Home at Last II: Gerbera hieracioides (Kunth) Zardini (Mutisieae, Asteraceae) is really a Chaptalia

Abstract Gerbera hieracioides (Kunth) Zardini of the Gerbera-complex (Mutisieae, Asteraceae/Compositae) is distributed in Ecuador and Peru. This perennial herb was first named as Onoseris hieracioides Kunth and was later recognised as Trichocline hieracioides (Kunth) Ferreyra. Now it is generally treated as Gerbera hieracioides (Kunth) Zardini but it has never been included in any section of Gerbera. In this study, the position of Gerbera hieracioides is assessed based on morphology and a molecular phylogeny that includes G. hieracioides and 28 other species from the Gerbera-complex. Morphologically, G. hieracioides bears leaves with the adaxial epidermal surface without stomates but with soft thin trichomes, bracteate scapes, trimorphic capitula and inner ray florets with the corolla shorter than the style. These characters suggest that the species is most closely related to Chaptalia rather than to Gerbera or Trichocline. Furthermore, the phylogenetic results based on two nuclear (ITS and ETS) and two chloroplast (trnL–trnF and trnL–rpl32) sequences strongly support the placement of G. hieracioides nested within Chaptalia. As both morphological characters and the molecular phylogenetic results support the transfer of G. hieracioides to Chaptalia, this enigmatic taxon is recognised as Chaptalia hieracioides (Kunth) X.-D. Xu & W. Zheng.


Introduction
this study. Most of the specimens were sampled from the United States National Herbarium (US) of the Smithsonian Institution (Tables 1, 2).
Adaxial leaf epidermal morphology. Lamina (0.5-1.0 cm 2 ) were placed with the adaxial side exposed on carbon tape over stubs for the scanning electron microscopy (SEM), without soaking the material in different solutions prior to SEM. The stubs bearing leaves were treated with gold-palladium to 16.6 μm thickness and were examined under a Philips XL-30 scanning electron microscope at the SEM Lab of the National Museum of Natural History (NMNH). The 22 samples were subsequently observed and photographed under SEM. Images of the leaves were captured using the proprietary software associated with the Philips SEM. Images of at least 15 different areas of the adaxial leaf surface were captured.
Floret morphology. The florets and scapes of 20 herbarium specimens were examined in the United States National Herbarium, Smithsonian Institution, using an optical microscope.
DNA extraction, amplification and sequencing. The molecular work was performed in the Laboratory of Analytical Biology (LAB) of NMNH (Smithsonian Institution). DNAs of 16 samples (15 species, including two samples of Gerbera hieracioides) were extracted using the AutoGen. Herbarium leaf samples, along with 1.0 and 2.3 mm diameter beads, were dipped in liquid nitrogen then immediately shaken for 30 seconds at 18000 rpm. About 500 ml of CTAB was added to the tubes, vortexed and incubated overnight (500 rpm at 45 °C). Then 300 μl of the supernatant was transferred to an AutoGen plate. AutoGen was run according to the manufacturer's default settings (AutoGen, Inc., Holliston, MA, USA).
The cycle sequencing programme was 30 cycles of 95 °C for 30 s, 50 °C for 30 s and 60 °C for 4 min. The resultant product was sephadex filtered and sequenced through an ABI 3730 automated sequencer (Applied Biosystems, Foster City, USA). The PCR reactions were performed in a Veriti PCR Thermal Cycler. The amplification protocols for all markers are summarised in Table 3. Sequences were aligned by using MAFFT (Katoh and Standley 2013) using Geneious 10.0.9. (Biomatters Ltd., Auckland, New Zealand) and checked manually. A total of 57 newly generated sequences from the 16 samples were deposited in GenBank (Table 2).     A total of 37 sequences of 16 species were retrieved from NCBI for the related taxa within the tribe Mutisieae (Table 2). Phylogenetic relationships were inferred based on the concatenated ITS+ETS+trnL-rpl32+trnL-trnF data with MrBayes v. 3.2.2 (Ronquist et al. 2012) by using the substitution model of GTR based on the best-fitting model determined using jModelTest 2.1.6 (Posada 2008), the chain length of 10,000,000, rate variation of gamma, gamma categories of 4, heated chains of 4, heated chain temp. of 0.2, subsampling freq. of 200 and burn-in length of 100,000. Tracer v. 1.5 (Rambaut and Drummond 2009) was used to confirm that the effective sample size (ESS) for all relevant parameters was > 200. After discarding the trees as burn-in, a 50 % majority-rule consensus tree and posterior probabilities (PP) for node support were calculated using the remaining trees.

Results
Adaxial leaf epidermal morphology. The results of the SEM work (Table 1) showed that the two tested samples of Gerbera hieracioides have no stomates but have soft, thin and appressed trichomes on the adaxial leaf surface ( Figure 1G). These adaxial leaf morphological traits differ from the Gerbera species: (1) they are different from Gerbera sections sampled [sect. Lasiopus (4 species), sect. Piloselloides (2 species) and sect. Pseudoseris (1 species)] which have stomates and stiff, straight, upright trichomes. Figure 1 has images of one sample for each section: G. ambigua (Fig. 1A), G. piloselloides (Fig. 1B) and G. perrieri (Fig. 1D), respectively. (2) they are different from the members of Gerbera sect. Gerbera which have stomates and soft, thin and appressed trichomes. Three species from South Africa were examined and represented by G. crocea (Fig. 1C). (3) they are different from the Asian Gerbera sect. Isanthus which have no stomates and no trichomes based on this study of five species of sect. Isanthus that were examined in the study and are represented by G. maxima (Fig. 1E): the authors' observations agree with Lin et al. (2008) for the Asian species G. delavayi. Additionally, the morphological traits of G. hieracioides differ significantly from those of the Trichocline species, which have many stomates with guard cells sunken on the leaf surface, illustrated by T. catharinensis (Fig. 1H). However, the two tested G. hieracioides samples share the same adaxial leaf epidermal characters such as soft, thin and appressed trichomes, epidermal cell shape and striations and absence of stomates, with the three examined Chaptalia species, as represented by C. pringlei (Fig. 1F). Therefore, based on the adaxial leaf epidermal morphology, G. hieracioides is most closely related to Chaptalia rather than to Gerbera or Trichocline.
Scape and floret morphology. The results (Table 1) showed that the two examined samples of Gerbera hieracioides have bracteate scapes and trimorphic capitula which have the inner rays with corollae shorter than the styles (Fig. 2G, H). The above morphological traits also differ from those of the Gerbera species: (1) Gerbera sect. Lasiopus, sect. Piloselloides and sect. Pseudoseris have ebracteate scapes and trimorphic capitula and the inner rays have corollae as long as the styles or longer. Gerbera jamesonii (Fig. 2A) and G. ambigua (Fig. 2B) belong to sect. Lasiopus and G. cordata (Fig. 2C) for sect. Piloselloides.
(2) they are different from Gerbera sect. Gerbera and sect. Isanthus, which have bracteate scapes but dimorphic capitula without inner rays of florets. Three South African species and five Asian species were examined and are illustrated by G. crocea (Fig. 2D), G. nivea (Fig. 2E) and G. gossypina (Fig. 2F). The two tested G. hieracioides samples share the traits of bracteate scapes and trimorphic capitula which have inner rays with corollae shorter than the styles with the three tested Chaptalia species, represented by C. meridensis (Fig.  2I) and C. mandonii (Fig. 2J). Therefore, based on the scape and floret morphology, G. hieracioides should be best considered as a species of Chaptalia rather than Gerbera.
Phylogenetic analysis. The MrBayes analysis of the combined nuclear markers and two plastid genes showed four clades of the sampled species of the Gerbera-complex, all with a strong biogeographic signal (Fig. 3): (1) the African and Australian species of the Gerbera complex (African Gerbera species are sister to the Australian Amblysperma), (2) the American genus Chaptalia and the South American Gerbera hieracioides, (3) the Asian Gerbera species and (4) the South American genus Trichocline. However, there is no well-supported resolution amongst the first three clades mentioned above, so no conclusions can be made about the monophyly of Gerbera at this time.
Both samples of Gerbera hieracioides were nested within the Chaptalia clade. Gerbera hieracioides is sister to Chaptalia pringlei; then the G. hieracioides-C. pringlei clade is sister to the other Chaptalia species with strong support (posterior probability of 1.00). Therefore, the molecular data also support the placement of G. hieracioides in Chaptalia.

Discussion
The molecular phylogeny of the Gerbera-complex showed that G. hieracioides did not group with Trichocline ( Fig. 3) but was nested inside Chaptalia. Furthermore, the leaf adaxial epidermis of G. hieracioides has no stomates, while that of Trichocline usually has many stomates (Fig. 1). In addition, Katinas (2004) presented a key to distinguish the genera of the Gerbera-complex and Gerbera and Chaptalia were found to share the same trait of achenes rostrate at the apex but this is not found in Trichocline.
The confusion about the placement of Gerbera hieracioides is no doubt the result of the morphology falling between that of Gerbera and Chaptalia. A good case concerning this point is the inner ray florets of the trimorphic capitula: Gerbera has a corolla as long as the style or longer and the staminodes are present, whereas Chaptalia has the corolla shorter than the style and without staminodes (Katinas 2004). As for G. hieracioides, the inner ray florets have moderately reduced stamens (Fig. 2G, H) which are different from both Gerbera and Chaptalia species. Although the stamen morphology of G. hieracioides is not identical to Chaptalia, their moderately reduced corollae (Fig. 2G, H) are similar to those of Chaptalia rather than those of Gerbera, according to Katinas (2004). Furthermore, the characters of leaf adaxial epidermis of G. hieracioides including the lack of stomates and the presence of soft thin trichomes, as well as bracteate scapes and cell shape and striations, all suggest that the species is closest to Chaptalia. Additionally, Hansen (1990) stated that the achene hairs of G. hieracioides are sub-inflated with a lower L/W-ratio than that of Gerbera. Therefore, the morphological data support the transfer of G. hieracioides to Chaptalia that was consistent with the molecular phylogeny (Fig. 3) based on both nuclear ITS and ETS and chloroplast trnL-trnF and trnL-rpl32. This transfer is in agreement with the geographic distribution (Fig. 3), because G. hieracioides is from South America and all the other Chaptalia species are from the New World (Nesom 2004b(Nesom , 1995. Chaptalia is a New World genus and contains about 70 species in the Americas . Although there are partial regional treatments, there is no comprehensive monograph of the genus (e.g. Burkart 1944, Cabrera and Nesom 2003, Nesom 2004a. Hansen (2006) argued that the most significant problem of the Gerbera-complex is the lack of a revisionary treatment of Chaptalia and argued for further studies to test whether Chaptalia is monophyletic. In the molecular analysis (Fig. 3), the nine Chaptalia samples (including G. hieracioides) grouped into two wellsupported clades. This result indicates that Chaptalia seems to be monophyletic when G. hieracioides is included. Chaptalia is typically characterised by differentiated and reduced rays (Hansen 1990): the inner ray florets with corolla strongly reduced, filiform (irregularly tubular, ligulate or bilabiate), shorter than the style and without staminodes (Katinas 2004). The inner ray florets of G. hieracioides with moderately reduced corollae and stamens suggest that the inner ray florets of trimorphic capitula may be a key morphological character for the further revisionary treatment of Chaptalia. As for Gerbera, this study showed that it falls into two distinct clades, one from Africa which is the sister group of the Australian genus Amblysperma and the other contains all the Asian Gerbera (Fig. 3). However, the two Gerbera clades are in a trichotomy with the Chaptalia clade. It is clear that, based on the sampling, the Asian taxa may be best separated out into a separate genus then Amblysperma is the sister genus of African Gerbera. If the two clades of Gerbera form a single clade, then Amblysperma will most likely be nested within that clade. The decision must wait for ongoing studies using additional data. However, it is clear that Gerbera hieracioides should be considered within Chaptalia.

Conclusions
The placement of Gerbera hieracioides within Chaptalia is strongly supported by both the molecular sequence data (two nuclear markers ITS and ETS and two chloroplast markers trnL-trnF and trnL-rpl32) and the morphology of the scape, capitula and the leaf adaxial epidermal surface. Therefore, Gerbera hieracioides has been transferred to Chaptalia and it is recognised as Chaptalia hieracioides (Kunth) X.-D. Xu et W. Zheng.