Introduction

The genus Drosera L. (Droseraceae, Nepenthales or noncore Caryophyllales) comprises c. 250 species of carnivorous herbs commonly referred to as sundews, which catch small invertebrate prey by means of adhesive flypaper traps formed by their tentacle-bearing leaves. There are four centres of species diversity (hosting more than 20 species) in the genus, which are all situated in the S hemisphere: the SW part of Western Australia, tropical N Australia, the SC Brazilian highlands and the Cape Floristic Region of South Africa. With the exception of tropical N Australia, these Drosera diversity hotspots lie in Mediterranean or subtropical climates, and are found in open, low shrubby vegetation over nutrient poor, usually sandstonebased, acidic soils (Fleischmann & al. 2018).

The mountain ranges of the Overberg District, Western Cape Province, South Africa, which range from Franschhoek to Kleinmond (covering Hottentots Holland Mountains and Kogelberg Mountains) host the highest plant biodiversity and highest number of species endemism of the Cape Floristic Region, in a unique but threatened vegetation type referred to as the Kogelberg Sandstone Fynbos (Rebelo & al. 2006; Mucina & Rutherford 2006). These mountains are also the centre of diversity of the genus Drosera in the Cape, with at least 14 species, of which four are endemic to the region. A floristically very similar vegetation type within the Fynbos Biome, the Overberg Sandstone Fynbos, is found in the mountains from Bot River in the west to Bredasdorp in the east, including the Babylonstoring Mountains, the Kleinrivier Mountains and Napier Mountain (Rebelo & al. 2006). The Fernkloof Nature Reserve in the Kleinrivier Mountains is particularly rich in Drosera: nine species and one natural hybrid have been recorded so far (Gibson & Green 1999). The narrow endemic D. slackii Cheek is confined to the Overberg, where it has been recorded from the Kleinrivier (Fernkloof and Phillipskop) and Kogelberg (Palmietriver valley) Mountains (see specimens examined; Bennett & Cheek 1990).

Some confusion arose about a peculiar, conspicuous Drosera with bright green, spatulate leaves and contrasting, bright red tentacles that grows on comparatively dry quarzitic soils in the coastal mountains of the Overberg District. These plants have been known for a long time, and have been occasionally collected and photographed, for example from the Fernkloof Nature Reserve near Hermanus (e.g. Gibson 1997, 1998). The name D. curviscapa T. M. Salter has long been applied erroneously to these plants. However, close examination of Salter's type material and description (Salter 1940) revealed that his D. curviscapa is conspecific with the widespread D. aliciae Raym.-Hamet (a name of which Salter was obviously unaware). The description of D. aliciae (Hamet 1905, published in September 1905 when that author was 15 years old) only briefly predated Diels's generic treatment for Engler's Das Pflanzenreich (Diels 1906, printed in March–June 1906, issued at the end of July 1906) and was probably unknown to Diels until his monograph went to print. Hamet likewise was unaware of Diels's forthcoming monograph (see Hamet 1907), so it is reasonable to assume that both authors did not correspond at the time. Hence, D. aliciae was not included in Diels's treatment, which was for long a trusted source of information for contemporary botanists. Note that, in Diels's monograph and prior to that (e.g. Sonder 1859), all flat-rosetted Drosera species from the Cape with stipulate, narrowly cuneate to cuneiform leaves had been treated as a single species, D. cuneifolia L. f., comprising, apart from D. cuneifolia, those species here recognized as D. admirabilis Debbert and D. aliciae. As a consequence, the name D. aliciae had frequently been overlooked in the early 20^th century, and the younger, heterotypic synonym D. curviscapa had been unknowingly applied to that taxon (e.g. by Salter 1950; Exell & Laundon 1956; Batten & Bokelmann 1966; Taylor 1978). It was not until 1970 (Kress 1970; Obermeyer 1970) that the forgotten name D. aliciae was resurrected and became rightly established for the taxon, as first circumscribed by Hamet (1905, 1907), and the previously applied D. curviscapa was sunk into its synonymy. Despite this, Salter's name D. curviscapa was sometimes still erroneously used in parallel for a notably distinct taxon of drier, quarzitic soils from the Overberg mountains (Debbert 1992; Gibson 1997, 1998; Rivadavia 2000). This sundew turned out to represent a distinctive and as yet unnamed species, which is described as new to science here.

Material and methods

Measurements were taken from dried herbarium specimens from B, BOL, G, K, M, NBG, PRE and WAG (herbarium codes following Thiers (2017+). Information on ecology and habitats was taken from herbarium label data, as well as from personal observations made in situ in the Western Cape in September 2006 (SANBI Flora Research Permit No. AAA005-00064-0028). The maps in Fig. 3 are based on an OpenStreetMap image for the Western Cape retrieved from  https://maps-for-free.com (accessed July 2017), colour-adjusted and edited with Adobe Photoshop; distribution data on the maps are from the specimens examined and georeferenced photographs from iSpot Nature ( https://www.ispotnature.org, accessed July 2017).

Preparation of herbarium material — Flat-rosetted Drosera species with decumbent, cuneate to oblong leaves are among the most taxonomically difficult in the genus regarding species identification (see, e.g., Rivadavia & al. 2014) because diagnostic foliar characters, such as stipule shape and abaxial leaf indumentum, are usually obscured by the compact growth. This is particularly true in pressed herbarium specimens, where the dense, spiral leaf arrangement often does not allow the examination of lower leaf surfaces and the characteristic, membranous stipule adnate to each leaf base. For identification purposes, careful dissection of a single leaf (comprising the entire leaf base including its stipule) from the rosette is recommended for all rosetted Drosera, although this can sometimes be difficult in brittle herbarium material. The dried detached leaf material can then be re-softened in boiling water, in order to spread the leaf and stipule for microscopic study, and can be re-pressed afterwards (ideally, the leaf preparation should be returned to the respective herbarium specimen). When collecting fresh material to make herbarium specimens of flat-rosetted Drosera species, to facilitate later identification, at least one mature leaf should be pulled from the rosette directly in the field (taking care to remove the entire leaf base with its stipule) and pressed separately with the collected plant. It is recommended to detach at least two leaves from each fresh rosette when preparing herbarium specimens — one to be fixed later to the herbarium sheet showing its adaxial (tentacle-bearing) side, the other showing its abaxial (lower) side.

Results

Drosera xerophila A. Fleischm., sp. nov. — Fig. 1, 2, 4A, 5B.

Holotype: South Africa, South Western Region [Western Cape], Caledon Division, NE of Somerset Sneeuwkop [Sneeukop] between Somerset, Sneeuwkop & Guardian, 3000–3500 ft [915–1067 m], sandy slopes, 6 Jan 1944, Esterhuysen 9802b (BOL201166! [4 plants marked “b”; Fig. 5B]).

Diagnosis — Related to Drosera esterhuyseniae (T. M. Salter) Debbert, but differs from that species in having broadly spatulate, decumbent, rosette leaves (leaves narrowly cuneate to rectangular, held upright at an angle of 70°–30°) and glabrous, emarginate to bifid stigmatic tips (stigmas entire, knob-shaped, papillate). Resembles D. aliciae Raym.-Hamet due to similar flat, rosetted habit and basally ascending inflorescence scape, but differs from that species in having well-pronounced, cuneate, eglandular petioles (3-)5-10 mm long (petiole 1-3(-5) mm long and hence leaves appearing almost sessile [“apetiolate”] in D. aliciae), in style arms entire or only shortly bifid in distal ⅓-¼ (style arms forking below middle, usually again bifid or tripartite in distal part) and fusiform seeds without terminal appendages (seeds narrowly fusiform with filiform appendage on micropylar and chalazal end).

Description — Rosetted perennial herb, aged specimens forming column-like, condensed stems to 2 cm tall, densely covered by persistent dead leaves; rosettes of flowering individuals (2-)3-5 cm in diam. (excluding long marginal tentacles). Roots terete, dark, sparsely branched. Leaves 14–25 mm long (excluding marginal tentacles), bright green to yellowish green with bright red tentacles, leaf vernation geniculate, active leaves 8–14, decumbent, broadly spatulate (rarely nearly cuneate), apex obtuse; petiole narrowly cuneiform, (3-)5-10 mm long, c. 1 mm wide at base, gradually widening to 2–3.5 mm, adaxial surface glabrous (sometimes covered with single white, eglandular hairs), lateral margins and abaxial surface densely covered with appressed, upward-facing, straight or only slightly curled, white-translucent, simple, eglandular hairs 2–3.5 mm long, hairs tapering toward acute apex; lamina (tentaclebearing part of leaf) spatulate, orbicular or transversely elliptic, 7–15 long, 2–3.5 mm wide at base, 5–10 mm wide at greatest width, apex rounded, adaxial surface covered with numerous red, carnivorous, capitate, radially symmetrical tentacles, marginal tentacles at leaf apex unifacial, to 0.8(-1) mm long, with ellipsoid gland head and narrowly triangular, flattened stalk, stalk base to 0.05 mm wide; lamina abaxial surface with appressed, simple, white-translucent hairs 0.5–1 mm long; stipule very narrowly triangular, 5–6 mm long, c. 1.5 mm wide at base, membranous, translucent white (drying pale brown), basalmost 1–1.5 mm entire, then divided into 3 unequal segments, 2 lateral segments entire, lineartriangular, tapering from c. 0.25 mm wide base to acute apex, 5–6 mm long, subequalling or exceeding median segment in length; median segment narrowly triangular, 3.5–4.5 mm long, c. 1 mm wide at base, apical ⅓ laciniate, divided into 2 or 3 fimbriae. Scape 1(or 2); peduncle (10-)12-20 cm long (including floriferous part), 1–2 mm in diam. near base, arcuate at very base, terete, reddish, glabrous (single, appressed, white, eglandular hairs on very few basal mm), floriferous part densely covered with small, short-stalked, reddish glands. Inflorescence a simple scorpioid cyme, bearing up to 9 flowers, each c. 20 mm in diam., opening consecutively, lasting only one day; bracts caducous, 1-3 × 0.1-0.2 mm, narrowly obovate to subulate, glandular pilose; pedicel 1–5 mm long at anthesis, to 10 mm long in fruit, sparsely glandular pilose. Sepals 3-4 × 1-1.5 mm, basally connate, persistent in fruit, obovate to elliptic, sparsely glandular pilose, apex obtuse to acute; petals 10-12 × 8-10 mm, broadly obovate, light pink to violetpink. Stamens 5, c. 4 mm long; filaments white, apically strongly dilated; anthers 3–4 mm long, bithecate, yellow. Ovary 3-carpellate, fused, c. 1 mm in diam., globose, greenish white; styles 3, each bifurcate from base, style arms c. 5 mm long (including stigmas), c. 0.2 mm wide, terete, dark pink, sometimes apically bifurcate; stigmas obtuse, emarginate or shortly bifurcate, translucent white, stigmatic surface smooth. Fruit a dry dehiscent capsule, ellipsoid, 3-valvate. Seeds fusiform, without terminal appendages, 1–1.2 mm long; testa black, reticulate.

Phenology — Collected and photographed in both flower and fruit in December and January.

Distribution — Drosera xerophila is endemic to the coastal mountains of Overberg District, Western Cape (local municipalities of Overstrand and westernmost part of Theewaterskloof), from Hottentots-Holland Mountains to Kleinmond (Kogelberg Mountains), frequent in the Kleinrivier Mountains near Hermanus and extending eastwards to the Napier Mountains (Soetmuisberg) (Fig. 3). This distribution pattern is shared with the often sympatric, closely related D. esterhuyseniae (T. M. Salter) Debbert, which however extends further to the northwest (Babyionstoring, Caledon Swartberg and Franschhoek), from where D. xerophila has not yet been recorded. A similar distribution range is shared by the Cape endemic genera Orothamnus Pappe ex Hook. (Proteaceae), Pillansia L. Bolus (Iridaceae), Retzia Thunb. (Stilbaceae) and Sonderothamnus R. Dahlgren (Penaeaceae) (Rebelo & al. 2006), and by Otholobium thomii (Harv.) C. H. Stirt. (Fabaceae) (Charles Stirton, pers. comm.).

Ecology and habitats — Drosera xerophila grows in Sandstone Fynbos vegetation over Table Mountain Sandstone, on well-drained ground in sandstone gravel, in cracks of sandstone rock, or in sandy soils with little organic matter (Fig. 2A, B).

A few other perennial Drosera species of the Western Cape share similar habitat preferences of only slightly moist soils in fynbos habitats, namely the stem-forming species D. ericgreenii A. Fleischm. & al., D. esterhuyseniae, D. glabripes (Harv.) Stein, D. hilaris Cham. & Schltdl., D. ramentacea Burch, ex Harv. & Sond. and the rosetted D. cuneifolia — with D. esterhuyseniae and D. glabripes often growing together with D. xerophila. During the dry season, these species slow down growth and formation of new leaves and, especially in the caulescent species, active leaves often desiccate and the plants survive the dry period as a dormant apical bud (see Fleischmann & al. 2008). Interestingly, none of these species dwelling on drier ground forms adventitious plantlets from the laminae of detached leaves. This mode of asexual propagation is common in those Drosera species preferring wetter habitats, such as perennially wet seepage areas in mountain wetlands (e.g. D. admirabilis, D. aliciae, D. rubrifolia Debbert and D. slackii). Therefore, the ability of sundews to reproduce asexually from leaves (a feature widely made use of as leaf cuttings by horticulturalists; see, e.g., Slack 1979) could represent an ecological adaptation to wet habitats (and often seasonally flooded conditions), where this mode of vegetative clonal growth also occurs naturally (pers. obs.). However, it is of low use in drier conditions, where soil moisture does not last long enough to establish leaf-derived plantlets, and hence the genetic ability to reproduce asexually from leaves might have been lost. Interestingly, the same is paralleled in Brazilian tetraploid Drosera, where wet-growing taxa will reproduce from detached leaves, in contrast to their congeners growing in dry soil (pers. obs.).

Fig. 1.

Drosera xerophila — A: habit; B: leaf adaxial surface, stipule detached and pictured to right; C: leaf abaxial surface, stipule reflexed; D: petal; E: stamens, left: face view, right: lateral view; F: gynoecium and styles. — A from Esterhuysen 9802b; B–F from Fleischmann s.n. — Drawing by A. Fleischmann.

Fig. 2.

Drosera xerophila — A: habitat in sandy fynbos vegetation in Fernkloof Nature Reserve; B: habit of plants in comparatively dry, quarzitic soil; C: leaf; note long marginal tentacles; D: rosette with developing scape; E: aged specimen with column of old leaf remnants; note burrow of a trapdoor spider in base of stem; F: inflorescence. — A–E: South Africa, Western Cape Province, Fernkloof Nature Reserve, 15 Sep 2006; F: in cultivation; all photographs by A. Fleischmann.

Fig. 3.

Distribution of Drosera xerophila and related taxa, based on the specimens examined and the georeferenced photographs from iSpot Nature cited in this paper.

Conservation status — Near Threatened (NT), sensu IUCN (2001). Although all known populations lie within the ranges of nature reserves, Drosera xerophila occurs in localized and sometimes small populations restricted to a very special habitat. Like most fynbos local endemics, populations can be negatively affected by inappropriate fire regimes (prevention of wildfires or too frequent burning), and human impact on soil hydrology (through drainage or channelling of habitat for urban water supply).

Etymology — The specific epithet denotes the fact that this species prefers to grow on well-drained, drier, only slightly moist ground (xerophilus = dry loving). This is not unusual for several species of Drosera from Mediterranean areas in the S hemisphere, but contrasts with the “conventional”, largely N-hemisphere-based experience of sundews as typical bog plants.

Discussion

Specimens that can be assigned to Drosera xerophila were treated under D. aliciae by Obermeyer (1970); however, they can be readily distinguished from the latter species as follows. In D. aliciae, which also occurs in the Fernkloof valley (but never together with D. xerophila, see Ecology and habitats), each of the six style arms is apically further divided into two or three stigmatic tips (vs the six style arms entire [or rarely shortly bifid] in D. xerophila; Fig. 1F, 4A, G). The leaves, although of similar overall shape and size in both species, in D. aliciae consist of a much longer, cuneate, glandular lamina and a barely noticeable, short petiole 1-3(-5) mm long (hence “tentacles on the entire leaf surface” or “leaves apetiolate”; Fig. 4G). In D. xerophila, in contrast, the very broadly spatulate (to cuneate) leaves consist of a more or less orbicular lamina (the glandular part of the leaf), which graduates continuously into a cuneate to narrowly cuneate petiole 5–10 mm long, which is devoid of carnivorous glands and makes up a large part of the leaf (hence “tentacles only in the upper half of the leaf”; Fig. 4A).

Drosera xerophila is closely related to D. esterhuyseniae. The close affinity between these two species was first noticed by Debbert (1992), who raised Salter's D. curviscapa var. esterhuyseniae (Salter 1944) to specific rank, as D. esterhuyseniae. However, he was erroneously assuming that Salter's autonymic variety, D. curviscapa var. curviscapa, was the dry-loving plant described here as D. xerophila. Yet Salter (1940) clearly described his D. curviscapa as “a marsh species and it is often locally abundant in damp peaty places […]” — habitat preferences that clearly fit D. aliciae, as well as the related D. admirabilis (Debbert 1987; Gibson 2002; pers. obs.) — and it was just for his “var. Esterhuysenae [sic!]” that Salter (1944) noted: “It appears to grow in drier situations than the typical form”. From observations made in situ, Gibson (1997, 1998) and others (Eric Green, pers. comm.) were also well aware of these two distinct, rosetted taxa present in the Fernkloof, and depicted two “variants” of what they commonly referred to as D. curviscapa, and also highlighted the different ecological preferences of both. However, Gibson's (1998) “Drosera curviscapa Hemanus-1” is actually the dry-loving species from sandy soils newly described here as D. xerophila, whereas his “D. curviscapa Hermanus-2” refers to D. aliciae. That D. aliciae prefers drier places, as concluded by Bennett & Cheek (1990), could not be confirmed from personal observations in situ or from herbarium label data.

Apart from different habitat preferences, Drosera aliciae can be easily distinguished from D. esterhuyseniae and D. xerophila by floral morphology (the style arms are twice bifid to multifid in the smaller flowers of D. aliciae, but undivided to just apically shortly bifid in the latter two species, which also have distinctly larger flowers; Fig. 4A, C, G) and by its seed shape (seeds fusiform, pale brown with a light brown filiform appendage on both the micropylar and chalazal ends in D. aliciae, compared to black fusiform seeds without appendages in D. xerophila and D. esterhuyseniae).

Although they often occur sympatrically and share several taxonomic characters, Drosera xerophila and D. esterhuyseniae remain stable and morphologically distinct in cultivation and breed true from seed (pers. obs.). Therefore, they cannot simply represent ecologically induced growth forms of the same taxon, and are accordingly treated here as two distinct species.

Fig. 4.

Comparison of leaves (left half illustrating adaxial surface, right half illustrating abaxial surface) and styles (one of three styles of a flower shown). — A: Drosera xerophila (from Fleischmann s.n.); B: D. rubrifolia (Debbert 132); C: D. esterhuyseniae (Debbert 123); D: D. cuneifolia (Froembling 453); E: D. slackii (Cheek 2015); F: D. admirabilis (Debbert 15); G: D. aliciae (Kress s.n.). — All at same scale as A. — Drawing by A. Fleischmann.

Fig. 5.

A: Lectotype of Drosera curviscapa var. esterhuyseniae T. M. Salter (≡ D. esterhuyseniae (T. M. Salter) Debbert), Esterhuysen 9802a (BOL135702). — B: Holotype of Drosera xerophila A. Fleischm., Esterhuysen 9802b (BOL201166). — Reproduction of herbarium scan with kind permission from Bolus Herbarium (BOL), University of Cape Town, South Africa.

Table 1.

Morphological comparison of the rosetted hemicryptophyte Drosera species from the Western Cape. Data from herbarium specimens and plants studied in situ.

Salter's (1944) description of Drosera curviscapa var. esterhuyseniae was most likely at least partly based on both taxa here recognized as D. esterhuyseniae and D. xerophila, as is evident from the underlying type gathering, Esterhuysen 9802, or at least from the duplicate conserved at BOL, which is a mixture of both species now curated as two specimens mounted together on one herbarium sheet: the eight individual plants of D. esterhuyseniae curated as Esterhuysen 9802a (BOL135702) and the four plants of D. xerophila as Esterhuysen 9802b (BOL201166) (Fig. 5). The duplicates at K and PRE exclusively consist of D. esterhuyseniae. However, in his description of the vegetative parts and his illustration of the leaf, Salter (1944) clearly depicted D. esterhuyseniae (sensu Debbert 1992), a specimen of which is therefore designated here as the lectotype for D. curviscapa var. esterhuyseniae (see under D. esterhuyseniae below). Debbert's amended description and material cited for D. esterhuyseniae (Debbert 1992) exclusively refer to that species, and do not include D. xerophila.

Drosera aliciae is a widespread species, extending along the S coast into the E part of the Western Cape Province (in the Langeberg Mountains in the north, e.g. McDonald 1645 [NBG!], east in Knysna District and in the Kammanassie Mountains, Oudtshoorn, e.g. Matthews 1035 [NBG!]) and even into the W part of the Eastern Cape Province as far as Port Elizabeth [Obermeyer 1970]). All other rosetted Drosera species treated here are restricted to the mountains in the W part of the Western Cape (Fig. 3). In the E part of its range, D. aliciae is usually white-flowered, and becomes difficult to distinguish from two other, often sympatrically occurring, rosetted Drosera species that are more widespread in the Eastern Cape: D. natalensis Diels and D. venusta Debbert. These three species share similar floral morphology (styles twice bifid to multifid with filiform stigmas), fusiform seed shape, and overall habit, but can be distinguished by the number, orientation and shape of the leaves (see Obermeyer 1970; Debbert 1987). In the Kogelberg and Kleinrivier mountains, individuals of D. aliciae are particularly large, with wider leaves and larger flowers, and stylar arms that are often multifid. Considering that D. aliciae is known to occur in different ploidy levels (Kress 1970; Bennett & Cheek 1990), these could perhaps represent polyploid populations.

Rosetted hemicryptophyte Drosera species from the SW Western Cape (municipalities of Cape Town, Overberg and Cape Winelands)

A key to the stem-forming species was provided by Fleischmann & al. (2008).