Abstract
Recent molecular studies (Plunkett et al., 1996b, 1997; Downie et al., 1999, 2001; Plunkett, 2001; Valiejo-Roman et al., 2002) on the family Apiaceae have shown that the existing classification system is largely artificial and that genera and tribes are in need of considerable re-arrangement. The fruit morphology and anatomy of 212 genera and 267 species of the family were studied. This study shows that a new interpretation of traditional morphological evidence will not only help to improve our understanding of the newly emerging molecular phylogeny of the Apiaceae, but also be useful for systematic studies of the family.
Fruit structure, the very wide commissure and thick mesocarp as is found in Araliaceae fruits especially, provides some support for the molecular systematic study (Plunkett, 2001) in which it was proposed to move Hydrocotyle L. from Drude’s (1898) Hydrocotyloideae Link. to be a sister of ‘core Araliaceae’. Fruit structural evidence from selected Araliaceae suggests that the Araliaceae and Apiaceae are not fundamentally different and that the two families, as currently circumscribed are linked by taxa sharing character states of both families. Hydrocotyle shows features of both Araliaceae and Mackinlayeae. Two genera, Myodocarpus Brongn. & Gris. and Mackinlaya F. Muell., should be moved from Araliaceae to respectively the Myodocarpeae R. Vig. and Mackinlayeae R. Vig., which cladistically are placed below the Apiaceae. The former has median wings and large vittae and the latter has crystals present in the innermost portion of the mesocarp adjacent to the endocarp. These features are usually absent in Araliaceae but may be present in Apiaceae. A study of fruit features of all other genera with laterally compressed mericarps of Drude’s (1898) Hydrocotyloideae (except for Hydrocotyle) suggests that it may be possible to move these taxa to Mackinlayeae because they share not only laterally compressed mericarps, but a similar crystal position and arrangement of ventral bundles (there are cells between the ventral bundles, but a true carpophore is absent), as with Mackinlaya.
In the present study the subfamily Mulinoideae [Plunkett et al. (2004) has now independently described this subfamily as the Azorelloideae Plunkett & Lowry subfam. nov. ined.] is proposed and used for Drude’s (1898) Mulineae DC. The Mulinoideae includes three groups, viz. the Klotzschia Cham., Azorella Lam. and Bolax Commerson ex Jussieu groups. The mericarps of the genera in the Klotzschia group may have five prominent ribs and a very wide commissure. In the Azorella group the lateral ribs are small and the marginal ribs may be prominent or winged and the commissure narrow. In the Bolax group the mericarps have more prominent lateral ribs or wings and the marginal ribs are small or undeveloped. The commissure is very narrow. Crystals may be present in the Klotzschia group. They are usually absent in the Azorella group and present in most genera of the Bolax group. These results partly support a molecular study (Plunkett, 2001), in which three groups: the Klotzschia, Azorella and Bowlesia groups, were indicated.
The subfamily Saniculoideae is unique in fruit structure. Intrajugal oil ducts are present and there are sometimes some small vittae scattered in the mesocarp. Crystals are dispersed in the mesocarp. Lagoecia L., which has vallecular and commissural vittae, should not be grouped in Saniculoideae but rather be positioned with Apioideae as was indicated by a molecular study (Plunkett et al. 1996; Downie et al., 2001; Valiejo-Roman et al., 2002). Several African genera, including Lichtensteinia Cham. & Schltdl., Marlothiella H. Wolff, Arctopus L., Polemanniopsis B.L. Burtt and Steganotaenia Hochst, have only large intrajugal oil ducts as in Saniculoideae (Steganotaenia also has some small vittae scattered in the mesocarp). A slightly lignified endocarp may be present in most genera (except in Marlothiella) and crystals are present in Lichtensteinia and Marlothiella as in Saniculoideae and absent in other genera. It is proposed that these taxa be moved from Apioideae to near the Saniculoideae as different groups: the Arctopus, Polemanniopsis and Lichtensteinia groups (the latter includes Marlothiella), in addition to the Saniculoideae sensu stricto. This corresponds to a large extent with DNA sequence data (Downie & Katz-Downie, 1999; Plunkett, 2001).
The ‘basal Apioideae’ is morphologically poorly defined but can be recognized by a combination of plesiomorphic characters, such as the woody habit, heteromorphic fruits, deciduous leaves and poorly developed carpophores. The most reliable difference between this group and the genera of ‘remaining Apioideae’ is the dispersed crystals in the mesocarp as in Saniculoideae. In ‘basal Apioideae’, DNA studies (Downie et al., 2001; Plunkett, 2001) indicated that the Heteromorpheae Downie & Watson include five endemic African genera: Anginon Raf., Dracosciadium Hilliard & B.L. Burtt, Glia Sond., Heteromorpha Cham. & Schltdl. and Polemannia Eckl. & Zeyh. The fruit study shows that in these genera laterally-winged mericarps (of heteromorphic fruits) and lignified endocarps and mesocarps may be present as well as crystals dispersed in the mesocarp and ventral bundles usually becoming free. Some other endemic African and Madogascar genera, viz. Andriana B-E. van Wyk, Cannaboidea B-E. van Wyk, Pseudocannaboides B-E. van Wyk, Pseudocarum C. Norman and Tana B-E. van Wyk, should also be placed in, or close to, the Heteromorpheae because they have crystals dispersed in the mesocarp and ventral bundles usually becoming free as in some Heteromorpheae. In the Annesorhiza Cham. & Schltdl. group (Annesorhiza and Chamarea Eckl. & Zeyh.), laterally-winged fruits (in heteromorphic fruits), a distinct hypoderm and a lignified endocarp may be present. The monotypic genus Molopospermum W.D.J. Koch is related to Annesorhiza (Downie, personal communication to B-E. van Wyk and P.M. Tilney); it has lateral wings on one of the two mericarps and an abundance of crystals in the mesocarp. The mericarps of the Bupleureae Spreng. may have irregular vittae as in Saniculoideae sensu stricto. All these fruit features are present in the ‘basal Apioideae’ and may be found in Saniculoideae and related groups.
Remaining Apioideae differs from the genera of ‘basal Apioideae’ in the absence of dispersed crystals in the mesocarp and various other characters that occur alone or in combination (see above). Eight tribes (including some subtribes) and seven additional clades in ‘remaining Apioideae’ were defined by Downie et al. (2001). The circumscription of some of the tribes, subtribes and clades distinguished by Downie et al. (2001) on the basis of DNA data receive support from fruit structure. These taxa include tribes Aciphylleae and Echinophoreae, subtribes Daucinae and Torilidinae, and clades the Arracacia, Conioselinum and Heracleum clades.
The present study shows that Drude’s (1898) system has significant taxonomic value, although many problems are present in his Hydrocotyloideae and Apioideae. These problems may have arisen because some fruit characters were not available, were overlooked, ignored or over-emphasised. Although some of Drude’s groups will be retained in a new system, a clear break from the Drude system is indicated by molecular systematics. This study suggests that taxonomically important fruit characters are wing configuration, presence or absence of intrajugal oil ducts and vittae and their size, presence or absence of crystals and their positions, possible lignification of the endocarp, and the arrangement and position of ventral bundles that may or may not form free carpophores. Therefore, evidence provided by fruit structure, if properly interpreted, can help to support and strengthen the newly emerging molecular phylogeny of the Apiaceae and facilitate understanding of the evolution of taxa in the family.
Prof. B.E. van Wyk