Morphological Phylogenetic Analysis of Seven Varieties of Ficus deltoidea Jack from the Malay Peninsula of Malaysia

This study is the first report to suggest a morphological phylogenetic framework for the seven varieties of Ficus deltoidea Jack (Ficus: Moraceae) from the Malay Peninsula of Malaysia. Several molecular-based classifications on the genus Ficus had been proposed, but neither had discussed the relationship between seven varieties of F. deltoidea to its allies nor within the varieties. The relationship between seven varieties of F. deltoidea is still debated due to the extreme morphological variabilities and ambiguous boundaries between taxa. Thus, the correct identification of these varieties is important as several morphological characters are variety-specific. To test the monophyly and further resolved the relationship in F. deltoidea, a morphological phylogenetic analysis was conducted based on herbarium specimens representing the seven varieties of F. deltoidea that were collected from the Malay Peninsula of Malaysia, by using related species of the genus Ficus; F. grossularioides, F. ischnopoda and F. oleifolia as the outgroups. Parsimony and neighbour-joining analyses indicated that F. deltoidea is monophyletic, in that the seven varieties of F. deltoidea nested into two clades; clade subspecies deltoidea (var. deltoidea, var. bilobata, var. angustifolia, var. kunstleri and var. trengganuensis) and clade subspecies motleyana (var. intermedia and var. motleyana).


Introduction
Ficus deltoidea Jack (Ficus: Moraceae) is a diverse species of subgenus Ficus, section Ficus and subsection Frutescentiae [1]; which contains 25-30 species found in the Sino-Himalayan and western Malesian region [2]. Ficus deltoidea is a native and widely distributed throughout Malaysia, Thailand, Sumatra, Java, Kalimantan, Sulawesi and Moluccas [3]. This plant is a small shrub up to 3 m tall, sometimes occurring as an epiphyte [4]. It can be found in abundance along the beaches, peat soils and in hilly forest up to 3000 m above sea level [5]. The Malays called F. deltoidea as 'Mas Cotek' due to the presence of golden dots on the upper surface of its lamina [5]. The seven varieties of F. deltoidea; namely var. deltoidea, var. bilobata Corner, var. angustifolia (Miq.) Corner, var. intermedia Corner, var. kunstleri (King) Corner, var. motleyana (Miq.) Corner and var. trengganuensis Corner that were found in the Malay Peninsula were described by Kochummen [6]. The plants are often recognized by the presence of golden dots on the upper surface of the lamina, dichotomous midrib, unique fig (syconia) with flowers hidden inside the syconia, leafy twigs and the milky juice [4], [7].
Ficus deltoidea is commonly cultivated as a houseplant for decorative purposes and traditional medicinal uses by the local people [4]. The var. bilobata, var. angustifolia, var. intermedia, var. kunstleri, var. motleyana and var. trengganuensis are commonly used in the Malay traditional medicine [8]. The dried leaves, stems and roots are often commercialized as herbal tea [9]. The decoction of the leaves is traditionally used by women after childbirth to help strengthen the uterus [5]. It is also believed to improve blood circulation, regain energy and enhance fertility naturally for both men and women [10,11]. These claims were supported by previous bioassay studies, demonstrating that the aqueous extract of the leaves contains antidiabetic [12,13] and antinociceptive activities [10]. The leaf extracts were reported to be rich of phenolic and flavonoid compounds which are comparable with black and green teas as well as fruit juices [11]. The flavan-3-ols and flavones were the main compounds that contributed to the total antioxidant activity [11], whereas isovetexin and vetexin were reported to be responsible for the antidiabetic activity [14].
Although F. deltoidea has been exploited in many different ways, the taxonomy of this species is still controversial at the varietal level. Historically, several botanical names of F. deltoidea have been reported; namely F. diversifolia Blume, F. motleyana Miq. and F. oleifolia King [4][5][6][7][8][9][10][11][12][13][14][15]. The extreme morphological variations and unclear boundaries between varieties create misleading identification of F. deltoidea varieties. The leaf characters are probably the most variable and showed heterophylly in the species [16]. Nevertheless, the young plant and the matured plant of the same variety often displayed different states of leaf characteristics. In this study, seven varieties collected from the Malay Peninsula of Malaysia, as mentioned by Kochummen [6], were selected, to investigate the monophyly of F. deltoidea and to differentiate intraspecific variation based on information contributed by overall morphological characters.

Herbarium Specimens
Morphological data were scored from 108 herbarium specimens, with prior permissions from four different herbaria; the Herbarium of the Universiti Kebangsaan Malaysia (UKMB), Herbarium of the Forest Research Institute of Malaysia (FRIM), Herbarium of Sarawak (SAN) and the National Herbarium of Singapore (SING). All measurements and observations were taken from the herbarium specimens except data for flowers, which were gathered from the literatures [2], [4][5][6][7], due to limited number of syconium presents on each herbarium sample.

11
Waxy gland beneath the lamina 0: two, 1: equal or more than three.

13
Gland at the subsequent dichotomies of the midrib0: absent or rarely seen, 1: commonly seen.

28
Color of ripening figs 0: yellow to orange to brownish red, 1: rose red to dark purple.

Characters
The characters were surveyed throughout the ingroup and outgroup taxa using criteria of putative homology or hypothetical homology [17]. Furthermore, the characters should consistent in occurrence or absence among the terminal taxa, which implies that they are not environmentally plastic [18]. As a basic principle, the characters that varied between terminal taxa were chosen, but not those that varied within the taxa. In total, 32 morphological characters including 29 binary and three multistate characters were prepared (Table 1). Character states were then polarized using outgroup comparison method [19]. If a character state was not available or not applicable in a taxa, it was designated as missing.

Phylogenetic Analyses
The phylogenetic analyses were performed using the maximum parsimony (MP) and Wagner approaches and trees were generated using the PAUP* version 4.0b10 software [20,21]. All character states were run as unordered with equal weight. The search for the most parsimonious tree was determined by exhaustive search and bisection-reconnection (TBR) branch swapping, with retention of multiple parsimonious trees (MaxTrees = 100). Branches were collapsed and polytomies were created when maximum branch length is zero. Optimization of characters was performed using the ACCTRAN (Accelerated Transformation Optimization) option. To test the support for each clade, bootstrap analysis [21] was performed with 1000 replicates of simple taxon addition and TBR swapping, with a limit of 10 trees kept per replicate. Bootstrap percentages (BS) of 50-70 was considered weak, 71-85 as moderate and .85 as strong [22]. The distance tree was estimated by the neighbour-joining (NJ) method [23] based on the formulae of Kimura [24]. Graphic outputs were produced using the TreeView X software [25] and characters were mapped onto a single tree [26].

Results
The resultant data matrix is shown in Table S1. The analysis of the data matrix, containing nine terminal taxa and 32 characters, produced two shortest maximum parsimony (MP) trees with a minimum length of 68 steps, a consistency index (CI) of 0.5147 and a retention index (RI) of 0.5976. A total of 31 parsimonyinformative characters and only one parsimony-uninformative character were found in the dataset. The only difference between the two trees topologies is the position of var. motleyana and var. intermedia, which was supported and characterized by tepals longer than ovary (#29) and a synapomorphy of smooth and slightly angular ovary (#31). There was no character found to support this group, thus collapsed into polytomy in the other tree. Other character transformation series within the whole tree were found to be almost identical in both trees. The results were then compared with the neighbour-joining (NJ) tree ( Figure 1). Noted that the topology was similar in MP and NJ trees, but the bootstrap supports (BS) of NJ tree were generally improved compared to the MP tree.
The placement of clade subspecies deltoidea was moderately supported with 75% and 79% bootstrap values in the MP and NJ analyses, respectively. It was defined by the following characters; leaf length equal or less than 5 cm (#1), midrib forked near the middle of the lamina (#3), leaf apex rounded to truncate and minutely emarginate to form bilobed (#5), leaf spatulate (#7), waxy gland beneath the lamina is equal or more than three (#11), gland present at the forked midrib (#12) and commonly seen at the subsequent dichotomies of the midrib (#13). Within subspecies deltoidea, var. kunsleri and var. trengganuensis formed a strongly supported clade, 90% and 91% in the MP and NJ trees, respectively. They were described by having obtuse leaf base angustifolia were defined by extremely low confidence in the MP analysis and therefore, relationship between these varieties were not certain. However, the NJ analysis support their relationship by 70% bootstrap support and they share the peduncle length that is equal or less than 1 cm ( The clade subspecies motleyana received weak bootstrap support of 51% in the MP analysis, but was moderately supported, 61% in the NJ analysis. It was distant to subspecies deltoidea by having midrib not forked to forked near the apex (#3), tepals longer than ovary (#29) and smooth or slightly angular ovary (#31). Within subspecies motleyana, var. intermedia was distinguished by having stipule length more than 0.

Discussion
Most of the previous classifications (Table 2) were based on intuitive morphology. The number of varieties was easily increased or reduced based on its morphological variation and locality [27]. Different authors had their own opinion in discriminating taxon, such as Corner [28] [4]. On the other hand, Kochummen [4] has divided F. deltoidea into seven varieties namely, var. deltoidea, var. bilobata, var. angustifolia, var. intermedia, var. kunstleri, var. motleyana and var. trengganuensis, which were available in the Malay Peninsula of Malaysia or formerly known as Malaya, and described them. Later on, two endemic varieties of Borneo, namely var. subhirsuta Kochummen and var. recurvata Kochummen were added [27]. Berg [2] and Berg and Corner [7] recently subdivided the species into two major morphological entities (subspecies deltoidea and subspecies motleyana) which seems to be more practical and satisfactory in handling the variation. They further mentioned that as strong phytogeographical and ecological support is lacking, the rank of variety proposed by others might be more correct, but the chosen rank allows recognition of varieties for regional use. Therefore, the sampling of this study focused on the earlier classification scheme of Kochummen [6] who studied seven varieties from Peninsular Malaysia  region. These varietal variations then might be useful for allocating plant material to varieties as required for therapeutic and pharmaceutical applications. In this study, Ficus deltoidea formed one major clade. Nevertheless, character variation in the dataset was insufficient to resolve all the phylogenetic relationships, especially among the internal branching of subspecies deltoidea. The divergence of var. deltoidea, var. angustifolia and var. bilobata received extremely low bootstrap support, thus the branches collapsed in the strict consensus tree. The placement of these varieties, as revealed by a previously published RAPD analysis, were also uncertain due to the variable banding patterns and hence showed that similar varieties were grouped into different clusters [9].
From the results obtained, the seven varieties of F. deltoidea of Malesian region can be placed into two subspecies: subspecies deltoidea consisted of var. deltoidea, var. bilobata, var. angustifolia, var. kunstleri and var. trengganuensis. Within the subspecies deltoidea, var. kunstleri was shown to be more closely related to var. trengganuensis than to the other varieties. The positions of var. deltoidea, var. bilobata and var. angustifolia received extremely low support in the MP tree, suggesting that the analysis did not assure the relationships between these varieties. However, they were genetically closely related to each other and were fairly supported in the NJ analysis. This should, however, be confirmed by a detailed molecular studies. The second, subspecies motleyana comprised var. intermedia and var. motleyana. Our results showed an agreement with Berg [2], and Berg and Corner [7] that proposed the two subspecies based on the forked and non-forked midrib. Noted that var. intermedia was excluded in their classifications and then was transferred to F. oleifolia subspecies intermedia, because of mixture in characters of forked and non-forked midrib, and leaf shapes ranging from spatulate to obovate to oblanceolate [2,7]. However, this study showed that the relationship between var. intermedia and var. motleyana was constantly supported and defined in both the MP and NJ analyses.
With respect to intergeneric relationship between F. deltoidea, F. ischnopoda Miq., F. oleifolia King and F. grossularioides Burm.f., F. oleifolia was found to be the closest ally to F. deltoidea, whilst F. ischnopoda and F. grossularioides were placed at the base of trees and did not get any bootstrap support ( Figure 1). Their relationships, however, continue to require explicit examination using a combination of molecular and developmental dataset. It is interesting to note that Rønsted et al., [29] groups F. deltoidea var. borneensis and F. oleifolia together with F. adenosperma Miq., F. ochrochlora Ridl., F. dammaropsis Diels, F. pumila L., F. erecta Thunb., and placed F. ischnopoda at the base of section Ficus/Adenosperma on the basis of several molecular datasets. Furthermore, Rønsted et al., [30] recently found that F. oleifolia and F. ischnopoda were the closer relatives to F. deltoidea var. borneensis in subsection Frutescentiae, whilst F. grossularioides more remotely in section Eriosycea. The sampling of subsection Frutescentiae and F. deltoidea varieties has been very limited in the previous phylogenetic studies, which makes it difficult to know what would be the close relatives and appropriate outgroup. Therefore, it is not possible at this stage to compared F. deltoidea to its allies because there is no comprehensive phylogenetic classification available for comparison.
This study was the first attempt to suggest a morphological phylogenetic framework for the seven varieties of Ficus deltoidea from the Malay Peninsular, which will provide a basis for future molecular, cytological or phytochemical as well as pharmaceutical investigations. Deeper understanding of the systematic relationships between the varieties will help to promote expeditious exploitation and sustainable uses of this plant as a whole.