Introduction

Vatica mangachapoi is a dipterocarp species famous for its timber and non-wood byproducts. Its timber is important commercially in international tropical timber markets (Appanah and Turnbull 1998) and its leaves and stems contain a number of secondary metabolites that are used in perfumes and medicines (Qin et al. 2011). This species is widely distributed in Southeast Asia (Blanco 1837), and the populations of the genus Vatica Linn. (Dipterocarpaceae) on Hainan Island of China were first identified as V. mangachapoi (Qu 1956). Later, Wang (1985) treated the Hainan populations as a new species, V. hainanensis HT Chang et LC Wang, and subdivided it into two varieties, V. hainanensis var. parvifolia HT Chang apud LC Wang and V. hainanensis var. glandipetala LC Wang. Subsequently Yang et al. (2002) considered all Hainan populations as a subspecies of V. mangachapoi (V. mangachapoi spp. hainanensis). However, all these treatments were not accepted by Chen (2005) in Flora of Guangdong or by Li et al. (2007) in Flora of China, and only V. mangachapoi was considered to be native on Hainan Island. Nevertheless, a new variety, V. mangachapoi spp. hainanensis var. wanningensis Fu and Yang (2008), was described for populations growing on coastal sandy soils. These taxonomic studies did not quantify morphological variations at population level. The objectives of the present study were to quantify the variations of morphological characters within and among the populations of the taxa mentioned above and to determine their taxonomic validity.

Materials and methods

Sample collection

Nine populations of the genus Vatica were sampled on Hainan Island between 2011 and 2013 (Fig. 1; Table 1). Some were populations from which the type specimens of the new species, subspecies and varieties described above were collected. Six to thirty-four mature trees were randomly sampled in each population with the distances greater than 100 m between sampled trees. In total 133 individuals were sampled. For all sampled trees, morphological traits of leaf and fruit were measured (Table 2). Voucher specimens were deposited in Research Institute of Tropical Forestry (RITF), Chinese Academy of Forestry. Eighty-three herbarium specimens of the taxa mentioned above in South China Agricultural University Herbarium (CANT), Sun Yet-sen University Herbarium (SYS) and Chinese Academy of Forestry Herbarium (CAF) were also checked, and 11 specimens with fully-developed leaves and fruits were measured.

Fig. 1
figure 1

Localities of nine natural populations of Vatica spp. on Hainan Island, China

Full size image
Table 1 Nine natural populations of Vatica spp. on Hainan Island, China
Full size table
Table 2 Traits measured in the present study
Full size table

Morphological measurement

Species of Vatica are generally identified by their fruit sepal size, lamina size, lateral vein number, base form, and petiole length (Li et al. 2007). Twelve morphological characters of leaf and fruit were measured for each sampled tree and for specimens deposited in herbaria (Table 2). Thirty fruits and 15 completely developed leaves were measured for each sampled tree. On herbarium specimens, all fruits and fully developed leaves were measured. The measurement accuracy was about 0.01 mm for length and width and 0.1° for angle of lamina base (Table 2). Ratios of the longest fruit sepal length to width, the shortest fruit sepal length to width, fruit length to diameter (width), and lamina length to width were calculated to roughly depict shapes of these organs (Frampton and Ward 1990).

Statistical analysis

Coefficients of variation (CV) for each trait were calculated within each population and among all nine populations to assess variability. The CV among populations was calculated based on the mean value of each population. Box and whisker plots were drawn for fruit and leaf measurements of each population to assess whether morphological variation was continuous. Software NTSYS version 2.1 was applied to calculate Nei’s (1972) distances between individuals and generate a dendrogram of all sampled individuals and specimens by the neighbor joining method. FigTree version 1.3.1 was used to draw annular dendrogram.

Results

Of the 16 leaf and fruit traits measured, the angle of the lamina base (ALB) and height of maximum lamina width point (HMLW) showed in general the highest coefficients of variation (CV) within populations, and the ratio of fruit height to width (RFHW) the lowest (Table 3). Lower variances within populations were observed for fruit traits than for leaf traits. CV values were >20% for lamina length (LL), lamina width (LW) and HMLW in P5 and ALB in P9, while they were <6% for RFHW in all populations. The variances among populations for fruit traits were also lower than those for leaf traits: the CV of RFHW was <2%. As a whole, four ratio traits were less variable than size traits within as well as among populations.

Table 3 Coefficients of variation for leaf and fruit traits in nine natural populations of Vatica spp. on Hainan Island, China
Full size table

For all samples of the nine populations, the variation ranges of LL, LW and HMLW were the greatest of the 16 traits and their maxima were above five times the minimum (Table 4). Meanwhile, RFHW varied least with its ratio of maximum to minimum being approximately 1.7. The number of lateral veins (NLV) was the least variable of 12 directly-measured traits. All nine populations overlapped considerably for each of 16 morphological traits. Variation of each morphological trait was continuous among individuals and populations (Fig. 2).

Table 4 Variation in leaf and fruit traits in nine natural populations of Vatica spp. on Hainan Island, China
Full size table
Fig. 2
figure 2

Box and whisker plots showing the range of variation in fruit and leaf traits in the nine populations of Vatica spp. on Hainan Island, China. The edges of the box are 25% and 75%; the vertical lines are drawn from the box to the most extreme point within 1.5 interquartile ranges; and the transverse lines in box are the medians. See Table 1 for abbreviations of traits

Full size image

Cluster analysis shows the relationships between individuals among and within populations. Individuals of any population were not clearly distinguished from those of other populations and specimens of each variety did not cluster (Fig. 3; Table 5). From this we conclude that the published subspecific taxa possessed no morphological characters that distinguish them from V. mangachapoi.

Fig. 3
figure 3

Dendrogram at individual level generated from neighbor joining analysis based on Nei’s (1972) distances for Vatica spp. on Hainan Island, China. Note 1–9, codes of the populations listed in Fig. 1, the same number means individuals from the same population; S1–S11, codes of specimens of taxa published; and particularly, S6–S11 were collected from trees in Guangdong Academy of Forestry, Guangzhou City (GAF) which were introduced from Bawangling, Changjiang County in 1960. See Table 5 in detail

Full size image
Table 5 Detailed description of Fig. 3
Full size table

Discussion

We quantified differences among individuals and populations for 16 morphological traits. Our results demonstrate that there are morphological variations among populations of the species. These might be caused by genetic variability, micro-scale adaptation or environmentally-induced differences (Schluter 2001; Alan 2002; Shi et al. 2009). We also documented by field survey that Vatica spp. showed abundant variation of leaf and fruit morphology at a range of sites from coastal sand to montane rain forest on Hainan Island.

To determine whether a species is a recognizable taxon, it is necessary to quantify its patterns of morphological variation and compare those to patterns displayed by its closest relatives (Hsu 1998). Nooteboom (1992) concluded that it is necessary to conduct field surveys and to study enough specimens ideally representing the entire geographic range of the subject taxon. Chang and Wang (Wang 1985) identified populations of Vatica on Hainan Island as a new endemic species, V. hainanensis, rather than part of the variation within the widespread V. mangachapoi. They considered that the species of Vatica on Hainan Island was characterized as follows: NLE, 10–13; PL, 8–15 mm; LSL, 40–55 mm; LSW, 12–20 mm; SSL, 12–20 mm; SSW, 5–8 mm (Wang 1985). These measurements substantially differed from those of V. mangachapoi (NLE, 7–10; PL, 15–20 mm; LSL, 30–40 mm; LSW, 10–15 mm; SSL, 8 mm; SSW, 4 mm; Blanco 1837). In the present study on morphological variations of leaf and fruit traits of 133 individuals from nine populations, we found that the variation ranges of these traits were much wider (NLE, 8–21; PL, 5–21 mm; LSL, 22–78 mm; LSW, 6–22 mm; SSL, 10–32 mm; SSW, 3–9 mm). These measurements indicated that only SSL of the species on Hainan Island was obviously greater than that of V. mangachapoi, the ranges of NLE, PL, LSL, LSW and SSW found in the present study overlapped those described in the literature with both taxa identified. Since the study at population level yields a more accurate estimate of the variance within a taxon and is much more reliable than conclusions based on a small number of samples in herbaria (Nooteboom 1992; Hsu 1998), the two taxa could not be separated from each other. Our results suggest that all native populations of Vatica on Hainan Island are referable to V. mangachapoi without occurrence of subspecies because they are not distinguishable from the species described in Flora de Filipinas (Blanco 1837).

Our study also showed that each morphological trait showed continuous variation with a wide range on Hainan Island (Table 4; Fig. 2). The range of variation overlapped the reported measurements for var. parvifolia (LL: 40–60 mm; LW: 15–20 mm; LSL: 30–40 mm; and LSW: 8–10 mm) and var. glandipetala (LSL: 22–50 mm; LSW: 5–12 mm; and NLV: 6–9) published by Wang (1985) and Yang et al. (2002), and var. wanningensis (NLV: 8–9; and LSL: 45–52 mm) reported by Fu and Yang (2008) for each natural population. It is notable that the type specimen of variety glandipetala was collected from trees in Guangdong Academy of Forestry which were introduced from Bawangling, Hainan Island. Additionally, the presence of glands on the inner side of petal bases (Wang 1985) and indumentum differences on small branches (Fu and Yang 2008) are hardly sufficient differences for identifying variety glandipetala and wanningensis, respectively. Cluster analysis revealed that individuals in any population did not cluster, and specimens of a described taxon were assigned to the other taxa. This made it impossible to assign a given individual to any population based on these morphological traits. In other words, the varieties were poorly separated by cluster analysis based on Nei’s (1972) distances. The three varieties, parvifolia, glandipetala, and wanningensis, could thus not be distinguished from one another or from the type specimens of V. mangachapoi.

Conclusion

The present study showed that all populations of Vatica on Hainan Island, China should be referred to Vatica mangachapoi with no subspecific categories being recognized. This conclusion supports the taxonomic treatment of the genus Vatica from Hainan Island as reported in Flora of Guangdong and Flora of China.

Taxonomic treatment

Vatica mangachapoi Blanco, Flora de Filipinas, ed.1, 401, 1837.

  • =Vatica hainanensis HT Chang et LC Wang, Acta Scientiarum Naturalium Universitatis Sunyatseni, 24 (3): 94, 1985; Vatica mangachapoi Blanco ssp. hainanensis (HT Chang et LC Wang) YK Yang et JK Wu, Chinese Wild Plant Resources 21(5): 1, 2002.

  • =Vatica hainanensis HT Chang et LC Wang var. parvifolia H.T. Chang, Acta Scientiarum Naturalium Universitatis Sunyatseni, 24 (3): 96, 1985; Vatica mangachapoi Blanco var. parvifolia (HT Chang) YK Yang et JK Wu, Chinese Wild Plant Resources 21(5): 2, 2002.

  • =Vatica hainanensis H.T. Chang et L.C. Wang var. glandipetala L.C. Wang Acta Scientiarum Naturalium Universitatis Sunyatseni, 24 (3): 96, 1985; Vatica mangachapoi Blanco var. glandipetala (LC Wang) YK Yang et JK Wu, Chinese Wild Plant Resources 21(5): 1, 2002.

  • =Vatica mangachapoi Blanco var. wanningensis GA Fu et YK Yang, Bulletin of Botanical Research 28(3): 259, 2008, syn. nov.