Molecular phylogeny helps to delimit Plectranthus hadiensis from its related morph occurring in Sri Lanka

: Plectranthus hadiensis is an important medicinal plant in Sri Lanka. It was considered a separate species, P. zeylanicus , endemic to the island until its inclusion, as P. hadiensis var. tomentosus, together with morphs from southern Africa in the revised species concept of P. hadiensis . However, there are morphological, chemical, and therapeutic differences between the African and Sri Lankan morphs . We used eight molecular markers in a phylogenetic study to clarify the species concept of P. hadiensis and to investigate whether it should include the Sri Lankan morph. We examined the position of the two P. hadiensis morphs in relation to eight other Plectranthu s species. The maximum likelihood tree revealed three clades: a weakly supported clade including P. calycinus, P. glabratus, P. fruticosus, and P. malabaricus ; a highly supported clade including P. amboinicus and African and Sri Lankan specimens of P. hadiensis ; and a highly supported clade formed by P. barbatus , P. caninus , and P. hadiensis var. tomentosus . The African P. hadiensis specimens form a highly supported subclade sister to a subclade containing the Sri Lankan P. hadiensis , suggesting that the subclades correspond to either two sister species or two subspecies. We propose that they are more likely to be sister species given the differences in morphology, chemistry, and chromosome number .


INTRODUCTION
Plectranthus hadiensis (Forssk.) Schweinf. ex Sprenger (Lamiaceae) is a medicinal plant whose native range extends from southern and eastern Africa to the southern Arabian Peninsula (Codd, 1985). It also occurs in Sri Lanka (Thwaites, 1864;Trimen, 1895;Cramer, 1981) and is known as iriweriya in Sinhala and valakan in Sanskrit (Jayaweera, 1981). In revising the species concept of P. hadiensis, Codd (1985) recognised three separate varieties of P. hadiensis and included the Sri Lankan morph in var. tomentosus, which he named P. zatarhendi var. tomentosus in an earlier treatment (Codd, 1975;Fig. 1a, b). Until Codd's revisions (1975Codd's revisions ( , 1985, the morph occurring in Sri Lanka was thought to be a separate endemic species, P. zeylanicus Benth., first described by Bentham (Labiatarum Genera et Species 36, 1832) based on the type specimen from the island. While maintaining its endemicity, Cramer (1978Cramer ( , 1981 reclassified the Sri Lankan morph as Coleus zeylanicus (Benth.) L.H.Cramer based on fused stamens, a trait originally used by Bentham to distinguish Coleus Lour. from Plectranthus L'Hér. Both Thwaites (1864) and Trimen (1895) found P. zeylanicus only as a cultivated plant in Sri Lanka and the latter did not believe that it grew wild on the island. Codd (1985) speculated that the Sri Lankan plant was an introduction from southern Africa where P. hadiensis var. tomentosus occurs naturally in dry woodland and rocky grassland.
In a molecular, morphological, and phytochemical phylogenetic analysis of the Ocimae, Paton et al. (2004) found that the Plectranthus species in their study separated into two clades, one clade that included Plectranthus species previously placed within Coleus and the genera Pycnostachys Hook., Holostylon Robyns & Lebrun, and Anisochilus Wall. ex Benth., and a sister clade including the remaining Plectranthus species and Tetradenia Benth., Thorncroftia N.E.Br., and Aeollanthus Mart. ex Spreng. All species in the Coleus clade had a sigmoid corolla tube and a horizontal anterior corolla lobe. The corolla tube of the Sri Lankan morph of P. hadiensis is distinctly sigmoid with a horizontal lower lip (Fig. 1c). In the South African morph, Codd (1985) described the corolla tube as bent, without specifying the degree of geniculation. In addition, the Coleus clade displayed fusion of all stamens, although in some species within the clade this trait was lost (Paton et al., 2004). In the Sri Lankan morph of P. hadiensis, Cramer (1978) described the filaments as fused somewhat less than half-way along their length. However, in Codd's (1985) description, the stamens of P. hadiensis are free to the base. There are also dissimilarities in leaf pubescence of the Sri Lankan and South African morphs of P. hadiensis ;Cramer (1981) described the leaves of the Sri Lankan morph as sparsely hirtellous (see Fig. 1d), whereas Codd (1985) used the densely tomentose nature of leaves of P. hadiensis var. tomentosus (Benth.) Codd and var. hadiensis as a diagnostic trait to distinguish them from var. woodii (Gürke) Codd.
The Sri Lankan morph of P. hadiensis has many therapeutic uses in Ayurveda and folk medicine, mainly for treatment of gastrointestinal disorders such as diarrhoea and dysentery (Jayaweera, 1981;Mehrotra et al., 1989;Arambewela and Wijesinghe, 2006;Waldia et al., 2011). In contrast, in South African Zulu medicine P. hadiensis var. tomentosus is employed as an enema (Hutchings, 1989(Hutchings, , 1996. Given that the Sri Lankan morph is an ingredient in over 20 Ayurvedic preparations (Arambewela and Wijesinghe, 2006), its correct identification is important. The differences in morphology, chemistry, and therapeutic use of the two morphs suggest that the species concept of P. hadiensis needs further consideration. Therefore, we undertook a molecular phylogenetic analysis using eight chloroplast markers to ascertain whether Codd's delimitation of P. hadiensis should include the Sri Lankan morph of the species.

Sampling, DNA extraction, and PCR amplification
In this study, several accessions of Plectranthus were included. Accessions for ten species, including one for P. zeylanicus, were obtained from GenBank (Table 1). DNA of one P. hadiensis accession (DNA Bank ID: 19767) was obtained from Royal Botanic Gardens, Kew, DNA Bank (apps.kew.org/dnabank/, last accessed 2018-04-25). Genomic DNA was extracted from c. 20 mg of silica geldried (Chase and Hills, 1991) leaves of four P. hadiensis samples from Sri Lanka and one P. hadiensis herbarium specimen (Herbarium ID: K000468025, provided by Kew Herbarium) using the DNeasy Plant Mini Kit (Qiagen, Hilden, Germany) following the manufacturer's protocol. To avoid degradation, material was frozen in liquid nitrogen and then ground to a fine powder using glass beads. Detailed information on accessions can be found in Table 1.

Sequence alignment and phylogenetic analyses
Sequences were assembled and edited using Geneious (version 8.0.5, Kearse et al., 2012). The alignment of sequences was performed in Geneious using the MAFFT plugin and inspected manually with BioEdit v7.0.4. Unsequenced regions were coded as missing data in the combined matrix. To infer phylogenetic relationships, maximum parsimony (MP) and maximum likelihood (ML) analysis were performed. MP analyses were conducted in PAUP version 4.0a149 (Swofford, 2016). For each data set, heuristic searches were conducted using 1000 replicates of random addition sequence, tree-bisection-reconnection (TBR) branch-swapping, and 'keeping multiple trees' (MulTrees) but saving only 20 trees per replicate. Clade support was estimated by the bootstrap (Felsenstein, 1985) with 1000 replicates, TBR branch swapping, and simple addition sequence. To explore the variability of each marker, nine matrices were analysed with MP: (1) ITS, (2) matK, (3) rps16, (4) rpoC1, (5) rpoB, (6) trnT-trnL-trnF, (7) ndhC-trnV, (8) rpl32-trnL, and (9) all regions combined. Information about the alignment characteristics and number of variable and potentially parsimony informative sites were obtained for each marker from PAUP. ML analysis was conducted using the combined data only. An ML rapid bootstrap analysis (1000 replicates) with search for bestscoring ML tree in one run was conducted in RAxML v8.2.0 (Stamatakis, 2014). The general time reversible (GTR+GAMMA) model with six substitution types (one for each pair of nucleotides) and gamma-distributed rate variation across sites and a proportion of invariable sites was used for the analysis. Mentha longifolia obtained from GenBank was used as the outgroup. Trees were visualized and edited in FigTree v1.4.1 (http://tree.bio.ed.ac.uk/ software/figtree/, last accessed 2018-06-14).

RESULTS AND DISCUSSION
Our main aim in this study was the clarification of the species concept of P. hadiensis, specifically investigation of whether it should include the morph of P. hadiensis from Sri Lanka which differs from the African morph not only in its morphology and chemistry, but also in its therapeutic uses. We also examined the position of the two P. hadiensis morphs in relation to other Plectranthus species.
We sequenced both plastid and nuclear regions for our phylogenetic analysis. Of the plastid regions, rpl32-trnL exhibited the highest percentage of variable characters (12.6%). The nuclear ITS region was the most informative region with 85 (9.2%) potentially parsimony-informative sites. All eight markers combined resulted in a 7525 bp alignment and included 526 variable characters (7%) of which 158 (2.1%) were parsimony informative. Further parsimony statistics are given in Table 3.
With respect to P. hadiensis accessions, the two African specimens form a highly supported subclade ( Fig. 2; clade 2, green: 100, 96). This subclade is sister to a subclade containing the four specimens of P. hadiensis from Sri Lanka newly sequenced in this study and P. zeylanicus obtained from GenBank ( Fig. 2; clade 2, red: 100, 99). The accessions of Sri Lankan P. hadiensis and P. zeylanicus form a polytomy. Our phylogenetic trees suggest that the subclade containing African P. hadiensis and the subclade containing Sri Lankan P. hadiensis specimens (and P. zeylanicus) may correspond to either sister species or subspecies. However, based on the morphological and chemical differences between the African and Sri Lankan P. hadiensis morphs we discussed earlier and the differences in chromosome number, we propose that it is more likely that the two subclades correspond to sister species rather than subspecies. P. zeylanicus is a tetraploid with a haploid chromosome count of n = 14 (2n = 28; x = 7; Thoppil, 1993) whereas African P. hadiensis is a hexaploid with n = 21 (2n = 42; x = 7; de Wet, 1958). Among African species of Plectranthus, the commonest chromosome number for  Thoppil (1993). All other chromosome counts are from the Chromosome Counts Database (CCDB, version 1.46, Rice et al., 2015). As the Sri Lankan morph of P. hadiensis was formerly known as P. zeylanicus, we have assumed that the chromosome count for the Sri Lankan specimens of P. hadiensis is the same as that for P. zeylanicus. Sequences obtained from GenBank are indicated with an asterisk.
the genus is 2n = 28, with a basic number of 7 (de Wet, 1958;Morton, 1962;see also Fig. 2 Rice et al., 2015) ranges from n = 7 to n = 42, with a median of n = 14 (44% of the species). Whole genome duplication (polyploidy) has an important role in the evolution of angiosperms (Soltis et al., 2014). Recent phylogenetic analysis suggests that polyploidy is a key mechanism for cladogenesis and for speciation within plant genera (Zhan et al., 2016). Polyploidy may also introduce phenotypic and ecological diversity in plant lineages leading to niche differentiation and enhanced responses to environmental stress (Soltis et al., 2014). It is likely that whole genome duplication has played a role in speciation within the large genus of Plectranthus (currently 300 species, including Coleus; Stevens, 2001 onwards).
Interestingly, the GenBank accessions of P. hadiensis var. tomentosus, within which Codd (1985) included P. zeylanicus, are separated from the other P. hadiensis specimens and P. zeylanicus in our phylogenetic analysis ( Fig. 2; clade 3, blue). However, our analysis was based on only two markers for P. hadiensis var. tomentosus. It is important that a larger number of individuals and markers are used to understand the phylogenetic affinities of P. hadiensis var. tomentosus and to ascertain whether its inclusion within the P. hadiensis species concept is justified.

CONCLUSION
Our results suggest that the Sri Lankan and African morphs of P. hadiensis are phylogenetically distinct enough to be considered either sister species or subspecies. However, given the differences in morphological, chemical, and cytological traits, it is more likely that they are sister species. We therefore believe that reinstatement of Cramer's nomenclature, Coleus zeylanicus (Benth.) L.H.Cramer, to the Sri Lankan morph of P. hadiensis (iriweriya) may be warranted. The genus Coleus is currently synonymized with Plectranthus. However, in their most recent phylogenetic study of the subtribe Plectranthinae, Paton et al. (2018) show that the Coleus clade, which includes P. amboinicus (the type of Coleus), Solenostemon, Pycnostachys, and Anisochilus, forms a well-defined sister group to the rest of the species in the subtribe. Therefore, they recommend recognising Coleus as a separate genus. The circumscription of the new Coleus genus would include P. hadiensis (see Fig. 1 in Paton et al., 2018). We believe that this new finding justifies restoration of the Cramer rather than the Bentham nomenclature to iriweriya.
Proper identification of P. hadiensis taxa is important for future researchers and medical practitioners since the pharmacological properties of the Sri Lankan plants may be wrongly attributed to P. hadiensis from southern Africa and vice versa. Our study looked at only molecular markers, with only two markers for P. hadiensis var. tomentosus. Our results demonstrate the need for further phylogenetic analyses involving chemical constituents (particularly those of medicinal value) and morphological and cytological features, in addition to molecular markers, to confirm whether the two morphs of P. hadiensis are the same or different species. Such an analysis should include multiple individuals collected from different regions of south and southeast Asia (including Sri Lanka and India) and Africa, as well as a greater number of individuals of P. hadiensis var. tomentosus than we have used in our study, for a more precise delimitation of the P. hadiensis species concept. C. zeylanicus may well be an introduction from Africa as is P. amboinicus, which is also used in Sri Lanka for its medicinal properties. Trimen (1895) stated that C. zeylanicus is morphologically similar to P. parviflorus Willd., a species native to Oceania. On the other hand, C. zeylanicus may be an example of speciation following migration from Africa, as is the case for some Asian Plectranthus species (Paton et al., 2018). Ideally, any future analysis should include P. parviflorus and other African and Asian Plectranthus/Coleus species with morphological similarities to C. zeylanicus, as well as members of the Ocimae in the Sri Lankan flora, to understand the phylogenetic affinities and putative origins of C. zeylanicus.