ger- stigma type is found in species from other Til- many. GRISEBACH,A. H. R. 1864. Bromeliaceae.Flora of the landsia subgenera. The same is also true for British West Indian Islands, 590-599. Lovell Reeve, plicate staminal filaments. The correlation of London. 1) included stamens equal to the petal claw in LovTRuP, S. 1978. On von Baerianand Haeckelianre- length;2) plicate filaments;and 3) simple-erect capitulation.Syst. Zool. 27: 348-352. stigmas may provide a base for continued rec- MEz,C. 1934. Bromeliaceae.In A. Engler,Das Pflanzen- ognitionof subg.A noplophytum. However, with reick 4 (32): 1-667. READ,R. W. 1988. Tillandsiabergeri,amany-splendored the discovery of plicate filaments in taxa from species. J. BromeliadSoc. 38: 147-149. subg. Allardtia,the potential exists for plicate RuzIN, S. E. 1979. Root contractionin Freesia (Irida- filament/simple-erect stigma correlations ceae). Amer. J. Bot. 66: 522-531. within other parts of Tillandsia.Thus, the dis- SMITH,F. H. 1930. The corm and contractileroots of tinction between subgeneraA noplophytum and Brodiaealactea. Amer. J. Bot. 17: 916-927. Allardtiabecomes more vague. SMITH,L. B. 1935. Studiesin the BromeliaceaeVI. Proc. Amer. Acad. Arts 70: 147-220. The occurrenceof plicate staminal filaments , ANDR. J. DowNs. 1977. Bromeliaceae(Tilland- can no longerbe viewed as a primarydiagnostic sioideae). Flora Neotropica Monogr. 14: 663-1492. characterfor Tillandsia subgenusAnoplophy- TAKHTAJAN, A. 1954. Quelques problemes de la mor- tum as it is currentlyrecognized, and mono- phologie evolutive des angiospermes.Essais de bo- phylesis of the entire subgenusseems unlikely. tanique. Vopr. Bot. 2: 763-793. Ongoing studies of tillandsioid floral mor- 1972. Patternsof ontogeneticalterations in the evolutionof higherplants. Phytomorphology 22: 164- phology and its variation are beginning to fill 171. some of the conspicuous morphological data THODAY,D. 1926. The contractileroots of Oxalis in- voids. As this data base grows, a taxonomy carnata. Ann. Bot. (London) 40: 571-583. that better reflectsphylogeny becomes more of 1931. Mechanism of root contractionin Bro- a reality. diaea lactea. P1.Physiol. (Lancaster)6: 721-725. TUCKER, S. C. 1984. Origin of symmetry in flowers. In R. White and W. Dickinson [eds.], Contemporary LITERATURE CITED problemsin plantanatomy, 351-395. AcademicPress, New York. BAKER, J. G. 1887. Synopsis of Tillandsieae. J. Bot. 25: WILSON,K., ANDJ. N. HONEY. 1966. Root contraction 212. in Hyacinthusorientalis. Ann. Bot. (London)n.s. 30: BEER,J. G.. 1854. Versuch einer Eintheilung der Familie 47-61.