4.2. Dimorphotheca

The current circumscription of the genus Dimorphotheca comprises three main taxonomic entities that were segregated by Norlindh, 1943). The first of these is the original, narrower circumscription of the genus, Dimorphotheca sensu stricto, defined by the production of dimorphic ray and disc fruits (e.g., Fig. 1k, l; taxa indicated by “(D)” after the taxon names in Fig. 4), and comprising only seven species, three of which are represented in our tree (Fig. 4; D. pluvialis, D. sinuata, and D. polyptera). The second entity is Osteospermum section Blaxium, placed by Norlindh (1943) in Osteospermum since all species produce only ray fruit, and comprising eight species. Three of these are included in our tree, viz. D. jucunda E.Phillips, D. ecklonis DC., and D. fruticosa (L.) DC, and indicated by “(OsB)” after the taxon names in Fig. 4. The third taxonomic entity is the former genus Castalis, comprising just three species and segregated from Dimorphotheca by Norlindh (1943) on the basis of sterile ray florets, a unique feature in Calenduleae . Castalis is represented in our tree by the single species C. tragus (indicated by “(Ca)” in Fig. 4). Lastly, D. pinnata (Thunb.) Harv. was transferred to Dimorphotheca by Nordenstam (2006). This is a morphologically unusual annual species with pinnatisect leaves and spiny ray fruits, placed by Norlindh (1943) in the monospecific section O. sect. Acanthotheca (DC.) T. Norl .. However, we were unable to include a specimen of D. pinnata in our tree, and it has never been examined for seed oil composition. Despite this lack of evidence regarding its generic position, D. pinnata most closely resembles the annual Dimorphotheca sensu stricto species in floral characters.

Several authors acknowledged strong similarities of involucre and floral characters, disc fruit structure, and phytochemistry, between Dimorphotheca sensu stricto and the other taxonomic entities described above (Candolle, 1838; Harvey, 1865; Norlindh, 1943). However, Norlindh (1943, 1978) was reluctant to expand the concept of Dimorphotheca to incorporate them as this would have removed any clear morphological synapomorphies for the genus. Norlindh (1978) instead discussed resurrecting the genus Blaxium Cass., clearly intending to maintain three separate genera for the species currently in Dimorphotheca . However, Nordenstam (1994b) considered these entities to be congeneric on the basis of a morphological, cytological and chemical parsimony analysis, and transferred both Osteospermum sect. Blaxium and Castalis to Dimorphotheca, thereby nearly tripling the number of species. Together with the reversal by Nordenstam (2006) of the earlier transfer of D. pinnata to Osteospermum by Norlindh (1943), this constitutes the currently accepted circumscription of Dimorphotheca (Manning and Goldblatt, 2012) . With the sole exception of D. polyptera, the enlarged circumscription of Dimorphotheca proposed by Nordenstam (1994b) is strongly supported in our tree, despite the lack of clear morphological characters defining the genus. Moreover, although only seven of the » 20 currently accepted species of Dimorphotheca are included in our analysis, these represent all three of the main historical taxonomic entities now included in the genus. The strongest un-contradicted synapomorphy for the genus is the presence of dimorphecolic acid in the seed oils, since this has been found in only twelve of many investigated species of Calenduleae, all of which are Dimorphotheca species (Smith et al., 1960; Earle et al., 1962; Barclay and Earle, 1965).

Dimorphotheca polyptera was included without question in Dimorphotheca sensu stricto by both Norlindh (1943) and earlier workers (e.g. Candolle, 1838; Harvey, 1865) because it has the defining characteristic of the genus, producing dimorphic fruits from both disc and ray florets. Norlindh (1943) noted however that D. polyptera differs from all other Dimorphotheca species by its short, yellow ray ligules. In this respect, it is more similar to many members of Osteospermum sensu amplo . The structure and size of the disc fruits produced by D. polyptera are similar to those of both G. bipinnatum and D. montana, being obovoidoblong rather than distinctly cordate, and their small size (3 — 4 mm long) is similar to those of both D. sinuata (3 — 4 mm) and G. bipinnatum (4 — 5 mm), but smaller than most Calenduleae disc fruits, which can be up to 20 mm long (e.g. in D. spectabilis). The ray fruits in D. polyptera are polymorphic, and Candolle (1837) distinguished three different shapes of the ray fruits, establishing a novel section Triplocarpae within Dimorphotheca to accommodate it. The winged ray fruits of D. polyptera are pigmented on the body, terete to (more commonly) trigonous, and with varying degrees of development of pale wing-like processes on the angles (Fig. 1i). Three-winged ray fruits are common in both Dimorphotheca and Osteospermum (Fig. 4), but they are rarely discolorous, and do not possess broad, pale wings divided into processes. This type of fruit is, however, found in one species of Osteospermum, viz. O. bolusii (Compton) Norl, which was placed by Norlindh (1943) and Manning and Goldblatt (2012) in Osteospermum sect. Trialatae Norl. While O. bolusii differs from D. polyptera in being a perennial shrublet with entire leaf margins, in having female-sterile disc florets, and in having monomorphic ray fruits, the two species have strong floral similarities, their ray ligules being short, narrow, and uniformly pale yellow. In features of the fruits, ray florets, and in phylogenetic placement, D. polyptera does not conform to Dimorphotheca and agrees most closely with species in Osteospermum sensu amplo, and particularly in sect. Trialatae. As far as we are aware, the phytochemistry of D. polyptera has never been investigated but we predict that it is unlikely to contain dimorphecolic acid. While elucidation of the closest relatives of D. polyptera within Osteospermum s.a. will require greater sampling, our analyses confirm that the species is phylogenetically distant from the remainder of Dimorphotheca using multiple accessions from across the geographic range of the species (Fig. 3; Table 2).

While the exact phylogenetic position of D. polyptera within Osteospermum is uncertain, its retention in Dimorphotheca clearly renders that genus non-monophyletic. Also, removal of D. polyptera from Dimorphotheca simplifies the morphological definition of the genus, since it now contains no species with short, narrow and uniformly pale yellow ray ligules. We accordingly transfer D. polyptera to Osteospermum based on its phylogenetic position, ray ligule morphology, and ray fruit structure, and provisionally assign it to O. sect. Trialatae. While D. polyptera is anomalous within Osteospermum s.a. in being the only known species with female-fertile disc florets, there is a precedent for such an anomaly in the genus Garuleum, where a single species produces disc fruits in addition to the usual ray fruits.