This account presents information on all aspects of the biology of Gunnera tinctoria (Molina) Mirb. (G. chilensis Lam.; G. scabra Ruix & Pav.; G. pilosa Kunth) that are relevant to understanding its ecological characteristics and behaviour. The main topics are presented within the standard framework of the Biological Flora of the British Isles: distribution, habitat, communities, responses to biotic factors, responses to environment, structure and physiology, phenology, floral and seed characters, herbivores and disease, history and conservation.
Gunnera tinctoria is a gynomonoecious, clonal, perennial herb that is naturalized in parts of Britain, becoming invasive in parts of Ireland and, more recently, Scotland. It occurs where winter temperatures are mild, and precipitation and humidity are high. Gunnera tinctoria is native to South America, predominantly in the Andean region of Chile and Colombia, and probably in parts of Argentina, ranging from sea level to c. 2000 m a.s.l.
Typical habitats in Britain and Ireland include stream and river banks, lake and pond margins, coastal cliffs, as well as disturbed areas, such as roadsides, quarries and ditches. In its native range it occurs predominantly on the banks of rivers and streams, on coastal cliffs and within canopy gaps or at the margins of temperate-humid rain forests.
Gunnera tinctoria occurs on a variety of substrates, mainly on alluvial or colluvial soils derived from volcanic material or on thin gley soils of marine origin. In Ireland, it occurs naturally on soils with a pH ranging from 4.6 to 6.2 and has been cultivated in soils with a pH up to 7. Soil moisture content and soil organic matter vary greatly, although it rarely colonizes highly organic soils such as peat. It is susceptible to even mild water deficits at all stages of development, and its seedlings are also sensitive to waterlogged conditions.
Gunnera tinctoria produces large numbers of seeds and also spreads clonally, by a horizontal rhizome system. It is wind pollinated, although insect pollination has been reported in New Zealand. Seeds are likely to be predominantly water and/or bird dispersed. In its invasive range, it can form a large and persistent soil seed bank. Recruitment from seeds seems to be important for its initial establishment, while vegetative propagation is the main means of expansion, leading to dense clonal stands. Long-distance seed dispersal seems to be central to the colonization of new areas, although the transport of vegetative propagules may also be important.
Gunnera tinctoria is a strong competitor in its invasive range, particularly in wet, humid environments. Its competitive ability arises from its large stature, the persistence of its seeds and rhizomes and a capacity for fixing nitrogen through a unique intracellular symbiosis with cyanobacteria (Nostoc) that may be particularly important for supporting the rapid growth of established plants early in the spring.
Giant rhubarb. Gunneraceae. Subgenus Panke. Gunnera tinctoria (Molina) Mirb. (G. chilensis Lam.; G. scabra Ruix & Pav.; G. pilosa Kunth) is a large perennial, polycarpic, hemicryptophyte herb, up to 2 m in height. Rhizomes 6–25 cm in cross-section (Silva, Tavares & Pena 1996; M. Gioria, pers. obs.), mainly horizontal, above-ground, up to 3.5 m long (Hickey 2002), covered with cataphylls and containing green cup-shaped structures (Osborne et al. 1991). Petioles reddish-purple, except in very young plants, 18–100 cm, polystelic. Leaves deciduous, 30–200 cm in diameter, orbicular and palmately lobed, with unicellular, glandular hairs on the upper leaf surface and with non-glandular prickles on the petioles and larger veins of the lower leaf surface (Wilkinson & Wanntorp 2007). Inflorescence a robust panicle, three to four per plant, up to 100 cm, predominantly hermaphrodite near the apex and female at the base (Webb, Sykes & Garnock-Jones 1988; Pena 1995; González & Bello 2009). Individual flowers totalling hundreds of thousands per plant, sessile, regularly distributed on inflorescences. Terminal, most sub-terminal and mid-level flowers bisexual, with two pairs of perianth segments; female flowers with a single perianth segment. Petals well developed in bisexual flowers, highly reduced or lacking in female flowers. Stamens 1–2; styles 2, shorter than the ovary; ovary inferior with one ovule. Fruit an oblong, red, fleshy, indehiscent drupe (2 × 3 mm), containing a single ovoid seed with a copious oily endosperm (Williams et al. 2005).
In Britain and Ireland, and more generally in its non-native range, G. tinctoria is often mistaken for G. manicata, which differs in having more open inflorescences, green rather than reddish-brown old flowers, a narrower diameter of the central part of the main inflorescence axis (3–3.3 vs. 4–4.5 cm for G. tinctoria), longer inflorescence branches (9.5–11 vs.5–7 cm) and a narrower diameter of the inflorescence branches (3–4 vs. 5–7 mm) (Sykes 1969; Clement 2003; Grant 2004). The leaves of G. manicata are larger, often > 2 m, and are pinnately lobed rather than palmately lobed. Despite sharing a number of features, including a large stature and similar habitat preferences, only G. tinctoria has formed invasive populations, while records for G. manicata tend to be confined to areas where it was introduced and naturalized populations are scarce (Sykes 1969; Preston, Pearman & Dines 2002). For further information on the origins and identity of G. manicata, see Clement (2003), Wanntorp (2003), Wanntorp, Wanntorp & Källersjö 2002b and Shaw (2007).
A popular ornamental species from South America, Gunnera tinctoria, has long been introduced to gardens in Britain and Ireland, from which it has escaped and become invasive under suitably mild and moist climatic conditions.
I Geographical and altitudinal distribution
An alien species to Britain, G. tinctoria occurs predominantly in western coastal regions (Fig. 1), although records are increasing from central and eastern regions. It is established in England and is considered invasive (sensu Richardson et al. 2000) in western parts of Cornwall (Fig. 1; Pilkington 2011). It has been recorded in London, Surrey and West Sussex (Pilkington 2011) and is established in Derbyshire (Moyes & Willmot 2002), while it is very rare in Staffordshire (Hawksford & Hopkins 2011). In Wales, G. tinctoria is principally found in large gardens (Wade, Kay & Ellis 1994; Chater 2010; T. Rich, pers. comm.). In Scotland, it is naturalized in the Inner Hebrides and the Isle of Bute. In the Outer Hebrides, it is invasive in parts of North Harris and in the Lewis Castle grounds at Stornoway and is also found at Carloway, Leverburgh, Flodabay, Finsbay and Geocrab (R. Reid, pers. comm.). Gunnera tinctoria is absent from the Orkney and Shetland Islands. It occurs on the Isles of Scilly (French 2009b), the Isle of Man and the Channel Islands (McClintock 1975). In Northern Ireland, the species has not formed any invasive population and occurs in the proximity of large gardens, such as at Tempo Manor, Florencecourt and Ederny, Co. Fermanagh (R. Northridge, pers. comm.), and at Shane's Castle Estate, Lough Neagh, Co. Antrim (Beesley 2006). In Ireland, G. tinctoria displays a pronounced westerly distribution, where it is primarily found close to sea level at low altitudes, typically <100 m a.s.l. (Hickey & Osborne 1998; Gioria & Osborne 2009). The majority of invasive populations are found in Co. Mayo and Co. Galway. It also occurs on Clare Island, Co. Mayo (Doyle & Foss 1986). Details of a number of Irish records can be found in Reynolds (2002).
Gunnera tinctoria is native to South America, predominantly in Chile, from Coquimbo to Magallanes, from sea level to 2000 m a.s.l., mostly north of 47° S (Osborne 1988); it is also considered to be native in parts of Argentina, and in the Andean region of Colombia, Venezuela, Peru and Ecuador (30°–52° S; Molina 1978; Schick 1980, 1981; Silva, Tavares & Pena 1996), although definite information is lacking. Outside its native range, the species is also naturalized in north-western France (Osborne et al. 1991) and parts of California (Marin County; Howell 1970). Its presence has been recorded in Spain, although it has not formed naturalized populations (Sanz-Elorza, Dana & Sobrino 2001). Duretto (2009) reported that the species is cultivated in Tasmania and may persist for a short period of time as a garden escape, although its naturalization status there is unclear (M. Baker, pers. comm.). It is invasive on São Miguel Island, in the Azores (Fig. 2), predominantly in the east of the island, at altitudes between 100 and 900 m a.s.l., mainly above 500 m (60% of the records; Pena 1995; Silva, Tavares & Pena 1996). In New Zealand, G. tinctoria occurs on all three main islands (North Island, South Island and Stewart Island; Williams et al. 2005; Heenan et al. 2009) and is naturalized in all thirteen Department of Conservation Conservancies (Owen 1997). It is invasive on Mount Taranaki, in the North Island, from sea level to 380 m a.s.l., while it has a scattered distribution in the rest of the country (Williams et al. 2005). It is also naturalized in the Chatham Islands (de Lange, Heenan & Rolfe 2011).
(A) Climatic and Topographical Limitations
Gunnera tinctoria occurs in 237 10-km squares (hectads) in Britain and Ireland of a total of 3907 (c. 5%) (Preston, Pearman & Dines 2002). It occurs on flat terrain as well as steep slopes (e.g. Osborne et al. 1991; Silva, Tavares & Pena 1996; Osborne & Sprent 2002; Williams et al. 2005; Gioria & Osborne 2009; Heenan et al. 2009). Water availability and temperature are the major factors limiting its distribution (Osborne 1989a,b; Osborne et al. 1991; Osborne & Sprent 2002). Globally, it is restricted to areas characterized by moderate temperatures, where precipitation is high and frosts are infrequent (Osborne 1989a,b; Wanntorp et al. 2001; Osborne & Sprent 2002). High humidity appears to become an important environmental factor in the colonization of areas with relatively low precipitation. The species grows under dry conditions at Howth Head, Co. Dublin (Doogue et al. 1998) and in Marin County, California (Howell 1970; Table 1), where humidity tends to be high due to coastal fog and mist.
Table 1. Climatic data from meteorological stations at or close to native and invasive populations of Gunnera tinctoria (adapted from Hickey 2002; based on data from Müller 1982). Koppen zone classification based on Rubel & Kottek (2010)
Height a.s.l. (m)
Mean annual temp. (°C)
Mean annual min. temp. (°C)
Absolute min. annual temp. (°C)
Mean annual relative humidity (%)
Mean annual precipitation (mm)
Stations for which data were added. Missing data were derived from WorldClim (2011).
In Britain and Ireland, it is naturalized in areas typically characterized by an annual rainfall > 1100 mm, with average winter temperatures ranging from 3 to 6 °C, and average summer temperatures between 12 and 15 °C. Its native range encompasses a temperate climatic zone with a predominantly Mediterranean climate and a tropical moist climate sub-zone (Williams et al. 2005), where annual rainfall is high (> 2000 mm; Molina 1978; Skewes, Rodríguez & Jaksic 2007; Table 1). In the Azores, average winter and summer temperatures are approximately 14 °C (min. 13 °C) and 21 °C (max. 30 °C), respectively, while annual precipitation averages 700 mm and annual average humidity is high (77%; Met. Office 2012). In New Zealand, it grows in areas characterized by either average annual temperatures > 410.7 °C and an average annual precipitation exceeding 1500 mm, or by an average annual temperature > 12.2 °C and an average annual precipitation of 1000–1500 mm (National Institute of Water & Atmospheric Research2011).
Gunnera tinctoria grows on a variety of soil types. In Britain and Ireland, it is typically found on sandy soils and relatively acidic, wet soils (Silva, Tavares & Pena 1996; Williams et al. 2005; Gioria & Osborne 2009; R. Reid, pers. comm.). On North Harris, Outer Hebrides, it occurs on peaty gley soils (R. Reid, pers. comm.). On Achill Island, Co. Mayo, G. tinctoria grows on shallow mineral gleys of marine origin overlying schist, and on minerotrophic peaty gleys (Hickey & Osborne 2001). In its native range, the species is typically found on alluvial and colluvial wet soils, mostly originating from volcanic ash where water availability is high (Williams et al. 2005). On São Miguel Island, it occurs in pumice, in gravel, in organic-rich soils and along the margins of, or rooted in, stream beds (Silva, Tavares & Pena 1996). In New Zealand, it has been recorded on soils with a large component of volcanic material and on a wide range of sedimentary rocks (Williams et al. 2005).
Soil pH measurements recorded in western Ireland, from a range of habitat types on Achill Island and the Curraun peninsula, Co. Mayo (drainage ditches, a stream bank, a quarry, a coastal grassland, pasture and a cliff face) ranged between 4.6 and 5.8 (Table 2). At two locations on Achill Island, pH during the growing season ranged between 4.9 and 6.2 (Hickey 2002). At five localities in Connemara, Co. Galway, Gunnera tinctoria occurs on soils of gravel or morainic origin with mean (±SD) pH values ranging from 6.3 ± 0.6 to 6.9 ± 0.2 (Skeffington & Hall 2011).
Table 2. Mean and range of concentrations of nutrient elements and pH in soil samples collected under stands of Gunnera tinctoria at eight habitat types/locations in Achill Island and the Curraun peninsula, Co. Mayo (n = 3). Inorganic ions were analysed using atomic absorption/emission spectrophotometry on acetic acid extracts, using air-dried soil. Nitrogen was determined using a micro-Kjeldahl technique. Soil pH was measured in a mixture of equal volumes of air-dried soil and distilled water, using a pH microelectrode (B. Osborne, unpubl. data)
K (mg 100 g−1)
Ca (mg 100 g−1)
Mg (mg 100 g−1)
Cliff face 1
0.06 – 012
Cliff face 2
Although adult plants are moderately tolerant of waterlogged soils (Campbell 1994), they are rarely found in areas where the entire rhizome is permanently under water. Gioria (2007) showed that young seedlings were unable to survive two consecutive days under waterlogged conditions. Adult plants grow under a wide range of soil moisture conditions. On Achill Island, average soil moisture content was 38% and 77% at a coastal grassland and a wet meadow community, respectively (Hickey 2002). For these sites, the organic matter content varied greatly (12–73%), while smaller differences in soil bulk density were reported (0.2–0.3 g cm−3 on average). Total soil N concentrations averaged 4.2 ± 1.5 mg g−1 and 14.1 ± 6.2 mg g−1. Soil concentrations averaged 12.3–43.5 N μg g−1, while averaged from 10.1 to 29.4 N μg g−1, indicating that ammonium is the somewhat more common form of available nitrogen in habitats occupied by this species (Hickey 2002). Nutrient concentrations and pH for soils collected from a range of localities and/or habitat types on Achill Island are presented in Table 2.
In British plant communities classified by Rodwell (1998a,b, 2000), G. tinctoria has become invasive in two woodland and scrub communities (Salix cinerea–Galium palustre woodland, W1; Rubus fruticosus–Holcus lanatus underscrub, W24), one mire community (Juncus effusus–Galium palustre rush pasture, M23), two mesotrophic grassland communities (Holcus lanatus–Juncus effusus rush pasture, MG10; Festuca rubra–Agrostis stolonifera–Potentilla anserina grassland, MG11), three sand dune communities (Carex arenaria–Festuca ovina–Agrostis capillaris dune grassland, SD12; Ammophila arenaria–Festuca rubra semi-fixed dune community, SD7; Festuca rubra–Galium verum fixed dune grassland, SD8), three maritime communities (Festuca rubra–Armeria maritima maritime grassland, MC8; Festuca rubra–Holcus lanatus maritime grassland, MC9; Festuca rubra–Plantago spp maritime grassland, MC10) and one calcifugous grassland community (Festuca ovina–Agrostis capillaris–Rumex acetosella grassland, U1) (Hickey & Osborne 2001; Gioria 2007; Gioria & Osborne 2009, 2010). In particular, the species is commonly associated with a range of ruderal or competitor-ruderal species (sensu Grime 1977), including Apium nodiflorum, Galium aparine, G. palustre, Persicaria maculosa, Stachys sylvatica and Urtica dioica (Gioria & Osborne 2010; Table 3). In coastal areas, it displays highly invasive behaviour on cliff faces dominated by Armeria maritima, Festuca rubra and Plantago species. Species such as Bellis perennis and Viola riviniana have been recorded on Gunnera rhizomes in the spring months (Gioria 2007; Gioria & Osborne 2009; Table 3).
Table 3. List of species in the soil seed bank and the vegetation in areas invaded by Gunnera tinctoria on Achill Island, Co. Mayo, Ireland (from Gioria 2007)
Taraxacum officinale agg.
Taraxacum officinale agg.
Taraxacum officinale agg.
Taraxacum officinale agg.
In its native range, G. tinctoria occurs in canopy gaps and in forests margins adjacent to wetlands (Molina 1978; Schick 1980; Figueroa 2003; A. Pauchard, pers. comm.). Among the members of these communities are Aristotelia chilensis (Molina) Stuntz (Elaeocarpaceae), Buddleja globosa Hope (Scrophulariaceae), Fuchsia magellanica Lam. (Onagraceae), Nothofagus dombeyi (Mirb.) Oerst. (Fagaceae), Laureliopsis philippiana (Looser) R.Schodde (Monimiaceae) and species belonging to the Myrtaceae, with an understorey vegetation dominated by rich moss, vine and epiphyte communities (Schick 1980; Figueroa 2003). Gunnera species, including G. tinctoria, are a prominent feature of the understorey of moist forest ‘Yungas’ in the Eastern cordillera and Andean valleys (Ortuño et al. 2011).
On São Miguel Island, G. tinctoria occurs in areas where other weedy or invasive species are present, including Clethra arborea (Clethraceae), Leycesteria formosa (Caprifoliaceae) and Rubus inermis (Rosaceae) (Silva, Tavares & Pena 1996). Pena (1995) reported that communities previously invaded by heather, Calluna vulgaris, and ginger, Hedychium gardnerianum (Zingiberaceae), were particularly susceptible to G. tinctoria invasions. In contrast, G. tinctoria did not seem to be invasive in pastures (Pena 1995; Silva, Tavares & Pena 1996; Silva & Smith 2004).
In coastal areas in the North Island of New Zealand, G. tinctoria has colonized communities characterized by a range of native species, including Coprosma repens A. Rich. (Rubiaceae), Cyperus ustulatus A. Rich (Cyperaceae), the evergreen Phormium tenax J.R. Forst. & G. Forst. (Xanthorrhoeaceae), Samolus repens (J.R. Forst. & G. Forst.) Pers. (Primulaceae) and Typha orientalis C. Presl. (Typhaceae), tree ferns such as Cyathea medullaris (G. Forst) Sw. (Cyatheaceae) and Dicksonia squarrosa (G. Forst) Sw. (Dicksoniaceae), and the endemic grass Cortaderia fulvida (Buchanan) Zotov (Poaceae) (Williams et al. 2005). It also occurs in disturbed areas, with other species alien to New Zealand including Calystegia sepium, Holcus lanatus, Plantago lanceolata, Ranunculus repens, Rubus fruticosus, Solanum nigrum and Sonchus oleraceus.
IV Response to biotic factors
Gunnera tinctoria is a strong competitor in habitats characterized by high humidity/high rainfall, particularly along water courses, on coastal cliffs and in wet meadows. Its competitive ability in these habitats arises from a range of traits including its large stature, perenniality, a capacity for fixing nitrogen, high relative growth rates and growth early in the season, dense leaf canopy and abundant litter. In Britain and Ireland, its distribution is also strongly associated with human activities. Its ruderal traits include the production of a large number of seeds and a tendency to become invasive in abandoned fields, where sheep and cattle have been removed, as well as on waste ground, along roadways and drainage ditches, and at sites associated with the construction of new houses and quarries (Osborne et al. 1991; Gioria 2007; Skeffington & Hall 2011; R. Reid, pers. comm.). On North Harris, G. tinctoria grows in native woodland plantings where soil has been exposed, grazing has been eliminated and where the presence of birds, which presumably use the small trees as perches, may facilitate the spreading of seeds (R. Reid, pers. comm.). It has also been recorded on peatland that had been reclaimed for agricultural purposes and subsequently abandoned (R. Reid, pers. comm.). A recent study (Gioria, Dieterich & Osborne 2011) reported on the displacement of long-term G. tinctoria stands by Japanese knotweed Fallopia japonica (Houtt.) Ronse Decraene at a coastal site near Achill Sound, Co. Mayo. On São Miguel Island, G. tinctoria coexists rarely with the Japanese cedar Cryptomeria japonica (L.f.) D. Don (Taxodiaceae), another major invasive species on the island (Pena 1995; Silva, Tavares & Pena 1996; Silva & Smith 2004).
V Response to environment
In the Outer Hebrides, Gunnera tinctoria has formed large stands of several hundred square metres, although in the majority of locations it is found as scattered individuals or forms sporadic linear populations along the sides of watercourses (R. Reid, pers. comm.). On Achill Island, G. tinctoria tends to colonize large areas or can form scattered clusters. Its density in 218 0.5 × 0.5 km grid cells (of a total of 729 surveyed on the island) was classified by Armstrong (2008) as 28.4% large colonized areas, 36% scattered clusters, 5.9% scattered plants, 26.6% single clusters and 2.8% single plants. In São Miguel, Silva, Tavares & Pena (1996) recorded c. 3000–18 000 panicles ha−1, from four locations.
(B) Performance in Various Habitats
In Britain and Ireland, Gunnera tinctoria occurs along stream and river banks, and lake and pond margins besides human-disturbed areas. On Achill Island, Armstrong (2008) recorded G. tinctoria at 1168 locations, with 47% records on largely unimproved agricultural land and 39% on peatland. There, a ‘density’ rank of habitats where the species is dominant was calculated: improved and semi-natural grassland (mean habitat rank, 3.4), waterways (2.2), disturbed ground (1.4), roadsides (1.3) and scrub (1.2) (Armstrong 2008). While invasive stands are recorded in agricultural land, only scattered clusters or scattered plants are found on peat bogs. The latter records are, however, somewhat misleading since these occurrences are almost exclusively associated with trackways or disturbed areas, with little colonization of pristine bog habitats.
(C) Effect of Frost, Drought, Etc
There is little evidence of adult mortality associated with frosts, although there are reports of plant leaves showing visual symptoms of damage associated with sub-zero temperatures during early growth (B. Osborne, unpubl. data). Within its range in Britain and Ireland, however, the number of days of ground frost is low, given its predominantly westerly distribution. Where frosts occur, the species is deciduous (Sykes 1969; Osborne et al. 1991; Chiu et al. 2005; Williams et al. 2005). Examination of G. tinctoria plants in Connemara, Co. Galway, in February 2010 (an exceptionally cold year in Ireland) showed signs of frost damage to the rhizomes, with the percentage of dead rhizomes progressively increasing away from the ocean (from 75% dead 20 km from the ocean to 20% at 1 km; Skeffington & Hall 2011). Anecdotal horticultural information from gardens in easterly locations in Ireland indicates that young leaves are susceptible to early frosts, while there is little evidence of leaf damage from late frosts.
Drought and waterlogging
Gunnera tinctoria responds strongly to variations in water supply and growth is severely constrained by reduced water availability. Hennessy (2009) showed that 4 days without water caused permanent wilting and desiccation of G. tinctoria leaves, with no sign of recovery of the outer edges of the leaves after rewatering. Experiments conducted in an experimental garden at the University College Dublin campus, Co. Dublin, Ireland, where precipitation is 30–100% lower than that experienced in the species’ invasive range (Scotland, western Ireland and Cornwall), showed significantly lower biomass production, and plants died after 4 years (Campbell 1994). Experiments with plants growing in standing water also elicit similar symptoms if the humidity is low (high vapour pressure deficit; B. Osborne, unpubl. data), suggesting inherent constraints on water uptake and/or transport in the root–shoot pathway (Osborne et al. 1991).
Gunnera tinctoria is naturalized on coastal cliffs in Britain and Ireland (Osborne et al. 1991; Gioria & Osborne 2009) and also grows in coastal locations in its native range (Darwin 1845; Molina 1978), on São Miguel Island (Silva, Tavares & Pena 1996) and in New Zealand (Williams et al. 2005; Heenan et al. 2009). Its invasiveness on coastal cliffs, down to the high tide mark, is indicative of tolerance to inundation with salt water and to exposure to salt spray, although individuals growing under those conditions are somewhat smaller (c. 1.5 m in height) than those found along watercourses or in wet meadows (c. 2 m), and they can show visible signs of leaf damage (Gioria 2007).
VI Structure and physiology
Gunnera tinctoria is polystelic, a peculiarity of a number of the larger species of Gunnera (Osborne & Bergman 2009), which have a number of separate vascular strands in the stems and petioles (Bergman, Johansson & Söderbäck 1992). The alternate leaves are orbicular, palmately lobed, with five to nine lobes and incise-serrate margins (Chittenden 1956; Cook 1968), ranging from 30 to 200 cm in diameter (Silva, Tavares & Pena 1996; Williams et al. 2005). New leaves on mature plants show a trichotomous mid-vein pattern with the right-hand (and more basal) branch beginning to outgrow the left-hand branch (Fuller & Hickey 2005), while leaves of young plants show palmate-pinnate venation, suggesting that the bifurcating mid-vein syndrome develops gradually during later organogenesis (Fuller & Hickey 2005). Over-wintering rhizomes develop mainly horizontally, largely above-ground and can be up to 3.5 m long (Hickey 2002). Measurements on naturalized populations in São Miguel showed that the diameter of the rhizome varies between c. 6 and 10 cm (Silva, Tavares & Pena 1996); for invasive populations in Ireland, the cross-section in mature plants was 20–25 cm. The stems and rhizomes are covered with a large number of conical bracts. These structures (also referred to in the literature as leaf-like ligules or stipules), which presumably protect the stems and rhizome tips (Mora-Osejo 1984), have been recently interpreted as cataphylls (highly reduced scale-like leaves), on the basis of phylogenetic analyses (Wanntorp, Wanntorp & Rutishauser 2003). In mature plants, all the roots are adventitious, with few root hairs, and contain large air spaces (aerenchyma). Most of the stomata are found on the lower leaf surfaces (c. 200 mm−2), with lower numbers on the upper surfaces (50 mm−2; Campbell 1994). Stomata are, however, found on other plant parts and are particularly frequent on unusual, chlorophyllous cup-shaped structures that occur on the rhizome (Osborne et al. 1991).
(B) Mycorrhiza and other Symbioses
Gunnera tinctoria forms significant symbioses with prokaryotes (cyanobacteria) and fungi. Gunnera roots from invasive populations in Ireland showed the presence of the characteristic vesicles and arbuscules of arbuscular mycorrhiza (B. Osborne, unpubl. data).
Gunnera is the only genus of angiosperm that is known to form an intracellular symbiosis with nitrogen-fixing cyanobacteria (Silvester & McNamara 1976; Osborne et al. 1991; Bergman, Johansson & Söderbäck 1992). For G. tinctoria, early studies identified the cyanobacterium involved as Nostoc punctiforme (Kutz.) Hariot (Silvester & Smith 1969; Bonnett & Silvester 1981; Osborne et al. 1991). This cyanobacterium is particularly common in the humid and wet habitats where G. tinctoria predominantly occurs (Osborne et al. 1991; Osborne & Sprent 2002). Nostoc infects the host plants via structurally unique gland-like structures (hereafter referred to simply as glands) located on the hypocotyl of all Gunnera species (Bergman, Johansson & Söderbäck 1992; Bergman & Osborne 2002; Osborne & Bergman 2009). The first sign of gland formation in G. tinctoria is apparent just after germination as a small mound of tissue on the hypocotyl, beneath an intact epidermal layer, which subsequently turns red due to the presence of anthocyanin pigmentation (Osborne & Bergman 2009). Subsequent growth of the gland results in the breakage of the surrounding epidermis through the growth of a central spike (see also VIII E). Colonization of the gland is only possible after the epidermis has been ruptured (Osborne & Bergman 2009). Cyanobacteria appear to move through channels in the gland prior to colonization of cortical tissue. Copious mucilage produced by the gland attracts cyanobacteria, and this may also be associated with developmental and metabolic changes as the cyanobacterium switches from an autotrophic to a heterotrophic existence (Osborne & Bergman 2009). Sugar analyses of this mucilage showed the presence of glucose, arabinose and galactose (Rasmussen et al. 1996; Khamar et al. 2010). Different sugars may have roles in controlling cyanobacterial differentiation and their ability to colonize host tissues, as well as maintaining the ability of established colonies to assimilate atmospheric nitrogen (Khamar et al. 2010).
Current interpretation of these glands suggests that they may have more affinity with adventitious roots than true plant glands (Osborne & Bergman 2009). Within the cortical tissue, cyanobacteria are characterized by a high heterocyst frequency and fix atmospheric nitrogen that is subsequently assimilated by the host tissue. In return, carbon, principally as malate, is provided by the host plant, and this supplies the C skeletons and ATP required for N2 fixation (Black, Parsons & Osborne 2002; Black & Osborne 2004). The cyanobacteria form distinct, slow-growing colonies within the cortical tissue that are clearly visible to the naked eye. The formation of clearly delineated and separate colonies indicates a restriction in their ability to colonize all host cells. Although changes in the structure of host tissues are rather small, unlike many other N2-fixing symbioses, they include an increase in plasma membrane production (Bruce 1997). Like other large species of the genus, G. tinctoria possesses clusters of red stipules that surround the Nostoc-colonized glands, the majority of which are located close to the base of each leaf petiole in mature plants. It has been suggested that the stipules aid cyanobacterial transfer to the colonized sites (glands) in mature plants (Benson & Margulis 2002). Chapman & Margulis (1998) have also suggested that the formation of these glands is a product of co-evolution between Gunnera and Nostoc. According to Chiu et al. (2005), the position of the glands is predetermined, whereas gland development is specifically induced in response to nitrogen deprivation. Osborne et al. (1992) made estimates of the allocation of biomass between the partners in the symbiosis that indicated that a cyanobiont biomass of c. 1% is sufficient to meet all the requirements of the host. However, a significant proportion of the cells of Nostoc colonies in a number of Gunnera species, including G. tinctoria, may be degenerate (Bonnett 1990), with little or no capacity for N2 fixation (Söderbäck, Lindbald & Bergman 1990), suggesting that a relatively small number of colonies can supply all of the nitrogen requirements of the host. The genetic diversity of Nostoc microsymbionts for G. tinctoria is discussed by Nilsson, Bergman & Rasmussen (2000) and Guevara, Armesto & Caru (2002).
(C) Perennation: Reproduction
Gunnera tinctoria is a polycarpic, perennial hemicryptophyte. In Britain and Ireland, the plant is deciduous, with leaves dying off completely, beginning in October. Most leaf tissues and the associated petioles decompose prior to the initiation of new growth in early spring, although the presence of undecomposed leaves and petioles has been observed as late as June. In New Zealand, the species is deciduous in the colder eastern and southern areas of South Island, losing its leaves with the first frosts, while it is incompletely deciduous or evergreen in the warmer parts of North Island (Sykes 1969). Under heated glasshouse conditions in Ireland, adult individuals retained most of their leaves throughout the year, although little growth occurred during the winter period, confirming that in regions with warmer winters the plant is unlikely to be deciduous (B. Osborne, unpubl. data).
Large, overwintering, apical buds may grow up to 25 cm when growth is resumed, and they are covered with cataphylls (Mora-Osejo 1984; see 'Morphology'). The plant reproduces sexually, by the production of a large number of seeds, and propagates by vegetative means. Hickey (2002) recorded a mean annual increase in rhizome length of 15 cm year−1 (range 2–24, n =4) cm in a wet meadow population in Achill Island, Co. Mayo. In situ measurements showed that rhizome biomass represented c. 59–93% of total biomass (range of mean ± SD rhizome biomass: 1.9 ± 0.5–4.6 ± 0.3 kg m−2, n =4; Hickey 2002). Similar values were reported for plants grown in an experimental garden at different levels of water availability (58–84% of total biomass; Table 4).
Table 4. Mean (±SD) allocation of biomass and leaf characteristics (LAQ, leaf area quotient; SLA, specific leaf area) of Gunnera tinctoria plants (n = 3) grown in an experimental garden in Dublin, Ireland, between 1988 and 1990. Plants were subjected to two treatments: normal (Dublin) rainfall, c. 1000 mm year−1; high rainfall, c. 1800 mm year−1 (from Campbell 1994)
2.11 ± 0.34
0.85 ± 0.19
2.60 ± 0.46
0.20 ± 0.01
2.65 ± 0.37
0.13 ± 0.02
0.84 ± 0.08
0.34 ± 0.05
1.10 ± 0.18
0.03 ± 0.01
0.83 ± 0.19
0.05 ± 0.02
0.84 ± 0.09
0.57 ± 0.09
0.90 ± 0.17
0.05 ± 0.01
0.75 ± 0.08
0.07 ± 0.02
0.27 ± 0.01
0.22 ± 0.01
0.45 ± 0.02
0.23 ± 0.19
0.26 ± 0.19
0.02 ± 0.02
8.30 ± 1.75
2.00 ± 0.24
0.90 ± 0.17
0.17 ± 0.01
Leaf area (m2)
0.8 ± 0.03
0.1 ± 0.01
0.8 ± 0.01
0.06 ± 0.01
34.3 ± 9.0
0.17 ± 0.03
33.6 ± 6.2
13.8 ± 4.6
Leaf thickness (mm)
335.6 ± 0.2
331.7 ± 0.5
328.6 ± 0.8
325.1 ± 1.2
LAQ (m2 kg−1)
0.05 ± 0.01
0.05 ± 0.02
0.05 ± 0.01
0.04 ± 0.01
0.04 ± 0.01
0.03 ± 0.01
SLA (kg m−1)
19.3 ± 0.5
20.2 ± 1.2
18.1 ± 0.7
22.4 ± 0.7
26.7 ± 0.4
29.2 ± 4.5
Mature plants generally flower after approximately 5 years, although this is dependent upon the environmental conditions (Osborne et al. 1991). Comparison of the genetic structure of invasive populations from Ireland, New Zealand and the Azores with native populations from Chile indicated that the majority of invasive populations were genetically distinct from native populations (Fennell, Gallagher & Osborne 2010). In particular, molecular analyses showed that genetic differentiation from a putative founding population in Ireland had occurred relatively rapidly and within a small geographic area, providing evidence for significant sexual reproduction (Fennell, Gallagher & Osborne 2010). However, the putative founding populations were more closely related to the native populations examined than to recent invasive populations. Once established at a site, vegetative spread and dispersal of plant fragments appear to be the most important factors for the persistence of this species (Gioria & Osborne 2009).
The chromosome number for Gunnera tinctoria is 2n = 34 (Dawson 1983). Dawson (1983) suggested that the difference between his count and an old chromosome count of 2n = 24 by Darlington & Wylie (1955) may result from interpretation of pairs of adjacent subterminal or terminal chromosomes as single bi-armed chromosomes. Dawson (1983) reported a subterminal chromosome with a secondary constriction in its long arm proximal to the centromere in G. tinctoria and another giant species of the subgenus Panke, G. manicata. Examination of material from Achill Island also indicates that 2n = 34.
(E) Physiological Data
Osborne (1989a) reported an optimum temperature of 25 °C for the maximum photosynthetic rate (Pm) in G. tinctoria, with depressions in Pm and the quantum yield for O2 evolution below this temperature that were typical of C3 photosynthesis.
Established plants have a considerable capacity for biomass production when precipitation exceeds 1000 dm3 m−2 year−1 (mm year−1) (Campbell 1994). Over a 4-year period (1988–91), Campbell (1994) investigated the effect of two levels of water supply on above- and below-ground biomass of plants that had been transplanted from Achill Island into an experimental garden in Dublin (n =3 plants): normal (Dublin) rainfall, 1050 ± 29 dm3 m−2 year−1, and augmented rainfall (similar to western Ireland), 1842 ± 41 dm3 m−2 year−1. The total above-ground biomass for the higher water treatment was six times higher than in the lower one, while total plant biomass was 11.5 times greater (Campbell 1994). In particular, plants grown under the higher rainfall treatment produced > 2 g shoot dry mass dm−3 of water received, while those grown with less rainfall had a lower water-use efficiency (c. 0.6 g dm−3; Campbell 1994). Plants grown under higher rainfall gained 8–13 times more carbon than those grown under lower rainfall. The length of the rhizome increased by 45% for the higher rainfall treatment, compared with no increase with the lower rainfall, and leaf area was eight times higher. A 2.75-fold increase in seed biomass, a 4-fold increase in inflorescence biomass, an 8-fold increase in leaf biomass and a 7.5-fold increase in petiole biomass were all associated with the increased rainfall. These results suggest that the competitive ability of this species may only be realized in wet and/or humid habitats (Campbell 1994).
Drying experiments showed that slow wilting of leaf discs did not inhibit photosynthesis until the relative water content was lower than 20–40%, confirming earlier indications of an inherent tolerance to dehydration of photosynthesis in G. tinctoria (Osborne 1988; Osborne et al. 1991; Campbell & Osborne 1993). Such a tolerance to dehydration was confirmed by the similar chlorophyll, carbon and nitrogen contents at a range of tissue water contents (Campbell 1994). Similarly, comparable carboxylation efficiencies and maximum CO2-saturated rates in the regular and high rainfall treatments in the experimental garden are indicative of a minimal effect of water availability on the activity of Rubisco or electron transport capacity. Relative water contents of the leaves were also similar in the different water availability treatments (Campbell 1994). What seems to be more important is the sensitivity of leaf growth to water deficits, with leaf expansion severely reduced even at relatively moderate water deficits.
The highest ex situ photosynthesis rates, based on O2 evolution measurements, were in May, when the Pm for plants in both watering treatments was not light saturated, although light saturation was recorded later in the year; this indicates a high photosynthetic capacity early in the growing season that was relatively independent of the watering regime (Campbell & Osborne 1993; Campbell 1994).
Diurnal measurements of CO2 exchange showed peak net photosynthesis in the early morning (generally before 10 am) coupled with a high leaf conductance that was independent of the watering treatment (Osborne 1988; Campbell & Osborne 1993; Campbell 1994). Diurnal decreases in these variables under adequate water supply suggest that water transport is restricted somewhere in the root–shoot pathway (Campbell & Osborne 1993). At the whole-plant level, growth and photosynthetic responses to irradiance were similar to values reported in the literature for C3 plants, either in the presence or absence of the symbiotic cyanobacterium (Osborne 1989b), indicating that these traits are specifically host related.
Osborne (1988, 1989a) found a weak correlation between stomatal functioning and rates of CO2 exchange in G. tinctoria, suggesting a low water-use efficiency. In general, the stomata of G. tinctoria remain open in the dark and show little response to irradiance or intercellular CO2 concentrations at low or medium humidity levels (< 70–80%; Campbell 1994). An inability closely to regulate CO2 uptake and water loss may also contribute to the preference of this plant to wet habitats.
Tissue carbon contents tend to be similar for different parts of the plant (c. 36–38% dry mass, n =4; Hickey 2002). Data on C, N, P and other nutrients, for different parts of G. tinctoria plants, are presented in Tables 5, 6 and 7. The C : N ratio varies seasonally. In December, the C : N ratio in the petioles (30 : 1), stipules (leaf litter biomass; 26 : 1) and in the inflorescence (27 : 1) was higher than in the rhizome (21 : 1) and in the litter (current year's growth; 20 : 1). Higher C : N values were found for the rhizomes (45 : 1) and for litter (47 : 1) in February, suggesting that some N is also lost from the rhizomes as well as the expected loss of N from the litter during decomposition processes occurring over winter.
Table 5. Mean (±SD) concentrations of total carbon and nitrogen in different parts of Gunnera tinctoria plants (n = 4) collected at one site in Achill Island, Co. Mayo, Ireland (adapted from Hickey 2002)
Concentration (% dry mass)
21 December 1997
19 February 1998
20 May 1998
39.1 ± 0.96
1.9 ± 0.19
36.7 ± 0.16
0.8 ± 0.21
37.6 ± 0.64
1.3 ± 0.34
35.2 ± 1.24
1.2 ± 0.10
34.1 ± 1.73
0.7 ± 0.05
36.1 ± 2.19
1.6 ± 0.17
41.5 ± 0.52
3.2 ± 0.20
39.2 ± 0.66
1.4 ± 0.03
41.0 ± 0.88
2.1 ± 0.23
Leaf litter biomass
35.6 ± 0.23
1.4 ± 0.13
34.7 ± 0.82
1.4 ± 0.10
33.7 ± 0.71
1.8 ± 0.09
40.7 ± 0.68
2.1 ± 0.38
39.7 ± 1.22
1.1 ± 0.59
38.7 ± 1.14
2.0 ± 0.46
Table 6. Mean ± SD) concentrations of nitrogen and phosphorus (% dry mass) in different parts of Gunnera tinctoria plants (n = 4) sampled over a 2-year period (1997–98) by Hickey (2002) at two sites on Achill Island, Co. Mayo, Ireland
1.39 ± 0.42
1.33 ± 0.48
0.18 ± 0.11
0.18 ± 0.09
1.04 ± 0.62
1.60 ± 0.60
0.13 ± 0.08
0.11 ± 0.07
2.58 ± 0.61
2.49 ± 0.54
0.22 ± 0.12
0.22 ± 0.13
2.02 ± 0.48
1.81 ± 0.30
0.28 ± 0.09
0.27 ± 0.07
Leaf litter biomass
1.70 ± 0.33
1.55 ± 0.39
0.17 ± 0.09
0.16 ± 0.09
2.21 ± 0.86
2.10 ± 0.84
0.11 ± 0.06
0.12 ± 0.10
Table 7. Mean and range of concentrations of nutrient elements (% dry mass) in adult individuals of Gunnera tinctoria at eight habitat types/locations in Achill Island and the Curraun peninsula, Co. Mayo (n = 3). Metals were analysed using atomic absorption/emission spectrophotometry on concentrated HNO3/H2SO4 extracts of oven-dried samples. Nitrogen was determined using a micro-Kjeldahl technique (B. Osborne, unpubl. data)
Cliff face 1
Cliff face 2
Similar productivities, nitrogen concentrations and photosynthetic rates were reported for plants grown in the presence or absence of nitrogen (as nitrate or ammonium) for several weeks under controlled conditions (Osborne et al. 1992). Minimal differences in biomass production were found between plants supplied with nitrate and those supplied with nitrate-free media or ammonium, although there was evidence for a decrease in leaf and root + rhizome dry mass at high ammonium concentrations (10 mm). Little variation in leaf area with different nitrogen additions was also reported, although a decrease in leaf area was found at the highest ammonium concentrations used (Osborne et al. 1992). Differences in plant nitrogen concentration in these experiments were also minimal. Osborne et al. (1992) did not find any increase in nitrate reductase activity in any part of the intact plants in response to nitrate supply. Similarly, increases in nitrate reductase activity associated with shoot tissues incubated in nitrate solutions ex situ were small, with the exception of those recorded with the stipules and the inflorescence branches. Nitrate reductase activity was, however, inducible in situ in young leaves 24 h after spraying with 10 mm nitrate, or in leaf discs floated on nitrate solutions (Osborne et al. 1992) indicating a limitation in nitrate uptake and/or transport in intact plants. There was also little difference in the rates of photosynthetic O2 exchange between plants receiving no nitrogen and those supplied with 1 or 10 mm nitrate or ammonium. The highest rates of photosynthetic O2 exchange were associated with plants supplied with 1 mm, while the lowest values were found for plants receiving 1 mm. These results are somewhat surprising given the association between the distribution of G. tinctoria and soils in which ammonium is likely to be the more prevalent form of available nitrogen.
Experiments in open-top chambers conducted over 4 weeks (Hennessy 2009) and aimed at testing the effect of elevated CO2 (680 μmol mol−1) on the species’ physiology showed a positive response of Pm. However, in longer-term experiments (105 days), the effect of CO2 enrichment on photosynthesis was not significant, although biomass production increased by 22%. In even longer-term (12 month) experiments, exposure to elevated CO2 did not significantly affect either photosynthesis or biomass production (Hennessy 2009). Significant correlations between Pm and water-use efficiency (r2 = 0.703; P <0.05, n = 4) and between Pm and intercellular CO2 concentration (Ci; r2 = 0.942; P <0.05, n = 4) were found in this experiment (Hennessy 2009). Ci was relatively unchanged with alterations in stomatal conductance (Gs) so that no correlation was found between Gs and Pm (Hennessy 2009). This supports previous evidence for poor stomatal regulation both in response to water availability and CO2 concentration (Osborne 1988).
Elevated CO2 concentrations altered biomass partitioning, with an increased allocation to roots and rhizomes (Hennessy 2009). Biomass partitioning was modified after exposure for 105 days, with a 23% reduction in leaf biomass, which was compensated for by a 23% increase in the rhizome and root system. Hennessy did not report any change in leaf area ratio (LAR) or leaf mass ratio (LMR) under elevated CO2, and leaf N and P concentrations were not significantly different between ambient and elevated CO2. Overall, the photosynthetic responses of G. tinctoria to elevated CO2 were somewhat lower than might been expected for a nitrogen-fixing C3 species, although there was no evidence that this was due to reduced leaf N or P concentrations. The effects of elevated CO2 on dark respiration were variable depending on the length of exposure (Hennessy 2009). Exposure to elevated CO2 concentrations over 12 months did not affect Rubisco activity, nitrogen-use efficiency or phosphorus-use efficiency. Stomatal density generally decreased initially at elevated CO2, although the differences between the two treatments were not significant, suggesting significant differences in cell/leaf expansion. Overall, these results indicate that G. tinctoria exhibits a rather smaller response to elevated carbon dioxide concentrations in comparison with many C3 species.
(F) Biochemical Data
The rhizomes have c. 9.3% tannin content (Hegnauer 1966). The anthocyanin flavonoids that are present in the mucilage-secreting glands of G. manicata (Chiu et al. 2005) have moderate anti-oxidant properties (de Cassia Mariotti 2010), which would also be expected for G. tinctoria. Dried entire plants contain steroids (daucosterol), uvaol and erythrodiol, and loliolide (Barboza et al. 2009). Interestingly, the production of the cyanobacterial toxin β-N-Methyl amino-l-alanine (BMAA) has been reported to increase in symbiotic tissues (Cox et al. 2005). This suggests that Gunnera invasions may have significant implications on human health, since BMAA has been indicated as a possible cause of myotrophic lateral sclerosis/Parkinsonism–dementia complex (Cox et al. 2005). BMAA may also act as a signalling molecule (Brenner et al. 2000), and this also warrants further investigation. de Medeiros et al. (2000) showed a high anti-thrombin activity of dichloromethane and methanol extracts from G. tinctoria (78% and 67%, respectively).
In Britain and Ireland, plants start to grow in early spring (March), usually before the native vegetation, with leaves emerging from an apical bud by April, and growth is completed by August/September. Most leaves emerge together with the inflorescences, and only a few leaves are formed in the following months (Campbell 1994). Seed germination begins in March and peaks in May, although only a small number of seeds produced annually emerge in the field (Hickey & Osborne 2001; Gioria 2007; Gioria & Osborne 2009). Senescence of the shoot begins in October (Hickey & Osborne 2001; Gioria 2007; Armstrong 2008).
In Britain, Ireland and the Azores, flowers and fruits appear from April to October (Silva, Tavares & Pena 1996; Gioria 2007), while in South America they are found from October to February (Silva, Tavares & Pena 1996). Silva, Tavares & Pena (1996) described the phenology and stand structure of G. tinctoria on São Miguel Island, where panicles begin to develop in February, but without differentiated flowers, while bisexual flowers develop in March. There, pollination occurs in April, while in May, anthers fall, styles desiccate and the endocarp is differentiated. Between July and October, flowers with anthers or styles decrease in number and the drupaceous fruits increase in size and turn from green to red.
VIII Floral and seed characters
(A) Floral Biology
Gunnera tinctoria is gynomonoecious, having both bisexual and female flowers (Webb, Sykes & Garnock-Jones 1988; Pena 1995; González & Bello 2009). Flowers are distributed along two to four large compound inflorescences (Williams et al. 2005; Gioria 2007; Gioria & Osborne 2009; Skeffington & Hall 2011). González & Bello (2009) observed gradual structural changes along the inflorescences between the terminal, bisexual flowers with two pairs of perianth segments, and the female flowers with a single perianth organ. Most sub-terminal and mid-level flowers are also bisexual. Individual flowers are sessile, with minute sepals, and only c. 1 mm long. Style length is slightly less than the ovary. The drupes are reddish, oblong, each containing a single ovoid and flanged seed (1.2 × 1–1.5 mm), weighing c. 4 mg (Cook 1968; Williams et al. 2005). The seed embryo is small, strongly curved and embedded in a copious endosperm made of oil, starch and an aleurone with crystalloids (Wilkinson & Wanntorp 2007) that is well differentiated (Molina 1978). The species is mostly or exclusively wind pollinated (Carlquist 1974; Wanntorp et al. 2001; Wilkinson & Wanntorp 2007), although Williams et al. (2005) reported that individual flowers can be insect pollinated, probably by bees (Hymenoptera).
González & Bello (2009) provided a detailed description of the floral biology of G. tinctoria (referred to as G. pilosa) and scanning electron micrographs of inflorescence and floral development. They found that the first-initiated proximal floral apices show a reduced set of floral organs compared to the terminal, more complete, flower (see also Wanntorp & Ronse De Craene 2005; Ronse De Craene & Wanntorp 2006). While petals are well developed in bisexual flowers, they are highly reduced or lacking in female flowers. Flowers of the subgenus Panke are highly reduced, which is presumably related to wind pollination and unisexuality. They show acropetal initiation of the floral apices and basipetal development of flowers. The order of reduction in flowers in Panke is basipetal (petals > stamens > sepals > carpels), with a gradual change from dissymmetric, bisexual flowers to monosymmetric, female flowers (González & Bello 2009).
The presence of floral hydathodes in Gunnera was suggested by Ronse De Craene & Wanntorp (2006) and confirmed by González & Bello (2009), who reported that the hydathodes on the sepals are structurally similar to those previously described in the leaves of Gunnera species (Wilkinson 2000; Wilkinson & Wanntorp 2007). These authors observed that several Nostoc-like filaments are found near the guard cells of these hydathodes. They suggest that young sepals could become infected during early phases of the development of the inflorescences, allowing Nostoc to disperse throughout the fruit. This is an additional or alternative route for the infection of seedlings by Nostoc, rather than via the well-documented stem-gland pathway (Osborne & Bergman 2009), but it requires further investigation.
The ovule is anatropous (completely inverted), bitegmic (possessing two integuments) and crassinucellate (with one or more layers of cells outside the embryo sac but distinct from the epidermis), and the micropyle is only formed by the inner integument (Wilkinson & Wanntorp 2007).
Gunnera pollen is highly distinctive, tricolpate, suboblate and spheroidal (see Jarzen 1980; Wanntorp, Praglowski & Grafström 2004; Fig. 3). The pollen grains of G. tinctoria are two-celled at anthesis (Wilkinson & Wanntorp 2007). Gunnera tinctoria possesses a plesiomorphic type of pollen (type 2; Wanntorp, Praglowski & Grafström 2004), which is characterized by round and equidimensional lumina encircled by muri that are generally thicker than the diameter of the lumina (Wanntorp, Dettmann & Jarzen 2004).
There are no records of G. tinctoria forming viable hybrids. Palkovic (1978) reported the occurrence of hybrids between two large Gunnera species, G. insignis and G. talamancana, in Costa Rica, indicating a potential for hybridization between the larger members of this genus.
(C) Seed Production and Dispersal
Gunnera tinctoria produces a large number of seeds, although estimates vary substantially. Osborne et al. (1991) estimated c. 750 000 seeds per plant in an Irish population, while Williams et al. (2005) estimated up to 250 000 fruits per plant, based on seed collected from 1-m-long inflorescences on the coast west of Mount Taranaki (c. 12 000–83 000 per inflorescence). The mean number of seeds per panicle estimated by Silva, Tavares & Pena (1996) at four locations in São Miguel Island was c. 30 000–130 000. Clearly, the actual numbers could vary considerably, depending on plant and inflorescence size, stage of development and environmental conditions, particularly water availability (see 'Physiological Data').
In Britain and Ireland, seeds are typically formed by June, and ripen between July and October. Seed maturation extends to late October, when dispersal begins, although the majority of seeds are retained on inflorescences attached to the parent plant for an extended period of time, after which most seeds fall in close proximity to the parent plant (Gioria 2007; Gioria & Osborne 2009).
Long-distance dispersal is believed to be predominantly hydrochorous, occurring naturally along rivers, stream and ditches, and zoochorous (Silva, Tavares & Pena 1996; Hickey 2002; Figueroa 2003; Williams et al. 2005; Gioria 2007). On the Outer Hebrides, birds are thought to be responsible for most of the seed dispersal in autumn (see 'Animal Feeders and Parasites'). Birds have also been found to spread seeds on Achill Island (Hickey 2002), São Miguel Island (Silva, Tavares & Pena 1996) and in New Zealand (Williams et al. 2005; 'Animal Feeders and Parasites'). Dispersal by farm animals is a possibility. Vehicular traffic and disposal of contaminated soil play a major role in propagule dispersal for this species. Skeffington & Hall (2011) reported its presence in the Connemara National Park, Co. Galway, probably as a consequence of the use of gravel in the park originating from the nearby Guy's quarry. They also reported the spread of this species along roadways constructed using contaminated material from this and other quarries.
The formation of a soil seed bank can be viewed as a form of dispersal in time and may contribute to the invasiveness of this species (Gioria, Pyšek & Moravcová 2012). A large study of the seed bank of this species at three invaded sites in Ireland, in May (after the germination of the majority of seeds in the field) and October (soon after seed dispersal), showed a capacity for G. tinctoria to form a persistent seed bank (sensu Thompson, Bakker & Bekker 1997), of the order of hundreds of thousands of germinated seedlings m−2 (Gioria 2007; Gioria & Osborne 2009, 2010). The largest seed bank was recorded from a wet meadow (mean ± SD: 49 427 ± 2767 and 76 191 ± 8771 m−2 in May and October, respectively), while the smallest was found along a coastal cliff (10 084 ± 1378 and 11 255 ± 2428 m−2, in May and October, respectively). The persistent character of the G. tinctoria seed bank can be inferred from the proportion of seeds that germinated from deep soil layers, with 20% of seedlings emerging from the 5- to 10-cm soil layer and 10% from the deepest soil layer (10–15 cm), and by the large number of seedlings emerging from samples collected in May. Williams et al. (2005) suggested that this species only formed a minimal seed bank in New Zealand due to the high percentage germination observed in situ. They reported that in the Pukeiti Rhododendron Trust gardens, on the lower slopes of Mount Taranaki, no seedlings emerged after 2 years of removal of the flower heads to prevent seeds being produced. These conclusions were, however, not based on direct estimates of the size and type of seed bank. Climatic differences between Ireland and New Zealand might also contribute to these differences (Gioria & Osborne 2009).
(D) Viability of Seeds: Germination
Seeds are non-dormant at maturity, or have a short-term, non-deep simple morpho-physiological dormancy (sensu Baskin & Baskin 1998) and lack any cold-stratification requirement for embryo growth (Gioria 2007). Gioria (2007) showed that the majority of seeds retain their viability and germination potential for a long period of time. Under comparable light and temperature conditions (20/10 °C, 12:12 h), 70% and 62% of 6- and 12-month-old seeds, respectively, germinated over a 4-week period, compared to 100% germination for fresh seeds (Table 8). Those seeds that did not germinate over this period were no longer viable.
Table 8. Mean (±SD) germination (%) of fresh, 6- and 12-month-old seeds, and stratified seeds of Gunnera tinctoria, after 4 weeks of incubation at alternating temperatures (20/10 °C; 12 h:12 h photoperiod), at three light fluxes (n = 4, 50 seeds per replicate). Differences in the effects of seed age, stratification and light on cotyledon germination percentage were tested using ANOVA and Student-Newman-Keuls multiple comparison tests. Values followed by the same capital letter indicate that the effects of seed age or stratification on germination percentage (radicle + cotyledon emergence) were not significant. Values followed by the same small letter indicate that the effect of light on the percentage of germination was not significant (from Gioria 2007)
In situ germination within invasive populations in Ireland is low (c. 100–160 seedlings m−2; Hickey & Osborne 2001). Gioria (2007) observed the emergence of G. tinctoria seedlings from coastal grassland communities, beneath the plant canopy or on rhizomes, but none from wet meadows and peatland. Germination is asynchronous and occurs throughout the year, excepting the winter months, with a peak of germination in March (Hickey & Osborne 2001). A similar pattern was reported for seeds grown under unheated glasshouse conditions (Gioria & Osborne 2009).
Response to temperature
Temperature plays a major role in regulating seed dormancy, with evidence indicating a germination requirement for alternating temperatures (and light), with fresh seeds entering dormancy if those conditions are not met. Gioria (2007) showed that virtually all fresh seeds collected from an invasive population in Ireland (four replicates of 50 seeds) germinated when incubated at 20/10 °C (12:12 h, light flux density of 115 μmol m−2 s−1). Similar results were found in New Zealand for seeds grown at ambient temperature (Williams et al. 2005). A lower germination percentage was reported by Figueroa (2003) for seeds collected from a temperate rain forest in southern Chile, where only 77% of fresh seeds germinated. No germination was recorded at constant temperatures of 15 °C or 10 °C, resembling the conditions experienced by seeds buried under the top soil layer at many sites where this species occurs in Britain and Ireland (Gioria 2007; Table 8). This could be indicative of the possibility of a minimum temperature requirement for germination, although more investigations are needed.
These germination responses to temperature are consistent with observations in the field of seedlings on sandy soils but not in wet meadows or peaty soils (Gioria 2007). The soil temperature of sandy soil is typically higher than that within wet meadows, and may be subject to large variations at more exposed and drier sites, such as coastal cliffs. This could also explain the substantially lower size of the seed bank of coastal areas compared to that of a wet meadow on Achill Island (Gioria & Osborne 2009, 2010).
Response to light
Light plays a major role in determining seed germination percentage for this species. Gioria (2007) found that all fresh seeds germinated when incubated at a light flux density of 115 μmol m−2 s−1 (20/10 °C; 12:12 h; four replicates of 50 seeds), while only 70% of seeds emerged at 70 μmol m−2 s−1. A similar pattern was identified for fresh seeds, 6-month-old seeds, 12-month-old seeds and moist-cold-stratified seeds (i.e. seeds incubated at 5 °C in darkness for 6 months; Table 8). In this study, radicles only emerged from c. 30% of seeds incubated in darkness at 20/10 °C (12:12 h). These were etiolated and showed no sign of ‘gland’ formation (Fig. 4). A light requirement for seed germination was also reported by Figueroa (2003), who showed that no seedlings emerged when sown in the understorey of a secondary forest (non-gap treatment), while the majority of seeds germinated in a canopy gap treatment. In New Zealand, Williams et al. (2005) reported a high germination percentage (100% over 30 days) in the light at ambient temperature for seeds collected in Taranaki.
Response to salinity
Gioria (2007) tested the effect of two NaCl concentrations similar to those that are experienced at coastal sites (200 and 400 mmol L−1) on seed germination. The addition of NaCl completely inhibited germination, independently of the combination of light and temperature conditions under which the seeds were incubated and/or the application of GA3. Nevertheless, seeds retained their viability and c. 90% of viable seeds germinated after being transferred to distilled water and incubated at 20/10 °C (12:12 h; four replicates of 50 seeds), indicating that salinity tends to inhibit germination osmotically. Thus, seeds are likely to be more susceptible to salinity than whole plants, a situation that is not uncommon even for halophytic plant species (Waisel 1972).
(E) Seedling Morphology
Germination is epigeal; the first true leaf arises from the axis between the cotyledons, after c.14 days. The primary root axis does not develop significantly, and subsequently, adventitious roots are initiated in close proximity to the stem glands (Osborne & Bergman 2009; Fig. 5). Seedling morphology is strongly dependent upon the light conditions. Average (±SD) seedling length was 4.5 ± 0.3 mm in four replicates of 50 seeds for seedlings incubated under 115 μmol m−2 s−1, 1.1 ± 0.1 mm for those incubated under 60 μmol m−2 s−1 and 0.8 ± 0.09 mm for those grown in darkness (Gioria 2007).
IX Herbivory and disease
(A) Animal Feeders and Parasites
The fox moth Macrothylacia rubi L. (Lepidoptera: Lasiocampidae) was observed feeding on leaves of G. tinctoria in Connemara, Co. Galway, and Achill Island (Skeffington & Hall 2011). Larvae of the angle shades moth, Phlogophora meticulosa (L.) (Lepidoptera: Noctuidae), were also observed on G. tinctoria leaves (I.-R. Barrs, pers. obs.). In Chile, the moth Eupithecia horismoides Rindge (Lepidoptera: Geometridae) perforates the petioles of G. tinctoria, and its life cycle is strongly dependent upon this plant (Ibarra-Vidal & Parra 1993). Mealybugs (Hemiptera: Pseudococcidae) were recorded on plants grown in glasshouses (B. Osborne, pers. obs.). In São Miguel, insects of the order Thysanoptera and polyphagous Lepidopteran larvae (caterpillars) were found on leaves and may damage young plants, while no insects were observed feeding on flowers or fruits, and no feeding damage was found on rhizomes and roots (Silva, Tavares & Pena 1996). A range of invertebrates, including those belonging to the Mollusca, Annelidae, Chilopoda and Blattodea, were found on the rhizomes of G. tinctoria (Silva, Tavares & Pena 1996).
In the Outer Hebrides, the starling Sturnus vulgaris L., the redwing Turdus iliacus L. and the song thrush, Turdus philomelos Brehm, have been reported feeding on and spreading fleshy fruits (R. Reid, pers. comm.). On Achill Island, the dunnock Prunella modularis (L.) and the crow Corvus corone L. were also found feeding on the fruits (Hickey 2002). Williams et al. (2005) recovered G. tinctoria seeds from dry bird faeces, probably of blackbirds, Turdus merula L., at Oeo in Egmont Ecological District, New Zealand. Silva, Tavares & Pena (1996) found seeds of G. tinctoria in faecal pellets of the chaffinch Fringilla coelebs L. and Turdus merula on São Miguel Island.
In the Outer Hebrides, cattle and sheep appear to damage seedlings and scattered adult plants (R. Reid, pers. comm.). In Ireland, young individuals may be subject to some grazing by sheep during early spring in coastal grasslands, although there is no evidence of leaf damage by sheep on adult individuals during the growing season (Gioria 2007). Evidence of rodents chewing inflorescences was provided by Skeffington & Hall (2011) in Ireland. On São Miguel Island, Silva, Tavares & Pena (1996) reported damage to leaves by rabbits and goats. In Chile, the rhizomes of G. tinctoria represent one of the main sources of food for introduced wild boar Sus scrofa L. (Skewes, Rodríguez & Jaksic 2007). There, seeds of G. tinctoria are eaten by rodents, in particular by Oryzomys longicaudatus philippi (Landbeck) (Rodentia: Muridae) and Akodon longipilis hirtus Thomas (Rodentia: Cricetidae; Reise & Venegas 1987).
(B) Plant Parasites
Information on plant parasites is scarce. The presence of the red cup fungus of the genus Scutellinia (Ascomycota: Pezizales), probably S. scutellata (L.) Lamb., has been noted on decaying material on Achill Island (Hickey 2002). A fungus of the genus Nectria (Ascomycota: Hypocreales) was also found on G. tinctoria rhizomes in the Azores (Silva, Tavares & Pena 1996).
Introduction and Naturalization History
Gunnera tinctoria was introduced into Britain in 1849 as an ornamental species, and evidence of its escape into the wild dates back to the early 1900s (Preston, Pearman & Dines 2002; Skeffington & Hall 2011). The history of G. tinctoria invasions in Britain and Ireland is relatively poorly documented, but it is likely to have followed a ‘naturalization-invasion continuum’, as described by Richardson et al. (2000) and Richardson & Pyšek (2006) for other invasive species (Gioria 2007; Gioria & Osborne 2009). The majority of hectads in which its presence was recorded (91%) were after 1987, with 40% after 1999, suggesting either an exponential growth after a long lag-phase and/or an under-recorded distribution prior to 1999, likely due to its alien status.
In the Channel Islands, it was introduced to Guernsey in 1851 and has formed naturalized populations for ‘well over a century’ (McClintock 1975; see Skeffington & Hall 2011). In western parts of Cornwall, where the majority of records in England are found (Hill et al. 2005) and where it is now considered naturalized (Pilkington 2011), the first record is relatively recent, dating back to 1959 (Margetts & David 1981), although, according to French (2009a), the first record dates back to 1950. Naturalized or invasive populations are also found where the species is only a recent introduction, such as in the Outer Hebrides, where it was introduced at Urgha about 1992 as an ornamental plant (Outer Hebrides Fisheries Trust 2010). In the Isles of Scilly, the first record of naturalized G. tinctoria is also recent (1992; French 2009b).
Gunnera tinctoria does not seem to have formed naturalized or invasive populations in Wales, despite being common in large gardens (T. Rich, pers. comm.), and records in the wild are rather recent. In 2008, it was observed being naturalized in a fen in Nant Cledlyn, Cardiganshire (Ellis 2009). In Ireland, despite extensive research aimed at reconstructing the history of G. tinctoria invasions, virtually all information available is anecdotal (Osborne et al. 1991; Hickey & Osborne 1998), apart from recent genetic investigations (Fennell, Gallagher & Osborne 2010). The first documented record in Ireland can be found in Praeger (1939), who mistook it for G. manicata. He described abundant naturalized stands on the south and north side of Killary Harbour below Leenane, Co. Galway, where, despite a recent eradication programme (2008–09), it can still be found (Marchant 2008; Wade 2009). A substantial increase in the number of records was observed between 1960 and 1988 (Rich, Beesley & Goodwillie 2001; Preston, Pearman & Dines 2002), probably as a consequence of the extensive abandonment of agricultural land, as well as urban and tourism developments, suggesting that G. tinctoria may have a much greater potential range expansion and a larger impact in the future. Webb & Scannell (1983) reported that G. tinctoria was widely naturalized in West Mayo and Galway. The majority of records in Ireland are also post-1999 (Preston, Pearman & Dines 2002), confirming that the exponential invasion phase is a relatively recent phenomenon, despite the possible underestimation of its early distribution. Molecular analyses showed that the potential founder population for G. tinctoria in Achill Island and the Curraun peninsula, where the largest invasive Irish populations are found, could be Curraun House, on the Curraun peninsula (Fennell, Gallagher & Osborne 2010), where the species was introduced as an ornamental at the end of the nineteenth century (J. Steger-Lewis, pers. comm.). Other invaded regions may, however, have other founder populations (Fennell, Gallagher & Osborne 2010; Fennell et al. 2012).
On São Miguel Island, G. tinctoria spread from Furnas gardens, where it had been introduced as an ornamental, and was first recorded in the wild in 1964 (Silva & Silva 1974; Silva, Tavares & Pena 1996); by 1968, it was regarded as naturalized (Silva, Tavares & Pena 1996; Silva & Smith 2004). In New Zealand, the plant was first recorded in the wild in 1958 (Williams et al. 2005). Wilson (1982, 1987) described G. tinctoria as a ‘semi-wild’ garden escape, either as a rare or local occurrence. Webb, Sykes & Garnock-Jones (1988) reported that G. tinctoria was naturalized in Hawke's Bay, Taranaki, Wanganui, Banks Peninsula, Dunedin and from all Conservancies and half of the Department of Conservation Areas (DOC 2011) a decade later (Williams et al. 2005). On Stewart Island, it was considered naturalized by 1988 (Webb, Sykes & Garnock-Jones 1988) and, by 2007, ‘mature and well-established’ individuals were found from over 30 sites on the island, in urban environments and within the indigenous vegetation (Heenan et al. 2009).
A fascination based largely on the visual appearance and size of G. tinctoria has long been reported. Darwin (1845) reported that: ‘On the large island of Tanqui there was scarcely one cleared spot, the trees on every side extending their branches over the sea-beach. I one day noticed, growing on the sandstone cliffs, some very fine plants of the Panke (Gunnera scabra), which somewhat resembles the rhubarb on a gigantic scale. The inhabitants eat the stalks, which are subacid, and tan leather with the roots, and prepare a black dye from them. The leaf is nearly circular, but deeply indented on its margin. I measured one which was nearly eight feet in diameter, and therefore no less than twenty-four in circumference! The stalk is rather more than a yard high, and each plant sends out four or five of these enormous leaves, presenting together a very noble appearance' (Fig. 6). The species is edible and it has been reported to have medicinal properties (Hedrick 1972; Ladio, Lozada & Weigandta 2007); this has to be reconciled with reports of elevated levels of the neurotoxin BMAA in symbiotic tissues ('Biochemical Data'). Young leaf stalks are reported to be slightly acid and refreshing (Reiche 2004) and eaten by the local communities in South America (Ladio & Lozada 2001) as salads (‘nalca’; Wilkinson & Wanntorp 2007). de Medeiros et al. (2000) recommended the use of G. tinctoria to search for new anticoagulants that could be used to prevent blood clotting in cancer patients and lower the potential spread of cancer cells, due to the anti-thrombin activity of its extracts ('Biochemical Data'). Leaves are astringent, depurative and antidiarrhoeal, and can be used to treat back pain, while unspecified parts are thought to be antitussive, alleviating or suppressing coughs (Barboza et al. 2009). Medications containing dried aerial parts are indicated for the treatment of hyperuricaemia and related medical conditions including gout in humans (Barboza et al. 2009).
Gunnera tinctoria is now placed in its own monogeneric family, the Gunneraceae, in the Halogarales (Takhtajan 1980; Cronquist 1981; Soltis et al. 2000, 2003, 2005), which comprises 40–60 species (Jarzen 1980; Bergman, Johansson & Söderbäck 1992; Wanntorp et al. 2001; Fuller & Hickey 2005; Soltis et al. 2005; Wilkinson & Wanntorp 2007; González & Bello 2009). The taxonomy of this genus has proved to be complex (Fuller & Hickey 2005), and the classification of the family has been subject to frequent revisions (Wanntorp et al. 2001; Soltis et al. 2000, 2003, 2005; Angiosperm Phylogeny Group (APG II) 2003). Due to phylogenetic affinities such as the simple epigynous (floral parts near the apex of the ovary) and dimerous flowers, as well as a preference for wet habitats (Wanntorp et al. 2001), Gunnera had traditionally been considered an anomalous member of the Haloragaceae (Schindler 1905; Doyle & Scogin 1988), but a number of phylogenetic placements have been suggested, including the Myrtales, Cornales, Umbellales, Urticales, Saxifragales and the Balanophoraceae (Wilkinson 1998; Fuller & Hickey 2005 and references therein). Recent evidence shows that Gunnera is part of an early radiation of tricolpate (eudicot) angiosperms (Wilkinson 1998; Soltis et al. 2003). There also appears to be a close relationship between the Gunneraceae and the Myrothamnaceae (Soltis et al. 2000, 2003, 2005). These families were strongly excluded from all other eudicot clades (Soltis et al. 2000), and they are now treated as a single order, Gunnerales (Angiosperm Phylogeny Group (APG II) 2003; Angiosperm Phylogeny Group (APG III) 2009), which was originally proposed as monofamilial by Takhtajan (1997; see González & Bello 2009). Molecular analyses (Soltis et al. 2003; Hilu et al. 2003) have shown that the Gunneraceae and Myrothamnaceae could be sister to all other core eudicots, although Wanntorp & Ronse De Craene (2005) and Ronse De Craene & Wanntorp (2006) argued that the Gunnerales cannot be ancestral to core eudicots based on morphological characteristics.
The genus Gunnera was named by Carl von Linné, who named some specimens coming from the Cape of Good Hope as Gunnera perpensa in honour of the Norwegian bishop and botanist Johan Ernst Gunnerus (1718–73) (Bergman, Johansson & Söderbäck 1992). Within the genus Gunnera, six subgenera have been recognized (Misandra, Milligania, Pseudo-gunnera, Perpensum, Panke and Ostenigunnera; Schindler 1905; Mattfeld 1933; Wanntorp, Wanntorp & Källersjö 2002a,b). Gunnera tinctoria belongs to the subgenus Panke (Molina) Schindler, which includes large species with at least some hermaphroditic flowers (Fuller & Hickey 2005). This subgenus is mostly restricted to the Andean cordillera (Wanntorp & Wanntorp 2003; Wanntorp, Wanntorp & Rutishauser 2003). Based on morphological and flavonoid analyses, G. tinctoria appears to be the closest ancestor to three Gunnera species of Panke found on the Juan Fernández Islands, located c. 600 km off the coast of Chile, where seeds from mainland Chile may have arrived via the digestive system of birds (Pacheco et al. 1993; Wanntorp, Wanntorp & Rutishauser 2003).
Gunnera is one of the oldest living angiosperm genera (Jarzen 1980), with a fossil history that is remarkably well documented and relatively continuous, from the Early Cretaceous to the present (Schrank 1984; Jarzen 1980; Jarzen & Dettmann 1989). The earliest pollen record of Gunnera (Tricolpites reticulatus) is from the Turonian (c. 90 Ma) in South America (Wanntorp, Praglowski & Grafström 2004). Wanntorp et al. (2001) and Wanntorp, Wanntorp & Rutishauser 2003) showed that Gunnera can be considered of Gondwanan origin and was widespread and abundant in North America from the Late Cretaceous to the Eocene (80–50 Ma), although its distribution has retreated over time, probably due to climatic and environmentally related changes (Jarzen 1980; Osborne & Sprent 2002; Osborne & Bergman 2009), particularly increases in aridity and the disappearance of inland seas and seaways that were common during the Cretaceous period (Osborne & Sprent 2002).
In Britain, the species is listed on Schedule 9 of the amended Wildlife and Countryside Act 1981 (WCA 1981), because of its documented negative effects on plant diversity (Pena 1995; Silva, Tavares & Pena 1996; Hickey & Osborne 2001; Hickey 2002; Gioria & Osborne 2008, 2009, 2010). Among the most ecologically valuable communities invaded by G. tinctoria in Ireland are those similar to species-rich Salix cinerea–Galium palustre woodlands (Rodwell 1998a), leading to the replacement of Salix cinerea L., and altering natural successional processes (Hickey & Osborne 2001).
Gioria & Osborne (2009, 2010) reported a capacity for this species to alter the soil seed bank of resident communities significantly. Under invasive stands at three sites, seedlings of G. tinctoria accounted for a large proportion of the total number of seedlings recorded (coastal cliff, 80.7–86.2%; wet meadow, 53.2–56.5%; coastal grassland, 78.3–81.9%). These studies highlighted a tendency for this species to homogenize the seed flora, by reducing the diversity of the resident seed bank and increasing the relative abundance of seeds of weeds and rushes. These effects were evident even in deep soil layers (to 15 cm depth), indicating that invasions by G. tinctoria can alter the transient and the more persistent component of the seed bank, with potential long-term implications on the composition of the native vegetation.
At the ecosystem level, additional impacts associated with G. tinctoria invasions include considerable increases in above- and below-ground biomass (Hickey & Osborne 1998; Hickey 2002), alterations in the quantity and quality of litter, changes in water and biogeochemical cycles (Gioria 2007). Such ecosystem alterations may, in turn, facilitate its own growth or promote invasions by other species (Gioria, Dieterich & Osborne 2011). Soil erosion also represents a major issue, particularly in coastal areas and along the banks of rivers and streams, where it can create large areas of exposed ground (Hickey 2002; Williams et al. 2005; Gioria 2007) and may block drainage ditches (Reid R., pers. obs.).
The ecological impacts of this species are particularly severe where it becomes established in areas of high conservation value, such as in São Miguel Island, where the species threatens native communities at Lagoa das Furnas, Lagoa do Fogo, Sete Cidades, Salto do Cavalo and Pico da Vara (Pena 1995; Silva & Smith 2004). In New Zealand, the species has been recognized as a weed of conservation land (Owen 1997; Williams et al. 2005; Heenan et al. 2009) and is included on the National Pest Plant Accord list (Williams et al. 2005; NPPA 2012). In western parts of the North Island, particularly in the Taranaki Region, G. tinctoria threatens species of national and regional significance, such as Crassula peduncularis (Sm.) Meigen and Myosotis pygmaea Colenso (Taranaki Regional Council 2001; de Lange et al. 2004; Williams et al. 2005). It is naturalized in indigenous forest remnants in the Wellington Botanic Garden (Mitcalfe & Horne 2003). In the area of Northland, the species is a potential threat to native coastal forests (Sullivan, Timmins & Williams 2005). The ecological impact of G. tinctoria on Stewart Island is less known compared to the North Island (Heenan et al. 2009). Due to the large number of indigenous species (585) and the large proportion of the island belonging to the Rakiura National Park (85% of the total surface area), its presence represents a major threat to the native flora (see Williams et al. 2005). Mature adults also threaten the flora of the Halfmoon Bay area, particularly within indigenous bush and along stream margins and coastal cliffs (Heenan et al. 2009). The plant is considered a serious environmental threat in the Chatham Islands (de Lange, Heenan & Rolfe 2011).
Garden cultivation of G. tinctoria represents a major source of propagules of this plant. Moreover, its seeds can be purchased over the internet from a range of gardens. The species received the Award of Garden Merit in 2006 by the Royal Horticultural Society (Royal Horticultural Society 2011). Despite their potential to escape and become naturalized and/or invasive, G. tinctoria (and G. manicata, although what is advertised as G. manicata is G. tinctoria) is proudly advertised as a giant, tropical species in gardens such as Dunster Castle (Somerset), the Abbotsbury Subtropical Gardens (Dorset) and the Trebah Gardens (Cornwall). Proper risk assessment analysis of existing living collections in Britain and Ireland is thus strongly recommended. Consideration of the effect of climatic changes on its distribution should also be made, since expansion of its range could be favoured by projected increases in winter rainfall and summer temperatures (Osborne 2007; Gioria & Osborne 2009).
A ‘weediness score’ of 30 (of 36) has been calculated for G. tinctoria in New Zealand, based on the following calculation (Owen 1997):
where EoS is the effect/impact on the system and BSR is the Biological Success Rating (BSR). An EoS = 6 was calculated (maximum score = 9) as the sum of Significance of Change (score = 2), Suppress Native Regeneration (score = 2) and Persistence (score = 2). The BRS score was calculated as the sum of Maturation Rate, Seedability, Seed bank Persistence, Dispersal Effectiveness, Establishment and Growth Rate, and Vegetative Reproduction, all of which scored 3, for a total BSR score of 18, equalling the maximum score for BRS. An additional score of 6 was calculated as the sum of Increased Fire Risk (score = 1), Competitive Ability (score = 3) and Resistance to Management (score = 2). A weediness score of 30 places G. tinctoria within the ‘Priority Group A’ species (weediness score 29–36), that is, those species for which eradication is advocated (Owen 1998).
Those ecological traits that have contributed to the success of G. tinctoria as an ornamental plant worldwide make its eradication a difficult task. Effective control is complicated by the characteristics of the habitats where it becomes invasive, including steep slopes in coastal areas and margins of water bodies, where the use of herbicides is restricted (Armstrong 2008). Its presence in protected areas is also problematical. Potentially, there are three main strategies to control G. tinctoria, that is, biological control, mechanical control, herbicidal control or a combination of these. No biological control for G. tinctoria is known (see also Williams & Hayes 2007).
Herbicidal control is feasible only over large areas, although its effectiveness is dependent upon the age of the plants. While young plants can be readily killed by herbicide applications, the control of mature plants requires the application of large amounts of herbicides (Williams et al. 2005). The timing of herbicide applications is also important (Williams et al. 2005; Armstrong 2008). Contrasting results have been reported in the literature, with herbicidal applications being most effective when undertaken at the beginning of the growing season in New Zealand, while those conducted prior to the formation of the majority of leaves showed a minimal effect on plant growth in Ireland (Armstrong 2008). In the Taranaki Region, New Zealand, helicopter spray trials using glyphosate have been undertaken on slopes ranging from steep to near vertical, where conventional applications are impractical (Williams et al. 2005). These trials were successful, leading to a 55% reduction in biomass when 1% glyphosate solution was used.
Eradication by mechanical removal has proved effective only where the entire rhizome system was removed together with standing biomass, to prevent any re-sprouting (Williams et al. 2005). It thus represents a feasible option only over a small area and to control young individuals (Armstrong 2008). A combination of mechanical measures and herbicidal application appears to be most effective, where feasible. This option is also preferred when the plant is located along watercourses, where it is necessary to minimize the risk of water pollution. In an investigation conducted in Ireland, Armstrong (2008) used two methods of applying glyphosate solutions: Cut and Paint (C&P) and Cut and Injection (C&I). C&P involved cutting the petioles and inflorescences and applying 2% glyphosate to cut surfaces using a sponge. C&I involved the removal of petioles and inflorescences and excavating one or more wells in the rhizomes, into which 5 mL of herbicide was injected. Both approaches resulted in the death of the standing biomass over the first year of application; nevertheless, after 2 years, signs of re-growth were observed, due to incomplete decomposition of the large rhizomes, indicative of the need for reapplication in any control protocol. The C&P technique was the most effective control method also on coastal cliffs in New Zealand. This was based on a combination of cutting of the leaves and flower stalks together with a 25% glyphosate application, and resulted in 17 re-sprouting plants of 376 growing tips (5%) (Williams et al. 2005). The re-treatment of re-sprouting plants and the removal and spraying of any seedlings that have germinated may, however, be required (Williams et al. 2005).
Its capacity to form a large, persistent seed bank and to alter the seed bank of resident communities limits substantially the efficacy of control measures (Gioria 2007). First, any disturbance caused by its removal could promote the germination of its buried seeds (Gioria & Osborne 2008, 2009). Second, the creation of gaps during clearing operations would provide an opportunity for the recruitment of seeds of undesirable opportunistic species, such as Juncus species and agricultural weeds (Gioria & Osborne 2008, 2009; Table 4). Third, invasion-related reductions in the richness and abundance of seeds of native species require the sowing of seeds of desirable species to restore invaded sites (Gioria 2007; Gioria & Osborne 2008, 2009).
We thank Louis Ronse De Craene, Robert Northridge, Robin Reid, Tim Rich, Luis Silva, Ann Thompson, Carol West and Livia Wanntorp for providing information on the characteristics and distribution of this species. We are grateful to Ann Cullen, Liam Kavanagh, Bridget Moran, Ray O'Hare and Eugene Sherry, at the UCD School of Biology and Environmental Science, for their assistance in the field and with laboratory experiments. We are also grateful to all those graduate, undergraduate and post-doctoral researchers whose work contributed substantially to generating essential information used in this account; these include Betsy Hickey, Garret Campbell, Jane Hennessy, Kevin Black, Germaine Leveille, Carol Bruce, Fiona Doris, Mark Fennell and Cristina Armstrong, as well as staff members Jan-Robert Baars, Gerry Doyle, Tommy Gallagher, Bernard Kay, Martin Steer and Graham Wilson. M. Gioria thanks the Irish Environmental Protection Agency (EPA) for providing the funding necessary to conduct field work and laboratory experiments and to support her research activities (ERTDI PhD Scholarship and STRIVE Fellowship) since 2004. Part of the input to this work was funded by the EPA, the Irish Council for Science and Technology and the former Botany Department/UCD School of Biology and Environmental Science, through Research Demonstratorships and other support. We are grateful to Chris Preston, Michael Proctor, David Streeter and Michael Usher for their useful comments on the text and additional information supplied. Sincere appreciation goes to Tony Davy for his helpful comments on earlier versions of this manuscript and for his detailed editing of the final manuscript.