Prescribed burning of northern heathlands: Calluna vulgaris germination cues and seed-bank dynamics
- First Online:
- Cite this article as:
- Måren, I.E., Janovský, Z., Spindelböck, J.P. et al. Plant Ecol (2010) 207: 245. doi:10.1007/s11258-009-9669-1
The European coastal heathlands are important habitats for international conservation. Today, these low-intensity farming systems are threatened by the cessation of traditional management regimes, such as grazing and prescribed burning. In natural systems, the effects of fire on germination responses are often explained by adaptation to fire over extended periods of time. However, Northern heathlands are semi-natural systems with only a limited fire history. We investigated whether and how the keystone species in this system, Calluna vulgaris, responded to prescribed burning, based on previous findings where Calluna germinable seed-bank densities showed a pronounced peak right after fire. Our main findings were (i) an ecophysiological response to smoke; (ii) a potential explanation for this pattern, revealed by a seed-bank experiment where we managed to re-create the germination pattern experimentally by using an aqueous plant-derived smoke solution; and (iii) a history of anthropogenic use of fire and the development of heathlands in the region documented through palaeoecological investigations.
KeywordsAnthropogenic disturbance Germination cues Palaeoecology Plant-derived smoke Secondary succession
Fire is a disturbance influencing plant communities in many parts of the world. However, fire differs from many other disturbances in the way that it ‘consumes’ complex organic molecules, and it has been argued that fire is more analogous to herbivory than to other abiotic disturbances (Bond and Keeley 2005). After fire, an open habitat is created and a flush of nutrients released (Evans and Allen 1971), creating, from a plant’s perspective, a unique opportunity for colonization and growth. A subsequent wave of germination from soil-stored seed-banks is a well-known phenomenon where seed germination is stimulated indirectly through responses to micro-environmental changes or directly through responses to high temperatures or chemical cues derived from smoke, charcoal or ash (Zackrisson et al. 1996; Brown and van Staden 1997; Reyes and Casal 1998; van Staden et al. 2000). The ecophysiological responses to such fire-related germination cues have been studied extensively on species in ecosystems with naturally occurring frequent fires such as the South African fynbos (van Staden et al. 2000; Brown et al. 2003), the Australian kwongan (Dixon et al. 1995; Roche et al. 1997; Tieu et al. 2001; Read et al. 2000; Thomas et al. 2007), the North American chaparral (Keeley 1987; Keeley and Fotheringham 1988, 1997, 1998) and the Mediterranean maquis (Crosti et al. 2006). A major active compound in the smoke response has recently been identified as karrikinolide (3-methyl-2H-furo[2,3-c]pyran-2-one), a derivate from the combustion of cellulose. This represents a universal and reliable cue signalling the combustion of plant material (Flematti et al. 2004a, b; Van Staden et al. 2004; Light et al. 2008), and it has been shown to stimulate germination and seedling growth in a number of species of different biogeographic and ecological affinities (Light et al. 2008). Many studies have also found ash-beds to favour seedling growth (Keeley 1987). However, the response of seed germination to ash is less well understood (Reyes and Casal 1998).
Anthropogenic use of fire, often in conjunction with grazing, is contributing to the creation and maintenance of semi-natural ecosystems on marginal lands worldwide (e.g. African savannas and Australian kwongan; Bond and Keeley 2005). In these systems, as in natural fire-prone ecosystems, germination tends to peak immediately after fire (e.g. Burns 1952). However, such semi-natural systems have typically only been exposed to fire for a few thousand years (e.g. Kaland 1986; Webb 1998; Prøsch-Danielsen and Simonsen 2000), and there is a limited understanding of the germination responses of species to fire-related cues in these systems. The North European coastal heathlands, dominated by the ericaceous shrub Calluna vulgaris (L.) Hull (hereafter called Calluna), are one such anthropogenic ‘fire-prone’ ecosystem. These heathlands have a limited and declining distribution internationally, and considerable cultural and aesthetic significance; consequently they are in focus for conservation and ecological restoration (Webb 1986; Gimingham 1972, 1992; Webb 1998; EU Habitats Directive 92/43/EEC). In northern heathlands, prescribed fires are still part of the recommended management protocol for heathland management and conservation; hence, the effects of fire on germination dynamics require further investigation, although general fire ecology of European heathlands has been studied extensively (Kenworthy 1963; Hobbs 1981; Mallik et al. 1984; Mallik and Gimingham 1985; Hobbs and Gimingham 1984a, b; 1987; Miller and Cummins 1987; Davies et al. 2008). The coastal heathlands of the North Hordaland region, Norway, are distributed on islands of varying size and proximity from the mainland. Heathland development took place over a long temporal scale (6000–100 years BP), creating heathlands of different ages.
In cyclic vegetation types such as heathlands, seed-banks are particularly important so that species can survive locally, since seed inputs from surrounding unburned vegetation are negligible compared to those from the soil seed-bank after fire (Hobbs and Gimingham 1984a, b). In order to understand the mechanisms and processes which select for persistent seed-banks, and also the dormancy-breaking mechanisms operating, descriptive seed-bank, studies need to be complemented by experiments. In a previous study (Måren and Vandvik 2009) patterns in Calluna seed-bank density after fire were studied using a 24-year chronosequence. Germinable seed-bank densities showed a pronounced peak in soils of newly burnt heathlands (samples collected in the same year of the fire, i.e. 2–3 months after fire and in the year after, i.e. 14–16 months). In this study, we combined two data sets to explore the effects of fire on Calluna regeneration and seed-bank dynamics: We (i) performed experiments on fresh Calluna seeds to test whether the post-fire seed-bank behaviour of Calluna can be accounted for by ecophysiological responses to fire-related germination cues, and (ii) conducted experiments on soil seed-banks to test wether the patterns observed in the previous study could be re-created experimentally. In particular, we predicted that if the seed-bank dynamics of these heathlands are regulated by fire-related germination cues (smoke and/or ash), then we should observe increased germination under experimental treatments with smoke and/or ash in soil seed-bank samples collected from mature heath but not in samples collected from newly burnt heath, which have recently been exposed to these cues in situ.
Materials and methods
The study was carried out at the islands of Lygra and Lurekalven which are situated at 60°42′N and 5°5′E, 40 km northwest of Bergen, Western Norway. Climate is oceanic with mean temperatures of 12.0°C in June and 1.0°C in January, a growing season of ca. 220 days and mean annual precipitation of 1600 mm distributed evenly throughout the year (www.met.no). The islands are narrow and relatively flat, and acid bedrock gives rise to shallow and nutrient-poor soils. Calluna heath dominates with mires and Salix shrubs in wetter areas and mixed grass-heaths on relatively nutrient-rich areas such as beaches and former cultivated fields. Common heath species are Erica tetralix, Vaccinium myrtillus, Vaccinium vitis-idaea, Vaccinium uliginosum, Vaccinium oxycoccus, Arctostaphylos uva-ursi and Empetrum nigrum. Traditional management includes grazing and prescribed burning (Øvstedal 1985; Kaland 1986; Øvstedal and Heegaard 2000; Vandvik et al. 2005). ‘Prescribed burning’ as quoted in Kayll (1974) is ‘the skilful burning of natural fuel under specific conditions of weather, fuel moisture, soil moisture, etc., such that the fire is confined to a pre-determined area and at the same time its intensity is sufficient to accomplish specific objectives of management’. Plant nomenclature is adopted from Lid and Lid (2005).
Species in focus: Calluna vulgaris
Calluna is a monotypic genus. The ecological amplitude of Calluna is fairly high. It is well known as an oligotrophic calcifuge species. Approximate soil pH limits are 3.2–7.0, but usually it is found within the 3.5–6.5 range. It is a low, branched and normally hemi-spherical shrub, evergreen and with minute leaves (Gimingham 1960). It produces copious, minute (0.7 × 0.3 mm) seeds (Nordhagen 1937; Legg et al. 1992; Barclay-Estrup and Gimmingham 1994), which form persistent seed-banks (seed-bank longevity estimate of 150 years in blanket peat; Cumming and Legg 1995). The Calluna soil seed-bank does not exhibit a pronounced seasonal peak, producing a Type IV seed-bank as described by Thompson and Grime (1979). Well over 90% of the buried seeds are found within the top 50 mm of the podzolic heathland soils (Putwain et al. 1982; Miller and Cummins 1987). European heathland species (Calluna and several Erica species) have fairly complex light and temperature requirements for germination, dependent on various intensities and exposure times (Pons 1989), 18°C being the optimum temperature for Calluna seeds (Grimstad 1985; Thomas and Davies 2002). The species has been present in western parts of Norway throughout the Holocene, which started at ca. 11500 cal. BP (Bondevik and Mangerud 2002) in this region, but the abundance increased considerably during the establishment of the coastal heath vegetation, 5000–1000 years BP.
Data sets and analyses
(i) Responses of freshly collected Calluna seeds to fire-related germination cues were tested experimentally. Infructescences from 15 individual plants, separated by at least 15 m, were collected at Lygra in Oct/Nov of 2007. The material was dried at 20°C for 2 days before the seeds were shaken out and separated from the remaining debris followed by storage at 15% relative humidity and 15°C before starting the germination experiments. Three of the harvested individuals contained no seeds at all, whereas for the other 12 the number of harvested seeds per individual varied between 1300 and 13800 seeds. The average seed mass ranged from 0.027 to 0.040 mg per seed and was not correlated to the number of seeds in each individual.
Linear model of total counts of Calluna vulgaris seeds from data set (ii), only significant (non-sample) terms displayed. Residual dfs = 46, residual SE = 28.31
Moss cover (high):age (old)
Moss cover (high):age (old):ash
Moss cover (high):age (old):smoke
The individual means were tested for normality with a Kolmogorov–Smirnov test and for significant differences with a mixed model ANOVA with mother plant as a random factor and treatment as a fixed factor. ANOVA analyses and all the following calculations under (ii) were performed using R version 2.7.1 (R; The R Foundation for Statistical Computing 2008).
(ii) We performed a germination experiment of soil seed-bank samples to test whether the observed germination peak in post-fire succession (Måren and Vandvik 2009) could be re-created experimentally. Soil seed-bank was sampled at Lygra in March 2008. Twenty soil samples, 20 × 20 × 5 cm deep, each of a volume of 2000 cm3, were collected. Ten samples were taken in mature to degenerate phases of heather growth (28 years old), called ‘old heath’, hereafter. Ten samples were taken in pioneer phase of heather growth, burnt the previous year, called ‘newly burnt heath’, hereafter. The original samples were split in four; each quarter was randomly assigned to one of the following treatments: (1) control (2) aqueous smoke solution, (3) ash and (4) smoke solution and ash. This resulted in a final design of 80 sample trays. We followed Leck et al.’s (1989) definition of soil seed-banks as all viable seeds in or on the ground, including litter; hence, both litter and humus layers were sampled. Samples were collected in March, after cold stratification had taken place naturally, and the germination recorded is, therefore, assumed to be representative of the total reservoir of buried seeds in this system. We used a seedling emergence technique; an indirect method for estimating the seed stock in the soil, and soil samples were collected and germinated in a greenhouse for seedling counts in accordance with Thompson et al. (1997). Within 48 h of collection, the soil was washed with water over a 0.4-cm meshed sieve to remove roots, twigs and stones, contributing to bulk reduction, in accordance with Ter Heerdt et al. (1996). Each sample was thoroughly mixed and spread out in a very thin layer of ca. 0.1 cm, so that seeds were exposed to light and suitable temperatures, on trays (30 cm × 60 cm) filled with 5 cm of sterile subsoil, consisting of equal amounts of sterile peat, perlite and growth soil. Trays were randomly placed in an unheated greenhouse and watered regularly from above with tap water. In addition, six control trays were randomly placed to monitor any airborne contaminants. Additional light sources were SON-XL high pressure sodium lamps of 400 W with a light regime of 8-h darkness and 16-h light.
The aqueous smoke solution used in the experiment was the same as for data set (i), see above for details. The concentrated smoke solution was diluted by 1:1000. We applied 0.05 l of the diluted solution to the sub-samples using a pressurised spray bottle. Ash was prepared from ca. 300 l of compacted mature heather cuttings, collected at Lygra, which were dried at 80°C for 4 h before burning them at about 600°C. This yielded ca. 1 l of ash. The ash was spread in a thin layer on the top of the sample trays to imitate conditions after prescribed heather burning.
Emerging seedlings were identified, counted and removed at approximately 2 week intervals. All other seedlings were also counted and removed from the trays. During the course of the experiment bryophytes also emerged in the trays. These were difficult to remove without damaging the very small emerging seedlings, and the moss layer was, therefore, left intact in the trays. We anticipated that the development of a moss layer could interfere with seed germination of a small-seeded species such as Calluna (Gimingham 1960). We, therefore, recorded moss cover on the trays, using a semi-quantitative scale with five levels (a: 0–1%, b: 1–5%, c: 5–25%, d: 25–75%, e: 75–100%). The covariate for the statistical analysis was constructed by counting the sum of the mid-values of each semi-quantitative class of moss cover over time. For each heath, age × treatment combination, replicates were split into low cumulative moss cover (lower than median) and high cumulative moss cover (higher than median). The experiment was terminated after 4 months when no more seeds germinated.
Our response variable was the total count of Calluna seedlings per tray, which was analysed by a linear model, with seed-bank samples accounted for as a factor with 19 degrees of freedom. The tested predictors were age of heath, ash and smoke treatments, with cumulative moss cover included as a categorical covariate. The model also contained all the interactions of predictors. The assumption of normality of the response variable was tested by using the Shapiro–Wilk test and homogeneity of variance was inspected by checking the model residuals.
(iii) We investigated the link between the temporal development of Calluna and the anthropogenic ‘fire-prone’ coastal heathland by means of pollen analysis. The pollen diagram depicting the historical use of fire in the region was obtained from a small lake, Grønevasstjørn Lake, situated on the island of Fedje, 8 km northwest of Lygra. The island of Fedje was selected because its large blanket bogs indicated an early deforestation. The Grønevasstjørn Lake was cored with a Livingston sampler of 110-mm diameter, and the plastic tubes were cut longitudinally in the laboratory. Microfossil analysis and dates from 14C radiocarbon analyses were performed on sediment samples, using standard methods of Fægri and Iversen (1989). Trees, herbs and graminoids, and Calluna were calculated on Σ pollen, while charcoal dust was calculated on Σ pollen + charcoal dust. The pollen sum varied between 600 and 800. Here, we present the relative pollen diagram with a selection of relevant taxa shown on the temporal scale from 6000 BP to present day.
(i) Calluna seeds and the response to aqueous smoke solution
(ii) Calluna seed-banks and the response to smoke and ash
(iii) The regional expansion of Calluna, formation of heathlands and the use of fire
Requiring specific cues for seed germination is a major adaptive response that allows seeds to ‘sense’ prevailing environmental conditions thereby providing a means of habitat selection (Harper 1994; Baskin and Baskin 1998). We found consistent positive effects of fire-related germination cues on the heathland keystone species C. vulgaris. While the ecophysiological response to smoke [data set (i)] parallels what is already known from other studies (Thomas and Davies 2002), our study is the first to combine this response with observational and experimental studies on the seed-bank [data set (ii)]. Combined, these data give insights into the ecological effects of the anthropogenic disturbance of prescribed burning on the dominant species of this heathland system.
Although germination cues can be very species specific (Thomas et al. 2003; Tierney 2006), smoke released from burning vegetation contains a chemical signal: karrikinolide, that triggers germination of a variety of species worldwide (Flematti et al. 2004a, b; Van Staden et al. 2004; Light et al. 2008), including Calluna (Thomas and Davies 2002). Interestingly, in a heathland context, smoke treatment has, for example, also been shown to stimulate germination responses in a range of Ericaceae species from the Cape Floral Region in South Africa (Brown 1993; van Staden et al. 2000).
European heathland species accumulate substantial soil seed-banks, which contribute to post-fire recruitment (Hobbs and Gimingham 1984a, b; Mallik et al. 1984; Miller and Cummins 1987), and yet, there has been little evidence that these species are directly stimulated by heat or fire-related chemicals (Hobbs and Gimingham 1984a, b; Mallik et al. 1984; Mallik and Gimingham 1985, Grimstad 1985), albeit, much study on the physical aspects of the fire itself has been carried out in Scotland by Gimingham and his students (Legg 1978; Hobbs 1981; Mallik et al. 1984; Mallik and Gimingham 1985; Hobbs and Gimingham 1984a, b, 1987; Davies et al. 2008). Two of the keystone species in the coastal heathlands of northern Europe, C. vulgaris and E. tetralix, have seeds that are light-stimulated and it has commonly been proposed that burning, as well as other disturbances, triggers recruitment by improving light conditions (Pons 1989). Accordingly, Miles (1974) found that pioneer and the mature phase stands showed the most contrasting effect on germination, while Whittaker and Gimingham (1962) found a highly significant increase in the number of Calluna seedlings emerging in burnt sites compared to unburnt sites. They argued that this could be attributed to heat stimulation and/or improved seed-bed conditions as a result of burning. However, it needs to be mentioned that their study pre-dates the knowledge of the cueing effect of smoke.
In our seed-bank study [data set (ii)], all the soil samples were spread thinly on top of sterile soil, and light and seed-bed conditions should be near optimal for germination (cf. Grimstad 1985). The near 100% increase in germination in the year 0 and 1 after fire (Måren and Vandvik 2009), therefore, cannot be explained simply by light and seed-bed conditions; expected seed-bank densities immediately after burning should resemble pre-fire conditions and, consequently, be similar to that of the mature/degenerate heath. Our experiments show that this significant increase in germinable Calluna seeds after fire may be attributed to fire-related germination cues: In the soil seed-bank samples collected from the mature/old heath, germination increased under experimental treatments with smoke and ash. In contrast, we did not observe the same trend in the samples collected from the newly burnt heath, in line with our prediction, as these sites have already been exposed to these cues in situ. The relatively weak response to the aqueous smoke solution in the seed-bank experiment (ii) in comparison to the fresh seed experiment (i) and the chronosequence study (Måren and Vandvik 2009) could be due to dilution differences; the active substance in the smoke solution could have been leached out by the watering regime in the greenhouses, before being effectively absorbed by the Calluna seeds. The aqueous smoke solution was applied once at the onset of the seed-bank experiment, and the effect might have been stronger if it had been applied repeatedly in the first couple of days.
Our study demonstrates the added value and importance of comparing results across methodologies; in other studies, the response of fresh Calluna seeds to watering with ash water has been shown to significantly reduce its germination (González-Rabanal and Casal 1995). This may be explained by the alkaline conditions of the ash water and Calluna’s strong pH dependence for germination (Gimingham 1960; Hobbs and Gimingham 1987). By treating the soil seed-bank samples with ash instead of soaking seeds in aqueous ash solution, we are able to mimic in situ post-fire conditions to a greater extent, and we see an enhanced germination in the old heath samples compared to the newly burnt samples.
The results from our study demonstrate that assessing the germination response of a species to fire can best be achieved through experimental studies that complement seed-bank species descriptions. Here, we show that the observed increase in germination of Calluna is brought about by cues associated with the fire itself, such as smoke, in addition to the already known factors, such as temperature, light and seed-bed conditions (Whittaker and Gimingham 1962; Pons 1989). One potential source of error in our comparison of seed-bank densities before and after fire is that the fire may have killed shallowly buried seeds. This would actually decrease the observed seed-bank densities after fire, and hence, create a pattern opposite to what we found.
Seed-banks serve as repositories of genetic diversity for many species, including Calluna (Mahy et al. 1999) with many seeds using cues such as temperature fluctuations, light quality, moisture fluctuations, soil nitrate and their own age to trigger germination and initiate vegetative growth (Mirov 1936; Philippi 1993). In order to cope with the lack of reliability of these proximate signals and the unpredictability of the post-germination environment, some species may have evolved ‘bet-hedging’ strategies, whereby only a certain fraction of the seed-bank germinates under favourable conditions (Venable and Lawlor 1980; Philippi 1993; Venable 2007). In addition, different populations of the same species may have different germination characteristics because of differences in the environmental factors acting on the maternal plant (Baskin and Baskin 1998). Species occurring in fire-prone habitats may be adapted to fire-cues because the post-fire environment offers increased recruitment opportunities in the form of safe-sites, enhanced light conditions and higher nutrient levels. Life-history approaches cued to post-fire conditions include maintenance of seed-banks cued by fire, fire-stimulated flowering (e.g. geophytes; second-year recruitment), and post-fire dispersal (Keeley and Fotheringham 1998). The widespread occurrence of smoke-induced germination in a variety of phylogenetically distant families supports the hypothesis that smoke-induced germination is the result of convergent evolution (Keeley and Fotheringham 1998). In Scandinavia, Zackrisson et al. (1996) believe wildfire to be an evolutionary important agent of disturbance in northern boreal forests, with a recurrence interval of 50–100 year. They revealed that charcoal has an important effect on enhancing site fertility in early post-fire successions of these forests. It is highly likely that the anthropogenic fire regime with a recurring occurrence of 10–30 years in the coastal heathlands of northern Europe is of great importance in habitat and biodiversity maintenance. Species comprising this semi-natural habitat have probably not evolved in this system, but they have greatly increased in distribution and abundance due to these particular management regimes. The pollen diagram from Grønevasstjørn Lake at Fedje (Fig. 3) shows that this is the case for Calluna, and this also holds true for other groups of species such as graminoids (Sykes et al. 1994), butterflies (Erhardt and Thomas 1991) and the endangered holoparasite Cuscuta epithymum (Meulebrouck et al. 2008).
Most studies have applied germination cues and physical fire measurements singly (even though previous studies probably confound the effects, e.g. heat treatment of soil samples to stimulate germination of heat-treated seeds may also release chemicals from soil organic fractions producing a smoke-treatment effect, as well). However, seeds in the soil experience multiple combinations of these cues and conditions during and after a fire. In some instances, the combination of cues may reduce germination, while in others it may enhance germination (Keeley and Fotheringham 1998). Proposed research topics for the future should focus on the relative importance of smoke, ash, heat, light and seed-bed changes for the germination success of C. vulgaris and other species of particular interest to natural or anthropogenic fire-prone systems.
Implications for conservation
The large soil-stored seed-banks of the heathland ecosystem bridge the temporal gap between seed production and seed germination. During early stages of secondary succession after fire, populations of species other than Calluna have an opportunity to expand, such as Pteridium aquilinum, E. tetralix, Deschampsia flexuosa, Molinia caerulea, Nardus stricta and Trichophorum cespitosum (Gimingham 1972). However, in well managed heathlands, Calluna recovers dominance after ca. three to four growing seasons (Måren and Vandvik 2009), leaving little time for any other species to establish. The timing of management fires, in combination with grazing regime, will determine whether the post-fire environment will act as a window for the expansion of unwanted species. We stress the importance of setting concise management goals and documenting conservation efforts along the way.
Alternative conservation approaches such as applying aqueous smoke solution or smoking the soil using smoke tents after the removal of non-native invasives have been suggested as a means of restoring native plants in areas, where controlled fires are too risky or unwanted (van Staden et al. 2000), such as in South Africa. However, our findings of smoke-stimulated germination in Calluna reinforce the important role of heather burning in encouraging Calluna regeneration, as smoke application does not remove all the biomass, and it does not create an open microsite for germination (unlike fire). Fire produces an open habitat with a flush of nutrients such as Na, K and Ca (Evans and Allen 1971). After a disturbance, such as prescribed fire, seedlings primarily recruit from the local seed-bank, rather than from the seeds dispersed from outside the burnt area (Hobbs and Gimingham 1984a, b). On the other hand, alternative biomass reduction strategies (e.g. mowing) may also be ineffective in stimulating Calluna germination and seedling establishment as they do not provide the smoke stimulus.
Prescribed heathland burning has been part of the management protocol for the coastal heathlands of northern Europe since, at least, the last five thousand years. This is also the case at our study site in North Hordaland, western Norway (see Fig. 3), and several other studies show similar patterns of decrease in boreal-tree pollen in combination with increased levels of Calluna pollen and charcoal particles (Prøsch-Danielsen and Simonsen 2000; Odgaard 1992) along the Atlantic coast of Norway.
The continuation of prescribed burning seems to be highly important for the sustainability of future heathland management and conservation, as fire can be seen to play an important role in maintaining diversity by providing a mosaic of varying stand structures/ages across the landscape. Fire temperatures and intensities have been shown not only to increase with stand age up to the mature stands and then to decline in the degenerate stands, as a result of changes in stand structure (Kenworthy 1963; Hobbs 1981), but also to depend on stand heterogeneity. Fire may, thus, influence the post-fire successional trajectory and, in turn, regeneration. At the northern Lygra site, small scale fires at intervals of ca. 15–20 years seem to be appropriate for post-fire establishment of C. vulgaris and characteristic heathland species.
We thank Neville Brown at Kirstenbosch Botanic Gardens, Marnie E. Light and Johannes van Staden at the University of Kwazulu Natal, South Africa, for supplying the smoke solutions. We also thank Zoë Cook for valuable help at Wakehurst Place, Royal Botanic Gardens Kew, UK, and Hana Pánková and Mari Jokerud for their help during the experimental set-up at the Arboretum at Milde. We thank Ella Blomsø Ødegård and the staff at the Arboretum for their excellent technical support during the greenhouse trials, and also John Birks, Alison Hester and an anonymous referee for their comments on an earlier version of the manuscript. Financial support was received from Bergen Myrdyrkings-forenings Fond, Ecological and Environmental Change Research Group (EECRG), University of Bergen, and Olaf Grolle Olsens Legat.