A K. linearis foliis lanceolatis vel anguste lanceolatis, inflorescentibus elongatis, bracteis floralibus ellipticis vel lanceolatis effusis, hypanthio late obconico vel campanulato differt. Etiam ordine rDNA ETS a K. linearis recedit.

(Fig. 40). New Zealand, Three Kings Islands group, Great Island, Lighthouse, 34°9'S, 172°8'E, 280 m a.s.l. ‘Forming a tree up to 8 m. Leaves with hairs fringing lamina, showing up as white margins on fresh material’ P. J. de Lange s.n., 4 Dec 1995, AK 226797! Isotypes. AD!

The specific epithet triregensis refers to this species being endemic to the Three Kings Island group. The recognition of Kunzea triregensis brings to 15 the number of endemic vascular plant taxa recognised for the Three Kings Island group.

(Figs 41, 42, 43). Growth habit mostly trees up to 18 × 3 m, forming a broadly rounded to somewhat spreading canopy with the lower 50–70% of the trunk usually completely devoid of branches. Trunk 1(–4), 0.10–0.60(–0.85) m d.b.h., mostly erect; basal portion of trunks covered with numerous semi-detached, long somewhat tabular lengths of rather corky-coriaceous bark. Bark early bark firmly coriaceous, grey or grey-brown, ± elongate, usually bearing a few transverse cracks (especially on branch flanges and decurrent leaf bases) otherwise remaining firmly attached, margins elongate sinuous, ± entire with scarcely any flaking; old bark similar though more distinctly corky-coriaceous, tessellated, firmly attached, detaching basally with age, and peeling upwards along trunk in broad, tabular strips, margins ± entire to weakly irregular; upper surface often deeply corrugated and cracked but not peeling; margins sinuous to lunate; early and old bark flakes firm, not crumbling in hand, snapping with a ± entire margin. Branches numerous, usually confined to the upper 30–50% of trunk; upright to somewhat spreading; branchlets numerous, slender, ± quadrangular to subterete, leaves ± evenly spaced along length; branchlets sericeous, indumentum copious; hairs long appressed, usually flexuose (220–)480(–520) μm long, hyaline to translucent (appearing white when young, maturing grey). Vegetative buds conspicuous; at resting stage 1.0(–2.2) mm diam., narrowly lanceolate; scales absent. Leaves sessile, well spaced along branchlets, spreading, patent to recurved; lamina (6.0–)10.0(–13.5) × (1.1–)1.8(–2.3) mm, dark glossy green above, paler beneath with leaf margins and midrib appearing distinctly white because of dense hair growth; lamina lanceolate to narrowly lanceolate; usually strongly recurved for about half of total length; apex acute to narrowly acute, base attenuate; adaxial surface usually deeply concave to weakly so, very rarely flat, oil glands not evident when fresh, conspicuous when dry, up to c.200; midrib slightly raised near base, otherwise not evident for rest of length, finely covered in antrorse-appressed, silky hairs in lower 50–70% otherwise glabrous; abaxial surface convex to v-shaped, glandular punctate, oil glands up to 200, more evident when dry; midrib raised for most of length, densely silky hairy to leaf apex, hairs weakly flexuose, antrorse-appressed, up to 0.8 mm long, hyaline to translucent, appearing as white to naked eye; lamina margin completely obscured by dense covering of antrorse-appressed hairs aligned in a thick, up to 0.6 mm wide, almost plumose, white band meeting at leaf apex and continuous down branchlets along decurrent leaf bases. Perules squamiform, ± persistent grading into pherophylls, (4.0–)8.2(–11.8) × (0.9–)1.6(–2.2) mm; dark glossy green, broadly oblong to oblanceolate, usually strongly recurved, weakly concave, oil glands not evident when fresh, conspicuous when dry, up to c.80, margins ± flat, margins and midrib densely covered in sericeous, appressed, hairs, midrib weakly keeled. Inflorescence an elongated (3–)10(–20)-flowered botryum up to 200 mm long, basal portion sometimes bearing compact, lateral 3-flowered corymbiform botrya, or with the basal and terminal portions occasionally bearing lateral elongate botyra; distal 70% often interrupted by sections of leafy perules between which are spaced further flowers; or interrupted by short floral shoots bearing elongated 3–6-flowered botrya up to 20 mm long; terminal portion often bearing undeveloped flowers and vegetative terminal growth. Inflorescence axis densely invested in antrorse-appressed, weakly flexuose, hairs. Pherophylls persistent, foliose, (6.0–)9.8(–12.8) × (0.9–)1.8(–2.2) mm, dark glossy green, elliptic, broadly lanceolate to lanceolate; strongly recurved, to about half of total length or flat; apex acute, base attenuate; adaxial surface usually deeply concave to weakly so, oil glands not evident when fresh, conspicuous when dry, up to c. 80 (usually fewer); midrib slightly raised near base, otherwise not evident for rest of length, finely covered in antrorse-appressed, silky hairs for whole length; abaxial surface deeply convex, glandular punctate, oil glands up to 100 (usually fewer), more evident when dry; midrib raised for most of length, densely covered in antrorse-appressed, silky hairs to apex, lamina margin obscured by dense covering of antrorse-appressed, silky hairs. Pedicels subsessile to pedicellate (0.4–)1.3(–3.7) mm long at anthesis, usually elongating slightly after anthesis, terete, copiously invested in antrorse-appressed, weakly flexuose, silky hairs. Flower buds double-conic to ovoid, calyx lobes prior to bud burst mostly not or scarcely meeting, held flat across bud surface, occasionally suberect with lobes ± meeting. Fresh flowers when fully expanded (6.3–)10.2(–12.3) mm diam., usually reducing in size toward end of flowering season. Hypanthium (1.6–)2.8(–4.4) × (2.0–)3.0(–4.6) mm, with free portion 0.6–0.8 mm long, dark green or red-green, drying green-brown or red-brown; hemispherical to broadly obconic, sometimes campanulate or rarely cupular, terminating in dark-green to red-green coriaceous rim bearing five persistent erect calyx lobes; hypanthium surface when fresh, smooth to faintly ribbed, faintly and sparingly dotted with pink or colourless oil glands, densely to sparsely covered in silky, appressed antrorse hairs; similar when dry though with the ribs more strongly defined and clearly leading up to calyx lobes. Calyx lobes 5, erect, coriaceous, (0.5–)0.9(–1.3) × (0.3–)0.5(–0.8) mm, persistent, deltoid to ovate-deltoid, green to red-green, prominently keeled, with keel usually slightly darker-coloured and densely covered in antrorse-appressed, hairs; margins pale green often flushed pink, glabrescent, surface somewhat glandular punctate, oil glands inconspicuous, ± colourless. Receptacle green at anthesis, consistently darkening to crimson after fertilisation. Petals 5(–6), (1.3–)2.8(–4.3) × (1.9–)2.8(–4.8) mm, white, orbicular to broadly ovate, apex rounded, margins ± finely and irregularly denticulate, often when fresh appearing to be finely folded or crimped 1–3 or more times, oil glands colourless. Stamens 30–46(–53) in 1(–3) weakly defined whorls, arising from receptacular rim, filaments white. Antipetalous stamens (2–)3(–5) sometimes petaloid, antisepalous stamens (3–)4(–6). Outermost antipetalous stamens incurved or weakly outcurved, on filaments 1.0–3.8 mm long, inner stamen if present, 0.9–1.8 mm, incurved, with a further 1–3 stamens, of similar length to inner stamen often present at the base of the outermost antipetalous pair. Antisepalous stamens usually shorter than outermost antipetalous stamens, but sometimes of comparable length, generally 0.9–3.8 mm, weakly to strongly incurved, rarely outcurved, usually in mixtures of both. Anthers dorsifixed, 0.05–0.10 × 0.06–0.08 mm, testicular-ellipsoid, latrorse. Pollen white (12.0–)13.8(–16.0) μm. Anther connective gland prominent, pink or golden-yellow when fresh, drying yellow to pale orange, spheroidal, finely to coarsely papillate. Ovary 4(–5) locular, each with 20–24(–38) ovules in two rows on each placental lobe. Style (1.9–)2.8(–3.1) mm long at anthesis, elongating after anthesis, white or pinkish-white; stigma broadly capitate, conspicuously wider than style, ± flat, greenish-white or pale pink, flushing red after anthesis, surface granular-papillate. Fruits long persistent, (1.9–)3.2(–5.2) × (2.0–)3.1(–4.9) mm, initially dark chestnut-brown to almost black, fading with age to grey, hemispherical, broadly obconic, campanulate to cupular; calyx valves usually prominently erect to suberect, rarely incurved, splits concealed by dried, erect, free portion of hypanthium. Seeds 0.50–1.00(–1.10) × 0.50–0.60(–0.80) mm, oblong, oblong-obovate, curved near apex, laterally compressed, 2–3-angled with convex to flattened faces, apex rounded to subacute, base oblique, ± flattened; testa semi-glossy, orange-brown to dark brown; surface coarsely reticulate. FL: (Oct–)Dec(–May). FT: Oct–May. Chromosome Number n = 11_II, 2n = 22 (see de Lange and Murray 2004).

(30 sheets seen). New Zealand, Three Kings Island group. North East Island, G. F. Buddle s.n., 31 Dec 1947, (AK 24092); Manawatawhi / Great Island, T. F. Cheeseman s.n., Nov 1889., (AK 5516); Manawatawhi / Great Island, W. R. B. Oliver s.n., 20 Feb 1939, (WELT SP029481); Manawatawhi / Great Island, North East Bay, Isthmus Summit, P. J. de Lange 1105, 16 Oct 1991, (AK 207160, 207317, Duplicates: AD, CHR); Manawatawhi / Great Island, Tasman Stream, M. J. Thorsen s.n., 8 Apr 2000, AK 289060-289063; South-West Island, G. T. S. Baylis s.n., 10 Jan 1950, (OTA 3806); West Island, P. J. de Lange 3180, 5 Dec 1996, (AK 231919, Duplicate: HO).

(Fig. 7). Endemic. Three Kings Island group (sea level – 296 m a.s.l.).

Kunzea triregensis a Three Kings Island group endemic, is the only New Zealand Kunzea to be truly allopatric.. It is recognised here at species rank through a combination of morphological, reproductive and molecular characters (de Lange et al. 2005; de Lange 2007; Table 1). Morphologically, the distinctive elongate botrya (Fig. 41A, 42D–E) of the species is seen otherwise only in Kunzea amathicola, a species from which Kunzea triregensis differs by its rDNA ETS sequence (de Lange 2007), homophyllous growth habit, and lanceolate to narrowly lanceolate leaves (Fig. 41B–F; Table 1). The peculiar ability of the Kunzea triregensis inflorescence to produce, albeit infrequently, additional lateral elongate or reduced corymbiform botrya from the base and terminus of the main botryum further distinguishes it from Kunzea amathicola. Prior to this treatment Cheeseman (1906, 1914, 1925) and some later authors who appear to have uncritically followed him (e.g., Poole and Adams 1963; Wardle and Platt 2011) had referred Kunzea triregensis to the Aotea (Great Barrier Island) endemic Kunzea sinclairii (Kirk) W.Harris. Cheeseman’s error is difficult to understand. Kunzea triregensis has no close morphological or molecular affinity to Kunzea sinclairii, and although Cheeseman never saw Kunzea sinclairii in the wild he was furnished with specimens by Thomas Kirk, and indeed had it illustrated from these in Cheeseman (1914). The confusion may have arisen because on Cheeseman’s visit to Manawatawhi / Great Island, the largest island in the Three Kings Island group, he saw what he described as ‘suberect’ plants growing on the ‘declivities leading down to the cliffs’ (Cheeseman 1914). Further, because the leaf margins and abaxial midribs of Kunzea triregensis are copiously covered in white to silvery-white plumose hairs, and Kirk (1899; p.158) had emphasised ‘white silky hairs’ in his description of the leaves of Leptospermum (Kunzea) sinclairii, this may have influenced Cheeseman in his decision. It is also clear that Cheeseman (1914; caption facing Leptospermum sinclairii, pl. 47) had doubts as to the validity of Kunzea sinclairii, though he did conclude that Three Kings Islands plants had ‘leaves [that] were slightly narrower than in the Barrier plant’. Whatever the reason for Cheeseman’s decision, Kunzea triregensis differs markedly from Kunzea sinclairii, indeed it was referred by Oliver (1948) to Kunzea ericoides.

Nevertheless, to clarify any further ambiguity, some distinctions between Kunzea triregensis and Kunzea sinclairii are here offered (see also Table 1). Kunzea triregensis is usually a forest tree rather than a scrambling shrub. Although prostrate forms of Kunzea triregensis are known from the wild, cultivation has shown that these are environmentally induced, unlike the genetically fixed scrambling condition of Kunzea sinclairii. These two species differ markedly by their leaf colour, shape and degree of investiture (in Kunzea triregensis the leaves are dark green, lanceolate to narrowly lanceolate with the lamina margins and abaxial midrib copiously covered in white hairs, and the intervening lamina glabrous; in Kunzea sinclairii the leaves are consistently grey-green to grey, broadly lanceolate, elliptic to oblanceolate and completely hairy). Another key difference is the inflorescence. In Kunzea triregensis these are consistently in the form of an elongate botryum with leafy pherophylls, while Kunzea sinclairii has corymbiform botrya with small, deciduous pherophylls (Table 1).

Kunzea triregensis has also been confused with Kunzea linearis. Kunzea tiregensis differs from Kunzea linearis by its more openly vegetated, less densely crowded branchlets, and by the leaves which in Kunzea triregensis are consistently lanceolate to narrowly lanceolate rather than linear (Fig. 41C–F). Further, in Kunzea triregensis the thick bands of marginal and abaxial midrib hairs meet at the leaf apex, whereas in Kunzea linearis the marginal hairs meet just short of the adaxial face of the apex (Fig. 41E) and the abaxial midrib hairs stop short of the apex. The inflorescence of Kunzea triregensis is consistently elongated and the flowers are usually widely spaced, only in stressed conditions becoming more crowded (Figs 41A, 43C, E). In contrast, the inflorescence of Kunzea linearis is usually a condensed, densely packed spiciform botryum. However, in shade forms or late season flowering specimens of Kunzea linearis the inflorescences may elongate considerably, in which case distinction between Kunzea triregensis and Kunzea linearis using inflorescence type is less clear. In shade specimens, the early season inflorescences of Kunzea linearis will still show the typically condensed spiciform condition, and the flowers are mostly sessile to subsessile rather than the mostly pedicellate to subsessile condition of Kunzea triregensis. The pherophylls of both species are also diagnostic; those of Kunzea linearis are consistently linear while those of Kunzea triregensis are elliptic or broadly lanceolate to lanceolate (Fig. 41A). While the the calyx lobes of mature flower buds of Kunzea triregensis may sometimes be suberect and touching, and therefore resemble those of Kunzea linearis, the usual condition is that the calyx lobes are held flat or curved across the domed bud surface, and do not touch (see Figs 41A, 43D). Cytologically the chromosome karyotype of Kunzea triregensis has none of the marked size differences typical of Kunzea linearis (see de Lange and Murray 2004). Further, the ETS sequence of Kunzea linearis is particularly distinctive and that species shows no obvious relationship to Kunzea triregensis (see Kunzea linearis) (de Lange 2007). It is because of these distinctions that allopatric Kunzea triregensis is treated at species rank.

Five collections of Kunzea from the two main islands of the Poor Knights Island group, Aorangi (e.g., L. B. Moore s.n. & L. M. Cranwell (AK 102471)) and Tawhiti Rahi (e.g., B. S. Parris s.n. (AK 128064); A. E. Wright 3970 (AK 155364); A. E. Wright 11413 (AK 201664); E. K. Cameron 10274 (AK 252512)), are morphologically similar to Kunzea triregensis. The Tawhiti Rahi specimens differ from Kunzea triregensis mainly by their extremely linear leaves which are densely crowded along the branchlets, and which range from being rather hairy to almost glabrous. Otherwise the plants have elongate botrya similar to those of Kunzea triregensis. However, the fruits of these specimens are mostly barrel-shaped to cupular and vary from glabrate to distinctly hairy. These are features of Kunzea linearis, from which they differ by their shorter (up to 8 mm long) mostly spreading rather than ascending leaves, and shortly pedicellate rather than sessile fruits. The sole gathering from Aorangi is even more like Kunzea triregensis in that it has much broader lanceolate leaves but the fruits differ in that they are glabrate, up to 5 × 5 mm, mostly barrel-shaped (with a very few broadly obconic), and more or less consistently long-pedicellate. As both Kunzea linearis and Kunzea robusta have been collected from the Poor Knights Island group, these five gatherings are most likely examples of an introgressed hybrid swarm involving these two species. The distinctly linear-leaved Tawhiti Rahi plants are closer to Kunzea linearis than Kunzea robusta, while the Aorangi specimen is closer to Kunzea robusta. Seed that I have germinated from Aorangi Island (Poor Knights Island group) examples of these plants suggested they are also hybrids, as the seedlings showed clear segregation to both the postulated parents. It is plants such as these that appear to be the basis for the erroneous statement by Hynes (1950) that Kunzea sinclairii was found on the Poor Knights Islands (see de Lange and Cameron 1999; p. 464).

The DNA sequence data placed Kunzea triregensis next to Kunzea robusta (de Lange 2007), from which it differs only by the presence of an indel within the ETS sequence at alignment position 269 (Table 2). The same indel is universal to all the Australian members of the Kunzea ericoides complex but was found in New Zealand otherwise only in a single sample from a Kunzea of uncertain status sampled from Lottin Point, East Cape (de Lange 2007; Table 2).

Kunzea triregensis is the dominant woody tree on the Three Kings Island group where it occurs from near sea level to the summits of North East, Manawatawhi / Great Island (Fig. 43A, B) and West Islands. On South West Island it seems to be naturally uncommon. The flora of Manawatahi / Great Island is an assemblage of what survived after at least 60 years of intensive goat (Capra aegagrus hircus (Linnaeus, 1758)) browse (Baylis 1948, 1951; Oliver 1948), thus it retains few truly tree-forming species, as these were mostly extirpated. Of the few that still exist, most have fruits that depend on large birds for dispersal, and, as these are now absent from the Three Kings, such trees remain trapped in places where the goats could not reach (mostly cliff refugia), and so are unable to spread. For this reason, following the eradication of goats from the island, Kunzea triregensis, which would more usually be a short-lived successional forest species, has formed what is probably a self-sustaining forest type (Fig. 43B).

As a result, Kunzea triregensis is the dominant tree on Manawatawhi / Great Island, being scarce or absent only from steep, sparsely vegetated coastal cliffs, and boulder beaches. It is also common on North-East Island, though there it is being slowly replaced by Meryta sinclairii (Hook.f.) Seem. forest. On the exposed wind shorn cliff tops of Manawatawhi / Great Island, Kunzea triregensis presents often as environmentally induced decumbent to semi-erect, widely spreading bushy shrubs

On Manawatawhi / Great Island, Kunzea triregensis has been recorded as the host species for the threatened polypore fungus Dichomitus newhookii P.K.Buchanan et Ryvarden (McKenzie et al. 2006). The bark of the mature trees also provides a refuge for an unnamed gecko (Woodworthia “Three Kings”) endemic to the Three Kings Island group (R. Hitchmough pers. comm.).

Kunzea triregensis, being allopatric from the other New Zealand members of the Kunzea ericoides group does not naturally form hybrids. However, experimental hybrids were readily produced using Kunzea triregensis as pistillate or staminate parent (de Lange et al. 2005, as Kunzea aff. ericoides (e)).

Although the hybrid Kunzea amathicola × Kunzea linearis was not synthesised (de Lange et al. 2005), putative wild hybrids show a great similarity to Kunzea triregensis, suggesting that Kunzea triregensis may be a stable hybrid between these two species. Future research into the possible past hybrid origin of Kunzea triregensis, including the synthesis of Kunzea amathicola × Kunzea linearis would be worthwhile.

Kunzea triregensis appears to have no specific Maori name.

Kunzea triregensis as Kunzea aff. ericoides (e) (AK 226797; Three Kings) is appropriately listed by de Lange et al. (2013b) as ‘At Risk/Naturally Uncommon’ qualified ‘IE’ (Island Endemic) and ‘OL’ (One Location) because the species is confined to one island group.