Giant eucalypts – globally unique fire-adapted rain-forest trees?


Author for correspondence:

David M. J. S. Bowman

Tel: +61 3 6226 1943




II.Giant eucalypts in a global context1002
III.Giant eucalypts – taxonomy and distribution1004
IV.Growth of giant eucalypts1006
V.Fire and regeneration of giant eucalypts1008
VI.Are giant eucalypts different from other rain-forest trees?1009


Tree species exceeding 70 m in height are rare globally. Giant gymnosperms are concentrated near the Pacific coast of the USA, while the tallest angiosperms are eucalypts (Eucalyptus spp.) in southern and eastern Australia. Giant eucalypts co-occur with rain-forest trees in eastern Australia, creating unique vegetation communities comprising fire-dependent trees above fire-intolerant rain-forest. However, giant eucalypts can also tower over shrubby understoreys (e.g. in Western Australia). The local abundance of giant eucalypts is controlled by interactions between fire activity and landscape setting. Giant eucalypts have features that increase flammability (e.g. oil-rich foliage and open crowns) relative to other rain-forest trees but it is debatable if these features are adaptations. Probable drivers of eucalypt gigantism are intense intra-specific competition following severe fires, and inter-specific competition among adult trees. However, we suggest that this was made possible by a general capacity of eucalypts for ‘hyper-emergence’. We argue that, because giant eucalypts occur in rain-forest climates and share traits with rain-forest pioneers, they should be regarded as long-lived rain-forest pioneers, albeit with a particular dependence on fire for regeneration. These unique ecosystems are of high conservation value, following substantial clearing and logging over 150 yr.

I. Introduction

Gigantic trees arouse fascination and awe given their great age, size and global rarity (Griffiths, 2001; Spies & Duncan, 2009). These trees have been heavily exploited for forestry and there is political friction about the management of remaining old-growth forests given their high conservation, carbon storage and commercial timber values (Luyssaert et al., 2008; Keith et al., 2009; Dean & Wardell-Johnson, 2010; Lindenmayer et al., 2011). Gigantism also provides insights into biological constraints on tree growth. Surprisingly, there are few global synopses of giant trees, and particularly comparative analyses of the biology of giant conifers and angiosperms.

While conifers from western North America have long been recognized as including most of the world's tallest trees (Eckenwalder, 2009), it is less widely known that some angiosperm tree species in Australia and Borneo attain comparable heights (Fig. 1; Supporting Information Table S1). Australia is a centre of giant trees as a consequence of the presence of exceedingly tall eucalypts (members of the genus Eucalyptus sensu stricto: excluding Corymbia and Angophora spp.; Slee et al., 2006) in relatively fertile, mesic areas of the continent (Hickey et al., 2000; Sillett et al., 2010). Such giant trees are at the extreme tail of the distribution of tree heights. Although any definition of gigantism is necessarily arbitrary, a practical threshold of 70 m maximum height captures this tail because it delimits c. 50 species (Table S1), representing < 0.005% of an estimated total of 100 000 tree species (Oldfield et al., 1998). Conifer and eucalypt species exceeding 70 m represent 6% and 2% of maximum potential height distribution within their respective taxonomic groups (Fig. 2).

Figure 1.

Global distribution of the tree species known to reach 70 m in height (see also Supporting Information Table S1). Most of the tallest species are either conifers from the west coast of North America (represented by blue stars for the top five species and light blue dots for the remainder) or eucalypts in Tasmania (red stars for the three tallest species and light red dots for the remainder), although one dipterocarp species from Borneo (yellow star) and one conifer from New Guinea (blue star) rank among the top 10. Other angiosperm species that can exceed 70 m (pale yellow dots) are found in Southeast Asia, especially Borneo. One tall conifer (pale blue dot) occurs in Eurasia.

Figure 2.

Maximum height class distributions of conifer (= 603 taxa; blue bars) and Eucalyptus spp. (= 783 taxa; red bars). All known taxa of conifers and eucalypts (including Eucalyptus subspecies) have been included. Height data for conifers were compiled from Farjon (2010) and The Gymnosperm Database ( and for eucalypts from EUCLID (Slee et al., 2006), and supplemented from data sources listed in Supporting Information Table S1. Two conifer taxa (Abies hildalgensis Debreczy, Rácz & Guízar and Dacrydium leptophyllum (Wasscher) de Laub.) were excluded because of a lack of height information.

Despite extensive literature on the ecology of various giant eucalypt forest types (see Ashton & Attiwill, 1994; Harrington et al., 2000; Wardell-Johnson, 2000), few studies compare the ecology of giant eucalypt forests across their geographic range. The preponderance of giant eucalypts in Australia is puzzling, given that Australia is the driest vegetated continent and, while giant eucalypts inhabit the mesic parts of Australia, there are similar mesic habitats in the Southern Hemisphere that have not evolved comparably tall angiosperm trees. Solving this apparent paradox might illuminate the evolutionary advantages of tree gigantism. This demands understanding of giant eucalypts in a global context, yet this quest is frustrated by terminological issues surrounding the classification of Australian rain-forest (Adam, 1992; Bowman, 2000a; Lynch & Neldner, 2000). Although giant eucalypts typically regenerate following fire disturbance (Jackson, 1968; Ashton & Attiwill, 1994), they are often emergent from a tree layer made up of rain-forest species able to regenerate without disturbance (Bowman, 2000a). Several theories have been proposed to explain the co-occurrence of these pyrophobic and pyrophilic tree species (Gilbert, 1959; Jackson, 1968; Warman & Moles, 2009; Wood & Bowman, 2012), yet it remains unclear whether giant eucalypts form plant communities that are alternative states to rain-forest, or are simply ‘fire weeds’ (Cremer, 1960) on the margins of rain-forests and therefore functionally a pioneer rain-forest plant. Exploring these issues in a global context is the primary motivation of this review. To do this, we focus on the ecological and phylogenetic distributions of giant trees, how eucalypts fit into this group of plants, what allows eucalypts to be giants, and their intimate relationship with fire, and finally we integrate this information to consider the unique ecological relationships of eucalypts with rain-forest.

II. Giant eucalypts in a global context

We were able to identify reliable records of 46 angiosperm and gymnosperm species with heights over 70 m in natural vegetation (Table S1; includes taxonomic authorities). Anecdotal evidence suggests that a few other species (such as Cupressus cashmeriana Royle ex Carrière (Farjon, 2010) and possibly Ceiba pentandra (L.) Gaertn.) may also reach such heights. Some of the species with giant trees can grow very tall across much of their range (e.g. Sequoia sempervirens and Eucalyptus regnans), but others show large variation in stature depending on environment and genotype (e.g. Eucalyptus globulus (Jordan et al., 2000) and Pseudotsuga menziesii (Farjon, 2010)). This review will focus on the giant forms.

Although giant trees grow in both tropical and temperate regions, they are very restricted geographically and phylogenetically (Fig. 1). All the known giant trees occur in mesic climates, but nearly all of them (members of 43 species) are found in three regions: western North America from California to British Columbia, Southeast Asia (especially Borneo) and eastern Australia (Fig. 1; Table S1). The remaining three species with giant trees are from temperate zones in southern Russia and southwestern Australia. The 17 species of conifers with giant trees are members of three families (Cupressaceae, Pinaceae and Araucariaceae), whereas almost all of the 29 angiosperm species with giant members are eucalypts or emergent tropical rain-forest trees of the family Dipterocarpaceae (i.e. dipterocarps; Ashton & Hall, 1992).

Eucalyptus regnans is the tallest flowering plant on Earth (Figs 1, 3a) with a living Tasmanian tree measured at 99.6 m and a convincing historical record of 114.3 m for a tree in Victoria (Mifsud, 2002; Table S1). In fact, these cool temperate regions of Tasmania and eastern Victoria are centres for giant eucalypts, containing the six tallest recorded species of the genus (Fig. 1; Table S1). Several other eucalypt species with giant trees are found in the mesic, subtropical zone of eastern Australian, and one of these species, Eucalyptus grandis (Fig. 3b), extends into the humid tropical forest zone of northern Queensland. In the highest rainfall parts of southwestern Australia, which has a mediterranean-type climate, Eucalyptus jacksonii and Eucalyptus diversicolor (Fig. 3c) attain comparable heights (Boland et al., 2006). Eucalyptus deglupta is the only extra-Australian giant eucalyptus species. This species occurs naturally in Mindanao, Indonesia and Papua New Guinea (Carr, 1972; Whitmore, 1998). Several eucalypt species also attain heights of over 70 m in plantations outside Australia, and have become the tallest recorded angiosperm trees in some regions. Thus, the tallest recorded angiosperm in Europe is a 72-m E. diversicolor in the Caucasus Mountains (Nicolle, 2011); the tallest tree in New Zealand is an E. regnans that was 69 m in 1984 (Burstall & Sale, 1984), and is still growing, and the tallest known tree in Africa is an 81.5-m-tall specimen of Eucalyptus saligna Sm. (Trabado, 2008). The tallest measured tropical angiosperm (Shorea faguetiana; Dipterocarpaceae) stands at 88.1 m (R. Dial, pers. comm.) but among dipterocarps such heights are the exception, with emergent dipterocarps typically < 60 m tall (Wyatt-Smith, 1964; Cao & Zhang, 1997; Whitmore, 1998). In addition to great heights, there are some other commonalities among eucalypts and dipterocarps, including the species richness of these clades, and the presence of species (e.g. Eucalyptus obliqua and Shorea spp.) that can both compete with broadleaf understorey species in unburnt settings (thus forming closed forests) and persist in frequently burnt communities with grassy understoreys (thus forming savanna; Stott, 1984).

Figure 3.

Characteristics of giant eucalypts. (a) The Centurion at 99.6 m (Eucalyptus regnans), the world's tallest flowering plant, Arve Valley, Tasmania. This tree overtops the main canopy by over 60 m. (b) Eucalyptus grandis, Mt Paluma, Queensland. (c) Mature even-aged stand of Eucalyptus diversicolor tall forest with sclerophyllous understorey, Porongorup National Park, Western Australia. Rain-forest existed in Western Australia until c. 3 millon yr ago (Dodson & Macphail, 2004). (d) Serotinous woody capsules of E. globulus, Hobart, Tasmania. (e) Radial longitudinal section of the outer part of the epicormic strand in the bark of E. regnans, an obligate seeding species. A meristemic strip which may function as an epicormic strand is arrowed at the right of the image. Several other meristem strips are partially shown on the left. (f) Eucalyptus grandis plantation in Cameron Highlands, Malaysia. Note the dense regeneration of native rain-forest in the understorey.

A global analysis of plant height identified the rainfall of the wettest month as the main predictor of height in plants, with plants generally being taller in the tropics (Moles et al., 2009). However, this predictor fails to explain the distribution of giant trees. Elucidating the climatic determinants of tree height is beyond the scope of this review, but a basic climatic analysis on annual potential evapotranspiration and annual precipitation axes show that giant trees occupy a broad climatic envelope (Fig. 4) that is also occupied by many forest types with no giant trees. The giant trees also occur across a wide range of thermal regimes, ranging from the tropical lowlands to cool temperate regions (Table S1). The tallest temperate conifers and eucalypts appear to be clustered within a relatively narrow-range climate envelope centred on 1000 mm annual potential evapotranspiration and 1000 mm mean annual precipitation, perhaps signifying convergence in habitat requirement (Fig. 4). Indeed, Sequoia sempervirens (coastal redwood), the tallest coniferous tree, and E. regnans, the tallest angiosperm, have been considered to be ecological analogues (Box, 2002). However, this view ignores the large differences in life history strategies between these species (Sillett et al., 2010).

Figure 4.

Distribution of 35 of the tallest tree species in the world along annual potential evapotranspiration and annual precipitation axes. The 10 species labelled are: 1, Sequoia sempervirens; 2, Pseudotsuga menziesii; 3, Eucalyptus regnans; 4, Sequoiadendron giganteum; 5, Abies procera; 6, Eucalyptus viminalis; 7, Araucaria hunstenii; 8, Eucalyptus delegatensis; 9, Petersianthus quadrialatus; 10, Eucalyptus obliqua. Tropical angiosperms (Petersianthus, Shorea and others) and temperate angiosperms (Eucalyptus spp.) clearly occupy different positions along evapotranspiration–precipitation axes, and Araucaria hunstenii is also climatically segregated from both groups. The remaining species that have been used in this analysis are indicated in Supporting Information Table S1.

Sequoia sempervirens regenerates in tree-fall gaps, grows very slowly and lives for over 2000 yr (Busing & Fujimori, 2002). The large size attained by S. sempervirens is believed to provide a buffer against environmental stress (especially for nutrients and moisture) and the extremely long average intervals between destructive fires and storms permit this conifer to outgrow co-occurring hardwoods with more limited stature and life spans (Waring & Franklin, 1979). By contrast, E. regnans does not depend on extreme longevity to gain great height. This species regenerates prolifically after intense fires, has extremely rapid growth in the first 100 yr of life and then senesces after c. 500 yr (Wood et al., 2010). After the first few months of growth, this species overtops its co-occurring community, which typically becomes progressively more dominated by rain-forest species (Gilbert, 1959). Within angiosperm-dominated forest systems, the two-tiered syndrome of a fire-dependent forest towering above a fire-intolerant forest is known only in the associations between eucalypts and rain-forest. Dipterocarps typically germinate and establish below a closed forest canopy and maintain seedling banks below closed forest (Whitmore & Brown, 1996). Although many giant dipterocarps, such as Parashorea malaanonan and Shorea johorensis (Whitmore & Brown, 1996), require the high light intensities of canopy gaps for growth, these species differ from eucalypts in being able to persist under shade (Meijer & Wood, 1964). All other large angiosperms (e.g. Koompassia and Ceiba) are scattered emergents in tropical rain-forest, dependent on gap-phase regeneration (Whitmore, 1998) and, unlike giant eucalypts or dipterocarps (Ashton, 1981a), do not become canopy monodominants.

Some giant conifers show a similar dependence on fire for regeneration to giant eucalypts. For instance, Pseudotsuga menziesii (Douglas-fir) is shade-intolerant and shows very rapid growth after landscape-scale fires induce regeneration. This species is therefore considered to be a pioneering species relative to the shade-tolerant and slower growing Tsuga heterophylla (western hemlock), with which it co-occurs (Bušina, 2007). Picea sitchensis (Sitka spruce) is also fast-growing and responds well to fire, but does not require fire to initiate regeneration to the same degree as eucalypts (Alaback, 1982). Some Southern Hemisphere forests with tall conifers (e.g. Agathis australis (D. Don) Loudon in New Zealand; Araucaria araucana (Molina) K. Koch and Fitzroya cupressoides I.M. Johnston. in southern South America; Araucaria bernieri Buchh. in New Caledonia; and Araucaria hunsteinii in Papua New Guinea) regenerate after large and infrequent landscape-level disturbances such as tectonic instability and volcanism but are not specialized to regenerate following fire. These species attain great height by virtue of great longevity, thereby persisting as emergents above close-canopied vegetation that subsequently develops beneath them (Lane-Poole, 1925; Jaffré, 1995; Ogden & Stewart, 1995; Veblen et al., 1995).

III. Giant eucalypts – taxonomy and distribution

The molecular phylogeny with best representation of species of Eucalyptus (Steane et al., 2002; Bayly & Ladiges, 2007) shows that giant eucalypts occur in at least seven different clades – three within subgenus Eucalyptus (E. jacksonii, Eucalyptus delegatensis, and the clade containing E. regnans, E. obliqua and Eucalyptus pilularis) and four within subgenus Symphyomyrtus (E. diversicolor, E. deglupta, E. grandis and section Maidenaria: Eucalyptus viminalis, Eucalyptus nitens, Eucalyptus nobilis and E. globulus; Fig. 5). While not attaining heights exceeding 70 m, a number of other tall eucalypts (e.g. Eucalyptus dunnii Maiden, Eucalyptus macta L.A.S. Johnson & K.D. Hill, and Eucalyptus subcrenulata Maiden & Blakely) and some species in closely related genera (Corymbia, Lophostemon and Syncarpia) exhibit maximum heights of 50–70 m and occur as emergents in rain-forests (Benson & Hager, 1993; Harrington et al., 2000; Keith, 2004; Harris & Kitchener, 2005; Boland et al., 2006).

Figure 5.

Eucalypt phylogeny showing the phylogenetic position of various giant eucalypts (> 70 m max height; blue lines), based on nuclear ribosomal interspacer (ITS) sequences (simplified from Steane et al., 2002). Red lines indicate giant eucalypt species that exhibit obligate seeding (Nicolle, 2006). Gigantism appears to have arisen independently at least seven times, and obligate seeding in giant trees has arisen independently from resprouting taxa at least four times.

Giant eucalypts are unable to cope with prolonged periods of drought and are therefore restricted to areas that receive at least 50 mm of rainfall in the driest month (Ashton, 1981a; Fig. 6; Table 1). Some giant eucalypts have very narrow environmental ranges requiring both high rainfall and fertile soils. For example, the most site-sensitive giant eucalypts, such as E. regnans, require deep and well-drained soils (Ashton, 1981a) and typically occur on very wet sites with high and reliable rainfall (in excess of 1200 mm per annum). Underscoring the narrow niche of this species are results from forestry growth trials that revealed little genetic variation in growth rates among provenances (Raymond et al., 1997). Some species with giant members, such as E. obliqua and E. viminalis, have ecotypes tolerant of a wide range of edaphic or other environmental conditions, and can also occur in diminutive forms in areas with drier climates, infertile soils, or both (Wells & Hickey, 2005). In eastern Australia, the distribution and climatic envelope of giant eucalypts overlap with rain-forest (Adam, 1992; Figs 6, 7; Table 1). However, two giant eucalypt species, E. diversicolor and E. jacksonii, occur in southwestern Australia (Wardell-Johnson, 2000), where rain-forest became locally extinct c. 3 million yr ago (Dodson & Macphail, 2004; Wardell-Johnson, 2000; Figs 6, 7). Most E. diversicolor forests occur in drier climates compared with E. regnans and E. grandis forests (Fig. 7). The only extra-Australian giant eucalypt, E. deglupta, occurs in rain-forest in New Guinea, Indonesia and the Philippines (Carr, 1972) under a hot tropical ever-wet climate of 2500–5000 mm of precipitation per year (Table 1).

Figure 6.

The distribution of tall eucalypt forest (red) and rain-forest (green) in Australia. Areas exceeding 1000 mm annual precipitation (data from Bureau of Meteorology, 2011) are indicated in grey. The main areas of tall eucalypt forest discussed in the text represent: (a) cool temperate Tasmania, where tall eucalypt forest extends widely in association with rain-forest; (b) the north Queensland Wet Tropics, where tall eucalypt forest occurs only as a marginal strip on the western edge of rain-forests; and (c) the mediterranean-climate Western Australia, where rain-forest does not currently occur. Note that the boundaries between rain-forest and tall eucalypt forest in eastern Australia are approximate. In reality, large areas of tall eucalypt forest may contain a rain-forest understorey. (Sources: Queensland – Department of Environment & Resource Management, 2011; Tasmania – Department of Primary Industries & Water, 2009; Western Australia – Western Australian Herbarium, 1998).

Figure 7.

Occurrence of Eucalyptus diversicolor, Eucalyptus regnans and Eucalyptus grandis tall forest and associated rain-forest types in Tasmania (TAS) and north Queensland (Wet Tropics) along a water balance expressed as precipitation:evaporation (evaporation data: Donohue et al., 2010). Higher values indicate wetter environments. Each box encompasses the 25th to 75th percentiles; the median is indicated by the boldest vertical line and the other vertical lines outside the box indicate the 10th and 90th percentiles. Dots indicate the outliers. In both north Queensland and Tasmania, the vegetation data used included all vegetation types mapped under wet eucalypt forest (which would include the giant eucalypts in this review) and rain-forest. Rain-forest is not currently present in Western Australia (WA) but was present in the region up to c. 3 million yr ago (Dodson & Macphail, 2004). (Sources: Queensland – Department of Environment & Resource Management, 2011; Tasmania – Department of Primary Industries & Water, 2009; Western Australia – Western Australian Herbarium, 1998).

Table 1. Rainfall envelope and altitudinal range of selected giant eucalypts and co-occurring rain-forest trees (except southwestern Australia)
aClimatic zone/speciesMean annual rainfall (mm)Altitude (m asl)
  1. a

    Sources of rainfall envelope data: Australia species: Boland et al. (2006); E. deglupta:; Pometia pinnata:

  2. asl, above sea level.

Tropics (Papua New Guinea)Eucalyptus deglupta Blume2500–50000–1800
Pometia pinnata J.R. Forster & G. Forster f.1500–50000–1700
Tropics (far north Queensland)Eucalyptus grandis W. Hill ex Maiden1000–3500< 0–1100
Flindersia pimenteliana F. Muell1100–3800< 0–1200
Subtropics (central coast of Queensland and New South Wales)Eucalyptus pilularis Sm.900–1750< 0–700
Ceratopetalum apetalum D. Don1000–2000100–900
Temperate (eastern Australia and Tasmania)Eucalyptus regnans F. Muell750–1700150–1100
Eucalyptus obliqua L'Hér500–2400< 0–750
Nothofagus cunninghamii (Hook.) Oerst.1100–2500< 0–1570
Atherosperma moschatum Labill.1000–2000< 0–1375
Mediterranean (southwestern Australia)Eucalyptus diversicolor F. Muell.900–1300< 0–300
Eucalyptus jacksonii Maid.1150–125050–150

The geographical distribution of giant eucalypts has varied considerably through time. Molecular phylogeographical data imply that, during the last glacial maximum, E. regnans, E. obliqua and associated rain-forest species were limited to multiple refugial areas scattered across the current ranges of these species (Nevill et al., 2009; Worth et al., 2009; Bloomfield et al., 2011). Likewise, pollen analyses from volcanic crater lakes in humid tropical Queensland show that rain-forest boundaries, and undoubtedly co-occurring giant eucalypt forest, oscillated during the Quaternary (Kershaw, 1976; Haberle, 2005). This is consistent with the comparative genetic uniformity of populations of E. grandis in humid Queensland compared with the more differentiated populations in southeastern Queensland and New South Wales (Jones et al., 2006). In southwestern Australia, there is evidence of rain-forest occurring in the region as recently as c. 3 million yr ago (Dodson & Macphail, 2004). The extinction of rain-forest species in southwestern Australia may simply reflect the development of drier and more fire-prone climates inimical for those taxa (Dodson & Macphail, 2004). Thus, it is conceivable that E. diversicolor forest has replaced rain-forest and now fills the role of an alternative stable state to more pyrophylic vegetation types in the region (Bowman, 2000b; see also Figs 6, 7).

Fire frequency and local environmental conditions influence the understoreys, which in turn influence the flammability of giant eucalypt forests. Sites that are frequently burnt, have infertile soils, or both are typically dominated by sclerophyllous shrubs, grasses, graminoids or ferns that have phylogenetic and floristic links with the understorey species of dry eucalypt forest (Florence, 1964; Adam, 1992). Sites that are less frequently burnt and/or have more fertile soils favour mesic shrubs and, if they occur in the regional flora, rain-forest trees. However, the spatial extent and understorey type of giant eucalypt forests differ markedly in different climate zones. In the southwestern and southeastern temperate zones of mainland Australia, giant eucalypts can form extensive forests above a shrub layer, or, in parts of the southeast, intergrade with Nothofagus rain-forest (Keith, 2004; Harris & Kitchener, 2005). Tropical and subtropical giants are typically restricted to narrow bands (< 4 km width) sandwiched between humid tropical rain-forest and eucalypt savanna (Harrington et al., 2000; Tng et al., 2012), whereas in Tasmania these often intergrade with Nothofagus rain-forest (Gilbert, 1959; Fig. 6).

An important control of the landscape-scale pattern of giant eucalypt forests and rain-forest is fire intensity. Temperate rain-forests sustain surface fires of very low intensity (Hill, 1983), yet E. regnans forests have the highest fire intensities (> 50 000 kW m−1) of any vegetation type in Australia (McCarthy et al., 1999), comparable to some Canadian and Alaskan boreal coniferous forests (Van Wagner, 1983). Such fires occur as a result of infrequent severe fire weather and antecedent droughts, as indicated by the high forest fire danger index (FFDI) values (Noble et al., 1980) of giant eucalypt forests (Fig. 8). While tropical E. grandis forests have a higher mean FFDI, these forests are exposed to less extreme FFDI events than their temperate equivalents in southeastern and southwestern Australia (Fig. 8). This is a potential explanation as to why E. grandis does not penetrate tropical rain-forest habitats to the same extent as E. regnans infiltrates cool temperate rain-forest environments. Similarly, it is also plausible that the higher frequency of high to severe FFDI events of E. diversicolor forest compared with both E. regnans and E. grandis (Fig. 8) may be related to the occurrence of E. diversicolor in drier habitats (Fig. 7) and/or the lack of rain-forest in the region. Indeed, there is evidence that the microclimate of humid tropical rain-forest understoreys renders the vegetation type less flammable than the adjacent and more open canopied E. grandis forests (Little et al., 2012), and this is probably the case for E. regnans forests (Jackson, 1968). It is possible that flammability of E. regnans varies with age of the trees (McCarthy et al., 2001); for instance Jackson (1968) believed that younger regrowth eucalypt forest had higher flammability than older mixed NothofagusE. regnans forest.

Figure 8.

Forest fire danger index of three regions where tall eucalypt forest occurs (Tasmania, Wet Tropics and southwest Western Australia: Fig. 6), calculated using daily weather records (Bureau of Meteorology, 2011). The McArthur forest fire danger index is a widely used measure of fire risk in Australia and can be calculated for any given area (Noble et al., 1980). The index is measured using climatic data such as maximum temperature, mean wind speed, minimum relative humidity, total rainfall and mean soil moisture. Each box encompasses the 25th to 75th percentiles; the median is indicated by the boldest vertical line and the other vertical lines outside the box indicate the 10th and 90th percentiles. Dots indicate the outliers. A fire danger index over 50, as indicated by the line in the graph, is considered severe.

IV. Growth of giant eucalypts

Understanding the evolution of gigantism in eucalypts requires a consideration of the characteristics of eucalypts that allow them to exceed the heights of their co-occurring species, and the specific features of giant species that allow them to reach such extreme heights. Eucalypts and some closely related genera (Corymbia, Lophostemon and Syncarpia) form all of the canopy or emergent trees of many vegetation types in Australia (Groves, 1999), and are considerably taller than all or almost all other species in these communities. This trend of ‘hyper-emergence’ extends across climates and clades, with eucalypts and related genera exceeding the heights of other co-occurring species in many other vegetation types, including heath, mallee, dry sclerophyll, subalpine and savanna communities (Groves, 1999). Some giant eucalypt trees are > 60 m taller than the underlying rain-forest canopy (Fig. 3a). This characteristic of hyper-emergence involves not only the capacity to deal with the mechanical and hydraulic limitations of height per se, but also the need for rapid growth to attain this height before the tree senesces or succumbs to disease. The nearly ubiquitous nature of hyper-emergence in eucalypts suggests that the trait of hyper-emergence is an ancestral feature of the eucalypt lineages, and if this is true, would have arisen > 60 million yr ago (see the dated phylogeny of Crisp et al., 2011). However, cross-matching the eucalypt phylogeny (Fig. 5) with the molecular dates of Crisp et al. (2011) suggests that all of the evolutionary transitions into giant trees occurred in the last 20 million yr, and it is possible that many, or even all of them, are much more recent. Most of the last 20 million yr has been associated with increasing aridification of the Australian continent (Bowler, 1982) which appears to have resulted in significant increases in the frequency of fire in rain-forests habitats (Kershaw et al., 1994). Thus, we propose that gigantism in eucalypts evolved opportunistically when members of this group of hyper-emergent plants were exposed to an environment that contained both the fire essential for regeneration and the environmental conditions that allowed the rapid growth necessary for a species to reach extreme height without extreme longevity.

The mechanisms that allow eucalypts to be much taller than other species in a wide range of habitats are poorly understood, and represent a fertile field of potential research. Such mechanistic explanations could consider how eucalypts deal with a series of problems relating to the difficulties of creating a trunk and root system that is biomechanically (Niklas, 1994) and hydraulically (Koch et al., 2004) adequate. The explanations could also consider what characteristics give the eucalypts (and other angiosperms) the high relative growth rates necessary to construct this trunk within the short lifespan of these trees compared with giant conifers (Stephenson & van Mantgem, 2005). Furthermore, evolutionary explanations of hyper-emergence in eucalypts should consider the adaptive costs and benefits of far exceeding the heights of competitors (Falster & Westoby, 2003). Little is known about these aspects of eucalypt biology, but some evidence has been presented regarding growth rates and hydraulics.

Although the physiological basis for rapid growth in eucalypts is unknown, there are at least two plausible theories for an adaptive advantage for very rapid height growth. Bond (2008) argued that very rapid early growth of trees, including savanna eucalypts, may allow them to escape a ‘fire trap’ such that this growth would allow saplings to reach heights that allow them to avoid the effects of high-intensity ground fires. Such processes could apply to eucalypts in general, but may be less applicable to giant eucalypt species, for which fire return intervals are typically very long (decades to centuries). In the giant eucalypts, shade intolerance combined with intense intraspecific and interspecific competition provides a strong selection pressure for rapid growth (Ashton, 1981b; Hardner & Potts, 1997; Falster & Westoby, 2003). For example, E. regnans can grow as quickly as 2 m yr−1 in the first decade (Ashton, 1981a) and attain half of its mature height within the first 25–35 yr (Jackson, 1968). During this phase there is intense self-thinning of the initial high seedling densities (Ashton, 1976; Jackson, 1968) and surviving stems form straight, branch-free trunks as a result of shedding of shaded branches (Jacobs, 1955). Hardner & Potts (1997) demonstrated that this selection had strong genetic effects by showing that inbred genotypes of E. regnans were rapidly eliminated, leaving only outbred individuals with rapid growth.

The few comparative studies of the functional ecology of giant eucalypts have suggested that eucalypts overtop their co-occurring rain-forest species. In the humid tropics, Duff (1987) studied a suite of nine species spanning a tropical rain-forest–tall eucalypt forest boundary and found that E. grandis grew faster and acquired more biomass than both pioneer and climax rain-forest taxa when grown under glasshouse conditions. Eucalyptus grandis exhibited patterns of resource allocation that were broadly similar to those of fast-growing rain-forest pioneers such as Alphitonia and Toona, and, given sufficient light and nutrients, could capitalize on the available resources more efficiently than the rain-forest species. Similar findings were reported by Barrett & Ash (1992), who compared the growth and carbon partitioning of rain-forest and eucalypt species occurring along a vegetation transitional sequence in south coastal New South Wales. They found that, under high irradiance, the mean plant biomass of eucalypts exceeded that of ecotonal species and rain-forest species, and concluded that the eucalypts maximized leaf area in proportion to plant mass for a given level of irradiance, presumably to maintain high growth rates.

Ryan & Yoder (1997) argued that hydraulics were major determinants of tree height, because greater height resulted in greater xylem resistance as a result of the greater distance over which water must be conducted and increased gravitational potential opposing the ascent of water in taller trees. Tall trees can deal with these effects (equal to 1 MP in water potential for 100 m in height) by some combination of constructing highly conductive xylem and operating at very low leaf water potentials. Furthermore, the risk of embolism in water-conducting tissue (Tyree & Sperry, 1989) increases not only with whole-plant water deficit but with tree height (Koch et al., 2004). These factors provide a possible explanation of why giant trees are restricted to mesic environments. There is some evidence that giant eucalypts have specific features that may help overcome hydraulic limitations associated with great height.

Compared with other hardwoods, E. regnans (England & Attiwill, 2007) and E. delegatensis (Mokany et al., 2003) have comparatively wide sapwood vessels (up to 278 and 270 μm, respectively). Likewise, E. regnans (Legge, 1985) and E. obliqua (Skene & Balodis, 1968) have comparatively long vessels (1.8 and 4 m). The pipe model theory proposes that, for a given tree species, the ratio of sapwood area (As) to foliage area (Af) should remain constant (Waring et al., 1982) or in fact increase as trees grow taller to compensate for the increased path length that water must travel to reach the leaves (Magnani et al., 2000). This mechanism of increasing As:Af ratios to cope with increasing heights has been demonstrated for a range of trees and specifically for the giant conifer Pseudotsuga menziesii (McDowell et al., 2002) and the tall eucalypt E. saligna (Barnard & Ryan, 2003). However, giant eucalypt species such as E. regnans (Vertessy et al., 1995), E. delegatensis (Mokany et al., 2003) and the related Eucalyptus sieberi L.A.S. Johnson (Roberts et al., 2001) exhibit decreasing As:Af ratios with increasing height, largely as a result of an increase in the specific conductivity of sapwood (Mokany et al., 2003; England & Attiwill, 2007). As E. regnans, E. pilularis, E. globulus and E. nitens trees mature, increased sapwood conductivity is achieved initially through increased vessel diameter, after which subsequent increases in conductivity result from increases in vessel density (Bamber & Curtin, 1974; England & Attiwill, 2007; Hudson et al., 1998). Petit et al. (2010) measured the vertical profiles of the conduit (i.e. vessel) dimensions and density of E. regnans trees of varying heights. They found that the way in which the xylem tapers in E. regnans is unusual and constitutes a highly effective strategy for compensating for the hydraulic limitations caused by increased tree height. They concluded that, relative to other fast-growing trees, E. regnans has evolved a xylem design that ensures a high hydraulic efficiency, enabling the species to rapidly attain heights beyond the maximum height (50–60 m) of most other hardwood trees (Petit et al., 2010).

V. Fire and regeneration of giant eucalypts

In their natural range, giant Australian eucalypts are generally known to be dependent on fire for regeneration (Ashton & Attiwill, 1994). In eucalypts, as with woody plants in general, there are two broad fire regeneration syndromes: obligate seeders and resprouters. Obligate seeders are usually killed by fire, but can have a competitive advantage over resprouters by growing more rapidly and maturing earlier than resprouters because they do not invest in protective structures and storage organs and regenerative tissues (Bond & van Wilgen, 1996; Knox & Clarke, 2005). Although almost all eucalypts exhibit strong resprouting responses after fire, several giant eucalypts, notably E. regnans, E. grandis, E. delegatensis and E. deglupta, are obligate seeders (Nicolle, 2006). Thus, E. regnans has an aerial seed bank in the form of woody capsules (i.e. Fig. 3d) that protect seeds from the heat of a fire (Ashton, 1981a), limited epicormic regrowth and no lignotubers (Nicolle, 2006; Waters et al., 2010). The large quantity of viable seed released after a crown-scorching fire saturates seed predators, allowing the survival of huge numbers of seedlings (Ashton, 1979; O'Dowd & Gill, 1984). Fire also releases nutrients and ameliorates soil conditions which would otherwise be unfavourable for seed germination and seedling growth of eucalypts such as E. regnans (Chambers & Attiwill, 1994). Growth of seedlings is further enhanced because the death of canopy and emergent trees releases seedlings from short-term competition for environmental resources (Dignan et al., 1998; Van Der Meer et al., 1999). However, intraspecific competition rapidly comes into play, further enhancing height growth (see section VI, ‘'Are giant eucalypts different from other rain-forest trees?'’).

Although severe fire typically results in the death of obligate seeding eucalypts, triggering massive regeneration and development in even-aged stands across large expanses of landscape (e.g. Fig. 3c; Ashton, 1975, 1981a; Wardell-Johnson et al., 1997), adults quite often survive in patches where fire is less intense (Gilbert, 1959; Vivian et al., 2008), creating mixed-aged stands (Simkin & Baker, 2008; Turner et al., 2009). Thus, Turner et al. (2009) found that almost half the stands of E. regnans in Tasmania were mixed-aged.

Nevertheless, obligate seeders such as E. regnans are less likely to form mixed-aged forests than resprouting species such as E. obliqua (Turner et al., 2009). These resprouting species can possess combinations of well-developed vegetative recovery mechanisms such as thick bark, epicormic buds and lignotubers, although some ecologically diverse species (e.g. E. viminalis) have less pronounced lignotubers in environments suitable for gigantism (Ladiges, 1974). Resprouter giant eucalypts also exhibit slower growth rates than obligate seeders such as E. regnans, probably as a consequence of the cost/benefit trade-offs of investing in lignotubers and/or thick bark (Ashton, 1981a). Smooth-barked eucalypt species have less fire protection than species with thick bark, but need to invest less in bark growth and can achieve small photosynthetic gains from chloroplasts in the bark (Cernusak et al., 2006).

The analysis of Crisp et al. (2011) suggests that an anatomical feature (deeply embedded cambial strands capable of generating epicormic stems) that enables prolific vegetative regeneration throughout the genus is an ancient adaptation allowing recovery from fire dating back to c. 60 million yr ago. Their analysis therefore implies that obligate seeding in E. regnans is a derived feature (Fig. 5) given that this species is deeply nested within the eucalypts (Ladiges et al., 2010) but still possesses the specialized cambial strands (Waters et al., 2010; Fig. 3e). The evolution of other fire-related traits in eucalypts is less clear. Thus, it remains unclear if eucalypts have specific adaptations to increase flammability, and hence increase their regeneration niche (e.g. Bradshaw et al., 2011; Keeley et al., 2011). For example, although oil-rich foliage is often claimed to be an adaptation to increase flammability, there is strong evidence that it acts as a chemical defence against invertebrate and vertebrate herbivores (O'Reilly-Wapstra et al., 2004). It is true that decorticating bark strips spread spotfires (Mount, 1979), but whether this feature is an adaptation for this purpose remains unproven (see Bowman et al., 2012).

VI. Are giant eucalypts different from other rain-forest trees?

We have demonstrated in the preceding review that Australian giant eucalypts are globally distinctive given (1) their dependence on fire to regenerate in rain-forest environments and (2) the development of an emergent canopy overtopping rain-forest. Although Schimper (1903) included giant eucalypts as rain-forest trees because they occurred in mesic environments, most Australian ecologists consider that giant eucalypts are not rain-forest trees because of their dependence on fire for regeneration. This has created ongoing controversy about the definition of ‘rain-forest’ amongst ecologists and environmentalists (e.g. Adam, 1992; Bowman, 2000a; Lynch & Nelder, 2000). Flammable eucalypt forests are accepted as being ecological distinct from pyrophobic rain forests, rendering the vast majority of Australian forests as having no global analogue, thereby frustrating international comparisons. Logging and burning a cut stand of giant eucalypts have been widely regarded as acceptable practices because these systems regenerate after fire disturbance, and because giant eucalypt forests are not classified as ‘rain forest’ they were not affected by the phasing out of rain-forest logging that has occurred in New South Wales and Queensland.

Rain-forest classifications excluding giant eucalypts on the basis of their fire dependence also run into a number of logical issues. First, lower statured and fire-sensitive rain-forest types in dry areas (oxymoronically described as ‘dry rain forest’) often have scattered eucalypt emergents (Sattler & Williams, 1999), commensurate with those in mesic rain forests with giant eucalypts. Secondly, non-forest fire-sensitive vegetation, such as alpine coniferous heaths, has ecological relationships with flammable vegetation analogous to those between rain forests and eucalypt forests (Bowman, 2000b). Thirdly, the extra-Australian giant eucalypt E. deglupta has always been accepted as a rain-forest tree (Carr, 1972). Likewise, no consensus has been reached regarding the question of whether a suite of other tall eucalypts (e.g. E. dunnii, Eucalyptus pellita F. Muell. and E. macta) and other analogous Myrtaceae (e.g. Lophostemon, Corymbia intermedia (R.T. Baker) K.D. Hill & L.A.S. Johnson and C. torelliana (F. Muell.) K.D. Hill & L.A.S. Johnson) which co-occur with rain forests are true rain-forest trees (Benson & Hager, 1993; Sattler & Williams, 1999). Vegetation with co-occurring giant eucalypts and Nothofagus rain-forest has been described as a specific plant community called ‘mixed forests’ (Gilbert, 1959). Jackson (1968), while acknowledging these forests to be successional, also considered them to be sufficiently stable to persist as a distinct vegetation type across the landscape (Wood & Bowman, 2012). In contrast, Warman & Moles (2009) suggested that giant eucalypt forests in northern Queensland are not actual plant communities, but rather a eucalypt-dominated unstable ecotone sandwiched between the two alternative stable states of pyrophobic tropical rain forest and pyrophilic savanna. In terms of function, giant eucalypts could then be considered rain-forest pioneers (Smith & Guyer, 1983). To address whether giant eucalypts should be considered as rain-forest species, we will consider how they fit into global views on what constitutes a rain-forest pioneer species, and where giant eucalypts fit along the pioneer–climax species spectrum.

Within a rain-forest, the pioneer–climax species spectrum refers to a continuum of species which have different tolerance to light or gap sizes (Turner, 2004). Climax species are typically extremely shade-tolerant, have the ability to regenerate continuously and subsequently grow or persist in a suppressed state under the dense shade of the forest canopy until released from this suppression by the influx of light caused by a tree-fall gap or other disturbances. By contrast, two characteristics are diagnostic of rain-forest pioneer species: seed germination that is dependent on the exposed conditions present in canopy gaps; and shade intolerance (Turner, 2004). Giant eucalypts conform to both of these features, albeit that the seedbeds and forest gaps are typically created by fires (Ashton, 1975; Ashton & Attiwill, 1994). Furthermore, rain-forest pioneers show r-selected reproductive strategies, with high reproductive output and rapid growth enabling them to complete their life cycle before being suppressed by slower-growing and more shade-tolerant trees (Swaine & Whitmore, 1988; Whitmore, 1998; Turner, 2004). Giant eucalypts employ similar r-selected reproductive strategies, with early reproductive maturity, prolific and often continuous production of small seeds and extremely rapid height growth that allows them to overtop slower-growing and more shade-tolerant trees in height by c. 50% (Ashton, 1981a). Another feature of many pioneer species is a persistent seed bank (Turner, 2004). In rain-forest pioneers this seed bank is commonly held in the soil, but the aerial seed bank of eucalypts can be argued to provide an analogous function.

Given that the life spans of giant eucalypts (c. 400–500 yr: Wood et al., 2010) are often equivalent to, or even greater than, those of co-occurring rain-forest species, the best analogue in the pioneer–climax paradigm that would apply to these eucalypts would be a subset of pioneer species known as long-lived secondary species (see Condit et al., 1998), or ‘large pioneers’ (Swaine & Whitmore, 1988). Examples of this guild of species include various Southern Hemisphere conifers discussed in Section II, ‘'Giant eucalypts in a global context'’, and some angiosperms such as Ceiba pentandra and Dipteryx panamaensis (Pittier) Record & Mell from tropical South America, and Weinmannia trichosperma Ruiz & Pav. from temperate Chile. While some of these species do not grow to extreme heights, they resemble eucalypts in being shade-intolerant and reliant on large infrequent disturbances (Condit et al., 1998; Lusk, 1999).

On the whole, giant eucalypts in mature forest show little or no sign that they actively inhibit the growth of developing rain-forest or late successional species in either the tropics or the temperate zone, consistent with the view that giant eucalypts are rain-forest pioneer trees. Thus, Tng et al. (2012) documented the expansion of humid tropical rain-forest into E. grandis forest in northern Queensland over the last 50 yr, and a similar process has been observed in E. grandis plantation within 27 yr on the central coast of New South Wales (Turner & Lambert, 1983). Similarly, the temperate rain-forest dominant Nothofagus cunninghamii colonizes the understoreys of unburnt Eucalyptus forests across a range of soil types (Ellis, 1985). In the absence of fire, the understoreys of E. regnans forest in central Victoria are being invaded by Pittosporum undulatum Vent., a broad-leaved rain-forest tree (Gleadow & Ashton, 1981).

While there is no experimental evidence showing that giant eucalypts facilitate rain-forest succession in natural settings, there are many possible ways that such facilitation could occur. For instance, overstorey trees in a regenerating rain-forest can improve soil water balance and give shallow-rooted plants such as rain-forest seedlings access to water through hydraulic lift (Phillips & Riha, 1994; Emerman & Dawson, 1996). Shade from the overstorey that can minimize photostress for regenerating plants, and reduce evaporative demand at times of water deficit (Messier et al., 1998), may also be important. Guevara et al. (1986) highlighted the importance of trees as perches for avian dispersers of rain-forest plants, and this is perhaps the most immediate and easily observable way in which giant eucalypts may act as facilitators of rain-forest regeneration. This mechanism of ecological facilitation is known as nucleation (Reis et al., 2010), whereby rain-forest trees, in particular bird-dispersed taxa, regenerate and exhibit a clustered distribution under pre-existing trees in the landscape. In northern Queensland, nucleation of rain-forest trees has been documented in eucalypt woodlands (Russell-Smith et al., 2004) and, likewise, bird-dispersed rain-forest taxa are also very common in the understoreys of E. grandis forest (D. Y. P. Tng, unpublished).

Ironically, the most detailed evidence for giant eucalypts facilitating the regeneration of rain-forest trees comes from extra-Australian studies of eucalypt plantations. Feyera et al. (2002) summarized data showing that the canopies of established plantation eucalypt trees can have facilitative or nurse effects on the regeneration of natural rain-forest. Other studies document uninhibited regeneration of rain-forest species under eucalypt plantations. For instance, native rain-forest species have been observed to regenerate in the understoreys of Eucalyptus grandis plantations in both Brazil (da Silva et al., 1995) and the Cameron Highlands in Peninsular Malaysia (D. Y. P. Tng, pers. obs.; Fig. 3f). Similar observations of native forest regeneration have also been documented in South Africa in the understorey of E. saligna plantations (Geldenhuys, 1997). This can be contrasted with cases where other exotic trees such as teak (Tectona grandis L.f.) can inhibit native vegetation regeneration (Healey & Gara, 2003).

In summary, we argue that, under optimal giant eucalypt regeneration, high eucalypt seedling density and intense competition for space and resources immediately following disturbance are inhibitive to rain-forest (as reviewed in Section IV, ‘'Growth of giant eucalypts'’). During the middle and later growth phases of the eucalypts, however, rain-forest regeneration is facilitated. This pattern is consistent with Finegan's (1984) schema where pioneer trees are defined by the ability to colonize, grow and produce seed in early successional environments. We therefore assert that there is a case for treating giant eucalypts as rain-forest pioneer trees, albeit with unique features relating to fire disturbance.

VII. Conclusions

Giant eucalypts are among the tallest plants on Earth, and include the tallest angiosperm. The giant eucalypt syndrome occurs in at least seven clades within eucalypts, among species occurring from tropical to temperate environments. Giant eucalypts can coexist with rain-forest trees on the margins of tropical rain-forests in Queensland, form large expanses of mixed Nothofagus rain-forest with emergent eucalypts in Tasmania and are the sole canopy tree in forests with shrubby understoreys in areas suitable for rain-forests in Victoria and southwestern Western Australia (Ashton, 1981a; Sattler & Williams, 1999; Wardell-Johnson, 2000; Harris & Kitchener, 2005). Although giant eucalypts require intense fire to regenerate and outcompete other rain-forest species, once established, adults do not significantly suppress, and possibly even facilitate, the development of continuously regenerating understorey made up of the same rain-forest species. The dependence of giant eucalypts on fire for regeneration, in contrast to rain-forest trees, has led Australian ecologists, with some exceptions (e.g. Smith & Guyer, 1983; Warman & Moles, 2009), to treat these forests as a distinct ecosystem. This approach to vegetation classification has created ongoing controversy about the definition of rain-forest in Australia (e.g. Bowman, 2000a; Lynch & Nelder, 2000) that has dogged Australian ecology and environmental politics for years. Further, it has stymied international comparative studies because of difficulties in relating Australian vegetation types to those on other continents. However, such problems disappear if we adopt the paradigm that giant eucalypts are functionally rain-forest trees, albeit globally unique pioneer species that depend on fire for regeneration.

Giant eucalypts conform to a general trend for eucalypts to act as hyper-emergents wherever they occur; the tallest eucalypt species may simply be those that can compete and/or survive in their habitat (i.e. the rain-forest habitat). An underlying capacity for gigantism may therefore have evolved once, with convergent evolution of other traits (e.g. rapid growth and obligate seeding) allowing the expression of extreme heights by providing these species with the capacity to occupy the relevant (rain-forest) habitats. The success of the giant eucalypts under contemporary conditions is variable and the available data show that past climates have influenced the distribution of giant eucalypt forests. For example, under glacial climates E. regnans was more restricted in range (Nevill et al., 2009). Climate has a very strong effect on fire activity, and it remains unclear how much the potential increase in flammability with the arrival of eucalypts changed the competitive balance with other rain-forest trees. A functional trait-based approach to understanding the ecological niche of giant eucalypts and the role of climate and eucalypt traits on fire activity is of considerable theoretical and applied significance. Phylogenetic research is required to explore whether there has been a co-evolutionary relationship between fire and eucalypts. Such knowledge is important because any significant increase in flammability created by eucalypts would have long-term implications for natural habitats in extra-Australian regions (e.g. Brazil, China and Portugal; da Silva et al., 1995; Malvar et al., 2011; Zhang et al., 2012) where eucalypt plantations are becoming increasingly important. If climate is the main arbiter of the competitive balance between rain-forest, giant eucalypts and fire activity, then warmer climate may see a further dominance of eucalypts. Monitoring the dynamics of giant eucalypts forests is a key step in understanding these temporal trends.


We thank the Wet Tropics Management Authority for provision of resources, and for hosting one of us (D.Y.P.T.) in kind, and Ed Frank, Ellen Weber, Steve Goosem, Jeremy Little, Sam Wood, Brad Potts, Tim Brodribb and Brett Murphy for stimulating discussions. We also thank Roman Dial for permission to use unpublished data, Farhan Bokhari and Geoff Burrows for providing images (Fig. 3c,e) and Louise Gilfedder and three anonymous referees whose comments have helped improve the manuscript substantially. This work was partly funded by the Australian Government to assist the Independent Verification Process established under the Tasmanian Forests Intergovernmental Agreement.