Biological flora of New Zealand 14: Metrosideros excelsa, pōhutukawa, New Zealand Christmas tree

Abstract

We review the biology and ecology of Metrosideros excelsa (Myrtaceae), an endemic angiosperm evergreen tree. Metrosideros excelsa belongs to a conspicuous and widely distributed Pacific Basin genus, with centres of diversity in both New Zealand and New Caledonia. Metrosideros excelsa is an iconic tree species that forms a significant component of northern New Zealand's exposed coastal headland and cliff vegetation. Where conditions are more favourable, M. excelsa forms tall coastal forest, ranging from simple young high-density stands to diverse mature forest. Inland, M. excelsa stands are confined to the margins of lakes and rivers on the Central Volcanic Plateau, where some may originate from early Māori plantings. Metrosideros excelsa is reliant on stochastic disturbance events (e.g. landslides, volcanic eruptions) to create open sites necessary for regeneration. Mass flowering (December–January), followed by abundant production of wind-dispersed seed maximises chance colonisation of such sites. Since human settlement in New Zealand, the distribution of M. excelsa forest has declined by c. 90% and the southern limit of the species has retreated north. Natural regeneration on the mainland is limited by the infrequency of large-scale disturbances and increased anthropogenic and herbivore pressures. Consequently, M. excelsa forest has become rare and localised on the mainland; monitoring and active management are fundamental to the species' long-term conservation.

Keywords:

• biological flora
• pōhutukawa
• Metrosideros excelsa
• ecology
• succession
• conservation
• New Zealand

Introduction

Metrosideros excelsa Sol. ex Gaertn. (Myrtaceae) is an iconic angiosperm tree up to 25 m tall forming a notable element of the coastal vegetation in northern New Zealand. Revered in both Māori and European culture, M. excelsa is known for its sudden and spectacular annual flowering episodes, a gnarly and sprawling growth form and an ability to survive in the most precarious and inhospitable coastal environments. The species’ tendency to form young pure stands, often with low plant species richness, is also a characteristic feature of M. excelsa (Simpson 1994; Bylsma 2012). As stands develop, species richness may increase due to the incursion of mid and late successional species (Clarkson 1990; Bylsma 2012). When flowering, trees are laden with red brush flowers that yield copious amounts of nectar for consumption by endemic nectar feeders. Trees also provide nesting/roosting sites for native birds, and the flaky bark and dense leaf litter provide habitat for a wide range of host-specific indigenous insects (Hosking & Hutcheson 1993; Simpson 1994; Taylor et al. 2007). In the past, M. excelsa forest has undergone serious decline, initially a result of burning by Māori, with further losses arising from nineteenth century land clearance by early European immigrants, followed by the expansion of pastoral farming and logging to supply timber to the ship-building industry (Dieffenbach 1843; Simpson 2005). Currently, land clearance, coupled with pressure from introduced herbivores continues to pose a serious threat to remaining M. excelsa stands on the mainland, which represent c. 10% of their past extent (Forest Research Institute 1989). Further deterioration will have implications beyond M. excelsa, also affecting the associated fauna and fauna that reply on M. excelsa forest communities for food and habitat (Hosking & Hutcheson 1993).

Metrosideros excelsa is well represented in New Zealand literature, including forest health studies (e.g. Forest Research Institute 1989; Hosking & Hutcheson 1993), historical and cultural reviews (Conly & Conly 1988; Simpson 2005) and comprehensive accounts of the species evolution, biology and management (e.g. Schmidt-Adam et al. 2000; Simpson 2005; Bergin & Hosking 2006). A more limited number of studies quantitatively describe M. excelsa forest dynamics and succession (Clarkson & Clarkson 1994; Atkinson 2004; Bylsma 2012). Here, we review the morphology, taxonomy, ecology and other topics relevant to the biology and conservation of M. excelsa, largely from primary publications, but also supported by unpublished work.

Morphological description

Metrosideros excelsa is a dicotyledonous, evergreen woody shrub or tree. Mature trees can reach 25 m in height and trunks can be up to 2.5 m in diameter (Allan 1961), often with a gnarly growth habit. Trees can be multi-stemmed, particularly in open sites, and are frequently wider than they are tall due to dense and sprawling canopies. In a forest environment however, shading results in a more upward growth form (Simpson 2005). Branching is sympodial and spreading, branchlets are stout, and young shoots on mature trees are often densely tomentose. Trees readily produce adventitious roots, with some trees producing large quantities of these, and others producing relatively few (Simpson 2005). Leaves are decussate, usually opposite, and are positioned on short, stout petioles. Lamina dimensions vary, but commonly range from 5 to 10 cm in length by 2.5 to 5 cm in width. Leaf shape can range from elliptic to oblong with either acute or obtuse leaf tips. Juvenile plants have glabrous leaves, however leaves on mature trees are coriaceous, thick and clad with tomentum; as leaves develop, the tomentum wears off the adaxial surface but persists on the abaxial surface (Fig. 1a). Inflorescences are composed of broad compound cymes terminating in groups of three flowers (Dawson 1968); a single inflorescence is composed of an average of 14.3 flowers (Schmidt-Adam et al. 1999). Pedicels are stout and tomentose (Allan 1961). Flowers comprise a ‘brush’ of stamens (Fig. 1b); these are crimson to scarlet in colour or occasionally yellow (Bergin & Hosking 2006). Each flower comprises c. 27 stamens, however, this is highly variable. Stamens are between 20 and 37 mm in length, and positioned around a hypanthium with a single central pistil (Wootherpoon 1993). Petals are insignificant, oblong and persistent (Dawson 1968). Capsules are 7–9 mm long, tomentose, distinctly exserted and loculicidally three-valved (Allan 1961). Seeds range from 3.0 to 3.2 mm in length by 0.2 to 0.42 mm in width, and weigh around 0.1–0.15 mg (Schmidt-Adam et al. 2002).

Figure 1 Some morphological features of Metrosideros excelsa. A, Metrosideros excelsa leaves densely clad with tomentum on both the upper and lower leaf surfaces. B, Metrosideros excelsa flower and inflorescence showing the arrangement of numerous scarlet stamens around hypanthium.

A small M. excelsa stand located near the species southern limit at Wai-iti (Taranaki) exhibits morphological characteristics distinct from M. excelsa found further north; this stand is suspected to be natural, rather than the result of a historic planting (Benson 1995; Simpson 1997). Metrosideros excelsa leaves and flower clusters in this population are frequently smaller than those typical of M. excelsa elsewhere, with leaves ranging from 4 to 6 cm long by 1 to 1.5 cm in diameter (Simpson 1997). Annual shoot growth increments are also smaller, at around 2–4 cm yr⁻¹ compared with 5–6 cm yr⁻¹ elsewhere in New Zealand. Morphological variation may represent local adaptation to the marginal conditions present at the leading or trailing edge of migration south since the post Pleistocene climatic amelioration (Simpson 1997).

Anatomy

Leaf

The leaf anatomy of M. excelsa has been described by Simpson (2005) and Bylsma (2012). Examination of leaf cross-sections show that the tissues are notably differentiated (Bylsma 2012) with features that aid survival in the saline and drought-prone coastal environment (Simpson 2005). Leaf tissues comprise upper and lower cuticle layers, which thicken near leaf margins; upper and lower epidermal layers, from which leaf hairs protrude; a layer of hypodermis (three to four cells thick), particularly thick around the mid-rib and main veins; a mesophyll layer comprised of vertically elongated palisade cells and spongy tissue, air pockets are generally not abundant. On the abaxial leaf surface numerous stomata are protected by a dense layer of tomentum (Simpson 2005).

Flower stigma and style

Following the classifications of Boland & Sedgley (1986), the stigma of M. excelsa is described as blunt and uniform. In cross-section the stigma is approximately circular and has an average diameter of 0.62 mm (Schmidt-Adam & Murray 2002). Unicellular papillae line the stigma and these lack a cuticle layer. The diameter of the style increases basipetally reaching c. 0.74 mm. Styles comprise an epidermis with outer cuticle (2–4 µm thick), parenchymous cortex and a central transmitting tissue/tract. The transmitting tract cells are loosely arranged and form a mosaic with large intercellular spaces filled with secretion products (Schmidt-Adam & Murray 2002).

Wood

The wood of M. excelsa has a swirled grain and is very dense. Annual growth rings are indistinct to slightly distinct (Meylan & Butterfield 1978). All members of the genus Metrosideros are renowned for the hardness and durability of their wood. The features of M. excelsa wood that underlie this trait include abundant parenchyma cells, wide rays arranged in radiating sheets, and a dominance of thick wall fibres comprising narrow elongated cells densely packed around water conducting vessels (Simpson 2005). The heartwood of M. excelsa has rich reddish-brown coloration, often with black or pinkish streaks. It has been suggested that the heartwood of M. excelsa is resistant to fungi, however, heartwood frequently has a decayed centre (Bergin & Hosking 2006).

Chemistry

Seven distinct anthocyanins have been identified in flower tissue of M. excelsa, where they are responsible for flower colour. These are delphinidin-3-glucoside, malvidin-3-glucoside, cyanidin-3-glucoside, petunidin-3-glucoside, peonidin-3-glucoside, cyanidin-3,5-diglucoside and malvidin-3,5-diglucoside (Andersen 1988). Anthocyanin pigmentation is also a feature of M. excelsa's adventitious roots. Anthocyanins in the root tissues tend to be cyanidin- and delphinidin-based and are present in cell layers of the root cap, epidermis, hypodermis and cortex. Light exposure may partially control the production of these pigments; consequently, they are more abundant in the outer-most layers, where they may protect against UV-B radiation. However, the function of anthocyanins in adventitious roots is still relatively unknown (Solangaarachchi & Gould 2001).

Five unusual C-methyl flavonoids with no B-ring oxidation have also been identified in M. excelsa tissues: 2′,4′-dihydroxy-3′,5-dimethyl-6-methoxychalcone; 2′,4′-dihydroxy-3′-methyl-6′-methoxychalcone; 2′,6′-dihydroxy-3′-methyl-4′-methoxychalcone; 2′-hydroxy-3′-methyl-4′,6′-dimethoxychalcone; and 5,7-dihydroxy-6,8-dimethylflavanone (Mustafa et al. 2005). The latter is a major constituent of M. excelsa leaves, whereas 2′,6′-dihhydroxy-3′-methl-4′-methoxychalcone is the dominant flavonoid in flowers, buds, twigs and seed. High levels of flavonoids found in flower buds may suggest that they function to defend reproductive organs from bacterial/fungal attack (Mustafa et al. 2005).

Changes during ontogeny

Metrosideros excelsa shows a smooth transition from juvenile to mature phenology (Clements et al. 1999). The vegetative phase change is characterised by a transition from juvenile plants with glabrous foliage to adult plants having leaves with dense pubescence on the abaxial surface (Kubien et al. 2007). Anatomical and physiological characteristics that accompany this transition from juvenile to mature phenology include: decreased photosynthetic rates, possibly due to a reduction in resources allocated to carbon gain; decreased stomatal limitation on CO₂ uptake, indicating that the control of water loss shifts from a physiological to a physical basis; and increased palisade and mesophyll depth (larger cells), which may reduce light penetration and prevent sun damage. Nitrogen levels were found to be much higher in juvenile leaves, indicating a reallocation of nitrogen in adult plants to areas where it may be used for defence and reproduction rather than photosynthesis (Kubien et al. 2007).

Quantification of the light environments occupied by M. excelsa juveniles indicates that the minimum light requirements of M. excelsa may also decrease as juveniles develop. However, this may not indicate an increased tolerance to shade, but alternatively may relate to the over-topping of M. excelsa juveniles by faster growing species (Bylsma 2012). Single mature M. excelsa trees are commonly observed growing beneath a closed canopy of taller species (Atkinson 2004).

Cytology

The chromosome number of M. excelsa is 2n = 22 (Dawson 2008).

Nomenclature

The genus name Metrosideros was first used by Rumphius in 1743 to identify a group of hard-wooded timber trees in Indonesia. In 1769, the name was re-applied to Metrosideros collina, a timber tree from Tahiti. The word Metrosideros is derived from the Greek terms metra meaning core or heart, and sideron meaning iron. Thus, Metrosideros species are often referred to as iron-hearted trees, alluding to the durability and hardness of their wood (Simpson 2005).

Metrosideros excelsa samples were collected by Banks and Solander during Cook's first voyage to New Zealand (1769). Sample descriptions were then published by Gaertner (1788), notes on the sample plate read: ‘Calyx tomentosus, quinquedentatus, capsulae ad medium usque adnatus. Capsula ovata, extra calycem prominens, pubescens, trilocularis’ (Oliver 1928). The term excelsa means high, lofty or outstanding, a likely reference to the species mass flowering habit.

The species has also been referred to as M. tomentosa (A. Rich.), as described in Essai d'une Flore de la Nouvelle-Zelande; a detailed account of the vegetation encountered by D'Urville on a voyage to New Zealand during the 1800s (Richard 1832). This name was given in reference to the densely clad underside of mature leaves (Dawson 1993). Metrosideros tomentosa is now considered an old synonym for M. excelsa (Oliver 1928), with naming authority given to Gaertner, for the first valid publication.

The Māori name ‘pōhutukawa’ is said to mean splashed by the spray, an acknowledgement of the harsh coastal environment for which this species is superbly adapted (Riley 1994). The name pōhutukawa is derived from the word hutukawa – a headdress of red feathers (Simpson 1994). Other Māori names for M. excelsa include kahika and rātā (Allan 1961). Metrosideros excelsa is commonly called New Zealand Christmas tree, as the flowering period coincides with Christmas and early European settlers used the flowers as a substitute for holly in their homes (Conly & Conly 1988).

Taxonomy and relationships

Metrosideros excelsa is a member of the genus Metrosideros, subgenus Metrosideros, and belongs to the family Myrtaceae. Other taxa within this family that are represented in New Zealand's endemic flora include members of Leptospermum, Syzygium, Kunzea, Lophomyrtus and Neomyrtus. The genus Metrosideros contains c. 50 described species and these form a conspicuous component of the Pacific Basin flora (Sreekantan et al. 2001; Gardner et al. 2004), with representatives in the Philippines, New Guinea, New Caledonia, New Zealand and many of the Pacific Islands. An outlying species is also present in South Africa (Andersen 1988). Metrosideros comprises two main subgenera (Dawson 1988); Metrosideros (the trees and shrubs) and Mearnsia (the vines). Common features of Metrosideros include the ability to occupy open rocky ground; preference for a warm temperate or subtropical climate; sensitivity to frost; the ability to form adventitious roots; and the ability to establish on young soil with a low nutrient status (Simpson 2005). New Zealand has 12 species of Metrosideros: six root climbing lianes (M. albiflora, M. carminea, M. colensoi, M. diffusa, M. fulgens, M. perforata), one shrub (M. parkinsonii) and five tree species (M. bartlettii, M. excelsa, M. kermadecensis, M. robusta, M. umbellata) (Dawson 1988).

Biogeography

The genus Metrosideros is one of the most widespread in the Pacific, with centres of diversity in the south (New Zealand; 12 spp.) and north (New Caledonia; 16 spp.). New Zealand is considered the origin and the source of the majority of Metrosideros taxa, being the only landmass where late Palaeocene/early Eocene fossil pollen grains have been recorded, and the primitive Metrosideros umbellata is considered the basal Metrosideros species (Wright et al. 2000). Metrosideros species fall into three monophyletic clades. The first clade includes seven New Caledonian species as well as three species in western Oceania. The second contains six taxa from Melanesia and Samoa. The third accounts for much of the total species range of Metrosideros subgenus Metrosideros, this includes New Zealand species (M. bartlettii, M. robusta and M. excelsa) as well as all of the Polynesia taxa (Wright et al. 2000).

Wright et al. (2000) suggest a minimum of four distinct dispersal events occurred in the history of the Metrosideros subgenus Metrosideros. An initial mid/late Tertiary dispersal from New Zealand to New Caledonia; a mid/late Tertiary dispersal from New Caledonia to the western Pacific; a mid/late Tertiary dispersal from New Zealand to the southwestern Pacific; and lastly, the most recent dispersal in the Pleistocene, when the Metrosideros lineage dispersed from New Zealand to a number of islands in Oceania and remote Polynesia. The latter dispersal has been interpreted as the result of climate change during the Pleistocene glaciations, as wind-dispersed species could have taken advantage of the westerly weather systems and migrated across the marine environment (Wright et al. 2000).

Geographic distribution and ecological tolerance

Metrosideros excelsa is primarily a coastal species (Fig. 2), limited to the warm temperate climate of northern New Zealand (Sakai & Wardle 1978; Simpson 2005). Metrosideros excelsa forest is currently a rare and localised ecosystem on the mainland and is rarely found more than 1000 m from the coast (Simpson 1994; Bylsma 2012). At the time of European settlement, however, the forest formed a continuous belt around the north of the North Island; stretching from Urenui on the west coast, to Gisborne or possibly Young Nick's Head (Te Kuri), on the east coast (Simpson 2005; Bergin & Hosking 2006). The southern limits have since retreated north and currently extend no further than Wai-iti on the west coast and Mawhai Point on the east coast. Metrosideros excelsa has an infrequent occurrence from Wai-iti to Manukau Harbour on the west coast, where soft highly eroded mudstone substrates dominate. Logging, accidental fire and land clearance have reduced the distribution of M. excelsa to small fragmented populations and scattered isolated trees, largely on cliff tops and coastal banks (Bergin & Hosking 2006).

Figure 2 Generalised natural distribution of Metrosideros excelsa in New Zealand.

Northern New Zealand islands provide the best examples of original M. excelsa forest, including natural plant and animal associations. Offshore islands where M. excelsa forest is prominent include Hauturu/Little Barrier Island (Hamilton & Atkinson 1961), the Poor Knights Island group (Simpson 2005), Rangitoto Island, Whakaari/White Island, Moutohora Island, Tuhua/Mayor Island and Motuotau Island (Clarkson 1990; Clarkson & Spring-Rice 1992; Clarkson & Clarkson 1994). Metrosideros excelsa has a strong affinity for volcanic substrates and grows naturally in proximity to geothermal activity (Clarkson et al. 1991; Clarkson & Clarkson 1994). Metrosideros excelsa forest on Whakaari/White Island is continuously disturbed by volcanic activity, including eruptions and toxic fumes. On Moutohora Island, stunted M. excelsa shrubs grow within metres of steaming springs and fumaroles (R J Bylsma pers. obs).

Inland M. excelsa stands are extant on the Central Volcanic Plateau, including on the shores of Lakes Rotorua, Ōkataina and Taupo (including Motutaiko Island), and along the Tarawera River (Simpson 2005; Bergin & Hosking 2006). In the 1840s, Colenso also recorded M. excelsa trees growing on a number of cliffs that fringed Lake Waikaremoana (Cheeseman 1906). Debate exists over the origins of inland stands (Clarkson et al. 1991). Some suggest that they originate from Māori plantings and have subsequently spread into the wider landscape. On the shores of Lake Taupo, M. excelsa co-occurs with archaeological/cultural sites and trees on Motutaiko Island are said to have been planted by Tūwharetoa ancestors (Stowe 2007). However, the presence of other normally coastal associates in the Rotorua Lakes area suggests that some remaining inland M. excelsa stands are natural (Clarkson et al. 1991). Metrosideros excelsa forest is likely to have been more widespread through inland localities, as a coloniser of new surfaces created by the late Pleistocene and early Holocene lava flows from the Ōkataina Volcanic Centre. Present-day sequences show strong competition from other broadleaved species (such as Weinmannia racemosa, Litsea calicaris and Beilschmiedia tawa), which confine M. excelsa forest to open or extreme sites (Clarkson & Clarkson 1994).

Metrosideros excelsa has been widely planted throughout New Zealand. In the South Island, M. excelsa is cultivated as far south as Jackson Bay on the west coast (Simpson 2005) and the Otago Peninsula on the east coast. There are also several M. excelsa trees on the Chatham Islands and cultivation occurs on Stewart Island (Dawson et al. 2010). Metrosideros excelsa has been introduced to many other countries including Australia, the USA and South Africa, where it has escaped from cultivation and is regarded as an invasive species (Yeates & Williams 2006; Dawson et al. 2010).

Genetic structure

Three studies have investigated the genetic structure of M. excelsa forests in New Zealand (Young et al. 2001; Gardner et al. 2004; Broadhurst et al. 2008). Using seed from 20 M. excelsa populations, Young et al. (2001) assessed genetic variation and interpopulation structure. Populations were genetically diverse and showed a high level of population differentiation. Three small isolated stands (Rotoiti, Rotorua and Paparoa reef) had particularly low genetic diversity compared with elsewhere and may originate from Māori plantings. Self-fertilisation was also more prevalent in small isolated populations. Gardner et al. (2004) applied chloroplast DNA sequencing to M. excelsa and other Metrosideros species from widespread New Zealand populations. Results indicated a greater diversity of haplotypes in areas identified as glacial refugia, including upper Northland, northwest Nelson and the Coromandel Peninsula. These findings are interpreted as the result of range contraction, giving rise to isolated refuges during periods of warmer climatic conditions, followed by range expansion during cooler climates. The persistence of populations in isolated refuges is suggested to have given rise to the greater haplotype diversity observed (Gardner et al. 2004). A later study (Broadhurst et al. 2008) examined the genetic structure between and within 10 populations of M. excelsa, using amplified fragment length polymorphism (AFLPs) on genomic DNA. Like Young et al.'s (2001) results, this study showed genetic differentiation among and within populations, with larger populations tending to have higher levels of genetic diversity, but no evidence for a relationship between genetic structure and geographical location (Broadhurst et al. 2008).

Hybridisation

Hybridisation between New Zealand's Metrosideros tree species is common. Metrosideros excelsa naturally hybridises with M. robusta (early reports of this include Kirk 1899; Carse 1927; Cooper 1954) because the species distributions and flowering times commonly overlap. This typically occurs inland from the coast and towards the end of flowering by M. excelsa (Dawson et al. 2010). Hybrid populations are present in the Rotorua and Tarawera Lakes areas and also on Rangitoto Island (Cooper 1954; Clarkson 1990; Dawson et al. 2010). Occasionally, hybrid individuals are also found among otherwise pure populations (Simpson 2005). Metrosideros excelsa is also able to form hybrids with M. umbellata and M. bartlettii; however, the distributions of M. excelsa and these two species less commonly overlap (Simpson 2005). Metrosideros excelsa and M. kermadecensis (Kermadec Island pōhutukawa) also readily hybridise when they occur together in cultivation. Hybridisation between New Zealand's Metrosideros species has raised concern around the genetic pollution of natural stands (Simpson 2005), however, genetic analysis has shown that hybridisation has been an important factor in the past evolution of the species, evidenced by numerous shared DNA sequences (Gardner et al. 2004). The extent of hybridisation in New Zealand's Metrosideros is comparable with that reported for Hawaiian Metrosideros (Harbaugh et al. 2009).

Fungal associates

A total of 209 species of fungi are known to be associated with Metrosideros species in New Zealand with only a few considered primary pathogens. However, secondary pathogens, such as wood-rotting basidiomycetes, may play a role in Metrosideros species die back following severe possum browse. A change in fungal pathogenicity after possum damage has yet to be investigated for M. excelsa. The majority of fungi identified on M. excelsa are saprobes; Leptomelanconium sp. is common on M. excelsa foliage and may cause leaf spots and discoloration. Although levels of fungal infection can be high on individual trees, the effect on overall tree health is unknown (McKenzie et al. 1999).

Nineteen endomycorrhiza have been identified within the inner tissues of M. excelsa roots. Seedlings are rapidly colonised by these fungi, which benefit the seedlings by increasing tolerance to environmental stress (e.g. drought and low nutrient availability) and increasing seedling growth rates, particularly when growing in infertile substrates (Simpson 2005). Metrosideros excelsa lacks ectomycorrhiza associations.

Reproductive biology

The reproductive strategy of M. excelsa has been described by Schmidt-Adam et al. (2000) as ‘wasteful but sufficient’. Metrosideros excelsa is a mass flowering species, and trees flower profusely for c. 2 wk during December to January. Each tree is thought to produce 13,000–40,000 flowers per annual flowering episode (Schmidt-Adam et al. 2000). Flowering episodes show high annual variation and this may promote cross-pollination between populations (Simpson 1994; Bergin & Hosking 2006). However, this annual variation is not well understood. Evidence suggests that floral induction takes place in autumn, in response to shortening day length. Following flower initiation, reduced temperatures may have an inhibitory effect on floral development (Sreekantan et al. 2001) and flower intensity may be affected by proximity of flower buds to developing seeds (Henriod et al. 2000). Six key morphological stages in M. excelsa inflorescence development have been identified and occur over a 10 wk period (Schmidt-Adam & Gould 2000): (1) initially, vegetative and floral buds appear identical, both covered with bud scales; (2) allometric changes result in inflorescence buds taking on an ellipsoidal shape for c. 8 d and bud scales senescing; (3) deciduous bracts subtending the secondary inflorescence axis partly cover inflorescence buds; (4) deciduous bracts abscise; (5) a pair of bracts and two pairs of bracteoles subtending lateral flower buds in each cymose inflorescence become visible; and (6) bracts and bracteoles abscise, individual flower buds become clearly separated and petals become evident (Schmidt-Adam & Gould 2000).

The breeding system of M. excelsa is the intermediate of facultative selfing and facultative outcrossing. Neither dichogamy nor herkogamy is important in preventing self-pollination (Schmidt-Adam et al. 2000). Flowers are hermaphroditic, displaying an initial female phase (mean duration 1.3 d) followed by a hermaphrodite phase (mean duration 4 d), which is then followed by a further female phase. Stigmas remain receptive for a minimum of 9 d; however in the final phase, pollination is rare because pollinator rewards are depleted. In earlier phases, individual flowers produce an average of 50 µL nectar d⁻¹. Pollen has high levels of viability (93.6%), however, pollen tube growth is slow compared with other angiosperms; this, coupled with relatively long styles, results in a 10–15 d delay before the pollen reaches the ovary (Schmidt-Adam et al. 1999).

All ovules (c. 933) within the ovaries appear to be potentially fertile, although the number of ovules that develop to form fertile seeds is low (c. 10.2%), irrespective of pollen availability (Schmidt-Adam et al. 1999). Once ovules are successfully fertilised, seed develops and is later released in March–April (Bergin & Hosking 2006). Despite low seed viability, overall seed production per tree is still sufficiently high (Schmidt-Adam et al. 1999). Metrosideros excelsa seed is readily dispersed by wind. Light winds in the range of 5–19 km h⁻¹ can disperse seeds over extremely long distances (Wright et al. 2000), providing a mechanism for gene flow between populations (Broadhurst et al. 2008). Metrosideros excelsa seeds lack endosperm tissue and have a thin two-layered seed coat, making them extremely vulnerable to desiccation. Consequently, M. excelsa seed can only persist in the soil for a short period and must germinate soon after dispersal (Schmidt-Adam et al. 2002).

Inbreeding and outbreeding

The floral biology of M. excelsa presents a paradox, the large floral display attracts pollinators and supports out-breeding and cross-pollination. However, mass flowering episodes coupled with a hermaphroditic phase and an absence of evolutionary traits to avoid pollen and stigma interference result in a high rates of self-fertilisation (Schmidt-Adam et al. 1999). Approximately 60% of seeds produced are self-pollinated. Controlled pollination experiments indicate that individual M. excelsa trees differ in their ability to set seed after self-fertilisation. Natural populations likely compose a mosaic of self-compatible and incompatible individuals (Schmidt-Adam et al. 1999). Inbreeding depression is late-acting in self-pollinated individuals of M. excelsa and has no effect on pollen tube growth, ovule/seed development or germination rate (Schmidt-Adam et al. 1999, 2000). However, following six months of seedling growth, inbreed individuals have markedly lower rates of shoot growth. It is probable that natural selection may ultimately eliminate inbred individuals before they reach reproductive maturity (Schmidt-Adam et al. 2000).

Plant–pollinator relationships

Metrosideros excelsa is associated with a broad range of pollinators (Table 1). The red flower coloration and copious nectar production are known to attract bird species (Godley 1979). Native birds are more important pollinators of M. excelsa than native bees because of their greater abundance, larger body sizes and foraging patterns which are more efficient at inducing seed set and promoting outcrossing. Similarly, introduced honeybees are more effective pollinators than native bees because their larger body size promotes stigma contact (Schmidt-Adam et al. 2009). Nocturnal visitation by New Zealand's endemic geckos (Whitaker 1987) and short-tailed bat (Mystacina tuberculata) has also been reported (Arkins et al. 1999; Pattemore & Wilcove 2012). On Hauturu/Little Barrier Island, short-tailed bats have significantly higher activity levels (indicated by echolocation calls) in the vicinity of flowering M. excelsa trees (Arkins et al. 1999).

Table 1 Recorded species visitation to Metrosideros excelsa flowers.

Flower visitor	Recorder
Bell bird (Anthornis melanura)	Schmidt-Adam et al. (2000); Pattemore & Wilcove (2012)
Gecko (Hoplodactylus spp.)	Bergin & Hosking (2006)
Chafinch (Fringilla coelebs)*	Schmidt-Adam et al. (2000)
German wasps (Vespa ger-manica)*	Schmidt-Adam et al. (2000)
Giant wētā (Deinacrida sp.)	Simpson (2005)
Hihi (Notiomystis cincta)	Schmidt-Adam et al. (2000); Anderson (2003); Bergin & Hosking (2006)
House sparrow (Passer domesticus)*	Godley (1979); Schmidt-Adam et al. (2000)
Insects (Hymenoptera spp., Coleoptera spp.)	Anderson (2003)
Introduced bees*	Schmidt-Adam et al. (2000)
Kākā (Nestor meridionalis)	Schmidt-Adam et al. (2000); Bergin & Hosking (2006)
Kākāriki (Cyanoramphus sp.)	Anderson (2003); Schmidt-Adam et al. (2000)
Myna (Acridotheres tristis)*	Schmidt-Adam et al. (2000); Anderson (2003)
Native bees (Leioproctus spp.)	Schmidt-Adam et al. (2000)
Saddleback (Philesturnus carunculatus)	Anderson (2003); Schmidt-Adam et al. (2000)
Ship rat (Rattus rattus)*	Pattemore & Wilcove (2012)
Short-tailed bat (Mystacina tuberculata)	Daniel (1976); Arkins et al. (1999); Pattemore & Wilcove (2012)
Silvereye (Zosterops lateralis)*	Schmidt-Adam et al. (2000); Pattemore & Wilcove (2012)
Starling (Sturnus vulgaris)*	Schmidt-Adam et al. (2000); Mc Cann (1952)
Tūī (Prosthemadera novaeseelandiae)	Schmidt-Adam et al. (2000); Pattemore & Wilcove (2012)
Whitehead (Mohoua albicilla)	Anderson (2003); Schmidt-Adam et al. (2000)

*Exotic species.

Since European settlement in New Zealand, a dramatic shift in bird pollinators has occurred (Schmidt-Adam et al. 2000), as many of the native nectar feeders, such as tūī (Prosthemadera novaeseelandiae), bellbird (Anthornis melanura) and hihi (Notiomystis cincta) are now locally extinct or rare on the mainland (Department of Conservation 2005). The original suite of pollinators, now only intact on offshore islands, has been largely replaced by non-native birds and honeybees. Owing to the ability of M. excelsa to use a wide range of pollinators, the reproductive system has been mostly resistant and the species does not suffer from enhanced inbreeding associated with pollinator shifts (Schmidt-Adam et al. 2000). However, Pattemore & Wilcove (2012) have indicated that mainland M. excelsa populations may be suffering from a degree of pollen limitation, though the invasive ship rat (Rattus rattus) and recent colonist bird silvereye (Zosterops lateralis) at least partially maintain pollination on the mainland.

As well as providing a nectar source, M. excelsa trees also provide valuable wildlife habitat. Shags (Phalacrocorax spp.) and herons (Egretta spp.) often nest and roost in M. excelsa trees due to the close proximity of the trees to the bird feeding grounds and habitat range (Simpson 1994; Heather & Robertson 1996). Guano from roosting shags can be responsible for dieback of M. excelsa individuals, often killing a tree slowly limb by limb as leaves are scorched by solutes in the excreta (Gillham 1960). Holes in M. excelsa trunks provide nesting sites for birds such as the saddleback (Philesturnus carunculatus), and the flaky bark and leaf litter provides habitat for a wide range of indigenous and host specific insects including moths, weevils, beetles, flies and scale insects (Hosking & Hutcheson 1993; Simpson 1994; Taylor et al. 2007).

Plant communities and associations

Metrosideros excelsa forest, particularly young stands (< 60 yr), can have low species richness when compared with more widespread forest types in New Zealand (Bylsma 2012). The simplest M. excelsa forest is located on Whakaari/White Island (Bay of Plenty), which comprises M. excelsa, Histiopteris incisa and Phormium tenax (Clarkson 1990). The low species richness observed in M. excelsa forests reflects habitat severity and also the tendency of M. excelsa to retard the rate at which a diverse community can develop, primarily the result of dense shading canopies and slowly decomposing leaf litter which inhibits seedling establishment (Atkinson 2004). Metrosideros excelsa commonly occupies exposed, saline and drought-prone coastal sites, where only a few other plant species are capable of establishment, including Pseudopanax lessonii, Kunzea ericoides, Piper excelsum and Pittosporum crassifolium. Further inland or along coastal valleys, where soil and moisture conditions are more favourable, M. excelsa can form tall coastal forest that hosts a diverse mix of coastal and lowland plant species. Common canopy associates include Corynocarpus laevigatus, Vitex lucens, Dysoxylum spectabile, Kunzea ericoides, Litsea calicaris, Knightia excelsa and Beilschmiedia tawa (Bergin & Hosking 2006; Bylsma 2012). Metrosideros excelsa forest composition varies with topographic landform, exposure and latitudinal position (Bylsma 2012). Examples of compositional associations (Table 2) from published and unpublished sources are summarised in six general geographic regions (Northland, Auckland, Coromandel, Bay of Plenty, East Coast and West Coast) in Supplementary file 1, to illustrate the pattern of M. excelsa dominated communities across the species natural range.

Table 2 Range of recorded vegetation types with Metrosideros excelsa as a dominant component. Metrosideros = Metrosideros excelsa.

Vegetation type	Author
Metrosideros forest	Clarkson & Clarkson (1994); Smale et al. (2009)
Metrosideros/broadleaved forest	Hamilton & Atkinson (1961)
Metrosideros/Pseudopanax lessonii forest	Beadel & Shaw (1988); Bylsma (2012)
Metrosideros/Melicytus ramiflorus–Pseudopanax lessonii forest	Bylsma (2012)
Metrosideros/Myrsine australis–Melicytus ramiflorus forest	Bylsma (2012)
Metrosideros/Myrsine australis–Coprosma macrocarpa forest	Bylsma (2012)
Metrosideros/Melicytus ramiflorus–Cordyline australis forest	Bylsma (2012)
(Kunzea ericoides) Metrosideros/Pseudopanax arboreus forest	Bylsma (2012)
(Knightia excelsa) Metrosideros/Myrsine australis – Melicytus ramiflorus forest	Bylsma (2012)
Metrosideros/Cyathea dealbata–Coprosma macrocarpa forest	Bylsma (2012)
Metrosideros–(Beilschmiedia tawa)–broadleaf forest	Regnier et al. (1988)
Metrosideros/Dysoxylum spectabile–Beilschmiedia tawa forest	Bylsma (2012)
(Cortaderia jubata) Coriaria arborea/Metrosideros–Coprosma robusta scrub	Bylsma (2012)
Metrosideros–Vitex lucens forest.	Regnier et al. (1988)
(Metrosideros)/Kunzea ericoides scrub	Regnier et al. (1988)
Metrosideros–Litsea calicaris forest	Beadel & Shaw (1988)
Metrosideros rockland	Beadel & Shaw (1988)
Metrosideros–Meryta sinclairii–Corynocarpus laevigatus forest	Bellingham (1984)
Metrosideros–Beilschmiedia tarairi–Corynocarpus laevigatus forest	Bellingham (1984)
Metrosideros–Dysoxylum spectabile–Corynocarpus laevigatus–Vitex lucens forest	Bellingham (1984)
Metrosideros–Cordyline australis forest	Regnier & Clarkson (1988)
(Metrosideros)–(Vitex lucens)/Leptospermum scoparium–Pteridium esculentum forest	Regnier (1987)
Metrosideros–Cortaderia splendens forest	Smale et al. (2009)
Podocapus totara–Metrosideros excelsa forest	Lux et al. (2007)
Metrosideros–Dacrydium cupressinum forest	Lux et al. (2007)
Metrosideros–Pittosporum crassifolium–Pseudopanax lessonii rockland	Goldwater & Beadel (2010)
Metrosideros–Beilschmiedia tarairi forest	Goldwater & Beadel (2010)
Metrosideros–Cordyline australis treeland	Regnier & Clarkson (1988)
Metrosideros–Dacrydium cupressinum forest	Lux et al. (2007)

Epiphyte associations

Epiphyte species are generally not abundant in M. excelsa forest (Bylsma 2012), primarily due to low humidity and high exposure. In sheltered or mature forest, epiphytes become more common. In a survey of M. excelsa forests in the Bay of Plenty, common epiphytic species included Pyrrosia eleagnifolia, Asplenium flaccidum, Microsorum scandens, Microsorum pustulatum, Blechnum filiforme, Peperomia urvilleana, Ichthyostomum pygmaeum, Astelia solandri and Collospermum hastatum. Climbers and lianas were less common but included Clematis foetida, Clematis paniculata and Ripogonum scandens (Bylsma 2012). Kirk (1872) reported M. excelsa trees growing on the shores of Lake Tarawera to be rich with epiphytic species, including Griselinia lucida, A. solandri and Pittosporum cornifolium. Pittosporum cornifolium has also been recorded growing in coastal M. excelsa forest, in the Tamaki Ecological District (CHR 497823). On the Poor Knights Islands (Northland), Xeronema callistemon has been recorded as an epiphyte on M. excelsa; it is unknown whether these originate from plants which have fallen from cliffs above and taken hold, or have established from seed (New Zealand Plant Conservation Network 2013).

Establishment and succession

The germination strategy of M. excelsa is tuned-in to dynamic landscapes and stochastic disturbance; seeds do not persist in the seed bank and germination is rapid (Atkinson 2004). In seed trials, germination occurred between 10 and 28 d (Burrows 2000), thus synchrony between disturbance, seed set and seed dispersal is necessary. Successful germination is constrained to sites with high light (typically > 16% canopy openness) and free of dense leaf litter (Wotherspoon 1993; Bylsma 2012). Seedlings also have a strong affinity for fissured rock and volcanic substrate (Bylsma 2012), although seedling establishment on sand dunes is also possible (Bergin & Hosking 2006); here, seedlings are usually associated with driftwood or debris and more common on consolidated back dunes and damp swales (Simpson 2005; Bergin & Hosking 2006). Upon germination, biomass is allocated to the developing roots, which seek out and cling to moist surfaces, and emerging cotyledons (1–2 mm long) with thick waxy cuticles to prevent desiccation (Bergin & Hosking 2006). A lack of natural M. excelsa regeneration reported on the mainland is attributed to herbivore browse, trampling and weed invasion (Wotherspoon 1993; Hosking & Simpson 2011), and forest initiation usually requires landscape-scale disturbances (e.g. fire or volcanic eruption), which do not occur frequently. Smaller scale disturbance events, such as landslides, however, are more frequent and provide localised regeneration sites (Bylsma 2012).

Once established on the landscape, M. excelsa undoubtedly alter the physical environment and ultimately facilitate the establishment of later successional species. On Rangitoto Island, M. excelsa is one of the first woody species to colonise the sterile volcanic substrate. Beneath the drip zone of individual M. excelsa trees, ground temperature is reduced, humidity is increased and a humus layer is formed, and this facilitates further species establishment and the development of vegetation aggregations (Clarkson 1990). Aggregation development on Rangitoto Island shows that aggregation size is a strong predictor of species richness, with aggregations of 0.1 to > 100 m² correlating with 1 to > 18 vascular taxa (BD Clarkson, University of Waikato, unpubl. data). Following canopy disturbance on White Island, however, direct succession of M. excelsa results from either the recovery and spread of an existing M. excelsa population through epicormic re-sprouting, or by colonisation of a newly emplaced surface by M. excelsa seed. Continual volcanic disturbance, isolation and extreme soil conditions prevent the incursion of later successional species here (Clarkson 1990).

Metrosideros excelsa forest succession and development have been quantified on the northern offshore islands (Atkinson 2004), Bay of Plenty islands and the Bay of Plenty mainland (Clarkson & Clarkson 1994; Bylsma 2012). Following disturbance, high-density, pure or mixed stands establish; juvenile M. excelsa densities as high as 3000 individuals ha⁻¹ have been reported (Matata Scenic Reserve, Bylsma 2012). In the absence of further disturbance, no additional M. excelsa recruitment occurs, giving rise to cohort populations (Fig. 3). However, on sites with frequent small-scale disturbances (e.g. coastal headlands), multi-aged stands are possible, with exceptionally large stemmed individuals representing relicts of prior colonisation events (Clarkson & Clarkson 1994; Bylsma 2012).

Figure 3 Diameter frequency distributions of Metrosideros excelsa stems in Bay of Plenty forests, across four stand age groups: 0–60, 60–180, 180–300 and > 300 yr. Saplings are > 50 cm tall and < 2 cm dbh. Age groups have been estimated from a diameter–age regression for Bay of Plenty trees (Bylsma 2012).

Self-thinning occurs throughout M. excelsa forest development (Fig. 4), resulting in mature forest (250–300 yr old) generally having < 300 M. excelsa stems ha⁻¹. Self-thinning is most prolific in young forest (< 50 yr old), although in the early stages this may have little effect on total canopy foliage because gaps are successively filled by surviving M. excelsa. Self-thinning of mature M. excelsa forest generally involves the sequential loss of tree limbs. This creates canopy gaps and allows later successional tree species to reach the canopy and replace M. excelsa (Atkinson 2004; Bylsma 2012). The decrease of M. excelsa stems throughout stand development is initially coupled with an increase in M. excelsa basal area, reaching c. 50 m² ha⁻¹ in 80 yr, but does not significantly increase thereafter (Bylsma 2012).

Figure 4 Metrosideros excelsa density versus approximated age of stands in the Bay of Plenty. Axes have been log transformed.

Replacement strategies among key canopy species involves establishment at different stages of forest development, and this reflects species differing shade tolerances. Ranking tree species by their minimum light requirements (10th percentile of light environment occupancy) approximates their order of arrival in forest development (Metrosideros excelsa > Dysoxylum spectabile > Corynocarpus laevigatus > Litsea calicaris > Beilschmiedia tawa; Bylsma 2012). Early colonists that establish with M. excelsa may include K. ericoides, Leptospermum scoparium, Coriaria arborea var. arborea, Cordyline australis and Hebe stricta var. stricta. After c. 30 yr of development, a suite of shade-tolerant understory species typically establish beneath M. excelsa, including Myrsine australis, Melycytus ramiflorus, P. lessonii, P. excelsum subsp. excelsum and large leaved Coprosma species (C. robusta, C. macrocarpa, C. lucida and C. grandifolia). Mid-successional, shade-tolerant canopy species typically arrive following c. 60 yr of forest development and include C. laevigatus, V. lucens, L. calicaris, K. excelsa and D. spectabile; most of which are bird dispersed. The next stage of forest succession generally involves the arrival of late successional tree species such as Beilschmiedia tarairi and/or B. tawa (Percy 1956; Atkinson 2004; Bylsma 2012). In favourable locations (such as Hen and Chicken Islands, Northland), these can replace M. excelsa in the canopy after 200 yr (Atkinson 2004). In the Bay of Plenty, however, B. tarairi is not common, and B. tawa generally replaces M. excelsa following a minimum timespan of 300 yr (Clarkson & Clarkson 1994; Bylsma 2012), although relict individuals may be capable of surviving for 1000 yr (Simpson 1994). Mid-successional tree species can also contribute to M. excelsa replacement, but often to a lesser degree because mid-successional species are more reliant on canopy gaps to initiate height growth (Smale & Kimberley 1983; Bylsma 2012). Successional trends are comparable to those in other Metrosideros-dominated systems, for example, Metrosidersos polymorpha on Hawaiian lava flows (Atkinson 1970; Clarkson 1997).

Diameter growth rates

The diameter–age relationship exhibited by M. excelsa stems in Bay of Plenty forests (Fig. 5) suggests that trees have high initial growth rates, which can exceed 4 mm year⁻¹ in the first 80 yr (Bylsma 2012). Once trees exceed 100 yr of age, diameter growth subsequently reduces to < 2 mm year⁻¹ (Clarkson & Clarkson 1994; Bylsma 2012). Average diameter growth reported for Whakaari/White Island trees is 2.7 mm year⁻¹ (Clarkson & Clarkson 1994); significantly lower than that reported for mainland trees and consistent with the harsh and frequently disturbed environment on the island (Bylsma 2012). Bergin & Hosking (2006) reported M. excelsa diameter growth rates for planted (> 100 yr ago) trees outside the species natural range as 0.9–1.8 mm year⁻¹. Pardy et al. (1992) also assessed M. excelsa growth rates in planted stands, finding that the mean diameter growth rate was 9.7 mm year⁻¹; much higher than that reported for natural stands.

Figure 5 Gompertz model of stem diameter (cm) and age relationship for Metrosideros excelsa stem discs collected from Mayor Island (n = 5), White Island (n = 10), Bay of Plenty mainland (n = 16).

Threats and conservation

Metrosideros excelsa is not recognised as a threatened species (de Lange et al. 2013). However, M. excelsa forest is a rare and localised vegetation community on the mainland (Hosking & Simpson 2011; Bylsma 2012). In the past, the greatest threat to M. excelsa forest was land clearance. Initially, this was the result of localised burning by Māori, followed by more extensive clearance by Europeans for settlement, the expansion of pastoral farming and tree felling to sustain the shipbuilding industry (Dieffenbach 1843; Simpson 2005). Fire was the primary tool used to clear land. Dense wood with a low moisture content, flaky bark, leaves containing flammable oils and dry leaf litter results in M. excelsa being extremely vulnerable to fire (Bergin & Hosking 2006). Despite the species having a strong ability to coppice and re-sprout epicormic shoots following damage by wind, herbivore browse or acid/toxic fumes, even a low-intensity fire can kill a mature tree (Clarkson & Clarkson 1994; Bergin & Hosking 2006). By 1989, it was estimated that 90% of the original M. excelsa forest was lost (Forest Research Institute 1989; Hosking 2000), with nearly 100% clearance in the west, where refuges from fire on steep cliffs and bluffs were infrequent. Remaining M. excelsa habitat along the northern coastline is highly desirable for residential development, thus ongoing land clearance continues to pose a serious threat (Simpson 2005; Bergin & Hosking 2006). Original M. excelsa forest on the mainland is now highly localised, isolated and fragmented (Hosking & Simpson 2011) and this has implications for the species’ long-term health and viability. Small, isolated populations are prone to increased rates of self-pollination and associated inbreeding depression in offspring (Schmidt-Adam et al. 2000), as well as increased edge effects (e.g. microclimactic changes in light, temperature and wind around the perimeter of the fragments). Contemporary pressures also include herbivore browse, trampling by stock, insect attack, accidental fire and a limited number of suitable regeneration sites (free of leaf litter, herbivore browse and competition from other vegetation; Wotherspoon 1993).

Possum and herbivore browse is a major cause of current stand deterioration (Simpson 1994; Bergin & Hosking 2006; Baigent-Mercer 2010), inhibiting the establishment and regeneration of M. excelsa and later successional species. In a Bay of Plenty survey, Orokawa and Homunga Bay stands (Waihi Beach) had severe browsing damage and a notable lack of palatable mid to late successional seedlings and saplings (e.g. L. calicaris), present in comparably aged stands elsewhere (Bylsma 2012). The leaves of M. excelsa are highly palatable. In winter, possums selectively browse old foliage, whereas in spring and summer the young leaves and vegetative buds are targeted. Following defoliation, trees reduce flower production and create many epicormic shoots. This alters the growth form of the tree, and as a consequence, possum-ravaged trees are more intensively grazed because of the plentiful young buds (Bergin & Hosking 2006). Once possums have removed c. 80% of a tree's foliage, tree death is inevitable (Wotherspoon 1993). Tree health can be improved if trees are protected from further possum damage, for example, by using possum control or by trunk banding which prevents possum access. Enhanced pest control, including the wider use of toxins and bait stations would undoubtedly benefit the long-term health and prosperity of M. excelsa (Wotherspoon 1993; Environment Bay of Plenty 2009; Baigent-Mercer 2010).

The foliage of M. excelsa is also grazed by a number of insects, including Costelytra zealandica (grass grub), Phasmatidae (stick insects), Neomycta rubida (weevil) and Eucolaspis brunneus (native bronze beetle) and inhabited by Psyllidea (sap suckers) (Bergin & Hosking 2006). Insect browse alone seldom causes lasting impacts on the health of mature trees (Bergin & Hosking 2006); however, E. brunneus has caused significant damage to M. excelsa trees on Little Barrier Island by destroying developing flower buds (G. Schmidt-Adam, pers. obs.). Psyllids can heavily infect and suppress growth of nursery-raised seedlings, however, plants rarely die and the foliage of mature trees is not targeted (Bergin & Hosking 2006).

During the late 1980s, widespread concern arose over the deteriorating state of New Zealand's M. excelsa forests, prompting a number of forest health studies, primarily undertaken by the Department of Conservation (DOC), Forest Research Institute (FRI) and later, Project Crimson (Hosking & Hutcheson 1993). The Project Crimson Trust, formed in 1990, is focused on the protection and enhancement of remaining M. excelsa forest. The trust has initiated hundreds of community-based conservation projects and funds scientific research and conservation management programmes. A re-assessment of the health of M. excelsa forests, 10 yr after the trust was established, showed that the number and extent of regeneration sites had increased, possum damage and browse had decreased and fencing around stands had increased (Bergin & Hosking 2006). However, re-establishment of M. excelsa through restoration plantings has had variable success, often related to a difficult establishment phase. Mortality and low growth rates of planted M. excesa seedlings are often related to plant form/size, substrate, draught, exposure and competition from faster growing vegetation (Bergin & Hosking 2006). A valuable set of guidelines for planting and management of M. excelsa is available (Bergin & Hosking 2006), with key considerations given to site selection and preparation, the use of high-quality and appropriately sourced stock and the continued maintenance and monitoring of populations. High genetic differentiation between natural populations and a complex geographic pattern suggests locally sourced material is important when using M. excelsa in restoration (Broadhurst et al. 2008). Where possible, seed should also be collected from larger populations, which are less prone to self-fertilisation (Young et al. 2001). However, if the management goal is to increase species diversity, the use of M. excelsa as a primary cover may not be appropriate due to its tendency to slow the rate at which a diverse community develops (Atkinson 2004). For example, M. excelsa was the main species used during initial re-vegetation of Tiritiri Matangi Island (Hauraki Gulf), although in subsequent yr, stands required extensive thinning and under-planting due to the limited nature of natural regeneration (Cooper 2010).

Utility, cultural and economic value

Conly & Conly (1988) review the cultural significance of M. excelsa and its place in Māori tradition and medicine. The current and past use and value of M. excelsa has been described by Simpson (2005) and Bergin & Hosking (2006). Metrosideros excelsa is regarded by Māori as a chiefly tree, featuring strongly in legend and spiritual ceremony, as well as having many medicinal uses (Simpson 1994). Traditionally, an infusion made from the bark, which contains the natural antioxidant ellagic acid, was made by the tohunga (chief priest) to remedy diarrhoea (Brooker et al. 1961; Riley 1994; Seeram et al. 2005). Inner bark was also bound to wounds to sooth inflammation and stop bleeding (Riley 1994), and chewed to alleviate toothache (Bergin & Hosking 2006). Nectar from M. excelsa flowers also has medicinal values; traditionally this was collected in large quantities and ingested to relieve throat pain (Riley 1994).

Early Māori used the hard wood of M. excelsa to make implements such as paddles, weapons and mauls. The high wood density meant it was also suited for flax-softening bats and pounders used to prepare rarauhe (Pteridium esculentum) rhizomes for consumption (Bergin & Hosking 2006). Māori frequently used M. excelsa timber for boat building (Bergin & Hosking 2006), particularly in the final fitting-out stage, because the strong and curved thwarts of M. excelsa were the preferred wood to brace and strengthen long and narrow canoes (Clarke 2007). Metrosideros excelsa flowers are also highly valued; their red coloration is considered chiefly and worn by individuals with high rank. Consequently, trees have been widely transplanted and often mark sacred sites (Bercusson & Torrance 1998).

The economic values of M. excelsa are numerous (Conly & Conly 1988). Before glues became readily available, the strong and durable timber was used extensively for knees in shipbuilding (Kirk 1874). The wood is also highly decorative and is commonly used for carving and turning (Conly & Conly 1988). Honey produced from M. excelsa nectar is highly valued for its flavour and soothing properties and is produced on a commercial scale (Simpson 2005). Metrosideros excelsa is grown extensively in cultivation making it readily available for use in landscaping and gardens (Dawson et al. 2010).

Conclusion

Metrosideros excelsa is a New Zealand icon, intrinsic to the cultural identity of many New Zealanders. The species has many ecological adaptations that enable it to thrive along northern New Zealand's rocky, saline and drought-prone coast, and also colonise raw volcanic substrates. During the late Pleistocene–early Holocene, M. excelsa was likely more widespread inland, as a coloniser of lava fields produced from the Ōkataina Volcanic Centre (Clarkson & Clarkson 1994). The reproductive strategy of M. excelsa is tuned in to these dynamic landscapes and stochastic disturbance events, flowering in mass and subsequently producing thousands of small wind-dispersed seeds with rapid germination. Once established, M. excelsa can dominate forest development for centuries (Bylsma 2012). Harsh coastal conditions result in M. excelsa typically forming pure young stands with low plant species diversity compared with other New Zealand forest types (Bylsma 2012). However, further from the coast and as forests develop, they become more diverse, partially due to the incursion of shade-tolerant shrubs and mid to late successional tree species; the latter may eventually overtop and outcompete M. excelsa.

Following the arrival of people in New Zealand, M. excelsa forests became severely depleted as coastal areas were cleared for settlements and agriculture, and remaining forests were subject to logging (Hosking & Simpson 2011). Forest losses were accompanied by a dramatic shift in M. excelsa pollinator relationships, as many native bird species become lost from the mainland and exotic birds and bees were introduced (Schmidt-Adam et al. 2000). The best examples of M. excelsa forest with intact native plant and animal associations are now restricted to offshore islands. The infrequent occurrence of large-scale disturbances, coupled with human and herbivore pressure, currently limits natural regeneration on the mainland and remaining stands continue to decline. Introduced herbivores, particularly the brushtail possum, likely represent the greatest contemporary influence on M. excelsa forests and forest succession, influencing tree health and limiting regeneration of M. excelsa and later successional species (Bylsma 2012). Deterioration and decline of M. excelsa has implications well beyond the species itself. These forests are dynamic (Bylsma 2012) and strongly integrated communities; further losses of M. excelsa will reduce the habitat and resource available to plant and animal associates, including host-specific indigenous insects (Hosking & Hutcheson 1993). Key considerations for the conservation and restoration of M. excelsa communities include appropriate site selection, fencing and pest management, seed sourcing for local provenance and long-term monitoring of tree health (Bergin & Hosking 2006). Although M. excelsa has been the focus of many scientific, practical and cultural publications further research and documentation is still necessary. Future research should advance existing guidelines that assist those establishing and managing M. excelsa forest and restoration sites. Emphasis should be on reducing establishment barriers and progressing understanding of M. excelsa population genetics within a restoration context, particularly the degree to which inbreeding depression, low seed viability and pollinator shifts affect natural regeneration and the long-term success of restoration and regeneration sites.

Supplementary file

Supplementary file 1: Compositional associations in six geographic regions.

Acknowledgements

This paper was part of an MSc project by Rebecca Bylsma at the University of Waikato and the authors thank Catherine Kirby, Dr Gabriele Schmidt-Adam and one anonymous reviewer for constructive comments on an earlier draft of the manuscript and Toni Cornes for providing the distribution map. This project was aided by funding received from the Environmental Research Institute (University of Waikato), the Whakatane Historical Society Trust and the Rotorua Botanical Society.