Printer Friendly

A consensus classification for the order Gentianales with additional details on the suborder Apocynineae.

II. Introduction

A comparison of the Gentianales was undertaken using the proposed classifications of Benson (1979), Cronquist (1988), Dahlgren (1983), Goldberg (1986), Heywood (1978b), Melchior (1964), Stebbins (1974), Takhtajan (1983), and Thorne (1992). These classifications were chosen because the authors involved used multifarious data sources and because they encompassed the whole of the dicotyledons. A consensus classification was then established based on how these authorities interpreted the data available for the order. Consensus is not used here in the cladistic sense (see Barrett et al., 1991); in fact, the reverse may be true, as the authorities chosen here have all used the principle of total evidence.

Only families with 50% to 100% consensus for inclusion are considered here to constitute the order Gentianales. It is hoped that this classificatory model of the Gentianales will be helpful to those teaching systematics at colleges and universities. Certainly it forms a useful base for students to build on or change as their knowledge and hands-on experience of the taxa involved grows. It may also prove useful to TABULAR DATA OMITTED workers employed in routine herbarium curation, plant identification, and the compilation of systematic lists of various kinds. Becker (1973) and Swift (1974) completed similar exercises when they compared six and eight systems of angiosperm classification, respectively, some unfortunately pre-dating the modern phylogenetic paradigms that dominate present macrosystematics. The Gentianalian model given here differs from Swift's in that 1960 has been used as a cut-off date for classifications used in the consensus. This has been done because these post-1960 macrosystematic classifications have used the vast store of multifarious data (including chemical) currently available. They are therefore a synthesis of information on a grand scale. However, older classifications should not be ignored on this account, and, although not forming part of the consensus, they are also briefly mentioned to give a more-or-less complete picture of the order.

There may be a major criticism concerning the consensus model presented in this paper, regarding the nine classifications that were chosen to obtain the necessary statistics. The classifications of Emberger (1960), Hutchinson (1969), Rouleau (1981), and Young and Seigler (1981) have not been used. If included they would alter the picture only slightly; in fact, the only major change would be the addition of the Menyanthaceae to the Gentianales by a consensus majority of 61%. However, as will be pointed out later, the available data does not support this family's inclusion. The other families would remain within the order with the following consensus figures (consensus figures from the nine chosen classifications are given in parentheses for comparison): Gentianaceae, 100% (100%); Loganiaceae, 58% (100%); Apocynaceae, 92% (100%); Asclepiadaceae, 77% (78%); Rubiaceae, 54% (56%); and Saccifoliaceae, 75% (75%).

Heywood (1974) makes a plea for establishing priorities in systematic research; the consensus model given here highlights families that should be targeted as priorities for detailed investigation using modern methods, namely, the Columelliaceae, Dialypetalanthaceae, Menyanthaceae, Oleaceae, and Salvadoraceae. Data accumulation for these taxa may uncover interesting evolutionary pathways, so improving present classifications and making them more predictive (Stuessy, 1993) and also more useful as tools for understanding and managing the earth's biodiversity. The predictivity of classifications is already important for locating chemicals, particularly those of medicinal or economic importance. Such predictivity may also become more important as a means of locating genes for use in the growing field of genetic engineering. Data accumulation must be done while enough of the species making up these families are still extant. These particular plant families may represent the last living pieces in an evolutionary jigsaw puzzle in which many of the pieces were already lost long before the development of taxonomy. We cannot afford further loss of diversity and its associated data, because even present gaps in our information make untestable the development of hypotheses concerning present relationships and past important phylogenetic trends (which depended on prehistoric modes of selection and speciation). We cannot base all our hopes on finding the fossil evidence necessary to complete this task.

Not included in the data for this consensus model, but nevertheless of historical importance, is the classification of Hutchinson (1969), who divided the dicotyledons into two major groups: the Lignosae (fundamentally woody plants) and the Herbaceae (fundamentally herbaceous plants). In the Lignosae could be found the Rubiales (consisting of the Dialypetalanthaceae and Rubiaceae), the Apocynales (consisting of the Apocynaceae, Asclepiadaceae, Periplocaceae, and Plocospermataceae), and the Loganiales (consisting of the Antoniaceae, Buddlejaceae, Loganiaceae, Oleaceae, Potaliaceae, Spigeliaceae, and Strychnaceae). The Gentianales (consisting of the Gentianaceae and Menyanthaceae) occur in Hutchinson's Herbaceae. Because of the rather artificial division into herbaceous and woody plants and the very narrow ordinal circumscriptions, it is difficult to include Hutchinson's otherwise historical classification in the present consensus model. However, where applicable, Hutchinson's ideas and concepts are discussed. To some degree, Benson's (1979) classification of the angiosperms into Thalamiflorae, Corolliflorae, Calyciflorae, and Ovariflorae is slightly artificial as well, but not so much so that it needed to be omitted for comparison in this study.

Unfortunately, Takhtajan's (1987) latest work is presently unavailable to the authors and it has not been possible to incorporate it into this consensus. As outlined in Brummitt (1992), Takhtajan's 1987 system differs from his 1983 system in recognizing the Saccifoliaceae, Desfontainiaceae, Plocospermataceae, and Spigeliaceae as distinct families. A newly included family is the Carlemanniaceae, usually included by consensus in the Caprifoliaceae. No families were reduced to synonymy or transferred out of the Gentianales as he defined it in 1983. Interestingly, Takhtajan (1987) seemed to restrict the circumscription of the Asteridae to the Campanulales and allies and the Asterales. The bulk of the orders that used to fall in the Asteridae now fall under the subclass Lamiidae, which includes the superorder Gentiananae--composed of the orders Gentianales and Oleales (this later being monofamilial). The Salvadoraceae can still be found in the Celastrales, the Columelliaceae in the Hydrangeales (suborder Escalloniineae), the Convolvulaceae and Cuscutaceae in the Convolvulales, and, finally, the Buddlejaceae and Retziaceae he placed in the Scrophulariales .The authors have also not had access to Young and Seigler (1981) and have used an account of their system given by Goldberg (1986). Brummitt (1992) also gives an account of Young's system. Here the Gentianales is composed of the Loganiaceae, Buddlejaceae, Rubiaceae (including Theligonaceae), Apocynaceae, Asclepiadaceae, Gentianaceae, and Menyanthaceae. The Columelliaceae may be found in the Hydrangeales and the Salvadoraceae, the Oleaceae (and Barbeyaceae) in the Oleales, and the Convolvulaceae (presumably including the Cuscutaceae) in the Solanales. No mention seems to be made of the Dialypetalanthaceae, Retziaceae, or Saccifoliaceae.

It is not within the scope of this paper to give detailed information on the morphology, anatomy, chemistry, etc., for each taxon involved, although the literature cited contains much of this data or cites literature that does. However, data and information considered by these authors to be significant are highlighted in the text of this paper.

Lastly, a computational analysis by Young and Watson (1970) highlights the taxonomic soundness of the order Gentianales. Their analysis, however, included only the families Apocynaceae, Asclepiadaceae, Gentianaceae, Periplocaceae, and Rubiaceae.

III. Historical Account of the Gentianales

Classifications, including those of higher taxa, will continue to change as our knowledge of these groups and their relationships changes. Based on the accumulation of additional data and the way in which the data are interpreted, taxa are added, shuffled, or deleted. Classifications are therefore largely eclectic in nature and we seem to proceed toward a clearer picture of the plant world, past and present, by successive approximations toward a (probably unattainable) state of total knowledge. Important also is that with each approximation the predictivity of the classificatory model increases, thus increasing the value of the classification to humankind. As a result, present-day concepts can be clearly understood only in the light of their own evolution. To obtain a useful taxonomic picture, workers must therefore undergo a certain amount of mental or intellectual recapitulation of the pivotal taxonomic events (taxonomic interpretations and decisions) and historic scientific paradigms that lead to the moulding of modern classifications.

The historical, angiosperm-wide classifications of the 1800s had a great influence on present-day concepts within the Gentianales. In particular, two historical classifications are important: Bentham and Hooker's Genera Plantarum (1862-1883) and various authors in Engler and Prantl's Die Naturlichen Pflanzenfamilien (1887-1915). These two classifications are still used as the basic phylogenetic systems on which most major herbaria are structurally based and on which their continuing daily operation depends. The system of Bentham and Hooker continues to be followed in Britain and much of the British Commonwealth; that of Engler and Prantl (as modified by de Dalla Torre & Harms, 1900-1907) is still used in much of continental Europe and countries historically associated with it. Lastly, mention should be made of the influential-American botanist Asa Gray, who took the natural classificatory systems of Europe and adapted them for use in North America (Gray, 1878). Gray's system influenced the arrangement of specimens in many early United States herbaria and continues to do so (Fernald, 1950). As herbaria are essentially large phylogenetic models (which we use for understanding, studying, and managing the phytodiversity of the world), most taxonomists have grown up using one of these historical classifications (or a modification) as a framework on which to work and build. As a result, we are all, to some extent, prisoners of our past learning and experience. In fact, Walters (1961) goes as far as suggesting that the present stability we see in many angiosperm families is due to taxonomists' unwillingness to change them, and family boundaries must therefore remain suspect. Whether this statement is true or not remains part of the ongoing systematic debate.

It is therefore important to discuss briefly the taxonomic and nomenclatural evolution of the Gentianales.


Historically it was Bartling (1830) who first brought together the families traditionally associated with the Gentianales and to which he applied the name Contortae. However, it was Lindley (1833) to whom the formal description of the order is attributed, he being the first author to consistently use the suffix -ales to denote the rank of order. Later, Lindley, having placed the Apocynaceae and Asclepiadaceae in the Gentianales, proceeded to place them apart (Lindley, 1845), the Apocynaceae being in the Gentianales and the Asclepiadaceae in the Solanales. In his 1867 account, Lindley gives a small nearest-neighbour diagram after each family description/discussion. If these diagrams are fitted together, a rather interesting picture emerges that not only tends to place the Gentianales together but allies them to the Cornales. This concurs, to some extent, with modern evidence and ideas. Melchior (1964) was the first authority to include the Rubiaceae in the Gentianales, although this idea gained wide acceptance only in the classifications of the 1980s.


In their Genera Plantarum (1862-1883), Bentham and Hooker defined the Gentianales as containing the Apocynaceae, Asclepiadaceae, Gentianaceae, Loganiaceae, Oleaceae, and Salvadoraceae. The Asclepiadaceae was considered to include the Periplocaceae as the subfamily Periplocoideae. The Loganiaceae included the Antonieae, Buddlejeae, Desfontainieae, Potalieae, Spigelieae, and Strychneae under the tribe Euloganieae, while the Plocospermateae was included under the tribe Gelsemieae. The Gentianaceae included the Menyanthaceae as the tribe Menyantheae.

Excluded families: The Rubiaceae was placed in the order Rubiales under the major division Inferae (due to its inferior ovary). The Convolvulaceae (including Cuscutaceae as the tribe Cuscuteae) was included in the Polemoniales. The genus Retzia ([is approximately equal to] monotypic Retziaceae) was placed in the Solanaceae (which was also included in the Polemoniales). The Columelliaceae fell under the Personales (along with the Scrophulariaceae). The Theligonaceae was placed as a tribe, the Theligoneae, under the Urticaceae. The Dialypetalanthaceae had not yet been discovered.


Various authors in Engler and Prantl's Die Naturlichen Pflanzenfamilien (1887-1915) dealt with the families of the Gentianales as follows. The Series Contortae is considered equivalent to Bentham and Hooker's Gentianales. Series Contortae was divided into two subseries. The first of these, the Oleineae, contained the families Oleaceae and Salvadoraceae (Knoblauch, 1892). The second subseries, the Gentianineae, contained the Loganiaceae, Gentianaceae, Apocynaceae, and Asclepiadaceae. The Loganiaceae included the tribes Antonieae, Fragraeae (which in turn included the Potalieae), Loganieae, Spigelieae, and Strychneae under the subfamily Loganioideae and the Buddlejaceae under the subfamily Buddlejoideae, while the monotypic Desfontainieae and Plocospermateae were placed at the end of the family with a mention of their uncertain taxonomic position (Solereder, 1892). The Gentianaceae included the Menyanthaceae as the subfamily Menyanthoideae (Gilg, 1895). The Asclepiadaceae included the subfamily Periplocoideae (Schumann, 1895).

Excluded Families: The Rubiaceae (Schumann, 1891) was placed in the Rubiales under the Sympetalae. The Convolvulaceae (including the Cuscutaceae as the subfamily Cuscutoideae) (Peter, 1891) was placed under the series Tubiflorae in the subseries Convolvulineae. Retzia was included under the family Solanaceae (Wettstein, 1891) but with comments on its close relationship to the Loganiaceae. The Theligonaceae was a synonym under the family Cynocrambaceae (Poulsen, 1893). The Dialypetalanthaceae had not yet been discovered. Reference should also be made here to Engler's Syllabus der Pflanzenfamilien (the 1912 edition assisted by Gilg and the 1936 edition co-authored by Diels).

IV. Comments on Classification above the Level of Order

Interestingly, the beginnings of the concept of the subclass Asteridae (to which the Gentianales belongs) can be seen as early as Jussieu (1789) and his Class VIII-Monopetalae, corolla hypogyna and de Candolle's (1813) Corolliflorae. In fact, the Corolliflorae used by Benson (1979) has its beginnings in the classification of de Candolle (1813-1819). Bartling (1830) was the first to use the term Contortae, this name being used for the last time by Wettstein in 1935, while the category Sympetalae, first used by Engler and Prantl (1887-1915) in 1897, was used right up until 1964 by Melchior. The use of the term Asteridae to denote this particular subclass seems to have first been used by Takhtajan in 1964. Cronquist (1988), Heywood (1978b), and Takhtajan (1983) have more or less retained this ([+ or -] sympetalous) circumscription of the Asteridae, while Dahlgren (1983), Takhtajan (1987; see Brummitt, 1992), and Thorne (1992) have abandoned it in favour of a number of separately and more narrowly defined subclasses. Takhtajan (1987) and Thorne (1992) have also tended to make more use of the superordinal and subordinal ranks than any of their predecessors, thus making suggested relationships within and between groups more apparent to users of macrosystematic classifications; see Wagenitz (1977) for further discussion of the subclass Asteridae.

In 1977, Dahlgren used the suffix -anae to denote these macrotaxa but subsequently changed this to -iflorae (Dahlgren, 1983). However, the suffix -iflorae is of limited value, in that it can only be applied to angiosperms, and is thus best avoided. The use of this suffix by R. Dahlgren has been changed to -anae by G. Dahlgren (1989). Thorne (1992) has also adjusted his system, and his subclasses now end in the suffix -anae.

Many of the groupings of Cronquist, Dahlgren, Takhtajan, and Thorne are similarly circumscribed, indicating a growing consensus, probably due to the holistic, systematic synthesis of all available data by these different workers. Unfortunately, much of the similarity of these systems is masked by inconsistently used basionyms for the macrotaxa. To establish clarity and order for these macrotaxa, urgent attention should be given by phylogenists to establish some sort of taxonomic priority, complete with basionyms, as is used for microtaxa. Cronquist's 1981 classification is nomenclaturally superior to its contemporaries, and it is therefore not surprising that it is now linked to the ideas, specimens, and literature of a continental flora (Kanis, 1981). As a result of this, Cronquist's (1981) choice of basionym has been followed in this work if two or more names compete for priority after 1753. Thus the name Gentianales (published 1833) has been accepted above the name Loganiales (also published 1833). Having said this, the classifications of Takhtajan (1987) and Thorne (1992) are believed by these authors to express relationships better than those of their contemporaries, although the merits of other contemporary classifications, particularly those of Cronquist (1981) and Dahlgren (1983), should not be ignored. Unfortunately, Thorne's (1992) work lacks synonyms, authorities, and literature citations; there is also a hesitancy on the part of modern taxonomists to accept this worker's often large family circumscriptions. Many of Thorne's subfamilies tend to correspond to the families of most other contemporary classifications, and his system may be viewed as a lumper's classification. At the other end of the spectrum, the classification of Hutchinson (1969), with its many small families, can be seen as a splitter's classification. In this paper the Asteridae is used in the sense of Cronquist (1988).

V. Phylogenetic Considerations

Besides being largely speculative, phylogenetic concepts are seriously hampered by nomenclature. Names of taxa carry concepts concerning the world as we symbolize it in speech. It is the problem of Popper's three worlds: the physical, the mental, and the symbolic as typified by speech (see discussion by Loevtrup, 1987). Whereas names and their symbology are static (Popper's third world) the taxa themselves have changed and are changing, often quite substantially, with geological time (Popper's first world). To apply present-day names such as the Loganiaceae to its early ancestors is therefore quite inappropriate. At the time of its emergence, this taxon (which may, using present paradigms, have been considered only a tribe or even genus of some extinct family) was a very different entity from the one we see today. The use of modern names to symbolize prehistoric taxa is fraught with danger. However, in the absence of a complete fossil record we are left with no choice but to use present-day concepts when we talk about these taxa. Where possible, we have used the prefix proto- to denote that these taxa would have been circumscribed very differently from their present-day progeny. Readers must bear this in mind during the following discussion.

Multifarious data suggest that the Asteridae may have arisen from what was an amorphous Cornales-Saxifragales-Rosales-like* complex of related taxa [*Cornales of Dahlgren (1983) or Rosanae-Cornanae of Thorne (1992)]. The Escalloniaceae, a present-day taxon of this complex, is of interest, in particular the genus Tetracarpaea, which has an apocarpous gynoecium and tetramerous flowers, this first character being associated with the Apocynaceae and the latter with the Scrophulariales via the Buddlejaceae. The genus Escallonia itself contains asperuloside, a heteroside of limited distribution but commonly occurring in the Rubiaceae (Takhtajan, 1983). The Brunelliaceae also has an apocarpous gynoecium, although in most other families of the Cornales-Rosales-Saxifragales complex the ovary is syncarpous. The Hydrangeaceae, which seems to have had a common origin with the Escalloniaceae (Takhtajan, 1983), has opposite rather than alternate leaves. Both kinds of leaf arrangements may also be found in the Asteridae. The family Alangiaceae has a flower structure and alkaloids reminiscent of the Rubiaceae (Takhtajan, 1983): furthermore, the genus Berenice of the Escalloniaceae appears to be a transition group between this family and the Campanulaceae. Another character in favour of the Cornales complex being a possible sister group to the Asteridae is the possession of iridoids, as are found in the less specialized members of the Asteridae, such as the Loganiaceae. Many species of the Gentianales (as defined in this paper) have been used as antimalarial treatments; it is thus interesting to note that the suspected sister group is known to have antimalarial alkaloids as well, in particular, species of Dichroa and Hydrangea, both in the Hydrangeaceae (Watt & Breyer-Brandwijk, 1962).

Fossil pollen (see Cronquist, 1988) would suggest that within the Asteridae, the Loganiaceae and Apocynaceae were possibly the first taxa to become distinct and recognizable (as we understand them today) within the subclass. The Rubiaceae may have begun to differentiate at an early stage as well, although this is not (as yet) borne out by the fossil record. However, the emergence of the Asteridae was probably not monophyletic and was possibly complicated by extensive reticulate evolution. Evidence that supports this complex reticulate, non-monophyletic evolution is the complex distribution of shared characters in extant Gentianalian-Cornalian taxa.

Data associated with the Columelliaceae (whose exact systematic position has been problematic) may suggest that it is the remains of part of that archaic web of taxa that once linked the early Gentianales (and possibly Scrophulariales) to its Cornalian-like ancestor. Interpretation of presently available data also suggests that the Desfontainiaceae (included in the Loganiaceae by consensus but possibly deserving family rank of its own) may also form another such bridging taxon.

The emergence of what we now call the Loganiaceae from what must have then been an ill-circumscribed Cornalian-Rosalian-Saxifragalian root, was itself probably amorphous. Still today there are continuing problems regarding its circumscription. The tribes making up the Loganiaceae form a central web of relationships that span and connect all the other Gentianalian families and (due to its early emergence) many other families of the Asteridae as well. It is interesting that within the Loganiaceae, the tribe Gelsemieae exhibits the greatest number of shared characters with the other members of the Gentianales. Bremer & Struwe (1992) also point out that this tribe is probably a sister group to the Rubiaceae.

Available data, particularly chemical, suggests that the Oleaceae may also be a relict taxon, bridging the proto-Gentianales and the proto-Scrophulariales. The Buddlejaceae and Retziaceae also form bridging taxa between the Scrophulariales and Gentianales, in particular with the family Loganiaceae (and possibly also the Solanales). Problems in assigning the monotypic Retziaceae to the Loganiaceae (Leeuwenberg & Leenhouts, 1980), Scrophulariales (Dahlgren, 1983), or Solanales (Cronquist, 1988) may be due to its origin at an early stage when these three orders were not as distinct as they are today.

Serological data suggest that the Caprifoliaceae may have begun to differentiate from a Cornalian-like ancestor along with the Loganiaceae (it also possesses a similar wood anatomy to our suggested bridging taxon Desfontainiaceae) and may have been near the proto-Rubiaceae, as it shares a number of morphological characters with the present-day Rubiaceae. The proto-Caprifoliaceae may in turn have given rise to the Dipsacales (Cronquist, 1981).

The many shared characters that the Rubiaceae have in common with the Loganiaceae, Apocynaceae, and Asclepiadaceae suggest that the Rubiaceae may have begun to differentiate at about the same time as the proto-Apocynaceae, although this is at present not supported by the fossil pollen record.

The many shared characters of the Loganiaceae, Rubiaceae, and Apocynaceae also suggest that the evolution of the Apocynaceae was probably not monophyletic. In particular, its very close association with the Plocospermateae in the Loganiaceae should be noted. As discussed later, the assumed monophyly of the Asclepiadaceae may also prove to be incorrect. Data suggest that the proto-Gentianaceae may have had a common ancestry with the loganiaceous tribe Potalieae.

A few comments need to be made about Figure 2, on which much of this discussion is based. In this figure shared characters have been given without weighting; however, such weighting could prove important in the final analysis because some characters, such as wood anatomy, may be phylogenetically more significant than leaf morphology, the latter being more phenotypically plastic. Also, some of the taxa in this diagram may share characters that are homoplasious and therefore of no phylogenetic significance. Homoplasy is probably widespread in these groups. Shared characters between the Rubiaceae and Gentianaceae have been omitted due to the forced, but practical, two-dimensional nature of the illustration. A three-dimensional model, if it could be created, would seem less confusing and be of more use as a symbolic representation of the shared character states. Lastly, this illustration does not show important, often unique, character states (apomorphies) that define the taxa themselves. In contrast, Figure 3 was obtained by weighting the characters. Note the isolated position of the Menyanthaceae, confirming its exclusion from the Gentianales and possibly the Solanales and Scrophulariales as well.

VI. Results

A consensus classification or consensus model for the Gentianales:

The Gentianales

Approximate number of genera = 1100

Approximate number of species = 13,050

I. Suborder Gentianineae

1. Gentianaceae (100% agreement for inclusion)

2. Saccifoliaceae (75% agreement for inclusion)

3. Loganiaceae (100% agreement for inclusion), including families Antoniaceae, Desfontainiaceae, Plocospermataceae, Potaliaceae, Spigeliaceae, and Strychnaceae

4. Rubiaceae (56% agreement for inclusion), including families Aparinaceae, Cephalanthaceae, Cinchonaceae, Coffeaceae, Galiaceae, Gardeniaceae, Guettardaceae, Henriqueziaceae, Lippayaceae, Lygodisodeaceae, Naucleaceae, Operculariaceae, Stellataceae, and Theligonaceae

II. Suborder Apocynineae

5. Apocynaceae (100% agreement for inclusion), including families Emeticaceae, Plumeriaceae, Vincaceae, and Willughbeiaceae

6. Asclepiadaceae (78% agreement for inclusion), including family Periplocaceae.

Families Excluded: Buddlejaceae, Carlemanniaceae, Columelliaceae, Convolvulaceae, Cuscutaceae, Dialypetalanthaceae, Menyanthaceae, Oleaceae, Retziaceae, and Salvadoraceae.

Figure 5 is a taxogram. It is based on the relationships of the taxa to each other (as inferred from their associated data) and nested according to accepted taxonomic hierarchy (Greuter et al., 1988). Although probably illustrating underlying evolutionary trends within the group, this is not what this figure represents as it is used here. The authors agree with Constance's (1964) warning that these types of illustrations become a danger only when their symbolic nature (used for easier understanding and communication) becomes confused with reality. Readers should be aware of the limitations and problems of such illustrations.

VII. Discussion


Lindley, Nixus pl. 1833.

Type family: Gentianaceae de Jussieu

Synonyms: Asclepiadaceae Brongniart (1843), Apocynales Hutchinson (1959), Caprifolia Endli cher (1836-1840) pro parte, Cinchonales Lindley (1833), Coffeinae Brongniart (1843), Contortae Bartling (1830), nom. illegit., Corolliflorae de Candolle (1813) pro parte, nom. illegit., Loganiales Lindley (1833), Rubiacineae Bartling (1830) pro parte, Rubiales Bentham & Hooker (1862-1883) pro parte, Seminiflorae Fries (1835) pro parte, nom. illegit., and Tubiflorae Fries (1835) pro parte, nom. illegit.

Gentianales: Lindley, 1867: 594; Bessey, 1915: 143; Melchior, 1964: 405-424; Cronquist, 1968: 284-286; Takhtajan, 1969: 230; Swift, 1974: 302-310; Heywood, 1978b: 222-227; Benson, 1979: 261-269; Cronquist, 1981: 859-882; Takhtajan, 1983: 196-197; Goldberg, 1986: 262. Apocynales: Hutchinson, 1969: 422-432. Cinchonales: Lindley, 1867: 756-765. Contortae: Wettstein, 1935: 918-919; Rendle, 1967: 459-460. Loganiales: Hutchinson, 1969: 412-422. Rubiales: Bessey, 1915: 161; Benson, 1979: 332-333; Cronquist, 1968: 303-305; Cronquist, 1981: 995-1001; Cronquist, 1988: 439-441; Fernald, 1950: xxix; Hutchinson, 1969: 432-454; Rendle, 1967: 550-562.

Description: Taxonomic statistics: approximately 1100 genera with about 13,050 species. Distribution: cosmopolitan. Habit diverse (including parasitic, saprophytic/achlorophyllous, epiphytic, and succulent plants), sometimes with laticiferous canals (Apocynaceae and Asclepiadaceae). Anatomy: vessels with simple perforations, seldom scalarifom perforations with few cross-bars; wood-rays usually 1-5 (-12) cells wide or lacking; internal phloem (intraxylary or interxylary) nearly always present, either as a continuous ring or as separate strands at the pith margin; superficial cork. Leaves opposite, occasionally whorled, rarely alternate, simple and entire, rarely compound, lobed or toothed; with (mainly but not exclusively Rubiaceae) or without stipules, interpetiolar stipules or these reduced to an interpetiolar line or colleters. Flowers usually regular, showy and sympetalous; actinomorphic, rarely slightly zygomorphic; bisexual, rarely unisexual. Calyx (2-)4-5(-12)-merous; connate or free; sometimes with basal nectary glands. Corolla (3-)4-5(-16)-merous but usually 5-merous occasionally 4-merous; usually convolute, occasionally imbricate or valvate; connate sometimes with corona. Androecium: stamens (1-)4-5(-16), but usually as many as and alternating with petals; free, connate or adnate to corolla or gynoecium. Pollen grains binucleate or trinucleate, often tricolporate and occasionally in tetrads. Gynoecium 2(-9) carpellate, but usually bicarpellate or occasionally 5-8-merous, rarely pseudomonomerous; syncarpous or apocarpous, (1-)2(-9) 1ocular; usually axile or parietal placentation, rarely free central, basal or apical; superior, semi-superior or inferior. Ovules 1 to many in each locule, anatropous to hemitropous, or amphitropous; unitegmic, simple, tenuinucellar or pseudocrassinucellar (in Asclepiadaceae), usually without an integumentary tapeturn; endosperm development usually nuclear, rarely cellular. Fruits usually a capsule or follicle, occasionally a berry, drupe or rarely a schizocarp. Seeds usually numerous, often comose or winged, endosperm well developed with or without oil. Chemistry: usually producing iridoids or alkaloids or both, as well as cardioglycosides, not strongly tanniferous, seldom cyanogenic or saponiferous, various calcium oxalate crystals present in some parenchymatous tissue cells. Chromosome number: n = (5-)11(-17), rarely polyploid. Economics: members of this order are used for dyes, fiber, food, horticulture, medicines, rubber, and timber; some are extremely toxic (Strychnos is the source of the poison strychnine), and a few have become noxious weeds.

Of interest in this order is the plasticity of the gynoecium. Not only does the number of carpels vary from 2 to 8 (the bicarpellate condition being the norm), but also the extent to which they become modified is significant. For instance, the genus Theligonum L. (Rubiaceae) has a pseudomonomerous ovary while in Asclepiadaceae the ovary is both modified and also fused to the staminal whorl to form the complex gynostegium. The ovary may also vary from superior to inferior, or from being free (apocarpous) to united (syncarpous). Even within families this plasticity is of interest (Endress et al., 1983). Few other orders within the angiosperms exhibit such gynoecial plasticity, especially since this character is usually considered to be fairly conservative in terms of evolutionary change. This diversity in the ovary is reflected in the diversity of fruit types. The Gentianales, as modeled here, contains approximately 8% of all known extant angiosperm genera and approximately 5.6% of all known extant angiosperm species. Of the 30 families known to have more than 100 genera (Brummitt, 1992), 3 belong to the Gentianales: Rubiaceae (5th largest), Asclepiadaceae (9th largest), and Apocynaceae (15th largest). This means that half of the families constituting the order are among the largest known, making the order itself of some interest.

2. Suborder Gentianineae

Type family: Gentianaceae de Jussieu

a. Gentianaceae

A. L. de Jussieu, Gen. pl. 141. 1789. nom. conserv.

Type genus: Gentiana L.

Gentianaceae: Lindley, 1867: 612-614; Bentham, 1876: 799-820; Gray, 1878: 384-390; Gilg, 1895: 50-108; Bessey, 1915: 144; Wettstein, 1935: 920-922; Marais & Verdoorn, 1963: 171-243; Melchior, 1964: 408-410; Rendle, 1967: 463-468; Hutchinson, 1969: 546-552; Swift, 1974: 303; Mabberley, 1978: 223-224; Benson, 1979: 262-263; Cronquist, 1981: 871-873: Goldberg, 1986: 269-271; Airy Shaw, 1985: 483; Brummitt, 1992: 572-573.

Description: Taxonomic statistics: approximately 74 genera with about 1000 species. Distribution: global, but more common in temperate zones, even occurring in arctic and alpine regions. Habit usually herbs, rarely shrubby to treelike and rarely halophytic or saprophytic-achlorophyllous. Anatomy: xylem usually forming a continuous cylinder; internal intraxylary phloem usually present as a continuous ring or as isolated bundles at the pith margin; vascular bundles bicollateral. Leaves opposite, rarely verticillate (Curtia) or alternate (Swertia), simple, entire or reduced to scales (Bartonia and the achlorophyllous genera); scattered mucilage cells often present in the epidermis and mesophyll; petiole without stipules but occasionally with interpetiolar line. Stems often winged. Inflorescence solitary or cymose (usually dichasial, rarely monochasial), seldom racemose; bracteoles may be present even large and foliar or connate and adnate to the calyx (Zonanthus). Flowers usually showy, dioecious, rarely monoecious, usually actinomorphic, rarely dimorphic (Hockinia); pollination by insects. Calyx [2 (Obolaria)-] 4-5(-12)-merous; imbricate; usually with a short but well-developed tube; sometimes with basal nectary gland; lobes rarely reduced or suppressed to appear 2-cleft; rarely winged (Canscora) or inflated and elongated to enclose corolla (Prepusa). Corolla 4-5 [-12 (Blackstonia)]-merous; connate, usually with a short to long tube (Tachiadenus), usually plicate at sinuses; occasionally produced into spurs (Halenia); usually convolute, occasionally sinistrorsely contorted (Halenia) or imbricate (Obolaria); often with scales or nectary-pits within; rarely with coronas. Androecium: stamens 4-5(-12), isomerous, alternipetalous, sometimes strongly zygomorphic and adnate to corolla tube or mouth, rarely some staminodal or even obsolete (Hoppea): anthers tetrasporangiate, usually introrse (extrorse in Gentiana) and dithecal; dehiscing via a longitudinal slit or rarely by terminal pores (Exacum); occasionally connate (Voyria); occasionally connective with one to two apical glands (Sebaea); rarely with long thread-like projections (Urogentias). Gynoecium bicarpellate, syncarpous, superior, usually unilocular and oriented in the antero-posterior plane; surrounded by nectary-disk or distinct glands; placentation parietal (this occasionally bifid) or rarely free axile, seldom bilocular with axile placentation: style simple and terminal, with entire or bilobed, papillate stigma, rarely style wanting and stigmas decurrent on ovary (Lomatogonium); ovules usually numerous and anatropous. Fruit usually a septicidal capsule or rarely a berry (Chironia). Chemistry: family rarely cyanogenic or saponiferous, iridoids present (especially gentiopicroside), accumulating xanthones and aluminum, calcium oxalate crystals present in parenchymatous tissues. Chromosome number n = 5-13, rarely polyploid. Economics: Some genera of horticultural importance (Gentiana, Sabatia), occasionally of medicinal importance (Gentiana, Centaurium, and others), and the source of yellow-coloured dyes (Blackstonia).

There is growing support for the idea that the Potalieae of the Loganiaceae should be transferred to the family Gentianaceae (Fosberg & Sachet, 1980). In addition to other evidence, this transfer is supported by chemical data (Jensen, 1992). Wherever it is placed, the Potalieae should be seen as a taxon linking the Loganiaceae and the Gentianaceae; for further comments see under the Loganiaceae.

The Menyanthaceae and Saccifoliaceae have been included under the Gentianaceae by some authorities. However, the consensus is that, although the Saccifoliaceae is probably related to the Gentianaceae it be considered a distinct family (see also Mabberley, 1978). Consensus is that the Menyanthaceae be excluded from the Gentianales. Amongst other characters, its alternate leaves, collateral vascular bundles, valvate aestivation, embryology, and chemistry make it not only anomalous in the family Gentianaceae but also in the order Gentianales. For further comments see the discussion under these two families.

b. Saccifoliaceae

B. Maguire & J.M. Pires, Mem. New York Bot. Gard. 29: 230-245. 1978. Type genus: Saccifolium Maguire & Pires

Saccifoliaceae: Cronquist, 1981: 873-876; Brummitt, 1992: 658-659.

Description: Taxonomic statistics: 1 genus, 1 species (Saccifolium bandeirae). Distribution: South America, Venezuela, Guyana Highlands. Habit a pulvinate sub-shrub. Anatomy: stems with well-developed strands of internal phloem. Leaves alternate, simple, sessile, closely crowded toward the branch tips, margins recurved, extrorsely saccate-vaginate distally; without stipules; several small glandular bodies in leaf axils. Flowers solitary in leaf axils, actinomorphic and dioecious. Calyx (4-) 5-merous, forming a short basal tube. Corolla 4 (-5)-merous, imbricate, connate, with well-developed tube. Androecium: stamens isomerous, alternipetalous and adnate to corolla-tube; anthers tetrasporangiate and dithecal, dehiscing via longitudinal slits; connective forming a small apical point. Gynoecium bicarpellate, syncarpous, superior, bilocular (with apical part of partitioning imperfect); style terminal with short bilobed stigma; ovules numerous and anatropous. Fruit unknown. Chemistry: produces iridoids. Chromosome number unknown. Economics: presently none known.

This family was first described in 1978 and is thus missing from many modern classifications. The Saccifoliaceae is represented by a single species, Saccifolium bandeirae Maguire & Pires, which is restricted to isolated mountains in southern Venezuela. Cronquist (1981) maintained that, in combination, its unique leaves, imbricate corolla lobes, and bilocular ovary with axile placentation make the Saccifoliaceae incongruous in the Gentianaceae. He did, however, tentatively place it near this family. This approach seems to be followed by Dahlgren (1983). Thorne (1992) likewise maintains the Saccifoliaceae as a separate family, having at first (Thorne, 1983) suggested it form a subfamily, the Saccifolioideae, under the Gentianales. Takhtajan (1980) suggests (with reservation, 1983) that the Saccifoliaceae be included in the Gentianaceae, but he later (Takhtajan, 1987) seems to maintain it as a separate family (Brummitt, 1992). If the Potalieae of the Loganiaceae is accepted into the Gentianaceae (as suggested by Fosberg & Sachet, 1980), it would be hard to justify keeping the Saccifoliaceae separate, too. Benson (1979) and Goldberg (1986) seem not to have included the Saccifoliaceae in their accounts.

c. Loganiaceae

R. Brown ex Martius, as Loganieae in Nov. gen. Spec. 2: 133. 1827a. nom. conserv.

Type genus: Logania R. Brown

Loganiaceae: Endlicher, 1838: 574 in 1836-1840; Meisner, 1840:257 in 1837-1843; Lindley, 1867: 602-605; Bentham, 1876: 786-799; Gray, 1878: 391-392; Solereder, 1892: 19-50; Bessey, 1915: 144; Wettstein, 1935: 919; Verdoorn, 1963: 134-171; Melchior, 1964: 406-408; Rendle, 1967: 460-463; Hutchinson, 1969: 412-421; Jones, 1978:222-223; Benson, 1979: 263; Leeuwenberg & Leenhouts, 1980: 8-91; Cronquist, 1981: 865-867; Goldberg, 1986: 262-264; Airy Shaw, 1985: 681-682; Brummitt, 1992: 605.

Synonyms: Antoniaceae Hutchinson, 1959: 375; Hutchinson, 1969: 416-417; Airy Shaw, 1985: 76. Desfontainiaceae Endlicher, 1839: 669; 1841: 336. Plocospermataceae Hutchinson, 1959: 379; Hutchinson, 1969: 423-425; Airy Shaw, 1985: 921. Potaliaceae Martius, 1827a: 89, 133; Hutchinson, 1969: 415; Airy Shaw, 1985: 76. Spigeliaceae Martius, 1827b: 124, 132; Hutchinson, 1969: 418-419; Airy Shaw, 1985: 1088. Strychnaceae G. Don, 1837: 64; Hutchinson, 1969: 419; Airy Shaw, 1985: 113.

Description: Taxonomic statistics: approximately 20 genera with about 100 species. Distribution: pantropical, occasionally occurring in temperate regions but rarely found at high altitudes. Habit usually woody, often large trees, sometimes herbaceous, occasionally lianas, rarely epiphytic (Fagraea). Anatomy: nodes usually unilacunar, occasionally multilacunar (Fagraea); conspicuous internal, interxylary, or intraxylary phloem. Leaves opposite, rarely verticillate, rarely ternifolius; simple and usually entire, rarely toothed (Desfontainia); occasionally 3-5 nerved from above the base (Strychnos); usually with stipules (which are either cauducous or persistent, sometimes foliar, sometimes interpetiolar or these reduced to an interpetiolar line) or colleters. Stems occasionally spiny (Strychnos) or with tendrils. Inflorescence terminal or axillary, solitary (Desfontainia) or cymose (usually a dichasia), rarely spikes, racemes and panicles. Flowers usually not large and showy; usually actinomorphic, rarely with one calyx or petal lobe enlarged (Usteria); usually bisexual, occasionally unisexual or polygamo-dioecious; rarely involucrate (Coinochlamys): Pollination is various and may even include bats (Fagraea). Calyx (2-) 4-5-merous; usually actinomorphic or rarely not so (outer lobe large and petaloid in Usteria); connate or free. Corolla 4-5 (-16)-merous; imbricate, contorted or valvate; connate; rarely with corona (Scyphostrychnos). Androecium: stamens [1 (Usteria)-]4-5(-16), isomerous, alternipetalous, adnate to corolla: anthers bisporangiate or tetrasporangiate, dithecal and introrse; dehiscing via longitudinal slits; usually isostylous, occasionally heterostylous (Gelsemieae); usually free, rarely connivent (Gardneria). Gynoecium 2 (-3-5) (pentacarpellate in Desfontainia); syncarpous, sometimes apically apocarpous or completely apocarpous (Mitreola); superior to semi-inferior (Mitreola); locules usually as many as carpels (except Anthocleista which has a false septurn) but sometimes with imperfect partitions in their upper portions: style 1 [-2(Cynoctonum)], terminal: stigma capitate (Strychnos) or usually shortly lobed (Plocosperma) or twice dichotomously branched: ovules usually numerous, rarely as few as 1-4, placenta usually peltate or parietal, rarely basal. Fruit usually a septicidal capsule, rarely a circumscissile capsule (Spigelia), seldom a drupe (Neuburgia) or berry (Potalia) [in some the pericarp of the berry becomes thick and woody (Strychnos)]. Seeds sometimes winged (Antonia) or with a coma (Plocosperma); rarely solitary (Plocosperma). Chemistry: family producing iridoids and alkaloids (of indole & oxindole types) notable amongst these being the tryptophanic alkaloids, rarely saponiferous, sometimes accumulating aluminum. Chromosome number: n = 6-12. Economics: some genera are a source of indo-alkaloid poisons (Gelsemium and others) and glycoside poisons (loganin in Strychnos); a few are of horticultural importance (Fagraea) or used for timber (Fagraea).

Within the Gentianales the circumscription of the Loganiaceae has been problematic for a long time, but due to the recent reassessment of existing morphological and chemical data, as interpreted with (Bremer & Struwe, 1992) and without (Jensen, 1992) cladistics, a consensus is slowly being reached on how the family should be defined.

Following traditional thought (viz. Bentham, 1876; Solereder, 1892), Jones (1978) divided the family into two subfamilies and seven tribes. The Loganioideae contained the tribes Antonieae, Gelsemieae, Loganieae, Potalieae, Spigelieae, and Strychneae, and the monotypic Buddlejoideae contained the Buddlejeae. More recently, but also following traditional thought, Leeuwenberg & Leenhouts (1980) recognized 10 tribes within the Loganiaceae: the Antonieae, Buddlejeae, Desfontainieae, Gelsemieae, Loganieae, Plocospermateae, Potalieae, Retzieae, Spigelieae, and Strychneae. Hutchinson (1969), breaking with tradition, treated some of these as distinct families, namely, the Antoniaceae, Buddlejaceae, Loganiaceae, Potaliaceae, Spigeliaceae, and Strychnaceae, all of which he then placed in his order the Loganiales, except the Plocospermateae, which he placed in his Apocynales. Of Leeuwenberg and Leenhout's 10 tribes, the Buddlejeae, Desfontainieae, Plocospermateae, Potalieae, and Retzieae remain in dispute; these are discussed below.

Tribe Buddlejeae (= Family Buddlejaceae): The overall consensus is that this tribe should be given family status, the Buddlejaceae, and that its true affinities are with the Scrophulariales and not the Gentianales (Airy Shaw, 1985; Cronquist, 1988; Dahlgren, 1983; Takhtajan, 1983; Wagenitz, 1977). Thorne (1983) originally placed it as a separate family in the Gentianales but later (Thorne, 1992) included the Buddlejaceae in his Bignoniales, which contained the Scrophulariaceae and its allies. The consensus is that this taxon be removed from the Gentianales.

Tribe Desfontainieae: This monotypic tribe occurs along the mountainous backbone of South America from Costa Rica to Cape Horn. It has often been given family status, the Desfontainiaceae (Dahlgren, 1983; Takhtajan, 1969) within the Gentianales. Later, however, Takhtajan (1983) went on to include the family, with reservations, in the Loganiaceae. Cronquist (1968) included it as a tribe in the Loganiaceae. Thorne in 1983 considered it a separate subfamily under the Loganiaceae, then in 1992 revived its family status but transferred it to his order Hydrangeales in the Cornanae. Data suggest that Desfontainia Ruiz & Pavon may be abridging taxon between the Cornanae and Loganiaceae and these authors see merits in including it, at family level, in either the Gentianales as defined in this paper or Cornanae sensu Thorne ([is approximately equal to]Rosales of Cronquist or Saxifragales of Takhtajan). Bisset et al. (1980) confirm the isolated or incongruous morphology of Desfontainia as compared with the rest of the Loganiaceae sensu Leeuwenberg & Leenhouts (1980), although it should be mentioned that they opt for keeping it in this family.

Tribe Plocospermateae: Hutchinson (1969) gave this taxon family status, the Plocospermataceae, and placed it in his Apocynales rather than his Loganiales. In fact, although most authorities continue to include this monotypic tribe in the Loganiaceae [Thorne (1992), accorded it sub family rank, Plocospermatoideae], its almost intermediate position between the Loganiaceae and Apocynaceae is generally acknowledged (Takhtajan, 1969, 1983; Thorne, 1976); see also Figure 3. Cronquist (1981) went as far as including this taxon in the family Apocynaceae rather than the Loganiaceae but gave it no rank of its own. Dahlgren (1983) made no mention of the taxon. The Plocospermateae illustrates rather effectively the taxonomic and nomenclatural problems posed by intermediate or "bridging taxa." Such taxa are often encountered and are always useful in understanding relationships and evolutionary trends. The one rather noticeable exception to the almost universal consensus regarding the nearest relatives of the Plocospermateae seems to be Jensen (1992), who suggests, on chemical grounds, that this monotypic tribe be placed outside the Gentianales near the families Scrophulariaceae or Oleaceae; but this does not concur with strong morphological data. The Plocospermateae consists of a single species, Plocosperma buxifolium Bentham, which occurs in southern Mexico and in Guatemala.

Tribe Potalieae: There is growing anatomical and chemical evidence to suggest the transfer of the Potalieae to the Gentianaceae (Fosberg & Sachet, 1980; Jensen, 1992); however, even if this is not done, the close connection between this tribe and the Gentianaceae is at present not disputed. This tribe consists of three genera: Potalia Aublet, found in northern South America, Anthocleista Afzelius in tropical Africa and Madagascar, and Fagraea Thunberg in Australasia.

Tribe Retzieae (= Family Retziaceae): Now usually given its own family, the Retziaceae. This taxon is variously placed depending on how associated data is interpreted. Following Wettstein (1891), Cronquist (1988) and Hutchinson (1969) considered the Retziaceae to be a separate family related to the Solanaceae and its allies. Dahlgren et al. (1979; Dahlgren, 1983) suggested an association with the Stilbaceae in his order Scrophulariales. Thorne at first gave this taxon subfamily status under the Loganiaceae (1983); then, in 1992, following Dahlgren and co-workers' line of thought, he transferred this subfamily to the family Stilbaceae in his Bignoniales. Takhtajan (1980, 1983) placed this family in the Scrophulariales near the Scrophulariaceae and Buddlejaceae. Although there is no consensus for its exact placement, agreement seems to be that the tribe Retzieae/family Retziaceae should not be placed within the Gentianales.

Tribes Antonieae, Gelsemieae, Spigelieae, and Strychneae (i.e., the residue of tribes) form the core of the family--viz. Loganiaceae sensu stricto. However, there are a few anomalies even here. The wood anatomy and fruits of Neuburgia (of the Strychneae) are unique in the Loganiaceae but more common in the Apocynaceae. This may suggest (along with evidence from the Plocospermateae) that the evolution of the Apocynaceae itself is paraphyletic or at least reticulate. Neuburgia also has stipules similar to those of the Rubiaceae (Leeuwenberg & Leenhouts, 1980). The Loganiaceae sensu consentaneus would also include the Desfontainieae, Plocospermateae, and Potalieae. By consensus the Buddlejaceae and Retziaceae are excluded from the Gentianales altogether. Only Leeuwenberg and Leenhouts (1980) make a strong plea for their continued inclusion in the Loganiaceae.

Phylogenists may have to look again at the inclusion of some of the tribes included in the Loganiaceae by consensus, because studies by Bremer and Struwe (1992) indicate that the evolution of the family (minus the Retzieae and Buddlejeae) is paraphyletic. They also suggest that the tribe Gelsemieae is a possible sister group to the Rubiaceae, and they discount the genus Mitreola L. ex Schaeffer (tribe Spigelieae) as being a possible connecting group between the Loganiaceae and Rubiaceae as suggested by Thorne (1976). It should be mentioned as a warning that the name Mitreola as used in the older literature was surrounded by some nomenclatural and taxonomic confusion.

Within the Gentianales, the Loganiaceae is unusual because of its possible evolutionary connection to three separate families, viz. to the Gentianaceae through the Potalieae, to the Apocynaceae through the Plocospermateae, and to the Rubiaceae through either the genus Mitreola (Thorne, 1976) or tribe Gelsemieae (Bremer & Struwe, 1992). There may even be a connection with this family to the Scrophulariales sensu Cronquist through the Buddlejaceae. The Loganiaceae (which is primarily woody and generally less specialized for insect pollination) would seem to be near the central base from which the order Gentianales arose.

d. Rubiaceae

Jussieu, Gen. pl. 196. 1789. nom. conserv.

Type genus: Rubia L.

Rubiaceae: Hooker, 1873: 7-151; Gray, 1878: 208; Schumann, 1891: 1-156; Bessey, 1915: 161; Wettstein, 1935: 933; Melchior, 1964: 417-422; Hutchinson, 1969: 432-455; Rendle, 1967: 552-562; Kupicha, 1978: 257-259; Benson, 1979: 333; Swift, 1974: 338-339; Cronquist, 1981: 995-1000; Goldberg, 1986: 267-269; Airy Shaw, 1985: 1008-1010.

Synonyms: Aparinaceae Hoffmansegg & Link. Cephalanthaceae Rafinesque. Cinchonaceae Batsch; Lindley, 1867: 761. Coffeaceae Batsch. Cynocrambaceae Nees. Galiaceae Lindley. Gardeniaceae Dumortier. Guettardaceae Batsch. Henriqueziaceae (Hooker f.) Bremekamp; Airy Shaw, 1985: 548, Lippayaceae Meisner. Lygodisodeaceae Bartling. Naucleaceae Wernham; Airy Shaw, 1985: 779. Operculariaceae Dumortier. Stellataceae Dulac. Theligonaceae Dumortier; Airy Shaw, 1985: 1145.

Description: Taxonomic statistics: approximately 600 genera with about 7500 species. Distribution: global, mostly tropical, occasionally temperate, rarely arctic (Galium). Habit: herbs (annual or perennial), shrubs or trees, rarely succulent (Theligonum) or epiphytic (Hydnophytum), some genera myrmecophilous (Myrmecodia and others). Anatomy: sometimes with anomalous secondary growth, lacking internal phloem. Leaves usually opposite or appearing whorled (Gardenia), rarely alternate by suppression of one member of a pair at each node (Didymochlamys), or upper leaves alternate and lower leaves opposite (Theligonum); simple; usually entire, rarely lobed; interpetiolar stipules present; these may become foliar in nature, increase in number, become connate to encircle the stem or become reduced to an interpetiolar line, less often stipules petiolar (Theligonum); usually with colleters. Stems rarely spiny (Catesbaea), tuberous (Myrmecodia) or leafless (Phyllacantha). Inflorescence mostly cymose (usually a small dichasia or a much branched cymose panicle), rarely solitary or sessile (Theligonum). Flowers actinomorphic or very rarely zygomorphic (Dorothea); bisexual or rarely unisexual; nearly always epigynous; often heterostylic (Cinchona): pollination zoophilous (mainly entomophilous, especially Diptera, Hymenoptera, and Lepidoptera) or rarely anemophilous (Theligonum). Calyx (2-) 4-5-merous, lobes often small, at times even obsolete or sometimes even enlarged and showy (Mussaenda). Corolla (3-) 4-5 (8-10)-merous or absent in staminate flowers (Theligonum); aestivation valvate, imbricate or convolute; connate; usually actino-

morphic or slightly zygomorphic to bilabiate (Ferdinandusa); rarely staminal flowers forming a pseudanthium (Theligonum). Androecium: stamens (2-) 7-12 (-30) (6-30 in Theligonum), usually isomerous, alternipetalous, adnate to corolla tube or corolla mouth: filaments rarely connate (Neurocalyx) or strongly unequal: anthers tetrasporangiate, dithecal, dehiscing via longitudinal slits or rarely by pores (Argostemma); rarely connivent (Argostemma); sometimes connective produced apically. Gynoecium 2(-3-9)-merous, usually bicarpellate, syncarpous, inferior, rarely semi-inferior (Synaptantha) or superior (Gaertnera and Pagamea); ovary with as many locules as carpels (in which case placentation is axile or a derivative), seldom unilocular (Gardenia) (in which case placentation is parietal) or pseudomonomerous (Theligonum): style usually one and slender or styles free (Galium), rarely basilateral (Theligonum), or even missing (Tetralopha): stigma capitate or lobed: ovules 1 to many in each locule; anatropous to hemitropous, or campylotropus (Theligonum); solitary, basal, without integumentary tapetum. Fruit a (septicidal or loculicidal) capsule, berry, drupe or dry and indehiscent or a schizocarp, sometimes dicoccous (Galium), rarely a nutlet (Theligonum). Seeds usually without a coma, rarely with, or very rarely arillate (Isidorea). Chemistry: family accumulates aluminum and produces triterpenes, iridoids, anthraquinones, various alkaloids (especially indole, but also quinoline, isoquinoline, and purine), and proanthocyanins, sometimes tanniferous and saponiferous, seldom cyanogenic. Chromosome number: n = 6-17, usually 11, less often 9. Economics: the source of coffee (Coffea), emetin (Cephaelis), dyes (Rubia), quinine (Cinchona), and drugs (Cephaelis); many are of horticultural interest (Gardenia and many others); the young shoots of Theligonum are used as a vegetable.

Due to the accumulation of significant evidence, there is growing consensus amongst macrosystematists that the Rubiaceae is as closely related to the Apocynineae as are its traditional relatives, namely, the Loganiaceae and Gentianaceae (Wagenitz, 1959). This closeness has been recognized in the classifications of Dahlgren (1980), Takhtajan (1969, 1980, 1983), and Thorne (1976, 1992), who included the Rubiaceae in the order Gentianales. Phylogenists such as Cronquist (1981, 1988), Hutchinson (1969), and Kupicha (1978), although recognizing the evolutionary association of these taxa, preferred to place the Rubiaceae in a separate order, the Rubiales. Cronquist (1981) was reluctant to place the Rubiaceae in the Gentianales because he felt that it could with equal justification also be placed in the Dipsacales, with which he said it had affinities. Takhtajan (1969), although not going this far, nevertheless maintained that the Gentianales, including the Rubiaceae, had a common origin with the Dipsacales. Wagenitz (1977), on the other hand, believed the Dipsacales to be nearer the Scrophulariales than the Gentianales, while Bremer and Struwe (1992) believed the similarities of the Rubiaceae to the Dipsacales to be superficial. Interestingly, Thorne (1992) placed the Dipsacales in his Cornanae rather than his Asteranae (which includes the Gentianales). This is not at odds with a great deal of available data which tends to suggest that the Cornanae and Asteranae had some common starting point. Lee and Fairbrothers (1978), combining multifarious data with serological information, go so far as to suggest that the Rubiaceae is more closely related to the Cornaceae and Nyssaceae than to the Gentianales and Dipsacales. Unfortunately, their study left out the Loganiaceae as well as representative elements of the Solanales and Scrophulariales, all of which are pivotal to understanding relationships within and between these related groups. Cronquist's comments (1981: 1009) concerning the Lee and Fairbrothers study (1978) should also be noted. The vast mass of evidence, however, continues to tie the Rubiaceae to the Gentianales, and the main contribution of Lee and Fairbrothers's serological results is that it strengthens the growing support for considering the Cornanae to be more closely related to the Gentianales than was previously thought.

The results of Bremer and Struwe's (1992) cladistic study seem to indicate that the tribe Gelsemieae (Loganiaceae) could be the closest extant relative of the Rubiaceae. This does not agree with suggestions by Takhtajan (1980) and Thorne (1976) that the genus Mitreola (Spigelieae: Loganiaceae) is somewhat intermediate between these two families. It should also be noted that two other genera of the Rubiaceae, Gaertnera Lamarck and Pagamea Aublet, which have superior ovaries, had erroneously been placed in the Loganiaceae at one time (Cronquist, 1988). It is such data that probably led Thorne (1976) to say that "the Rubiaceae seem little more than inferior ovaried loganiaceous cousins," a statement with which anyone who has had to construct an analytical key to the two families would concur. However, some important differences (anatomical and pollen morphology in particular but also, to a degree, chemical) do exist between these two families and should not be undervalued.

The infrafamilial classification of the Rubiaceae is controversial (Kupicha, 1978) and nomenclaturally complicated (Darwin, 1976; see also Bremekamp, 1966; Robbrecht, 1988; Verdcourt, 1958). There is almost total agreement by present-day phylogenists and experts in the family that the following families should be considered synonymous with the Rubiaceae: Cephalanthaceae, Cinchonaceae, Coffeaceae, Gardeniaceae, Guettardaceae, Lippayaceae, Lygodisodeaceae, and Operculariaceae. In contrast, there has been some recent debate on the status of the families Henriqueziaceae, Naucleaceae, and Theligonaceae. However, consensus is that even these be included in the Rubiaceae. By consensus, another, usually separate family, the Dialypetalanthaceae, has been excluded.

Dialypetalanthaceae: Cronquist (1968, 1981), Dahlgren (1980), and Hutchinson (1969) placed the Dialypetalanthaceae near the Rubiaceae. Cronquist (1988) later changed his mind and placed the family in his Rosales. Goldberg (1986) and Stebbins (1974) suggested that the family be placed in the order Myrtales, while Dahlgren (1980), Takhtajan (1983), and Thorne (1983) placed the family in the Gentianales. However, present consensus seems to be that it is not part of the Gentianales. The fact that the Dialypetalanthaceae show characteristics found in the Myrtales, Cornales, and Gentianales suggests that it may represent a particularly archaic and somewhat isolated group (all its closest relatives now being extinct) with connections to all three of these modern orders.

Henriqueziaceae: This family was split off from the Rubiaceae by Bremekamp (1957); however (although its irregular corolla and lack of both endosperm and colleters do make it unusual within the Rubiaceae), his ideas are generally not followed by most modern phylogenists (Dahlgren, 1983; Hutchinson, 1969; Takhtajan with reservation, 1969; Thorne, 1992). Airy Shaw (1985) agreed with the separation of this family, stating that it was erroneously included in the Rubiaceae because it possessed stipules and that its closest relatives are the Bignoniaceae, Pedaliaceae, and Thunbergiaceae (this latter usually included in the Acanthaceae). On the other hand, Cronquist (1981) pointed out that the genus Gleasonia Standley tends to bridge the gap between Henriquezia Spruce ex Bentham and the rest of the Rubiaceae, making separation unfeasible. Thorne (1983) made something of a compromise by giving the taxon subfamily status, Henriquezioideae, under the Rubiaceae, but later (Thorne, 1992) sank it back into the subfamily Rubioideae. The two genera involved, Henriquezia and Platycarpum Humboldt & Bonpland, are found in northern tropical South America, in particular, Brazil.

Naucleaceae: This family, first proposed by Wernham in 1912, is usually accepted into the Rubiaceae by most phylogenists (Cronquist, 1981; Hutchinson, 1969; Takhtajan, 1969); however, none gives reasons for doing so. Others, such as Dahlgren (1983), Stebbins (1974), and Thorne (1992), make no mention of the taxon. Interestingly, members of this taxon have a slightly combretaceous appearance (Airy Shaw, 1985). The Combretaceae form part of the Myrtales, which seems to have had a distant but common origin with the Gentianales from a proto-cornalian-like ancestor. In light of the myrtalian connection of the Dialypetalanthaceae (often associated with the Rubiaceae), the possible myrtalian connection of the Naucleaceae (usually included in the Rubiaceae) could be significant. Alternatively, parallel evolution could be at work. Whichever is true, this group of 10 genera requires more thorough investigation.

Theligonaceae (Syn.: Cynocrambaceae): Wunderlich (1971), who completed a thorough revision of Theligonum L., suggested it be placed in the Rubiaceae in its own tribe, the Theligoneae. However, despite this work, the rank and taxonomic position of this taxon has continued to vary greatly. Cronquist (1981, 1988) treated this monotypic family as separate from the Rubiaceae but included it in his bitypic Rubiales. Dahlgren (1983) kept it as a separate family within the Gentianales, as did Takhtajan (1980, 1983), although Takhtajan did question the validity of this family status. Thorne (1976, 1992) included the Theligonaceae in the Rubiaceae, specifically sinking Theligonum into the subfamily Rubioideae. Bremer and Struwe (1992) report the Rubioideae to be monophyletic, but unfortunately they do not state whether they included the Theligonaceae (and therefore associated data) as a synonym under this taxon. Heathcote (1978) and, with reservations, Stebbins (1974) placed the Theligonaceae under the order Haloragales (along with Haloragaceae, Gunneraceae, and Hippuridaceae). Airy Shaw (1985) also mentioned the possible connection of this taxon to the Caryophyllidae, especially the family Portulacaceae, while Goldberg (1986) and Hutchinson (1969) actually placed it in the Caryophyllales. Historically, Bentham (1880) placed this genus in the Urticaceae. By consensus this taxon is now included in the Rubiaceae, although a good case could be made for retaining this as a separate family in the Gentianales allied to the Rubiaceae. Theligonum contains three species and has an odd, disjunct distribution: It is found on the Canary Islands in the Atlantic Ocean, in the Mediterranean, and then (after a gap of thousands of miles) in southwestern China and Japan. Theligonum is myrmecochorous.

3. Suborder Apocynineae

Benson ex Rosatti, J. Arnold Arbor. 70: 307-401. 1989a. Type family: Apocynaceae de Jussieu Apocynineae: Benson, 1979: 124-356. Discussion: The Apocynineae consists of the Apocynaceae and Asclepiadaceae (the latter family including the Periplocaceae). Within the Gentianales, the families of the Apocynineae are unique in being the only ones to have a totally apocarpous ovary. This is unusual because this suborder is strongly characterized by the synorganization of its flowers; in fact, the Asclepiadaceae has the most synorganized flowers within the dicotyledons. The Asclepiadaceae sensu Rosatti (1989a, 1989b) also stands out because of the numerical constraints placed on its flower and fruit; this should be contrasted with the other family in the suborder, the Apocynaceae, in which the flower has fewer constraints (Endress, 1990) and the fruit is diverse. The different floral, numerical constraints between these two otherwise closely related families is not often emphasized as a major disjunction between these two taxa.

Taxonomically, the Apocynaceae, Asclepiadaceae, and Periplocaceae have been treated in different ways by various authorities. Bullock (1956), Dyer (1975), Hutchinson (1969), and Schlechter (1905, 1924) have all maintained these as separate taxa at the family level, while Cronquist (1981), Dahlgren (1980), Rosatti (1989a, 1989b), and Takhtajan (1983) have included the Periplocaceae under the Asclepiadaceae as the subfamily Periplocoideae. Safwat (1962), Stebbins (1974), and Thorne (1976, 1992) proposed that all three be included in an expanded Apocynaceae--an older classificatory concept apparently with growing modern support (Rosatti, 1989a).

All these classifications recognize the differences between these three taxa; they differ only in the way they express these differences using the presently accepted system of hierarchy. Whereas workers such as Schlechter maximize the differences, other workers such as Thorne minimize them. While modern literature is cited here, this particular dilemma dates back to the 1800s (see comments by Rosatti, 1989a; Woodson, 1930), and, in fact, R. Brown (1810) himself stated that on this matter botanists are "almost equally divided." It is interesting that some 180 years later this can still be said to be true.

Robert Brown (1810), who proposed the Asclepiadaceae [thus splitting the Apocynaceae of de Jussieu (1789) into two almost equal taxa], considered both families to be part of a continual or natural series. It is this unity that Thorne (1976, 1992) is attempting to express in his classification. In 1989, when Rosatti proposed the suborder Apocynineae (following Benson, 1979), he did so to emphasize the close relationship between the Apocynaceae and Asclepiadaceae without compromising the few differences between them. However, at the family level, Rosatti's classification is congruent only if the taxa involved are monophyletic.

If the monophylism of these taxa, as suggested by Rosatti (1989a) and Safwat (1962) (viz. following the evolutionary line: Apocynaceae [right arrow] Periplocoideae [right arrow] Secamonoideae [right arrow] Asclepiadoideae) is true, then the integrity of the taxa is not in dispute and only the level at which they are accorded recognition is open to discussion. Therefore, at present, the debate is not centered on hard empirical data but on personal interpretation of the use of hierarchy and the way taxonomists mentally define taxa (Clayton, 1974; Loevtrup, 1987; McNeill, 1979; Stevens, 1990; Walters, 1961). However, if the taxa of the Apocynineae are paraphyletic, then the nature of the debate must change (Bremer & Wanntorp, 1978). The staminal, pollinial, and coronal data given by Safwat (1962) which has been widely and uncritically used by other workers is open to differing interpretation and does not necessarily support a monophyletic line from the Apocynaceae through the Periplocoideae and Secamonoideae to the Asclepiadoideae.

Although not the norm, some Apocynaceae and some Periplocoideae are characterized by successive meiotic divisions of the two pollen mother cells, as in the Asclepiadoideae, while in Secamone R. Brown it is always simultaneous. This could be argued as evidence for paraphyletic rather than monophyletic evolution amongst the taxa. Likewise, the differing locular maturation in the anthers of the Plumerioideae and Apocynoideae could support paraphyly. That is, the equal locular development TABULAR DATA OMITTED of the Plumerioideae could have given rise to the four-pollinial state in Secamone while the trend towards the underdevelopment of the two adaxial locules in the Apocynoideae (Demeter, 1922) could have resulted, if taken to its ultimate conclusion, in the two-locular state found in the Asclepiadoideae. In contradiction to this, however, Woodson (1930) points out that in Apocynum L., Poacynum Baillon, and Trachomitum Woodson (Apocynaceae) the partition of adaxial and abaxial locule pairs ruptures to give the appearance of a bilocular anther, a condition he says could suggest affinities to the bilocular Asclepiadaceae. The odd stigmatic morphology of Secamone is also not quite congruous with a monophyletic theory, for, despite Safwat's (1962) interpretation, this structure remains anomalous in the Apocynineae. Within the Asclepiadaceae sensu Cronquist (1981), the nature of corpusculum attachment to the style-stigma head in Secarnone (Kunze, 1991) is also anomalous.

Other datum, not quite agreeing with a monophyletic evolution for the group, is the upwardly directed sterile placenta margins of Secamone, a character apparently shared with Mandevilla Lindley of the Apocynoideae (Mandevilla is also interesting because it produces unusual disk-like glands that resemble the translator apparatus of the Periplocoideae). In most other Apocynaceae and Asclepiadaceae these sterile margins are downwardly directed (Safwat, 1962; Woodson & Moore, 1938). Also, both superior and sub-inferior carpels are found in the Apocynaceae, while in Secamone they are sub-inferior and in Asclepiadaceae they are superior. This conflicting data may also tend to argue in favour of a paraphyletic evolution.

The fact that the Periplocoideae, Secamone, and Fockea Endlicher all lack structures that could be precursors of the caudicles or translator arms found in the rest of the Asclepiadaceae also lays a monophyletic evolution open to some doubt. In fact, Kunze (1993), in a thorough study of the translator of the Periplocoideae and Asclepiadaceae, states: "The occurrence of functionally similar structures in the more highly evolved Asclepiadaceae is most probably a case of parallel evolution." Having said this, he then suggested a monophyletic evolution for the group (based on translator structure and ontogeny) as a whole, with the evolutionary line as follows: Periplocoideae [right arrow] Secamoneae [right arrow] Fockea [right arrow] Asclepiadaceae. However, this trend, although of considerable merit, does not fit well with pollinial data, in particular, the four-pollinial state of Secamone, which is anomalous in the line suggested.

Also, depending on interpretation, the jump from sticky tetrads to well-defined pollinia (with no intermediate stages known) could be considered quite large by some. Such an all-or-nothing character could support an evolutionary disjunction between these two taxa. Further, although the absence of the periplocaceous pollination spoon in Secamone can be interpreted as being lost through reduction (monophyly), it can also be interpreted as never having been present in the evolution of Secamone (paraphyly); the absence of even a vestigial remnant of this structure tends to argue in favour of the latter.
COPYRIGHT 1994 New York Botanical Garden
No portion of this article can be reproduced without the express written permission from the copyright holder.
Copyright 1994 Gale, Cengage Learning. All rights reserved.

Article Details
Printer friendly Cite/link Email Feedback
Author:Nicholas, A.; Baijnath, H.
Publication:The Botanical Review
Date:Oct 1, 1994
Previous Article:The role of seed coats in seed viability.
Next Article:Seed germination of the Australian desert shrub Eremophila (Myoporaceae).

Terms of use | Copyright © 2018 Farlex, Inc. | Feedback | For webmasters