This paper concerns two poorly known Malesian vittarioid ferns usually called Haplopteris angustissima (Holttum) S.Linds. and Monogramma capillaris Copel. The generic delimitation of vittarioids has changed substantially in recent years due to insights from molecular phylogenetic research (e.g. Crane et al. 1995; Ruhfel et al. 2008; Schuettpelz et al. 2016). Here we follow Schuettpelz et al. (2016) and Chen et al. (2016), and therefore treat the above two species under Haplopteris.

Holttum (1947) described Vittaria angustissima based on Carr's collection from Fraser's Hill in Pahang, Peninsular Malaysia. He mentioned its similarity to Vittaria angustifolia Blume [now Haplopteris angustifolia (Blume) E.H.Crane] and Vittaria parvula Bory, but distinguished it by the smaller scales and fronds, and the fact that many fronds have sori only on one edge of the lamina. After its establishment, subsequent authors have maintained Holttum's species (e.g. Parris and Latiff 1997), but no new argument has been made for its systematic relationship. Lindsay (2010) transferred it to Haplopteris following Crane (1997).

Copeland (1911) described Monogramma capillaris [Haplopteris capillaris (Copel.) C.W.Chen, S.Linds. & K.T.Yong comb. nov., see below] based on Merrill's collections from the island of Negros in the Philippines. He compared it with the congeneric species, Monogramma dareicarpa Hook. [now Haplopteris dareicarpa (Hook.) S.Linds. & C.W.Chen] and distinguished it from that by the more closely spaced and narrower fronds. Monogramma capillaris was later transferred to Vaginularia Fée by Christensen (1934) and Pleurofossa Nakai ex H.Ito by Ito (1936). Since then, this species has seldom been mentioned in the literature and was even omitted by Copeland himself in his Fern Flora of the Philippines (Copeland 1960). Price (1990) synonymized this species under M. dareicarpa, and his treatment was followed by Pelser et al. (2011 [onwards]). Recently, Kuo et al. (2017) included a specimen of H. capillaris (as Monogramma capillaris) from Sabah in their molecular phylogenetic analysis and discovered that it had a close affinity to an independent gametophyte population from Taiwan.

Morphologically, H. angustissima and H. capillaris are extremely simplified and very similar (Fig. 1); they are diminutive plants with hair-like fronds, ca. 1 mm wide and less than 15 cm long. These two species have not been studied systematically, probably due to the scarcity of materials, and so their relationship is unknown. Excluding the type collections, H. angustissima is only known by nine other collections from Peninsular Malaysia, and H. capillaris is only known by seven other collections from Borneo. Despite their apparent rarity, we recently found both species during collecting expeditions in Peninsular Malaysia, Borneo, and the Philippines. These new materials covered most of the known localities of both species (including the type locality of H. capillaris) and proved critical to deciphering both the relationship of H. angustissima and H. capillaris, and their systematic position within the vittarioid clade.

Materials and Methods

Plant Materials—Four populations of H. angustissima and H. capillaris were sampled between 2015–2017 in Borneo (Sabah & Sarawak), Peninsular Malaysia, and the Philippines (Chen Wade4360, YKT 813, Chen Wade4825, and Chen Wade4959; Table 1). Due to their morphological similarity, we first identified our specimens based on the type locality, i.e. Chen Wade4825 from Peninsular Malaysia as H. angustissima and Chen Wade4959 from the Philippines as H. capillaris. The two specimens from Borneo were identified later (as H. capillaris) from the results of the molecular phylogenetic analysis (see below). As well as our new collections, specimens of H. angustissima and H. capillaris deposited at K, KLU, KYO, L, MICH, SING, TAIF, and US (including the lectotype of Monogramma capillaris at MICH and the holotype of Vittaria angustissima at SING) were used for morphological analyses (Table 1) and/or the preparation of taxonomic descriptions.

FIG. 1.

Habit, rhizome, and sorus of the two Haplopteris species studied here. A–C. Haplopteris angustissima (based on Chen Wade4825). D–F. Haplopteris capillaris (based on Chen Wade4959).

Molecular Phylogenetic Analysis—To infer the phylogenetic relationship of H. angustissima and H. capillaris, a phylogeny of 22 Haplopteris taxa, including our four recent collections was reconstructed based on seven chloroplast DNA regions (atpA, chlL, matK, ndhF, rbcL, rpoA, and trnL-F) with Radiovittaria [Radiovittaria gardneriana (Fée) E.H. Crane] as outgroup. Voucher information and GenBank accession numbers are listed in Appendix 1. Extraction of genomic DNA was performed using a modified CTAB-Qiagen column (Valencia, CA, USA) protocol from fresh, silica-dried, or herbarium specimens following Kuo (2015). The PCR primer sets used in this study followed Schuettpelz et al. (2016) and Chen et al. (2017). The PCR amplifications and sequencing reactions were performed following the protocol of Chen et al. (2013). Sequences were aligned using default options in Muscle (Edgar 2004). This alignment is available through Dryad (Chen et al. 2019) and TreeBASE ( http://www.treebase.org, TreeBASE study # 23706). A maximum likelihood (ML) analysis was conducted using GARLI v. 2.0 (Zwickl 2006) with the substitution models derived in PartitionFinder v. 2.1.1 (Guindon et al. 2010; Lanfear et al. 2012, 2017), and the genthreshfortopoterm option set to 20,000. Branch support was assessed with 1000 bootstrap replicates under the same criteria in the CIPRES Science Gateway (Miller et al. 2010). Branch support was also assessed in a Bayesian framework following Chen et al. (2017) with the partitioning scheme and substitution models determined by PartitionFinder (as above).

Morphological Analyses—In total, 15 specimens (eight of Haplopteris angustissima and seven of H. capillaris) were used for morphological analyses (Table 1). Morphological characters including rhizomes scales, terminal cells of soral paraphyses (simply referred to as paraphyses hereafter), frond venation, epidermal silica bodies, and spores were examined and measured with a light microscope (DMR, Leica, Wetzlar, Germany) and/or a tabletop scanning electron microscope (TM-3000 Hitachi, Ibaraki, Japan), following the methods of Kao et al. (2008), Vasco et al. (2014), and Chen et al. (2017). The size (length and width) of 30 paraphyses, the thickness of 30 cell walls of rhizome scales, the length of 30 epidermal silica bodies from adaxial laminae (more abundant than abaxial surface), and the length of 30 spores were measured for each specimen. The ANOVA with p adjusted by TukeyHSD was used to compare the differences in cell wall thickness in the rhizome scales and the differences in spore size among the specimens. Clustering of the width and length of paraphyses of both taxa was tested using k-means clustering analysis. All statistical analyses were performed using R v. 3.4.0 (R Core Team 2017) with ggplot2 package (Wickham 2009) to visualize the results.

Table 1.

Morphological traits of Haplopteris angustissima and H. capillaris.

Results

Molecular Phylogeny—The final concatenated dataset comprised 7586 aligned sites (1733, 879, 1035, 1208, 1195, 576, and 960 from atpA, chlL, matK, ndhF, rbcL, rpoA, and trnL-F, respectively) and included 12.9% (25,479 characters) missing data and 15.1% (1151 sites) parsimony-informative sites. The optimal partitioning scheme based on the corrected Akaike information criterion (AICc), as inferred by PartitionFinder, included 15 data subsets (their corresponding models are listed in parentheses): atpA codon position 1 + chlL codon position 1 (GTR + I + G); chlL codon position 2 (K81uf + I); chlL codon position 3 (TVM + G); matK codon position 1 + ndhF codon position 3 (TVM + G); matK codon position 2 + ndhF codon position 1 (TVM + G); matK codon position 3 (TVM + G); ndhF codon position 2 (TVM + I + G); trnL-F (TVM + G); rpoA codon position 1 (GTR + G); rpoA codon position 2 (TVM + G); rbcL codon position 3 + rpoA codon position 3 (TVM + G); atpA codon position 2 (GTR + I + G); atpA codon position 3 (GTR + I); rbcL codon position 1 (TIM + I); rbcL codon position 2 (TIMef + I + G).

Fig. 2.

Phylogram of 22 Haplopteris taxa from maximum likelihood analysis of the concatenated plastid dataset (seven loci; 15 data partitions). Maximum likelihood bootstrap support values and Bayesian posterior probabilities are shown for nodes with high support (i.e. BS ≥ 70) and solid circles at nodes indicate bootstrap = 100%, posterior probability = 1. Haplopteris angustissima and Haplopteris capillaris are in bold-face type. Branch length of the outgroups is not to scale for better visualization of ingroup relationships. The sequence of independent gametophytes of Haplopteris capillaris from Taiwan is derived from two samples (A and E30, Appendix 1).

The reconstructed phylogeny of the 22 Haplopteris taxa is shown in Fig. 2. Overall the relationships within the genus are well supported and congruent with previous studies (Schuettpelz et al. 2016; Chen et al. 2017). Of the 22 resolved nodes, 19 received good support (i.e. BS ≥ 70). Three samples of H. capillaris were resolved as monophyletic. The independent gametophyte from Taiwan was resolved within the H. capillaris clade and differed by only five bp from the sporophytic sample from the Philippines. Despite their morphological similarity, H. angustissima and H. capillaris were resolved in different clades.

Morphological Analyses—The results of the morphological analyses are presented in Tables 1–2 and Figs. 3–7. Almost all the fronds of each species have single sorus, but the type specimen of H. angustissima has ca. 10 fronds with double sori, and the type specimen of H. capillaris has two very small fragments with double sori (Table 1). Considerable variation was observed in the size of paraphyses, rhizome scales, and spores such that there was no clear distinction between the two species (Figs. 3, 4, 5). For example, clustering analysis on the size of paraphyses showed that more than 23% of the paraphyses of H. capillaris were clustered with those of H. angustissima (Table 2). However, the shape of the silica bodies and the venation patterns were found to be useful for species delimitation. Specifically, the ends of the lobes of the silica bodies in H. angustissima are pointed whereas those in H. capillaris are blunt (Fig. 6); and fronds (with single sorus) of H. capillaris contain only a midvein (Fig. 7C) whereas those (with single sorus) of H. angustissima have two very close parallel veins connected by a few oblique cross-veins (thereby creating a few very long narrow areoles) (Fig. 7A).

Table 2.

K-means clustering analysis on the length and width of paraphyses of Haplopteris angustissima and H. capillaris. Eleven and 49 data points of H. angustissima and H. capillaris were misclassified, respectively.

Discussion

Species Delimitation—The phylogenetic analysis resolved H. angustissima and H. capillaris in different clades that are well-separated from each other (Fig. 2), providing support for their distinctiveness. Haplopteris angustissima was clustered with H. hainanensis (C.Chr.) E.H.Crane and H. sikkimensis (Kuhn) E.H.Crane despite their morphological and geographical dissimilarity. Morphologically, H. hainanensis and H. sikkimensis can be distinguished from H. angustissima by always having fronds with double sori. Additionally, H. hainanensis is a much larger species with fronds ca. 10–30 cm long (Zhang et al. 2013) compared to ca. 2–13 cm long in H. angustissima. Geographically, H. angustissima is known only from Peninsular Malaysia, H. hainanensis is known only from southern China and northern Vietnam, and H. sikkimensis is known only from the Himalayan area and northern Indochina (Zhang et al. 2013).

To reveal the discriminative characters of H. angustissima and H. capillaris, we conducted a detailed morphological comparison and included not only those characters that are commonly used in Haplopteris (i.e. paraphyses, rhizome scales, and spores) (e.g. Chen et al. 2013, 2014, 2017), but also two other less often used characters (i.e. silica bodies and venation). Our results show that only silica bodies and venation can be used to unambiguously delimit the two species.

Fig. 3.

Paraphyses (upper panel) and rhizome scales (lower panel) of the two Haplopteris species studied here. A–D. Haplopteris angustissima. E–K. Haplopteris capillaris. Scale bars = 100 µm for the paraphyses and 1 mm for the scales.

Our discovery of different venation patterns in H. angustissima and H. capillaris is based only on fronds with single sori. We were able to determine the venation pattern in one frond of H. angustissima with double sori (Fig. 7B), but could not compare this with H. capillaris due to the size and condition of the relevant frond fragments. However, we expect those fragments to contain at least two veins because, as far as we know, two sori never arise from the same vein in Haplopteris.

Silica bodies appear to be a synapomorphy for all vittarioid ferns, and high variation has been found at both the generic and the species level (Ito 1936; Bhandari and Mukhopadhyay 2009; Sundue 2009). Here we have demonstrated that silica bodies can be used for species delimitation. Future studies with broader sampling are needed to further explore their systematic value.

Convergent Evolution—Biseriate areolate venation, supporting two linear sori per frond, is found in most Haplopteris species (Schuettpelz et al. 2016); however, some of the smallest species, which have fronds only a few centimeters long, have only a single vein (midvein) supporting a single sorus. These include H. capillaris, H. dareicarpa (Hook.) S.Linds. & C.W.Chen, and H. graminea (Poir.) comb. ined. Despite the similarity of these three species, the phylogenetic data shows that they are only distantly related (Fig. 2). This indicates that extreme morphological simplification has likely evolved independently multiple times in the history of the genus.

Independent Gametophyte—Kuo et al. (2017) included a specimen of H. capillaris from Sabah in a molecular phylogenetic analysis and found that it had a close affinity with an independent gametophyte population in Taiwan. They suspected the independent gametophyte was conspecific with H. capillaris but refrained from publishing this conclusion due to the systematic uncertainty of H. capillaris. In this study, we were able to confirm the distinctiveness of H. capillaris by morphological observations (including examination of the type specimens) and molecular phylogenetic analysis. In addition to the two specimens sequenced by Kuo et al. (2017), we were able to include two other specimens of H. capillaris in the molecular phylogenetic analysis, one from Mt. Kanlaon, the type locality in the Philippines, and one from Sarawak. We found that all four samples form a well-supported clade and there is only a five bp difference between the sporophytic specimen from the Philippines and the independent gametophyte from Taiwan. Our results confirm the suspicions of Kuo et al. (2017) that independent gametophytes of H. capillaris occur in Taiwan.

Fig. 4.

Rhizome cell wall thickness (A) and spore size (B) in specimens of Haplopteris angustissima (dark grey) and Haplopteris capillaris (light grey). Median, 25%, and 75% quartiles are shown; whiskers indicate 5% and 95% percentiles. Specimens sharing the same letter are not significantly different (p < 0.05, Tukey HSD post-hoc test for significant differences among means).

Fig. 5.

Scatter plot of length and width of the end cells of the soral paraphyses of Haplopteris angustissima (solid circles) and Haplopteris capillaris (open circles). The 95% confidence ellipses are shown.

Taxonomic Treatment

Haplopteris angustissima (Holttum) S.Linds., Gard. Bull. Singapore 62: 119. 2010. Vittaria angustissima Holttum, Gard. Bull. Singapore 11: 274. 1947. TYPE: MALAYSIA. Pahang: Fraser's Hill, 1219 m [4000 ft], Mar 1929, C.E. Carr s.n. (holotype: SING! [SING0144295]).

Epiphytic or lithophytic. Rhizome short-creeping, densely scaly, ca. 1 mm diam including scales, bearing fronds close together. Rhizome scales narrowly lanceolate, gradually narrowing from the base towards a long-tailed and ultimately filiform apex, 0.8–1.8 mm long and 0.2–0.3 mm wide (at base), clathrate, toothed at margin, the walls thickened and dentate. Fronds sessile, simple, linear, 2.0–13.2 cm long, 0.5–0.8 mm wide, widest above the middle, gradually narrowing towards both ends, apex acute or (occasionally) truncate, laminae coriaceous, glabrous, margins entire, flat on both surfaces. Venation difficult to see even with transmitted light, in fronds with single sori, midvein branching near frond base into two very close parallel main veins and reuniting near the frond apex, a few oblique lateral veins between the two main veins, forming a few long, narrow areoles; in fronds with double sori (our observation is based on the examination of one frond), midvein branching into three almost parallel main veins (midvein and two lateral veins) near frond base, very sparse oblique lateral veins present only on one side of midvein. Epidermal silica bodies linear, 0.1–0.7 mm long, present on both sides of the lamina, margin lobed, ends of lobes pointed. Sori linear, uninterrupted, following one main vein, slightly covered by a soral flap and open to one side, or rarely, fertile on both sides, each following a lateral main vein. Soral paraphyses copious, stalks uniseriate, filiform, colourless; apical cells large, funnel-shaped, length-to-width ratio of 1.1 ± 0.23, yellowish. Spores bilateral, bean-shaped, monolete, yellowish, surface smooth.

Fig. 6.

Illustrations (left panel) and enlarged margins (right panel) of silica bodies on the adaxial surfaces of Haplopteris angustissima (A–E) and Haplopteris capillaris (F–I). A. Carr s.n. (holotype). B. Chen Wade4825. C. Burkill & Holttum 8835. D. Molesworth Allen 3941. E. Smith 899. F. Chen Wade4360. G. Chen Wade4959. H. Kato et al. B-10358. I. Yong 813. Scale bars = 300 µm (left panel) and 50 µm (right panel).

Distribution—Peninsular Malaysia. Known from only three areas on the Main Range (Fraser's Hill, Gunung Kerbau, and Cameron Highlands).

Ecology—Epiphytic on mossy tree trunks in damp forests or lithophytic on quartz outcrops near the ridge of mossy forests. The lowest recorded elevation is 1200 m and the highest is 1730 m.

Additional Specimens Examined—Malaysia.—PENINSULAR MALAYSIA: Pahang, Cameron Highlands, 1524 m [5000 ft], 4 Mar 1948, Molesworth Allen, B.E.G. 1061 (K! [K000637847]); Pahang, Cameron Highlands, slope of G. Jasar, 1615 m [5300 ft], 20 Apr 1958, Molesworh Allen, B.E.G. 3928 (K! [K000637844]); Pahang, Cameron Highlands, [See note below concerning discrepancies in locality data], underside of sloping trunk hanging over stream in tall dark forest. A mossy place. Only one patch (but found in other localities), rare, 1524 m [5000 ft], 22 Apr 1958, Molesworth Allen, B.E.G. 3941 (K!, SING! [SING0085700]); Pahang, Cameron Highlands, Many fronds now dead, others medium green. In small clumps, very dwarfed, looking like Monogramma trichoidea. Abundant locally on vertical sites, in tiny crevices on the actual rock. Quartz outcrop [sometimes called “Castle Rock”] beyond Brinchang Village., 1615 m [5300 ft], 9 Feb 1959, Molesworth Allen, B.E.G. 4135 (K!, 2 sheets [one with barcode K000637843], the other without barcode, US! [US01483064]); Pahang, Cameron Highlands, Brinchang Village, trail to Coral hill, lithophyte on quartz outcrop, rare, 4.495282°, 101.3955°, 1730 m, 5 Jul 2017, Chen, C.-W. Wade4825 (KLU!), TAIF! [TAIF513482]); Pahang, Cameron Highlands, on Jacob's ladder, on underside of root base tangle, in very damp and dark position, 1524 m [5000 ft], 25 Oct 1965, Evans 118 (KLU!); Pahang, Fraser's Hill [“Bukit Fraser”], 1219 m [4000 ft], Jun 1922, Smith, E. 899 (K! [K000637846], SING! [SING0085650]); Pahang, Frasers Hill, upon the Selangor border, epiphyte on tree stump among moss. 1219–1331 m [4000–4370 ft], 26 Sep 1922, Burkill, I.H, & Holttum, R.E. 8835 (SING! [SING0085634, SING0085635]); Pahang, Fraser's Hill, 1280 m [4200 ft], 20 Apr 1957, Molesworth Allen, B.E.G. 3347 (US! [US1485955]); Perak, Gunung Inas, 1158 m, 7 Dec 1899, Yapp, R.H. 417 (K! [K000637845]); Perak, Gunong Kerbau, 1371 m [4500 ft], May 1909, Haniff, M. 3938 (SING! [SING0085673]).

Notes—1) The two sheets of Molesworth Allen 3941, from the Cameron Highlands show the same elevation data, collection date, and ecological information, but the one at K is labelled “slopes of G. Kemmiting” whereas the one at SING is labelled “Forest above Walkerburn Cottage.” 2) The locality of Chen Wade4825 is possibly the same locality as Molesworth Allen 4135.

Fig. 7.

Venation of Haplopteris angustissima (A from Chen Wade4825, B from Carr s.n. (holotype)) and Haplopteris capillaris (C from Chen Wade4959). The open margins of the soral flaps are indicated by dotted lines. Scale bar = 5 mm.

Haplopteris capillaris (Copel.) C.W.Chen, S.Linds. & K.T.Yong, comb. nov. Monogramma capillaris Copel., Philipp. J. Sci. C. 6: 147. 1911. TYPE: PHILIPPINES. Negros. Mt. Kanlaon [“Canlaon”], Apr 1910, E.D. Merrill 6961 (holotype PNH (lost); lectotype MICH! [MICH1309835], designated by Price (1990); isolectotypes P [P00608464], NY [NY00127501]).

Vaginularia capillaris (Copel.) C.Chr., Index Filic., Suppl. Tert. 193. 1934.

Pleurofossa capillaris (Copel.) Nakai ex H.Ito, J. Jap. Bot. 12: 408. 1936.

Epiphytic. Rhizome short-creeping, densely scaly, ca. 1 mm diam including scales, bearing fronds close together. Rhizome scales narrowly lanceolate, gradually narrowing from the base towards a long-tailed and ultimately filiform apex, 0.7–2 mm long and 0.2–0.3 mm wide (at base), clathrate, toothed at margin, the walls thickened and dentate, sometimes reddish. Fronds sessile, simple, linear, 2.7–4.5 cm long, 0.6–0.8 mm wide, widest above the middle, gradually narrowing towards both ends, apex acute or (occasionally) retuse, laminae char-taceous, glabrous, margins entire, flat on both surfaces. Venation difficult to see even with transmitted light, simple, with only a midvein (in fronds with single sori). Epidermal silica bodies linear, 0.1–0.64 mm long, present on both sides of lamina, margin lobed, ends of lobes blunt. Sori linear, uninterrupted, following the midvein, covered by a soral flap and open to one side, very rarely fertile on both sides. Soral paraphyses copious, stalks uniseriate, filiform, colourless; apical cells large, funnel-shaped, length-to-width ratio of 1.7 ± 0.23, yellowish. Spores bilateral, bean-shaped, monolete, yellowish, surface smooth.

Distribution—Only known from 10 collections. The type and Chen Wade4959 are from Negros, the Philippines, seven collections are from Borneo (Sabah, Sarawak, Kalimantan, Brunei), and one collection is from an independent gametophyte population in Northern Taiwan.

Ecology—A low epiphyte on mossy tree trunks in damp forests. The lowest recorded elevation is 1100 m and the highest is 1600 m.

Additional Specimens Examined—Brunei.—TEMBURONG: Ulu Belalong, 540 m, 17 Jan 1994, Dransfield, J. 7350 (K! [K000507730]); NE of G. Retak, 1125 m, 10 Mar 1991, Johns, R.J. 6637 (K! [K000705731]). Indonesia.—EAST KALIMANTAN: Gunung Batu Harun, 1150–1650 m, 25 Jul 1981, Kato, M., Okamoto, M., & Walujo, E.B. B-9856 (KYO!); Gunung Batu Linanit, 1100 m, 31 Jul 1981, Kato, M., Okamoto, M., & Walujo, E.B. B-10358 (KYO! & L! [L3641722]). Malaysia.—SABAH: Mt. Kinabalu, ca. 1676 m [5500 ft], 15 Feb 1957, Molesworth Allen, B.E.G. 3217 (US! [US01485956], K! [K000699438]); Tenompok, 1432 m [4700 ft], 10 Nov 1931, Holttum, R.E. 25715 (SING! [SING0040889]); Tenompok, 1676 m [5500 ft], 5 May 1932, Clemens, J. & M.S. 29574 (G! US! [US01503549]); Mt. Trus Madi, Jaafar Nyiro Cabin, epiphyte on tree fern trunk, 1400 m, 10 May 2015, Chen, C.-W. Wade4360 (SAN! [SAN130433], TAIF! [TAIF468793]).—SARAWAK: Miri Division, Upper Baram, Tama Abu Protected Forest, patches of virgin and > 20 yr old logged over forest in adjacent to Sungai (River) Baleh, between Long Beruang and Pa Dali. Transect no. 7. Kerangas forest near the transect end, 980 m, 22 Aug 2017, Yong, K.T. 813 (TAIF! [TAIF513503]). Philippines.—NEGROS: Murcia, trail from Wasay DENR Ranger Station to Mt. Canlaon, 10.47358°, 123.1242°, 1300 m, 26 Oct 2017, Chen, C.-W. Wade4959 (CMUH! [CMUH00010961], TAIF! [TAIF512395]).