Palynological characters and their phylogenetic signal in Rubiaceae.
I. Abstract II. Introduction III. Material and Methods IV. Results A. Pollen Morphological Characteristics of Rubiaceae 1. Dispersal Unit 2. Pollen Size 3. Pollen Shape 4. Apertures a. Number b. Position c. Type d. Protruding Onci and Pollen Buds 5. Sexine 6. Nexine Ornamentation 7. Stratification of the Pollen Wall B. Heterostyly and Pollen Dimorphism C. Orbicule Characteristics of Rubiaceae D. Fossil Pollen Record for Rubiaceae V. Discussion A. Systematic Significance of Pollen Morphology in Rubiaceae 1. Subfamily Level 2. Tribal Level 3. Generic and Infragenerie Level. a. Generic Level b. Subgeneric and Specific Level c. Infraspecific Variation B. Systematic Value of Pollen Data Compared with Other Morphological Characters C. Orbicules and Phylogeny D. Coding Pollen Characters for Cladistic Analyses VI. Future Research VII. Acknowledgments VIII. Literature Cited IX. Appendix 1: Template for Pollen Descriptions X. Appendix 2: List of Genera in the Pollen Database, with Reference to the Literature XI. Appendix 3: List of Genera according to Tribes
The predominantly tropical family Rubiaceae includes approximately 12,000 species in 650 genera (Delprete, 1999a), making it one of the five largest families of flowering plants. The publication of Robbrecht's Tropical Woody Rubiaceae (1988a) boosted systematic interest in the family by providing a synthesis of the available morphological data and a comprehensive, worldwide classification system. Since then, many investigators have concentrated their research efforts on Rubiaceae, and a considerable amount of new morphological and molecular data have become available. Mainly due to the advance of molecular systematics, unexpected relationships were often suggested, and the infrafamily classification has changed dramatically to translate the new insights. One of the four subfamilies recognized by Robbrecht (1988a, 1993)--i.e., Antirheoideae--collapsed, and many tribes were redefined (e.g., Isertieae, Naucleeae, Spermacoceae). This process of refining our knowledge of the evolutionary history of Rubiaceae is constantly assisted by morphological observations and continues to date, resulting in shifting genera and tribes, in redelimitation of genera, and in description of new taxa.
Pollen morphological data have proved to be particularly informative in elucidating evolutionary relationships within Rubiaceae. Erdtman's (1952, reprinted in 1971) comparative pollen study of 230 species in 120 genera of Rubiaceae reflected some of the variation. Bremekamp (1952), however, was the pioneer in using pollen characters at a large scale to infer the phylogeny of the family. This is not surprising, because he was also a student of the eurypalynous family Acantaceae, in which he used pollen features to clarify many difficult issues, such as the relationship between Acanthopale C. B. Clarke and Dischistocalyx T. Anders ex Benth., to cite one. Subsequent studies dealing with the systematics of the family often included pollen morphological evidence to support or reject taxonomic decisions (e.g., Bremekamp, 1966; Lewis, 1965a, 1965b; Verdcourt, 1958). Since then, several taxonomists have documented the pollen morphology of their study group, and a few palynological articles have been published at the tribal level, e.g., Morindeae (Johansson, 1987a), Gardenieae (Persson, 1993).
In spite of these efforts, Rubiaceae remained one of the blank spots on the pollen morphological map of flowering plants (Muller, 1981). In the last two decades, however, the input of new observations increased. These new pollen data are scattered over systematic and palynological journals and are often difficult to compare. Moreover, some contributions are not widely accessible, because they are part of unpublished theses (e.g., Es, 1999; Vanthournout, 2002). Because a new synthesis of the pollen morphological knowledge of Rubiaceae was also badly needed, we compiled a database at the Laboratory of Plant Systematics (Catholic University of Leuven), containing key pollen features of Rubiaceae genera with reference to relevant literature. In the present article we portray and discuss the known variation of the main pollen characters. Gaps in our pollen knowledge are localized, and features that deserve more attention are pinpointed. We critically discuss coding issues and measure the power of the phylogenetic signal by evaluating the data on up-to-date evolutionary hypotheses. We also hope that this article will encourage further palynological research in the family and enhance the use of pollen data in phylogenetic analyses.
III. Material and Methods
In order to synthesize the pollen morphology of Rubiaceae, we gathered all of the literature listed in the Seventh Bibliographic Index to the Pollen Morphology of Angiosperms (Tissot & van der Ham, 1994) and added all modern literature. For each genus, we retained the most recent or the most reliable and relevant pollen morphological works. From these, after checking all taxa against the most recent classifications, we distilled the pollen morphological characters at the generic level and stored them in a database with reference to the literature consulted. The pollen database is linked to a systematic database holding basic information such as species number, distribution, and the systematic position of each genus. The database is available online at <http://www.kuleuven.ac.be/bio/sys/pollen>. The quality requirements for the works retained were set as follows: (1) Clear illustrations (preferably SEM and LM photographs) are present; (2) the authors have reliable taxonomic knowledge of the species investigated; (3) pollen descriptions are present or, if not, the main pollen characters can be deduced from the illustrations. Geographically oriented palynological contributions (e.g., pollen floras) or general works on Rubiaceae with scanty notes on pollen morphology were therefore often omitted, but reference to them is made in the database. Genera for which pollen data found in literature do not fulfill the quality requirements set above are present in the database, but they are not used to prepare the pollen description of the tribes to which they belong.
In some cases, how to convert the raw data in an article into pollen features for genera was not evident, especially when the results section was limited to the descriptions of pollen types and no species or generic pollen data were provided. In order to avoid these problems and to facilitate the exchange of pollen data among scientists, we propose a pollen description template for Rubiaceae in Appendix 1.
Table I summarizes the pollen data at the tribal level. Exceptional observations are not taken into account in the table, but they can be found in the pollen database. All genera included in the database with reference to the pollen works used are listed in Appendix 2. The tribal position of the genera is given in Appendix 3.
The phylogenetic value of pollen characters at the tribal level is evaluated on the basis of a summary cladogram reflecting the currently accepted relationships in Rubiaceae. This tree is constructed with MacClade, and the backbone is based on the study by Bremer et al. (1999: rbcL, ndhF). In this study, three subfamilies were recognized: Rubioideae, Cinchonoideae, and Ixoroideae. Later in this article we use Cinchonoideae s.str, and Ixoroideae s.l., because these two subfamilies were respectively reduced and enlarged vis-a-vis Robbrecht's (1988a, 1993) earlier concepts with the same name. Relationships within the three subfamilies as drawn in the summary cladogram are based on the following articles: Andersson and Rova (1999: rps16intron), Bremer and Manen (2000: rbcL, atpB, rbcL, rps16-intron), Piesschaert et al. (2000a: rps 16-intron), and Dessein (2003: rps 16-intron; ITS) for the Rubioideae; Rova (1999: trnL-F, trnL-F spacer, rps16-intron), Razafimandimbison and Bremer (2001:ITS, rbcL, trnT-F), Rova et al. (2002: trnL-F) for the Cinchonoideae s.str.; and Rova (1999: trnL-F, trnL-F spacer, rps16-intron), Bremer and Thulin (1998: rbcL), Andreasen et al. (1999: ITS, rbcL), Andreasen and Bremer (2000: ITS, rbcL), Dessein et al. (2001b: rps16-intron, rbcL), and Rova et al. (2002: trnL-F) for the Ixoroideae s.l.
Terminology follows, unless stated otherwise, the online edition of Punt et al. (1994) at <http://www.bio.uu.nl/~palaeo/glossary/index.htm>. For shape classes (P/E) we adopted the wide definitions of Nilsson and Praglowski (1992); that is, oblate (0.50-0.75), suboblate (0.750.88), spheroidal (0.88-1.14), subprolate (1.14-1.33), and prolate (1.33-2.00).
Our literature review resulted in 486 pollen morphologically documented genera, which represent about 75% of the total number of estimated rubiaceous genera (Appendix 2). For 423 genera (65%), pollen data were reliable enough to be included in this article, the following genera being excluded: Antirhea Comm. ex Juss., Amaracarpus Blume, Aulacocalyx Hook. f., Bobea Gaudich., Bouvardia Salisb., Callipeltis Stev., Calycosia A. Gray, Canephora Juss., Ceuthocarpus Aiello, Chapeliera A. Rich., Coccochondra P. Br., Cosmocalyx Standl., Craterispermum Benth., Cremaspora Benth., Cubanola Aiello, Deppea Cham. & Schltdl. Didymaea Hook. f., Dolichometra K. Schum., Duidania Standl., Feretia Delile, Fernelia Comm. ex Lam., Flagenium Baill., Galiniera Delile, Gallienia Dubard & Dop, Gillespiea A. C. Sin., Heinsenia K. Schum., Heterophyllaea Hook. f., Hydnophytum Jack, Hypobathrum Blume, Hyptianthera Wight & Am., Kraussia Harv., Machaonia Humb. & Bonpl., Morindopsis Hook. f., Nesohedyotis (Hook. f.) Bremek., Oreopolus Schltdl., Paraknoxia Bremek., Pentaloncha Hook. f., Pentanopsis Rendle, Phitopis Hook. f., Phuopsis (Griseb.) Hook. f., Pittoniotis Griseb., Placopoda Ball. f., Pouchetia A. Rich., Psathura Comm. ex Juss., Pseudonesohedyotis Tennant, Rachicallis DC., Readea Gillespie, Rhopalobrachium Schltr. & K. Krause, Robynsia Hutch., Sarcopygme Setchell & Christoph., Schmidtottia Urb., Siemensia Urb., Spiradiclis Blume, Steenisia Bakh. f., Stenostomum C. F.Gaertn., Streblosa Korth., Striolaria Ducke, Tamridaea Thulin & B. Bremer, Temnocalyx Robyns, Thecorchus Bremek., Thogsennia Aiello, Urophyllum Jack ex Wal., Warszewiczia Klotzsch, Zuccarinia Blume. For 94 genera (14%), data on orbicules were found. The pollen documentation for most genera is far from complete, which is illustrated by the fact that only an estimated 15% of all Rubiaceae species have been investigated. Moreover, in some writings pollen descriptions are concise or incomplete. Conflicts between pollen data found in different publications were sporadically noticed and mainly concerned shape in equatorial view and size. Differences in use of pollen terminology also remain an obstacle for a correct interpretation of the observations. We want to stress again the benefits of the general use of the international glossary by Punt et al. (1994).
A. POLLEN MORPHOLOGICAL CHARACTERISTICS OF RUBIACEAE
1. Dispersal Unit
Rubiaceae pollen grains are generally dispersed as monads (Fig. 1), but permanent tetrads (Figs. 2-3) are present in 13, mainly African, genera of Gardenieae (Robbrecht, 1988a) and in the distant genus Gleasonia Standl. in Henriquezieae (Rogers, 1984). Randia L. is the only genus known with both monads and tetrads (Persson, 1993), but the generic boundary of this genus is in flux (Gustafsson & Persson, 2002). The occasional presence of tetrads in Lasianthus W. Jack as reported by Huang (1972) needs confirmation. Massulae, pollen agglutinated in larger bodies of a few hundred grains and dispersed as a unit (Fig. 4), occur in a single genus of Gardenieae, Massularia K. Schum. (Persson, 1993). The observations by Igersheim (1989) of massulae in a few Psydrax Gaertn. and Keetia E. Philips species need confirmation.
[FIGURES 1-4 OMITTED]
The configuration of the permanent tetrads is sometimes variable. Ganguelia Robbr., for instance, shows 30% decussate and 70% tetrahedral tetrads in the same specimen; a single rhomboidal uniplanar tetrad has also been observed in that genus (Robbrecht et al., 1996). Persson (1993) described tetrads of 11 genera from the Gardenieae, but he gave no detailed information about configuration; all tetrads shown by him are tetrahedral.
2. Pollen Size
Pollen size after acetolysis (Figs. 5-7) ranges from 6 [micro]m in Danais cernua Baker (Huysmans, 1998) to 133 [micro]m in Diodella teres Small (Dessein, 2003). The majority of species have pollen grains in the 20-40 [micro]m range. Wodehouse (1935) showed that pollen grains within this size range are ideally adapted to be released from the anthers, transported by moderate winds, and attached to insect vectors and to stigmata. Observation of very large or very small pollen grains may therefore also point to specific pollination syndromes.
[FIGURES 5-7 OMITTED]
Although pollen size is often stable at the species level, we occasionally noticed a considerable intraspecific size variation (e.g., the equatorial diameter (E) varies from 19 to 41 in Spermacoce fabiformis Harwood). Sometimes this variation can be explained in terms of infraspecific taxa (e.g., Dessein et al., 2002a), polyploidy (e.g., Puff, 1986a), growth conditions, flower/anther size (e.g., Puff, 1986a), or heterostyly (see the section on "Heterostyly and Pollen Dimorphism" below). More often, the variation is difficult to interpret. In Spermacoce fabiformis, for example, individuals with larger pollen are often found in drier areas, exactly the opposite of what one would readily expect (Dessein et al., 2005b). Even within one individual, size variation can be considerable. In Hedythyrsus thamnoideus (K. Schum.) Bremek., the polar axis (P) varies from 14 to 30 [micro]m in the same specimen, whereas in all species of the genus Ernodea Sw. two size classes of pollen were observed in one anther (Negron-Ortiz, 1996). Negron-Ortiz figured out that in Ernodea large pollen grains are more abundant than are small pollen grains and that viability is associated with the larger grains. Two size ranges were also found in pollen of Hintonia latiflora (DC.) Bullock, where it seems to be correlated with differences in exine thickness (Ochoterena, 2000); in the latter case nothing is known about the viability of the pollen grains. Whether size dimorphism is a general feature of Rubiaceae pollen is unclear at present. Mathew and Philip (1983) listed 35 taxa in which they found significant size differences, but they did not state whether this variation was observed within one specimen or several specimens of one taxon.
At the family level, there is no correlation between flower size and pollen size. Within certain lineages, however, this correlation has been observed (Dessein et al., 2002a; Rogers, 1984).
The fact that pollen size may be variable within a single species, together with the strong influence of preparation and observation techniques on size and shape of pollen grains (see Reitsma, 1969), makes comparisons of size measurements across different publications tricky. Nevertheless, in our opinion, pollen size can be a marker distinguishing related genera or characterizing some tribes. The Rubieae and Naucleeae s.l., for example, have predominantly very small to small pollen grains (<30 [micro]m), whereas Palicourea Aubl. differs from most other Psychotrieae in having very large pollen grains (65-125 [micro]m). In eurypalynous groups such as the Spermacoceae s.str., however, pollen size is useful only at the species level (Dessein, 2003).
3. Pollen Shape
Pollen shape of hydrated grains in equatorial view (P/E) ranges from oblate (Fig. 8) to prolate (Fig. 10) but is most often spheroidal (Fig. 9). The outline of hydrated pollen grains in polar view (amb) is mostly circular (Fig. 11), often slightly lobed due to sunken colpi (Figs. 14-15). In a few species, such as Dictyandra arborescens Welw. ex Hook. f., a quadrangular shape is typical (Robbrecht, 1984), whereas triangular shapes (Figs. 12-13) are observed in, for example, the genera Tapiphyllum Robyns and Psydrax Gaertn. (Lens et al., 2000).
[FIGURES 8-15 OMITTED]
The shape of the grains depends heavily on harmomegathic effects: Dehydration often leads to more prolate pollen by invagination of the ectocolpi. The systematic value of shape characters is questionable in Rubiaceae, but, for instance, in Spermacoce stenophylla F. Muell. and S. inaperta F. Muell., pollen is characteristically oblate and triangular in polar view (Dessein et al., 2005b).
Triaperturate pollen grains are most common in Rubiaceae and exhibit the plesiomorphic condition in the family (Figs. 12-13). In numerous genera, tetraaperturate and, more rarely, pentaaperturate pollen grains are found in addition to the triaperturate ones. In the more derived herbaceous groups, Spermacoceae s.str, and Rubieae, pluriaperturate grains are the rule (Figs. 14-17). In Spermacoceae s.str, the number of apertures varies between 3 and 30 (Dessein et al., 2003), whereas the maximum number so far observed in Rubieae is 13 (Huysmans et al., 2003). Pluriaperturate pollen grains are also sporadically observed in a few species or genera of other tribes, such as Theligoneae (Theligonum L.), Spermacoceae s.1. (Gomphocalyx Baker, Kohautia Cham. & Schltdl., Neanotis W. H. Lewis, Phylohydrax Puff), and Gardenieae (Rosenbergiodendron Fagerl. and Randia). In the paleotropical specimens of Geophila repens (L.) Johnst., the apertures are as numerous as 150 (Vanthournout, 2002).
[FIGURES 16-17 OMITTED]
In some species of Antirhea (Chaw & Darwin, 1992), Chassalia Comm. ex Juss. (Piesschaert et al., 1999a), Nertera Banks & Sol. ex Gaertn. (Robbrecht, 1982b), Palicourea Aubl. (Robbrecht, 1988a), Psychotria L. (Johansson, 1992), and Rudgea Salisb. (Jung-Mendacolli, 1984; Piesschaert, unpubl.), apertures seem to be absent (Fig. 11). It is unclear whether this pollen is really inaperturate or rather omniaperturate. In the heterostylous species of Hymenocoleus Robbr., brevistylous flowers seem to produce nonaperturate pollen grains, whereas longistylous flowers produce 3-5-porate pollen grains (Robbrecht, 1977). Biaperturate pollen grains are reported in 16 genera, including Aidia Lour. (Puttock, 1992), Atractocarpus Schltr. & K. Krause (Puttock, 1992), Coussarea Aubl. (Jung-Mendacolli, 1984), Myrmecodia Jack (Robbrecht, 1988a), and Schradera Vahl (Puff et al., 1993a). Only for the monospecific genus Leucocodon Gardner might it serve as a diagnostic character (Puff & Buchner, 1998); in all other genera 2-aperturate pollen grains are found together with 3-aperturate ones. Chaw and Darwin (1992) describe pollen of Antirhea as inaperturate or 1-porate; D'hondt et al. (2004) report monoaperturate pollen grains in Blepharidium Standl.; Huang (1972) described pollen of Hedyotis diffusa Willd. (= Oldenlandia diffusa (Willd.) Roxb.) as 1-5 colporate, and Huxley and Jebb (1993) observed 1-3-porate pollen grains in Myrmecodia. These observations of monoaperturate pollen grains in Rubiaceae need confirmation, and their viability should be investigated.
The number of apertures may vary considerably within a genus, which is illustrated for Spermacoce L. (Dessein et al., 2002a), in which a weak positive correlation has been found between pollen size and number of apertures. If the pantoaperturate species are excluded from the test, the correlation is strong. Also, within Danais Comm. ex Vent. there seems to be a correlation between pollen grain size and the number of colpori. Taxa with invariably 3- or 3(-4)-colporate grains tend to have smaller grain sizes, and species with 4(-5)-colpori have larger grains (Buchner & Puff, 1993).
In most Rubiaceae, pollen is angulaperturate--i.e., the apertures are situated at the angles of the outline in polar view (Fig. 12)--with the apertures situated only at the equator (zonoaperturate). Planaperturate pollen--i.e., apertures situated at the sides when seen in polar view (Fig. 13)--is observed in a few species only; for example, Rondeletia odorata Jacq. (El-Ghazaly et al., 2001) and Spermacoce stenophylla (Dessein et al., 2005b). Pantoaperturate grains with the apertures evenly spread over the pollen surface (e.g., in Geophila repens, Spermacoce phyteuma Schweinf. ex Hiern, and S. annua Verdc.) occur sporadically in the family (Fig. 18). Two remarkable, previously unreported, organization patterns were recently observed in Spermacoce (Dessein et al., 2002a). The first type has several short colpori arranged in a looplike pattern (Fig. 16), similar to the line on a tennis ball (S. filifolia (Schumach. & Thonn.) J.-P. Lebrun & Stork, S. octodon J.-P. Lebrun & Stork, and S. tenuissima Hiern). In the second type (Fig. 17), the colpori are arranged in a spiral pattern (S. terminaliflora Good and S. thymoidea (Hiern) Verdc.). To our knowledge, these two types of aperture position have not previously been observed in angiosperms. These types can be considered intermediate forms between zono- and genuine pantoaperturate grains. In Coffea L., spiraperturate and syncolporate grains may occur, probably because of hybridization events (Chinnappa & Warner, 1981, 1982; Stoffelen et al., 1997).
[FIGURE 18 OMITTED]
The aperture type of Rubiaceae pollen is a major character for identification and often useful for inferring phylogeny. Compound apertures are the rule; that is, two or three (non)congruent apertures on top of each other, situated in different layers of the wall. A full set of apertures thus consists of an ectoaperture, which is a thinning or hole in the sexine, a mesoaperture situated in the foot layer, and an endoaperture in the endexine and/or membranous granular layer MGL (see below for details). The first comprehensive analysis of the structure of Rubiaceous apertures was made by Lobreau-Callen (1978). Much confusion is caused by the terminology of pollen classes such as colporate or pororate, because they are used mostly on the basis of external SEM observations only. When a grain has undifferentiated ectocolpi (without mesoapertures) in external equatorial view, it is usually named "colpate." This is the case, for example, in Chiococca phaenostemon Schlecht. (Huysmans et al., 1999). However, on broken grains or in LM, endocolpi are visible with diverging ends perpendicular to the ectocolpi. According to the definition in Punt et al. (1994), Chiococca P. Br. is thus colporate. Consequently, the thinnings at the inside of the pollen wall merit closer examination (see under endoapertures).
We propose to follow the terminology of Punt et al. (1994) and hence to use the term "colporate" for the combination of an ectocolpus with any other aperture, whether a mesoaperture is visible at the outside, an endoaperture is visible only at the inside of the grain, or both. For cladistic analyses, however, the character aperture type should be coded in a different way (cf. below).
Grains with an apertural system consisting of three compound apertures are believed to be plesiomorphic in Rubiaceae. A restricted number of genera--for instance, Galium L. (truly colpate)--have pollen with less-differentiated, single apertures.
Ectoapertures, apertures in the outer layer of the sporoderm, can be colpi (Figs. 19-21) or pori (Figs. 22-24). The length of the colpi is very variable and sometimes demarcates genera or groups of related species. Colpus endings are equally variable. They can be truncate, acute, obtuse, or, more rarely, fishtail shaped. The colpus membrane can be smooth or granular (Fig. 19).
[FIGURES 19-24 OMITTED]
Opercula, or sexinous structures that cover part of the ectoaperture, were only observed in a few species with pantoaperturate grains: Geophila repens (Fig. 23; Vanthournout, 2002), and in at least two species of Spermacoce (Fig. 24; Dessein et al., 2002a). Opercula-like structures are also shown, but not discussed, for Sarcopygme pacifica (Reinecke) Setchell & Christoph (Darwin, 1979).
Mesoapertures, if present, are pores (Fig. 20). Thickenings around these pores are common, either visible at the outside (aspis) or at the inside (here called "costa"). The combination of an ectocolpus with a perpendicular endocolpus may result in a square hole at the equator (Fig. 21). It is not a well-defined mesoporus but merely the intersection of the two other apertures.
The variation in endoapertures in Rubiaceae is large and has significant systematic value. Many Rubiaceae, including their earliest derived members, show an endoaperture perpendicular to the ectocolpus. Endocolpi are the predominant endoaperture in Rubiaceae (Fig. 28). The endocolpus ends vary from acute to obtuse or fishtail shaped. Next to endocolpi, endocinguli are also frequently observed (Fig. 27). There are strong arguments to assume that endocolpi may fuse into an endocingulum, because many intermediate forms are observed (Dessein et al., 2000). Moreover, in many tribes both endocinguli and endocolpi occur. Some porate species have endopores surrounded by circular costae (Fig. 25). When ectopores and endopores are congruent, we, and many other authors, use the term "porate" to describe the pollen grains. In this case, endoapertures are considered to be absent, although the condition may derive from pollen grains with incongruent endo- and ectoapertures.
[FIGURES 25, 27-28 OMITTED]
Endoapertures can become very complex by the presence of extensions or by additional thinnings within, for example, an endocingulum. The tribe Spermacoceae s.str, exhibits intricate endocingula with two opposite triangular extensions in each mesocolpium and spindle-shaped thinnings underneath each ectocolpus (Fig. 27; Dessein et al., 2002a). In the Portlandia-Exostema-Catesbaeeae-Chiococceae-group (hereafter referred to as the "PECC-complex"), four distinct types of nexine ornamentation were described in spite of their uniform external pollen morphology. The most pronounced type in the group is the combination of an endocingulum with a single, broad extension in each mesocolpium. These extensions fuse at the poles, forming a star-shaped thinning. Near each ectoaperture, four smaller extensions are present, two at each side of the endocingulum, which fuse with the mesocolpial bands near the poles. In the cutaways, a smooth, perforated layer crops out (Huysmans et al., 1999). In Scandentia E. L. Cabral & Bacigalupo, Cabral and Bacigalupo (2001) observed several endoapertures for each ectocolpus.
d. Protruding Onci and Pollen Buds
In many angiosperm species, a lens-shaped structure that is not resistant to acetolysis and occurs beneath the aperture is formed. In Rubiaceae, these onci sometimes protrude through the apertures, forming papillae (Fig. 1). This feature has been discussed for Oldenlandia nudicaulis Roth (Farooq & Inamuddin, 1969), Stephegyne parviflora Korth. (? = Mitragyna sp.) (Ramam, 1954), and Mitriostigma axillare Hochst. (Hansson & El-Ghazaly, 2000). In many other publications, similar structures are seen in illustrations of unacetolyzed pollen grains of other taxa but are not discussed in detail; as, for example, in Atractocarpus (Puttock, 1992), Gardenia Ellis (Puttock, 1992), Scyphiphora C. F. Gaertn. (Puff & Rohrhofer, 1993), and Strumpfia Jacq. (Igersheim, 1993a).
In SEM, these protruding onci look very similar to pollen buds, which were reported in Mussaenda L., Ophiorrhiza L., Pseudomussaenda Wernham, and Schizomussaenda Li (Chennaveeraiah & Shivakumar, 1983; Igersheim & Weber, 1993; Mathew & Philip, 1987; Philip & Mathew, 1975; Puff et al., 1993b). The protoplasmatic vesicles seen in Myrmecodia (Robbrecht, 1988a) are probably also pollen buds. Pollen buds are larger than the protruding onci and differ in the fact that they possess a vacuole. According to Igersheim and Weber (1993) and Weber and Igersheim (1994), they also differ in separating from the pollen grains before shedding. Tilney and Van Wyk (1997) report protrusions of intine, often in addition to protoplasm in Canthium Lam., Keetia, and Psydrax. Although their work includes TEM observations, they do not distinguish protruding onci from pollen buds.
Rubiaceae show a wide array of morphological variation in sexine patterns, including psilate (Fig. 29), perforate (Fig. 30), foveolate, (micro)reticulate (Figs. 31-34), rugulate, rugulate-clavate, articulate, hamulate, striate, and double reticulate (Fig. 32). The majority of the species have a tectum perforatum or are (micro)reticulate. In Mycetia javanica (Blume) Reinw. ex Korth., the thick tectum is very irregular, with elongated or rounded processes beset with unequally spaced microspines (Fig. 35). Another strange sexine is observed in Gouldia terminalis (Hook. & Am.) Hillebr., in which the bacula are interconnected and beset with microspines (Fig. 36).
[FIGURES 29-36 OMITTED]
A double or complex reticulum characterizes the tribe Coccocypseleae (Piesschaert et al., 2000b). It also occurs in Pavetteae (De Block & Robbrecht, 1998), Spermacoceae s.1. (Dessein, 2003; Groeninckx, 2005; Pire, 1997a; Pire & Cabral, 1992), and the genus Metabolos (Puff & Igersheim, 1994a). A double reticulum consists of a smooth "suprareticulum" and a spinulate "infrareticulum" situated at a level slightly below the suprareticulum (Fig. 32). At present, it is unclear whether all structures named "double reticulum" in Rubiaceae are structurally identical. An in-depth TEM study is needed to verify this hypothesis.
Atectate grains are rare but do occasionally occur in Psychotrieae and in Versteegia Valeton (Pavetteae; De Block & Robbrecht, 1998). Versteegia cauliflora (K. Schum. & Lauterb.) Valeton has free-standing bacula on the foot layer.
Supratectal elements are absent in the majority of Rubiaceae. A number of tribes have pollen both with and without supratectal elements; for example, Coffeeae, Pavetteae, and Gaertnereae. Only in a few taxa--e.g., Rubieae (Huysmans et al., 2003) and all 24 genera of the PECC-complex (Huysmans et al., 1999, unpubl.)--the sexines are invariably echinate. The supratectal elements can be (micro)spines, microgemmae or microverrucae, or tiny striae. In several Australian Gardenia species, pollen grains with small gemmae intermingled with very large ones have been observed (Puttock, 1992).
Distinct margines of supratectal elements are rather exceptional. In Spermacoce natalensis Hochst., for instance, spines occur in an elliptic field around the ectocolpus but the rest of the sexine is psilate (Dessein et al., 2002a).
6. Nexine Ornamentation
The inside of acetolyzed Rubiaceae pollen is often granular. The nature of this granular layer was recently studied ontogenetically and histochemically by El-Ghazaly and Huysmans (2001). They found that it is a distinct wall layer with a particular mode and timing of development, different from both the endexine and the intine. Histochemically, this layer differs from the endexine in having fewer lipids and more proteins and is distinguished from the intine in containing more pectin and less acidic polysaccharides. The granular layer also resists acetolysis and was called the "membranous granular layer" (MGL). El-Ghazaly and Huysmans (2001) describe a MGL in four dicots (including Rondeletia odorata) and one monocot. The thinnings often observed at the inside of acetolyzed Rubiaceae pollen are located in the MGL and not in the endexine. For an unambiguous understanding, we prefer to use the circumscription "inner surface of the nexine" in what follows.
The systematic importance of nexine characters at the inside of pollen grains is repeatedly stressed in Rubiaceae (Jansen et al., 1996a; Van Campo, 1978), and we are fully convinced of their potential as phylogenetic markers (Huysmans et al., 1998a, 1999). The ornamentation of the nexine is mostly granular (Fig. 37); few genera have an entire smooth inner nexine surface (e.g., Chazaliella E. M. A. Petit ex E. M. A. Petit & Verdc.). In Coptosapelta Korth. the nexine is bumpy, like water droplets on an oily surface (Fig. 39; Verellen et al., 2004).
[FIGURES 37, 39 OMITTED]
Irregular grooves in the inner nexine surface, called "endocracks," are common in Rubiaceae (Fig. 38). They sometimes reduce the granular layer into isolated patches, as clearly seen in Isertia Schreb. (Huysmans et al., 1998a), Capirona Spruce, and Gaertnera Lam. (Fig. 40; Jansen et al., 1996a).
[FIGURES 38 & 40 OMITTED]
7. Stratification of the Pollen Wall
In general, Rubiaceae pollen corresponds to the basic pattern of pollen-wall stratification in angiosperms: rectum, columellae, foot layer, endexine, intine (Figs. 41-42, 44). The relative proportions of the respective layers are variable, even within one pollen grain. In a few genera (e.g., Coptosapelta, Durringtonia) columellae are reduced or totally lacking (Fig. 43). Atectate pollen has so far been observed only in Versteegia cauliflora (De Block & Robbrecht, 1998). Free-standing bacula in the lumina of a reticulate sexine (Fig. 34) occur in several species belonging to at least 21 genera scattered all over the family. It seems to be a typical character for many genera of the Morindeae (e.g., Prismatomeris Thawaites, Caelospermum Blume).
[FIGURES 41-44 OMITTED]
B. HETEROSTYLY AND POLLEN DIMORPHISM
Barrett (1992: 1) defined heterostyly as "a genetic polymorphism in which plant populations are composed of two (distyly) or three (tristyly) morphs that differ reciprocally in the heights of stigmas and anthers in flowers." Heterostyly is very common in Rubiaceae (cf. Anderson, 1973; Bir Bahadur, 1968b; Bremekamp, 1952, 1966), especially in genera of Psychotrieae and Spermacoceae s.1. (Robbrecht, 1988a). Given that congeners of heterostylous species are often homostylously ancestral, it must have evolved repeatedly in the family (Anderson, 1973). Pollen size dimorphism is often correlated with heterostyly. As a rule, pollen of the brevistylous morph is slightly larger than pollen in flowers with long styles (Bir Bahadur, 1963, 1966, 1968a, 1968b, 1970; Bir Bahadur & Rama Swamy, 1993; Bremekamp, 1963; Jung-Mendacolli & Melhem, 1995; Naiki & Nagamasu, 2003; Pailler & Thompson, 1997). However, the opposite situation is reported in Pentas schimperiana Vatke (Dessein et al., 2000), Pseudosabicea arborea (K. Schum.) N. Halle (Huysmans et al., 1998a), and Hindsia longiflora (Cham.) Benth. (Di Maio, 1996). In these species the larger pollen grains are observed in the longistylous flowers. However, one must bear in mind that size differences are usually slight and become obscured if different populations are compared (Puff, 1988).
Two other cases of pollen dimorphism correlated with heterostyly are frequently reported: variation in aperture number and sexine ornamentation. Baker (1956), for example, reported "spiniferous" pollen grains in short-styled flowers of Rudgea and "smooth" pollen grains in long-styled flowers. Similar cases are reported for several other genera scattered all over the family; for example, for the Danais-Payera-Schismatoclada complex (Buchner & Puff, 1993). In several species of Coccocypseleae, a tribe characterized by a double reticulum, it has been observed that the infrareticulum is better developed in the brevistylous flowers than in the longistylous flowers. Pire and Cabral (1992) made a similar observation for the genus Galianthe Griseb. (Spermacoceae s.str.). In most cases, however, the difference between the exine of the brevistylous and longistylous morphs is less pronounced: The dimorphism is often limited to a coarser reticulation in the thrum grains (Bir Bahadur, 1968b). The pollen morphs of Sabicea capitellata Benth. and S. orientalis Wernham differ morphologically only in the ratio of 3- and 4-aperturate grains. Longistylous flowers of Sabicea capitellata have considerably fewer 4-aperturate grains than do brevistylous flowers; only in the brevistylous morph do 5-aperturate grains occasionally occur. In Sabicea orientalis the opposite was observed: Brevi-pollen was mostly 3-aperturate, sometimes 4-aperturate, whereas the ratio of 3-/4-aperturate grains in longipollen is approximately 1 (Huysmans et al., 1998a). A more extreme example of difference in aperture number correlated with heterostyly was reported by Robbrecht (1977) in Hymenocoleus. Brevistylous individuals apparently produce nonaperturate pollen grains, whereas the longistylous flowers contain 3(-4-5)-porate grains. A SEM investigation of this genus should be undertaken, however.
In Pseudosabicea arborea, apart from size differences, pollen dimorphism is conspicuous in several other characters (Huysmans et al., 1998a). The sexine pattern is reticulate in long-styled flowers and perforate in short-styled flowers; apertures are larger in longistylous flowers; and the exine is twice as thick in longi-pollen ([+ or -] 2 [micro]m) as in brevi-pollen ([+ or -] 1 [micro]m).
C. ORBICULE CHARACTERISTICS OF RUBIACEAE
Orbicules are tiny sporopollenin particles up to 3-4 [micro]m in diameter that line the inner tangential and radial walls of the tapetal cells (Huysmans et al., 1998b, 2000). Thus orbicules coat the inner locular surface and are in close contact with the pollen grains. Orbicules may have an electronlucent core, but they are always acellular structures. They are often, but not always, present in Rubiaceae.
The first report of the presence of orbicules in Rubiaceae was by Andronova (1984). She reported orbicules in several unrelated species of the family. Because her work was published in Russian, however, little attention has been paid to this study. Consequently, Pacini and Franchi (1993) did not include any Rubiaceae in their listing of angiosperm species with secretory tapetum reported to have orbicules. In the same year, Igersheim (1993a) reported the presence of orbicules in the monospecific Caribbean genus Strumpfia. Since then, two subfamilies of Rubiaceae have been screened for the presence of orbicules: Huysmans et al. (1997) investigated 14 genera of Cinchonoideae s.str., and Vinckier et al. (2000) studied 32 genera of Ixoroideae s.1. Orbicules have also been reported in several other genera. In Table II all present data are summarized at the generic level.
Six orbicule types can be described in Rubiaceae (Huysmans et al., 1997; Vinckier et al., 2000):
I. Spiny orbicules (Figs. 45-47). Orbicules are up to 3 [micro]m in diameter and are not embedded in tapetal membrane (although exceptional semi-embedded forms do occur). The orbicular wall is smooth but beset with a few to numerous spinules; one or a few perforations may occur. A core is absent or present.
[FIGURES 45-47 OMITTED]
II. Microrugulate orbicules (Fig. 48). Orbicules are less than 2 [micro]m in diameter and are embedded in the tapetal membrane. Their shape is more or less spherical, but they seem to consist of several elongated elements.
[FIGURE 48 OMITTED]
III. Smooth orbicules (Figs. 49-51). Orbicules are less than 1.4 [micro]m and are rarely embedded in the tapetal membrane. Aggregates of two or more orbicules may occur. The orbicular wall is smooth. Subtype IIIa has rounded, oblate orbicules with small perforations in the wall; subtype IIIb has more flattened, circular orbicules with a single central indentation (doughnut shaped).
[FIGURES 49-51 OMITTED]
IV. Irregularly folded orbicules (Figs. 52-53). Orbicules are irregular in shape and not spherical because of a folded surface. Perforations may occur, but a core is absent. They are not or only slightly embedded in tapetal membrane.
[FIGURES 52-53 OMITTED]
V. Granular orbicules (Figs. 54-55). Orbicules are flattened and irregular, 1-2 [micro]m in diameter. The wall has several perforations, and small sporopollenine granules occur on the surface. A core is present, and they are often compound. They are not embedded in tapetal membrane.
[FIGURES 54-55 OMITTED]
VI. Embedded orbicules (Figs. 56-57). Orbicules are flattened, regular or irregular in shape and 0.5-1.6 [micro]m in diameter. Perforations and small sporopollenin granules may occur on the surface. Compound orbicules with several cores occur. The core-wall interface is not electron dense. They are embedded in tapetal membrane.
[FIGURES 56-57 OMITTED]
D. FOSSIL POLLEN RECORD FOR RUBIACEAE
Fossil data--both for microfossils and macrorests--are scarce for Rubiaceae. Muller (1981) provided the most recent overview of pollen records for the family. The oldest Rubiaceae pollen grains are recorded from the upper Eocene. The grains are similar to the tetrads observed in existing Gardenia species. During the Oligocene, Macrosphyra-type, Mitragyna-type, Faramea-type, and Coprosma-type pollen grains appeared. These were followed by Scyphiphora-type, Sabicea-type, Morinda-type, Timonius-type, and Morelia-type pollen in the Lower Miocene. In the Upper Miocene, pollen of the more derived genera, such as Borreria G. Mey. and Rubia L., emerge in the fossil record. This pollen record is too fragmentary to draw solid conclusions about the evolution of pollen types in the family. Using the appearance of pollen types as calibration points for dating phylogenetic trees based on DNA sequence data should be done carefully.
|Printer friendly Cite/link Email Feedback|
|Author:||Dessein, Steven; Ochoterena, Helga; De Block, Petra; Lens, Frederic; Robbrecht, Elmar; Schols, Peter|
|Publication:||The Botanical Review|
|Date:||Jul 1, 2005|
|Previous Article:||Systematics and leaf architecture of the Gunneraceae.|
|Next Article:||Palynological characters and their phylogenetic signal in Rubiaceae.|