192 S. Afr. 1. Bol. 1998, 64(3) 192-197

Ovule orientation, curvature and internal structure in the Aloaceae

E.MA Steyn* and G.F. Smith Research Directorate, National Bolanicallnstitute, Private Bag X101 , Pretoria, 0001 Republic of South Africa

Received 8 December 199i; revised 10 March 1998

This scanning electron and light microscope study, conducted on mature ovules of arbitrarily chosen species of the genera Aloe lo, Astro/aba Uitew., Choriolirion A.Berger, Gasteria Duval, Haworlhia Duval and Poelfnitzia Uitew" indicated that the Aloaceae are exclusively hemitropous. This report is, seemingly, the first account of a consistently hemitropous monocotyledonous family. At generic rank, addilional unifying embryological characters were found in the orientation of th e ovules, the organisation of the integuments, the posilion of the remaining nuceU ar tissue and the structure and arrangement of the embryo sac elements.

Keywords: Aloaceae, aril, embryo sac, hemitropy, ovule morphology.

*To whom correspondence should be addressed.

Introduction this taxon to be hemitropous and not anatropous as reported by The Aloaceae, here regarded as comprising the genera Aloe L. Kativu (1996). Steyn and Smith (1997) suggested that the early [including Aloe sect. LOlllolophyl/lIm (Willd.) GD.Rowley], development of th e broad appendage (aril) to Ihe short fun icle Astroloba Uitew., Chortolirion A. Berger, Gasteria Duval, has prevented the ovule from reaching the anatropous position. Hall'orrhia Duval and Poellnitzia Uitew., is an essentially Old Si nce all members of the Aloaceae and Asphodelaceae seem­ World family of rosulate, succulent-leaved and petaloid mono­ ingly have ovules with primordial arils, it was hypothesized that cotyledons. The taxa were formerly included in the Liliaceae and anatropy should be absent in the two families. This hypothesis have since 1985 often been regarded as members of the subfam­ certainly seems to hold true for the Asphodelaceae; only orthot­ ily Alooideae of the Asphodelaceae (Hutchinson 1973; Dahlgren ropy (Asphodeills L. and Asphodeline Rchb.) and hemitropy el al. 1985; Smith & Van Wyk 1991 ). The Aloaceae comprise (Bu/bine, Eremurus, Kniphofia Moench and Trachyandra) have approximately 550 species centred in southern Afri ca with only been reported for the family (Stenar 1928; Di Fulvio & Cave Aloe and, to a lesser degree, Chortolirion and Haworthia repre­ 1964; Steyn & Smith 1997, but see Kalivu 1996 for supposed sented outside the Flora of southern Africa region. Although anatropy in Trachyandra and Jodrellia Baijnath ). As far as the opini ons on intra-familial delimitation and, es pecially, generic genera of the Aloaceae are concerned, a study of embryological and species concepts have often been controversial (see Cron­ literature revealed a dearth of information regarding ovule mor­ quist 1981 ; Brummitt 1992; Smith el al. 1995 and references phology. Data on ovule curvature (i.e. the final pos iti on of the cited therein; Rowley 1996), members of the family form a natu­ component ovular parts with regard to each other) are , for ra l group (Smith & Van Wy k 199 1). Apart from morphological instance, only available for Aloe, which seems to have similarities, particularly their succulent leaf consistency, evi­ hemitropous as well as ortho- or campylotropous ovules (Joshi dence of unitying characters was established in phytochemical, 1937; McNaughton & Robertson 1974; Dahlgren el al. 1985). cytogenetic, anatom ical and embryological studies. These char­ Although the latter two ovule types are not in confli ct with our acters include, amongst others, the following: the presence of aforementioned hypothesis, th e occurrence of such types in Aloe anlhrone-C-glycosides in Ihe leaves and l-methyl-8-hydroxyan­ did not seem feasible, The numerous unirying characters of the thraquinones in the roots (Beaumont el al. 1985; Van Wyk el al. family, as mentioned above, suggest that ovular form should be 1995); a basic diploid karyotype (2n ~ 14) with four pairs of long constant among the genera of the family--embryologicaJ charac­ chromosomes and three much shorter pairs (B randham 1983 ; ters are usually remarkably conservative (Davis 1966; Dahlgren Smith 1991); the widespread occurrence of secondary thi ckening 1991). In addition, a study of embryological literature revealed growth ; vascular bundles containing a parenchymatous and cap­ that ortholropy is generally coupled wilh uni-ovuly and campy­ like inner bundle sheath; septal nectaries and, with the exception lot ropy with curved embryos (Bouman 1984), while our own of Aloe sect. Lomatophyllum. loc ulicidal capsules containing observations have shown that even the small-flowered aloes (e .g. ari ll ate seeds (Dahlgren el al. 1985). A. bOll'iea Schult. & J.H.Schult., and A. minima Baker) have sev­ The presence ofarillate seeds is a character th e Aloaceae share eral ovules per locule and that A loe seeds have straight embryos. with all members of the closely related family. Asphodelaceae During the present study we investigated ovule characters in (s ensu Brummitt 1992; Asphodeloideae senSll Dahlgren et al. representatives of Aloe, HalVorthia and Gasteria as well as in the 1985). In representatives of the Asphodelaceae, e.g. Eremurus lesser known genera, namely Astroloba, Chortolirion and M.B ieb., Blilbine Wolf and Asphodeills L. (Stenar 1928) and in Poel/nitzia. Our findings on ovular form , internal st ructure and Aloe (Joshi 1937; McNaughton & Robertso n 1974), the aril ini­ the orientation of the ovul es in the locules of these taxa are given tiates from the distal part of the funicle during the early develop­ in the present contribution. ment of the ovule primordium. During prefertilizati on stages, while the ovule primordium is curving towards anatropy, the arH Material and Methods grows into a conspicuous, annular structure at the ventral side of Open fl owers were coll ected on the third day after anthesis during the ovule. Further growth of the aril around th e ovul e is post­ the 1996--1 997 flowering season from plants growing in the nurser­ poned until fertilization has occurred (Stenar 1928; Steyn & ies at the Pretoria National Botanical Garden. These plants included Smith 1997). representati ves of Aloe bowiea, Asrroloba deltoidea (Hook. f.) Du ring a recent stud y of ovule structure in Trachyandra Kunth Uitew., Chortolirion angolense (Baker) A. Berger, Gasteria carinata (Asphodelaceae), Steyn and Smith (1997) found th e ovules of (M ill .) Duval, Haworthia Iimifolia Marloth and Poe/lnitzia S. Afr. 1. Bot. 1998, 64(3) 193

Figure l A-l Scanning electron micrographs depicting ovule orientation and morphology in representatives of the Aloaceae. Ovules are shown in dorsal view (A-D) and in lateral view (E-T) of a locule, after the dorsal wall and septa of the lacu le had been removed: A. Aloe bowiea. Note hypotropy in two lo\vest (most proximal) ovules B. Haworthia limifolia. C. Astra/aba deltoldea. D & E. Aloe suzamtae. F. Pue/lnil::ia rllbriflora. G. Chortolirion ango/ense. H & I. Gasleria carinata. Note placental papillae (black arrowheads) in E, G and Hand pollen tubes (while arrows) in l Abbreviations: G, aril ; s, position of septum before removal. Bar 100 ~m. A-G, 50 H & I. rubriJlora (80Ius) Uitew. Flowers of Aloe africana Mill. were takcn After clearing in a 50% (v/v) solution of household bleach (Jik), the from a herbarium specimen (FOllrcade -1 011, kept in the National ovules were thoroughly rinsed in distilled water and mounted in a Herbarium at Pretoria). Open flowers of Aloe suzannae Decary were 40% aqueous solution of calcium chloride (Keating 1996). Ovaries obtained from herbarium material fixed and stored in Carnay since intended for eventual sagittal sectioning of ovules were dehydrated, March 1994, \vhen a plant, raised from seed collected in Madagascar infiltrated and imbedded in glycol methacrylate (GMA) according to during the late 1960's, came into flower in a private garden in Preto­ the methods of Feder and O'Brien (1968). By following the protocol ria. Ovaries were di ssected from the flowers and, with the exception of Steyn and Smith (1997), median sagittal sections, 2 j.un in thick­ of th ose belonging to Aloe su=amwe, fixed overnight in a phosphate ness, were obtained of the ovules of all species, except A. Sllzannae buftl!red solution (pH 7.4) containing 2.5% glutaraldehyde and 4% and A. africana. In such sections the zygomorphic ovules could be formaldehyde. The dry ovaries of Aloe africana were rehydrated in viewed in their single plane of symmetry and thus be compared in haili ng water before fixation. structure and degree of curvatu re. Selected sections were stained A scanning electron microscope (SEM) study was conducted on with the periodic acid/Schiff reaction (PAS) and 0.5% (w/v) toluid­ the ovules of all species except A. africana. Fixed ovaries were ine blue (Feder & O'Brien 1968). dehydrated in a graded acetone series and critical point dried accord­ ing to conventional procedures. At this stage the dorsal ovary wall Results and Discussion and sepIa between the locu les were removed. The ovary segments were then sputtered with gold and viewed in a lSI SX 25 SEM. Several attempts to dehydrate the ovules in the ovaries of Aloe The SEM study of A. suzannae ovules was supplemented by a africana to a stage suitable for serial sectioning or SEM studies, light microscope (LM) investigation of partially cleared ovules. were completely unsuccessful. Photographic evidence of ovule 194 S.Afr.J.80t.1998.64(3)

Figure 2A-F Ovul es of members of the Aloaceae depicted in median sagittal view to illustrate ovule curvature and internal structure. Note in aJl micrographs, the direction of the longitudinal axis of the short funicule (long black arrow): A. Aloe bowjea. B. Chortolirion angolense. C. Haworthia linll/olia. D. Poeiinilzia rubriflora. E. Gasteria carinata. F. Astrolaba deltoideo. Note in F the three antipodals directly below the central cell nucleus. Abbreviations: a, aril; c, central cell nucleus; e, egg cell and II, hypostase. Bar 50 }1111. s. Mr. J. Bot. 1998, 64(3) 195

structure in this species CQ uld , therefore, not be obtained during the present investigation.

Orientation of ovules in the locules The ovaries of all species investigated were trilocular and multi­ ovu lar with ovules alternately arran ged in two longitudinal rows on axi le placentae. The septa dividi ng the locules ran longitudi­ nally downwards from the apex to the base of the ovary. When the dorsal wa ll and the septa between the loc ul es were removed, the ovules could clearly be seen in dorsal view. They varied in numbers per locule from six or seven in Chor/oUrion angolense, Gasleria cOI'inala, Haworlhia !imifolia and A IDe bowiea, to 20 or 22 in Poellnil=ia rubriflora and 24-30 in Aloe Sllzannae and Astra/aba de//oidea. In Aloe africana the number of ovules per locu le could not be exactly determined, but was more than 12. SEM studies showed that all ovules in the same longitudinal row, except the two most proximal, had thei r micropylar-chala­ zal axes orientated in such a way that the micro pyles were turned outwards, fac ing the same and nearest septum (Figure 1 A-C). The most proximal ovules were usually hypotropous, i.e. their Figure 4 Oblique section of the hem itropous ovule of Aloe bOlV­ micropyles faced the base of the locule (Figure IA). With the iea as seen in a transverse section oflhe locule. Abbreviations: a. notable exception of A. suzanl1ae, the distal ovules in the Iocu les aril ; em, embryo sac; pn, p~ rsi stent chalazal nucellus. Bar 50 ~ m. of all species were placed obliquely on the placentae, i.e. the inclination angle of the micropylar-chalazal axes of the ovules papillae in the narrow space between the ovule bases and the with regard to the longitudinal axis of the placenta was approxi­ base of the septum (F igu re tI). mate ly 45 degrees (Figures I A-C). The ovules of A. suzannae In a lateral view ofa loc ul e with septa removed. the arils of the were nearly horizontally orientated on the placenta and some of ovules were readi ly visible (Figures 1E - H). The funicles were them were pushed out of alignment (Figure I D). Another irregu­ short and broadened by the development of the ari ls wh ich larity, namely two ovules enclosed in the same outer integum ent, formed podium-like, boat-shaped structures and lifted the ovules was also encountered although less frequently (Figu res IE). proper above the level of the placental papillae (Figures IE, G These aberrations probably occurred in adjustment to th e co n­ and H). The ovules of all spec ie s studi ed were clearly bitegmic fined space within the loc ul e. In the multi-ovular loc ules of with th e inner integument protruding from the rim of th e outer Astroloba delloidea and P. rubrijIorQ similar aberrations were integument. Th e shape of the ovu les proper varied. Th ese struc­ not seen. tu res were ovoid in H. limifolia (Figure I B), Aloe suzan/we (Fig­ The orientation of the ovu les in the locules brought the mi cro­ ure I D and F), C. angolense (Figure 1G) and C. CGrinllta (F igure pyles very cl ose to groups of placental papillae occurring at the tH), but in A. bOlViea (Figure IA), Astraloba deltoidea (Figure bases of the ovu les, adjacent to the nearest septum (Figures ! E 1C) and P. rubrijIora (Figure I F) the ovo id ovules were con­ and G). Poll en tubes growing longitudinally along the septa and stricted in the chalazal and micropylar reg ions. over the papillae would , therefore, be in close proxim ity to the micropyles. rn pollinated ovu les ofG. carinata bundles of pollen Ovule curvature tubes were indeed found obscuring the groups of placental Median sagittal secti ons of ov ules were obtained of all species, except Aloe africana and A. su:annae. In th ese secti ons of ova­ ries, fix ed at the same deve lopm ental stage (Le. three days after anthesis), the ovules revealed no obvious intergeneric differ­ ences, in fact the ovules were so similar in structure that th ey could hardly be distinguished from one another (Figures 2A-F). As far as ovule cu rvature (i.e. the pos it ion of the co mponem ovu­ lar parts with regard to one another) was concerned, we fo und that th e micropylar-chalazal (l ongitudinal) axes of th e ovules were approximately perpendicul ar to the longitudinal axes of th e short funicles. None of th e ovules representing th e six genera of the Aloaceae could, therefore, be regarded as orthrotropous or anatropous. Furthermore, th e mi cropylar embryo sac elements (egg cell an d two synergi ds) lay in the same vertical plane as the chalazally placed centra l ce ll nucleus and three degenerating antipodals. This plane coi ncided with the longitudinal axis of th e ovule. The embryo sacs thus displayed 11 0 degree of curvature, which ruled out campylot ropy. All ovu les could therefore only be regarded as hem itropous. Fixing and prolonged storage in Carnoy was so detrimental to the ovul es of A. suzannae that properly stained, median sagittal Figure 3 Partly cleared hemitropous ovu le of Aloe suzannae. sections were unobtainable. The ovules were, therefore, partly Note the di rection of the xylem elemen ts (long black arrow), indi­ cleared to ascertain their degree of curvature. Although the cating the longitudinal axis of the funicle. Abbrev iations: G, aril, h. embryo sac elements were not seen, the horseshoe-shaped, hypostase, m, position of mi cropy le. Bar lOa ~m. darkly stained hypostase ti ssue in the chalaza! nucellus was read- 196 s. I\lr. J. 1301. 1998.64(3)

Figure SA-C Micrographs showing the lypical structure ofrhe embryo sac and nuce llus in a matur!! ovule of the Aloaccac: A. Egg cdl (e) and adjacent synergid. B. Synergids. C. Embryo sac covered by nucellus epidermis (n) and containing a conspicuous central eel[ nucleus (c) directly above three ephemeral antipodals (white arrow). Note the darkly stained hypostase tissue (11) in the pc.::rsistent chalaza I nucdlus (pl/). Bar 25 Iotm. ily visible (Figure 3) and indicated the position of the central cell depicted by McNaughton and Robertson (1974: 79, Figure 7a-c) nucleus and ephemeral antipodals (compare Figures 3 and 2F). and it can erroneously be concluded that the embryo sac is The chalazal elll bryo sac elements were, therefore, in line with curved. This could have been the reason for regarding A. alri­ the micropyle canal. Since this plane represented the longitudi­ cana as campylotropous. nal axis of the ovule and was perpendicular to the longitudinal axis of the funicle, A. slizannae was clearly hernitropous like the Inlernal structure of ovules other spec ies. Mature ovules of all representatives of the A loaceae treated in In monocotyledons anatropy is of general occurrence, whil e this investigation were bitegmic. On the dorsa! side of tile ovule hemitropy is found scattered among the families (Dahlgren el al. the outer integument consisted of three cell layers. On the ventral 1985 ). According to Davis (1966) thirteen angiosperm families, side this integument was squashed between the inn er integument all dicotyledonous, are exclusively hemitropous. The consistent and the developing aril and often not clearly delimited from the presence of hem it ropy in a monocotyledonous fam ily has, to our latter structure. The inner integument alone formed the micro­ knowledge, not yet been reported and this character can, there­ pyle canal. It was two-layered at the base, but consisted of more fore, be regarded as an important marker at the famity level. than two layers in the micropylar region and usually extended Tn the light of our observations, the report that the ovules ofA. beyond the rim of the outer integument (Figures 2A- F). africana are campylotropous (McNaughton & Robertson 1974) In a comparative embryological study of several genera of the has to be questioned. Were the conclusions of these authors per­ Liliaceae sensu lalo, Schnarf and Wunderlich (1939) investi­ haps based on improperly sectioned ovules? It was previously gated inter alia, the mature embryo sac of several species of Aloe pointed out that the degree of curvature experienced by an ovule and one species of Gasteria. These authors reported that the during development can only be assessed in median sagittal sec­ structure of the female gametophytes is completely similar in tions of ovules (Steyn & Smith 1997). In botanical literature such these two taxa and that the cells of the egg apparatus do not con­ sections are often referred to as longitudinal and they are not form to the usual angiosperm type: The chalaza! part of the con­ easily obtainable of the obliquely placed ovules of the aloes and spicuous synergids 'contains dense cytoplasm instead of the their allies (Stenar 1928; Steyn & Smith 1997). The photo­ usually large vacuole, the nucleus is seemingly ephemeral and graphic evidence presented by McNaughton and Robertson the striated filiform apparatus (FA) does not stain readily (1974) consists of two figures (F igures 1 and 7). The first is (Schnarf & Wunderlich 1939). In addition, the egg cell does not described as 'A longitudinal section through the mature ovule of extend below the synergids and its nucleus lies in the centre of A africana .... .'. This illustration resembles our Figure 4, depict­ the cell, surrounded by weakly stained cytoplasm. These authors ing the hemitropous ovule of A. bOll'iea in an oblique section. also mentioned and depicted the conspicuously large central cell Such sections are obtained when the ovary is sectioned trans­ nucleus in the chalazal embryo sac region, directly above the versely and are unusable for assessing ovule curvature. When ephemeral antipodals and the hypostase in the persistent chalazal sectioned in this plane, the micropylar and chalazal embryo sac l1ucellus. Subsequently th e aforementioned characters of the elements cannot be seen in the same section, but in succession as embryo sac elements were confirmed in extensive ultrastructural S. Me. J. Bot. 1998,64(3) 197

and histochemical studies on the pistil of Gasteria l'errucosa DAHLGREN , R.M.T ., CLIFFORD. I LT. & YEO, P.F. 1985. The lami­ (Willemse & Kapi l 1981 ; Willemse 1996 and references cited lies of Ihe monocolyledons: stnlcture. evolution. and taxoll0my. pp. therein). In these studies on G. verrucosa it was consistently 179- IB2. Springer Vl.!rlag. BI.!J'lin. found that the large FA reacts positively to PAS, indicating the DAHLGREN, G. 1991. Steps towards a natura! system o f the dicotyle­ presence of polysaccharides, and seems capable of secreting dons: embryological Chaf'clClers. Aliso 13: 107-165. large amounts of micropylar exudate, a possible attractant for DAVIS, G.L. 1966. Systematic embryology of the angiospenns. p. 15. approaching pollen tubes. John Wiley & Sons. London . In the present study we could establish that the aforemen­ Dr FULVIO, 1.E. & CAVE, M .S. l1J64 . Emhl)'ology of lJIalldjiJrdill tioned structural characters of the embryo sac and nucellus were lIobi/i.~ Smith (Liliaccae). with special reference to it..; taxonomic posi­ exhibited by representatives of all six genera of the Aloaceae. At lion. Phylomorphology 14: 487--499. li ght microscopic level, this typical structure of an aloaceous FEDER, N. & O·BRIEN. T.P. 196B. Plant microtechnique: some princi­ embryo sac with adjacent nucellar tissues is shown in Figure 5 ples and new methods. Am. 1. BOI. 55: 123-142 A-C, representing fI. limifolia. HUTCHINSON, 1. 1973. The families of flowering plaIlts. :\ITangcd according to a new system based on their probable phylogeny. 3rd Conclusions cdn., pp. 732-757. Clarendon Press. Oxlord. JOSHI, A.C. 1937. Megasporogenesis in Aloe vera Linn. J llldi(/n but This study, conducted on arbitrarily chosen species of Astra/aba, Soc. 15: 297-300. A/oe, Chorlolirioll, lfalVorl/Jia, Ga.Heria and Poellnit::ia, KATIVU. S. 1996. A study on microsporogenesis and ovule morphol­ revealed an array of ovular characters that was constant for the ogy in tropical African Anthericaceae and Asphodclaceae. In: The family. Most importantly, all genera were found to be hemitro­ biodiversity of African plants. ed. LJ.G. van der Maesen. pp. 447- pous. This familial character has considerable diagnostic value, 480. K1uwer Academic Publishers. Dortrecht. since an exclusively hemitropous monocotyledonous fami ly has KEATING, R.C. 1996. Anther investigations: a review of methods. In: not yet been reported in the literature. Our hypothesis that the The anther: [onn, function and phylogeny. eds W.G. D'Arc)' & R.C. early development of the aril (which occurred on the ventral si de Keating, Ch. 12. pp. 255- 271. of all ovules) prevents anatropy (Steyn & Smith 1997), certainly McNAUGHTON, J.E. & ROBERTSON. B.L. 1974. Megasporogencsis seems to be valid. Additional unifying generic characters were and megagametogcnesis in Aloe oJricana Mill. J S. AJr. BOl. 40' 75- detected in the orientation of the ovules in the locules, the com­ 79. position of the integuments and the formation of the micropylar ROWLEY. G. D. 1996. The bcrried aloes: Aloe section Loma/uphyllum. canal, the remains of the nucellar tissue and the structure of the Excelsa 17: 59--62. embryo sac elements. SCHNARF, K. & WUNDERLICI-L R. 1939. Zur vergleiehendcn Embryologie der Liliaceae- Asphodeloidcac. Flora 133: 297-327. Acknowledgements SMITH, G.F. 1991. The chromosomes of Chorwlirion and Poellnil::ia We are indebted to the Research Herbarium Support Services of (Asphodelaceae: Alooideae). BOlhalia 21 : 171-175. the National Botanical rnstitute for providing the infrastructure SMITH, G.D. & VAN WYK, B.E. 1991. Generic relationships in thc to execute this study and we especially wish to thank Or Sarie Alooideae (Asphodelaceac). Tawl1 40: 557- 581. Perold and Mrs Adela Romanowski for excellent technical sup­ SM ITH G.F., VAN WYK , B.E., MoSSMER. M. & VIUOEN, A. 1995. port. The taxonomy of A/oinella. Cuiflauminia and Lemeea «Aloaccae). Taxoll44: 513- 5I7. References STENAR, II. 1928. Zur Embryologic der A~phodelil1t!-Gruppc. Svensk IlEAUMONT. J., CUTLER, D.F., REYNOLDS, T. & VAUGHAN. J.G. bOi. Tidskr. 22: 145-159. 1985. The secretory tissues of aloes and their allies. Isr. J. BOI. 34: STEYN, E.M.A. & SMITH, G.F. 1997. Ovule structure in TraciJyandra 265-282. salrii (Asphodelaccae). S. Air. 1. Bot. 63: 223-226. BOUMAN. F. 1984. The ovule. In: Embryology of angiospcm1s, ed. VAN WYK, B-E., VILJOEN, A.M. & DAGNE, E. 1995. Chomota,o­ D.M. Johri. Ch. 3, pp . 123-157. Springer-Verlag. Berlin. nomic studies in African Aloaceae and AsphodcJaceae. Proc. 6th BRANDl-TAM, P.E. 1983. Evolution in a slable chromosome system. In: NAPRECA Symp. naL Prod., Kampala, Uganda, pp. 15-IB. Kew Chromosome ConferVascular plant families and genera, pp. 696- 9: 348-352. 697. Royal Botanic Gardens, "cw. WTLLEMSE, M.T. M. & KAPIL R.N. 1981. Antipodals of Gasleria CRONQUIST. A 198 r. An integrated system of classification of flow­ verrucosa (Liliaceae): an ultrastructural study. Acta bOI. Neer!. 30: ering plants. pp. 1215- 1217. Columbia University Press. New York. 25- 32.