Do plant traits explain tree seedling survival in bogs?

Authors


Summary

  1. Moss-dominated peat bogs store approximately 30% of global soil carbon. A climate induced-shift from current moss-dominated conditions to tree-dominated states is expected to strongly affect their functioning and carbon sequestration capacity. Consequently, unravelling the mechanisms that may explain successful tree seedling establishment in these ecosystems is highly relevant.
  2. To assess the role of drought on early tree seedling establishment and the relative importance of plant traits in tree seedling survival, we conducted a factorial glasshouse experiment with seven conifer species.
  3. Our results show that drought inhibits moss growth, thereby increasing survival of tree seedlings. Survival success was higher in Pinus than in Picea species, ranking Pinus banksiana Pinus sylvestris > Pinus nigra > Picea mariana > Picea glauca, Picea sitchensis > Picea rubens. We found that those species most successful under dry and wet conditions combined a fast shoot growth with high seed mass.
  4. We conclude that plant traits contribute to explaining successful early tree seedling establishment in bogs.

Introduction

Boreal peatlands are widely recognized for their disproportionally big carbon stores: almost 30% of all terrestrial soil carbon is locked away in metres of peat in the northern hemisphere (Turunen et al. 2002). Most of this peat has been built by Sphagnum mosses in bogs (Overbeck 1975). Bogs are open peatlands with waterlogged, anoxic, acidic and nutrient-poor conditions that hamper the establishment and growth of vascular plants (Vanbreemen 1995), especially trees (Ohlson et al. 2001). Despite these harsh environmental conditions, some trees manage to survive, leading to sparse tree cover landscapes (Scheffer et al. 2012). Tree occurrence on pristine bogs is mostly restricted to continental climates and has been related to the degree of water-table draw-down during the growing season, with more trees and better tree growth with relatively drier conditions at deeper water-tables (Rydin & Jeglum 2006). In wetter climates, or wetter environmental settings, tree occurrence is restricted to elevated microtopographical sites, called hummocks, where soils are not permanently waterlogged and moss growth is slower (Hornberg, Ohlson & Zackrisson 1997; Gunnarsson & Rydin 1998).

Recent tree and shrub encroachments in drained (Pellerin & Lavoie 2000; Linderholm & Leine 2004) and pristine bogs (Aberg 1992; Berg et al. 2009; Kapfer et al. 2011) have been attributed to warmer or drier climate, as well as to increased fire frequencies correlated with these climate conditions. Also, field experiments show that warmer and drier conditions can stimulate tree seedling survival and growth under nearly flooded conditions (Wolken et al. 2011). These results suggest that tree and shrub encroachments are likely to continue in the future because climate change scenarios for the northern high latitudes indicate a further increase in average air temperature and more frequent drought events (Meehl & Tebaldi 2004). Unravelling the mechanisms underlying tree recruitment in bogs is of pivotal importance because bogs show discontinuous tree distribution patterns that suggest abrupt nonlinear responses to changes in precipitation and temperature, with far stretching implications for ecosystem resilience (Scheffer et al. 2012). Early seedling survival is the most severe bottleneck for tree recruitment in bogs because more than 90% of all germinated seedlings do not survive the seedling stage in the field (Ohlson et al. 2001).

Summer droughts may open windows of opportunity for tree seedlings, sensu Scheffer et al. (2008), to effectively escape the mosses and establish. During drought, moss growth slows down as the photosynthetically active apical tissue dries out (Robroek et al. 2009), postponing the time it takes for moss to overgrow small tree seedlings. Also, as the water-table draws down, the water logged peat becomes oxygenated, stimulating decomposition and nutrient release (Fenner & Freeman 2011), potentially promoting seedling growth. Certain plant traits, such as fast intrinsic growth rate and shoot biomass allocation, during the seedling stage may be particularly important for tree seedlings to win the deadly race with the mosses during these dry periods. Although the significant role of plant traits in explaining ecological patterns and processes is widely recognized (Chapin et al. 2000; Lavorel & Garnier 2002), the use of the concept has remained notoriously absent in peatland ecology. Yet, identifying key traits that facilitate successful tree seedling establishment can contribute to understand the conversion of open peatlands into tree-dominated landscapes and predict future tree range expansions in these ecosystems.

We conducted an experiment to explore the role of morphological traits in explaining early tree seedling survival in drying peat moss. We hypothesized that (i) drought increases survival of tree seedlings on peat moss by hampering moss growth more strongly than seedling growth, and (ii) tree seedling survival increases with seedlings growth rate and investment in shoot biomass because this would allow seedlings to escape from being overgrown by moss. To our knowledge, we are the first to apply the plant trait concept to explain seedling survival in bogs.

Materials and methods

Experimental Design

We conducted two experiments under controlled glasshouse conditions. Experiment 1 assessed survival of conifer species on moss under contrasting water conditions and Experiment 2 assessed morphological traits of the conifer species used in Experiment 1.

Experiment 1

We set up a factorial glasshouse experiment with seven conifer species kept in peat-moss-filled pots subject to two contrasting water treatments, simulating either regular water availability or an extended dry period with limited rain. Each combination of treatments was replicated ten times. This resulted in a set-up with 140 pots arranged in 10 replicated blocks of 14 pots each. Treatments were randomly assigned to the experimental pots and implemented in the glasshouse. To reduce any systematic location effects, both the blocks and the pots within a block were randomly moved once a week. The experiment lasted 2 months, from 19 November 2010 till 21 January 2011.

Two different watering regimes were implemented: a wet treatment where we added 60 mL (8·5 mm) of water to each pot every other day and a dry treatment where we added 60 mL of water to each pot every other week. Water was given from above with the use of a syringe to simulate a natural rain event. For the water treatments, we used demineralized water amended with nutrients and spore elements (Smolders et al. 2001), simulating water chemistry in Dutch peatlands. Water content in the pot was monitored gravimetrically as pot weight once a week, and volumetrically at final harvest with a Theta probe (Delta-T). Pot weight remained fairly constant in the wet treatment, varying from 92% to 110% of the initial weight on average. In the dry treatment, pot weight dropped to 60% of the initial weight over the first 2 weeks, after which it gradually declined to 40% of the initial weight at the end of the experimental period. At harvest, volumetric water content in the top 5 cm of the dry treatment varied between 2% and 10%, corresponding to dry microsites 26–18 cm above the water-table, measured in the same species in the field in July 2008 (F. Preston & J. Limpens, unpublished data, Haaksbergerveen bog reserve).

Double, round, transparent plastic pots were used, where one pot fit into a second pot. The first pot had four or five holes (diameters 5 mm), allowing free drainage of water to the second pot, simulating the well-drained conditions of hummock microhabitats. None of the pots were closed or covered; the peat moss was in contact with the air all the time. Dimensions of the individual pots were 9·5 cm in diameter and 15 cm in height, and the dimensions of the entire construction (two pots placed into each other) were 9·5 cm in diameter and 19·3 cm in height. In every single pot, one pregrown (4 weeks old, see 'Plant Material') seedling was inserted into the peat moss in the centre of each pot with special care for the roots.

Experiment 2

Morphological traits were assessed independently of Experiment 1. We grew tree seedlings under optimal conditions by planting pregrown 4 weeks old tree seedlings (see 'Plant Material') into the centre of a (10 cm wide) pot, using a density of one seedling per pot. The pots were filled with sterilized organic soil, watered daily and kept under the same glasshouse light and humidity conditions as Experiment 1. Pots were arranged in five replicated blocks. Both the blocks and the pots within a block were randomly moved once a week.

Glasshouse conditions

The compartment used for both experiments had glass windows on all sides that remained closed during the course of the experiment to prevent natural precipitation and wind from influencing the water content in the pots. Sprinklers were switched off for the same reason, resulting in a relative humidity of 50–75%. Natural light was augmented by artificial light (SON-T Agro 400 Watt, Philips, Eindhoven, The Netherlands), ensuring a photoperiod of 16 h a day. The lights were turned off between 6 pm and 2 am. Day temperatures were maintained at 20 °C and at 15 °C during the night (5·30 pm to 0·30 am).

Plant Material

Seeds and seedlings

Seven conifer species were selected based on their differences in growth rate to account for a wide trait range (Table S1, Supporting Information). The species selected are native to Europe (Pinus sylvestris L., Pinus nigra Arnold) or the United States of America (Pinus banksiana Lamb., Picea sitchensis (Bong.) Carrière, Picea rubens Sarg., Picea glauca (Moench) Voss, Picea mariana (Mill.) Britton, Sterns & Poggenb.). They are all able to grow in nutrient-poor and acidic environments and have been reported from peat soils (Dickson & Savill 1974; Ohlson & Zackrisson 1992; McLeod & MacDonald 1997). Seeds were stored in a fridge at 5 °C until further processing. All seeds were scarified by 24 h soaking prior to sowing to improve the overall germination rate.

We sowed 80 seeds per species on trays filled with a shallow layer of standard potting compost (Lentse potgrond) covered with vermiculite in a glasshouse. The trays were covered with plastic foil and a light white fleece and watered regularly to avoid desiccation of the substrate. Foil and fleece were removed after seed germination. Four weeks after sowing, seedlings were gently extracted from their germination substrate, shortly submerged in demineralized water with their roots to dislodge soil particles, measured and used in the experiments.

Peat moss

We used Sphagnum fallax, a species common to peatlands of Europe and North America that tolerates a relatively wide range of nutrient and water availabilities (Gabka & Lamentowicz 2008) and creates a surface where tree recruitment is known to occur in the field (Limpens 2013). Under favourable conditions, S. fallax has a high growth rate compared with other peat moss species (Grosvernier, Matthey & Buttler 1997), allowing treatment assessment in a relatively short time frame.

Peat moss was collected in the field from the Haaksbergerveen (52°07′ N, 6°46′ O), the Netherlands, at the beginning of November 2011, about 1·5 weeks before the start of the experiment. To prevent drying out, the mosses were covered with tissues soaked in demineralized water until processing. Prior to the experiments, mosses were cut at one length (5 cm), leaving the moss apex (capitulum) intact, thereby creating a similar starting point across treatments. Care was taken that the growing tips were all at the same height, thereby preventing unequal growth and drying out of moss individuals sticking out of the carpet.

Before treatments were assigned, the peat moss was saturated with a standard water solution for peat moss cultures (Rudolph, Kirchhoff & Gliesmann 1988) to obtain an equal starting point in moisture and nutrient levels in all pots. The moss was submerged in 0·8 L of Rudolph solution for 15 min, after which moss and excess water leaking from the moss were removed.

Measurements

Moss growth

Moss surface height change was used as a proxy for growth and measured twice a week relative to the pot edge at three fixed points per pot by placing the edge of a ruler flush with the apex (capitulum) of the Sphagnum mosses. Moss surface height increased with 4·6 ± 1.0 SD cm in the wet treatment and 0·4 ± 0·5 SD cm in the dry treatment over the experimental period. The height increments fall in the ranges reported for several peat moss species characteristic of moist and dry microhabitats in various peatlands (Ohlson & Okland 1998; Limpens, Berendse & Klees 2004; Laiho et al. 2011). Height increment of the moss surface showed a strong linear relationship with time (R2 = 0·99, linear regression) in the wet treatment, but a weak response in the dry treatment mostly restricted to the first 2 weeks (not shown).

Tree seedling survival

The distance between the top of the tree seedling and the moss surface was measured at weekly intervals. At final harvest, the survival percentage was calculated per species in each experimental treatment (n = 10 seedlings). A seedling was considered dead when overgrown by moss, and the distance between the top of the highest needle and the moss was reduced to zero, or when the seedling had dried out, turning yellow with shrivelled needles.

Plant traits

We distinguished nine morphological and derivative traits (Table 1). All traits were measured on seedlings in both experiments to assess trait plasticity, but only the traits assessed in Experiment 2 were used to relate to seedling survival in Experiment 1.

Table 1. Overview of nine traits and their definitions determined for seven conifer species. Values were based on measurements on individual seedlings. All morphological traits were measured on seedlings in both experiments to assess trait plasticity, but only the morphological traits assessed in Experiment 2 were used to relate to seedling survival in Experiment 1. n = 10 seedlings for Experiment 1 and n = 0–10 seedlings for Experiment 2
TraitDefinition
Seed mass (mg seed−1)Average based on total mass of 50 seeds
Initial shoot height (cm)Top longest needle to stem–root transition before planting
Relative height growth of the shoot (mm day−1)Loge (shoot heightharvest-start)/10 (63 days)
Final shoot biomass (mg)Biomass of needles and stem at harvest
Initial root length (cm)Stem–root transition to end longest root before planting
Relative length growth of the root (mm day−1)Loge (root lengthharvest-start)/10 (63 days)
Final root biomass (mg)Biomass of all roots at harvest
Deep/shallow root ratioRoot biomass (bottom 5–13 cm)/(top 0–5 cm) at harvest
Shoot/root biomass ratio(Shoot biomass)/(Root biomass) at harvest

Seed mass was taken into account because it determines the initial seedling size (Campbell & Rochefort 2003), and thus initial height above the moss. Seed mass was measured for every species and was determined by weighing 50 seeds separately and calculating mean and standard error of the seed mass. To assess germination, we introduced seeds to Experiment 1 in the third week, corresponding to the time when pot water contents had stabilized. Three seeds were placed around the seedling of the same species, on the capitulum of a moss individual, using 3 × 140 = 420 seeds in total. Germination was checked twice a week until harvest, 5 weeks later. We considered a seed germinated when the integument had broken and a ‘shoot’ of at least 1 mm had emerged from the seed.

Seedling shoot height and root length were measured at the beginning and end of the experiments (Table 1). The initial height and length were determined just before planting the seedlings in their respective pots. Seedlings were placed on a wet tissue to avoid root desiccation while measuring. The final measurements were carried out at harvest. Seedlings were gently extracted from their substrates, spread out and measured before biomass assessment.

Biomass of shoot and roots were determined destructively at harvest. Seedlings were divided into three parts: shoot (i.e. needles and stem until the stem–root transition), upper roots (5 cm) and lower roots. We used the ratio between the upper and lower part of the roots [deep/shallow (DS) root ratio] to assess differences in root allocation pattern between the species (Table 1).

Infection of roots by ectomycorrhizal (EM) fungi was assessed to separate effects of morphological traits from those of EM-improved nutrient acquisition on seedling survival. At harvest, all seedling roots were visually checked for signs of EM infection (i.e. thickening of root tips). None of the seedlings in Experiments 1 and 2 showed signs of infection (yet). A check of seedlings of similar size in the field yielded the same results: infection of tree roots by EM fungi seems to take place in older seedlings in bogs.

Derivative plant traits

Derivative plant traits comprised relative height growth of the shoot (RHG shoot), relative length growth of the roots (RLG root), the shoot/root (SR) biomass ratio and the DS root ratio (Table 1). The DS root ratio was calculated to determine whether roots are mainly located shallow or deep and to see whether a species has a fixed rooting pattern. This can have consequences for the ability of a seedling to respond to changes in water availability.

Statistical Analysis

All data were checked for data distribution (Shapiro–Wilkinson test on residuals) and equality of variances (Levene's test). Where possible, the data were transformed to achieve a normal distribution; otherwise, nonparametric tests were used. Water treatment effects on seedling survival in Experiment 1 were tested per species using a chi-square (χ2) test. Species differences between traits (Experiment 2) were assessed with one-way anova's using species as independent treatment and trait as response variable. Correlations between seedling survival (Experiment 1) and traits (Experiment 2) were assessed using Spearman correlation analysis. Where appropriate, we further explored colinearity among traits using partial correlations.

Results

Water Content Effect on Germination and Survival

In Experiment 1, three of seven species germinated better on the moist moss beds of the wet treatment than on the dry moss beds: for the other species, including all three Pinus species, water content of the substrate did not significantly affect germination (Appendix S2, Supporting Information). On average, Pinus species had a higher germination percentage than Picea species, both under wet (84% Pinus vs. 66% Picea) and dry conditions (63% Pinus vs. 30% Picea), with the best performing species being P. nigra.

Seedling survival peaked in the dry treatment, irrespective of species: 84% of all seedlings survived on dry moss, whereas only 24% survived on wet moss (Fig. 1a). Most survivors in the wet treatment were from the genus Pinus. The difference in survival between dry and wet moss was most significant for Picea glauca (inline image = 9·9, < 0·01), Picea rubens (inline image = 13·3, < 0·001), Picea sitchensis (inline image = 9·9, < 0·01) and P. nigra (inline image = 8·6, < 0·01). For the other species (P. mariana, P. banksiana, P. sylvestris), this difference was not significant. Survival in the wet treatment was negatively correlated with moss growth, suggesting moss overgrowing the shoots as main cause of seedling mortality (Fig. 1b). For the dry treatment, seedling survival was not related to moss growth (Fig. 1b).

Figure 1.

(a) Seedling survival (%) on dry moss (open bars) and wet moss (filled bars) after 8 weeks (n = 10). (b) Relationship between seedling survival (%) and the part of the shoot overgrown by moss. Filled symbols indicate wet conditions; open symbols refer to dry conditions. Each symbol represents one conifer species. Triangles indicate species from the genus Picea; bullets represent the genus Pinus. gla = Picea glauca (n = 5), mar = Picea mariana, rub = Picea rubens, sit = Picea sitchensis, ban = Pinus banksiana, nig = Pinus nigra, syl = Pinus sylvatica.

Plant Traits

In Experiment 2, species differed significantly in all morphological traits, except the length growth of the roots (RLG root, Table 2). The Pinus species had the highest seed mass and the largest seedlings with high initial shoot height and initial root length. Within genus, the differences between species were less pronounced and most apparent for shoot height growth (RHG shoot), final shoot biomass and initial root length within Pinus, and RHG shoot, final root biomass and shoot/root (SR) biomass ratio within Picea (Table 2). Some morphological traits were strongly correlated with each other (Table 3). The highest colinearity was between initial shoot height and initial root length. In turn, both traits correlated most with seed mass and slightly less with shoot and root biomass at harvest time. Shoot biomass at harvest time was also correlated with shoot height growth (RHG shoot).

Table 2. Means (±1 SE) for nine traits for seven conifer species. Seed mass was based on the mass of 50 seeds. Other means were based on measurements on individual seedlings grown on standard potting soil (n = 10)
Trait Picea Pinus F d.f. 6
glamarrubsitbannigsyl
  1. gla = Picea glauca (n = 5), mar = Picea mariana, rub = Picea rubens, sit = Picea sitchensis, ban = Pinus banksiana, nig = Pinus nigra, syl = Pinus sylvatica. F = F-values of one-way anova testing significant differences between species for each trait.

  2. Different letters in superscript after the SE indicate significant differences between species.

  3. ns> 0·05, ***< 0·001.

Seed mass2·01·73·62·64·616·68·0
Initial shoot height3·1 ± 0·4a3·7 ± 0·2ab3·8 ± 0·2ab4·5 ± 0·1b8·0 ± 0·2c8·2 ± 0·3c7·4 ± 0·3c89***
Relative height growth of the shoot0·25 ± < 0·01a0·36 ± < 0·01bc0·33 ± < 0·01b0·35 ± < 0·01bc0·42 ± < 0·01d0·35 ± < 0·01bc0·38 ± < 0·01cd16***
Final shoot biomass42 ± 0·13a119 ± 0·11bc95 ± 0·7abc161 ± 0·17cd346 ± 0·26e221 ± 0·24d260 ± 0·36de20***
Initial root length2·8 ± 0·9a3·3 ± 0·5a3·7 ± 0·3a4·1 ± 0·6a13·7 ± 2·1b17·0 ± 2·5b8·0 ± 0·8a17***
Relative length growth of the root0·16 ± 0·010·24 ± < 0·010·18 ± < 0·010·20 ± < 0·010·15 ± < 0·010·17 ± < 0·010·19 ± < 0·01ns
Final root biomass11 ± 3a38 ± 8b48 ± 6b202 ± 44c267 ± 50c116 ± 10c141 ± 26c22***
Deep/shallow root ratio0·1 ± 0·1a0·4 ± 0·1a0·4 ± 0·1a0·4 ± 0·1a1·3 ± 0·2b3·7 ± 0·4c2·4 ± 0·4bc23***
Shoot/root biomass ratio3·6 ± 0·4a3·8 ± 0·4a2·2 ± 0·3bc1·0 ± 0·2b1·6 ± 0·2bc1·9 ± 0·1bc2·3 ± 0·4c10***
Table 3. Colinearity between traits derived from seedlings of seven conifer species grown on soil
Trait123456789
  1. Spearman's rho correlation coefficients between nine traits.

  2. *< 0·05; **< 0·01, two-tailed.

Seed mass (1)1 0·89 ** 0·250·68 0·89 ** −0·390·54 0·79 * −0·46
Initial shoot height (2) 0·89 ** 10·50 0·86 ** 1 ** −0·290·75 0·82 * −0·64
Relative height growth of the shoot (3)0·250·501 0·86 ** 0·50*0·110·710·54−0·21
Final shoot biomass (4)0·68 0·86 ** 0·86 ** 1 0·86 ** −0·18 0·86 ** 0·75−0·50
Initial root length (5) 0·89 ** 1 ** 0·50* 0·86 ** 1−0·290·75 0·82 * −0·64
Relative length growth of the root (6)−0·39−0·290·11−0·18−0·291−0·14−0·070·25
Final root biomass (7)0·540·750·71 0·86 ** 0·75−0·1410·390·82*
Deep/shallow root ratio (8) 0·79 * 0·82 * 0·540·75 0·82 * −0·070·391−0·11
Shoot/root biomass ratio (9)−0·46−0·64−0·21−0·50−0·640·250·82*−0·111

Survival and Traits

Under dry conditions, seedling survival correlated most with seed mass (rs = 0·95, < 0·01; Table 4) and, to a lesser extent, with initial shoot height (rspearman = 0·88, < 0 0·01) and initial root length (rspearman = 0·88, < 0 0·01). When controlling for seed mass, the correlations between initial shoot height (rpartial = 0·70, Ns) or root length (rpartial = 0·62, Ns) with survival were no longer significant, suggesting seed mass as the overarching explaining factor under dry conditions (Fig. 2a).

Table 4. Relating traits to survival. Spearman's rho correlation coefficients between traits of seven conifer species based on Experiment 2 and seedling survival (%) on wet and dry moss based on Experiment 1
Trait seedling Experiment 2Survival Experiment 1 (wet)Survival Experiment 1 (dry)
  1. Only traits showing significant (P < 0·05) differences between species (Table 2) are shown.

  2. *< 0·05; **< 0·01.

Seed mass0·52 0·95 **
Initial shoot height0·67 0·88 **
Relative height growth of the shoot 0·81 * 0·24
Final shoot biomass 0·88 ** 0·69
Initial root length0·67 0·88 **
Final root biomass0·560·52
Deep/shallow root ratio0·740·73
Shoot/root biomass ratio−0·13−0·47
Figure 2.

Relationship between seedling survival% on moss (n = 10) and (a) seed mass and (b) final shoot biomass. Filled symbols indicate wet conditions; open symbols refer to dry conditions. Each symbol represents one conifer species. Triangles indicate species from the genus Picea; bullets represent the genus Pinus. gla = Picea glauca (n = 5), mar = Picea mariana, rub = Picea rubens, sit = Picea sitchensis, ban = Pinus banksiana, nig = Pinus nigra, syl = Pinus sylvatica.

Under wet conditions, seedling survival was significantly correlated with shoot biomass (rspearman = 0·88, P < 0·01) and shoot height growth (RHG shoot, rspearman = 0·81, < 0·05), the latter two traits being strongly interrelated (Table 3). When controlling for shoot biomass, the correlation between survival and RHG shoot was no longer significant (rpartial = −0·28, Ns), suggesting shoot biomass as the driving trait under wet conditions (Fig. 2b). Final shoot biomass or RHG shoot was not related to moss growth, not even within species, suggesting absence of light competition between moss and seedling up until the moment of being overgrown.

When analyses were run for the species within a genus (i.e. Picea or Pinus), shoot biomass and seed mass remained among the traits with the highest correlation coefficients. Using stepwise linear regression, shoot biomass and seed mass were selected as the only covariates in the model for wet and dry conditions, respectively (not shown).

Trait Plasticity

To assess the plasticity of the morphological traits, we compared the morphological traits based on seedlings from Experiment 2 with values measured on surviving seedlings in Experiment 1. Most trait values were highly correlated between the experiments (rspearman > 0·70), indicating that for these fixed traits, relative differences between species were sustained between the experiments (Appendix S3, Supporting Information). Exceptions to the above were root length growth (RLG root), final root biomass and SR biomass ratio. For these plastic traits, the species ranked differently between the experiments: species with the highest value in Experiment 2 did not have the highest value in Experiment 1, reducing the predictive value of these traits. Trait plasticity did not correlate with seedling survival (not shown).

Discussion

Drought had a strong positive effect on tree seedling survival on experimental peat moss beds. Under dry conditions, moss growth was impeded, facilitating tree seedling establishment. This is consistent with field observations (Ohlson & Zackrisson 1992) and paleoecological studies (Mcnally & Doyle 1984; Mighall et al. 2004) and suggests that drought events may accelerate recruitment of tree seedlings in open boreal bogs, provided that the seedlings survive the return to wetter conditions after the drought (Gunnarsson & Rydin 1998). The transient nature of droughts may be more beneficial for tree recruitment than persistent drought, as seed germination peaks under wet conditions (Ohlson & Okland 1998), whereas seedling survival peaks under dry conditions. In our experiments, the species that performed well under dry conditions, did also well under wet conditions, suggesting that certain sets of traits could be beneficial under both conditions.

The species most successful under both dry and wet conditions combined a fast shoot growth with high seed mass. Large seed mass allows seedlings to grow rapidly without depending very strongly on the surrounding environmental conditions (Poorter et al. 2008). Under the wet conditions that generally favour seed germination, resources available in large seeds can be invested in shoot biomass (initial height, RHG shoot, final shoot biomass) allowing tree seedlings to reach a safe size, minimizing the chance of being overgrown and finally killed by fast-growing mosses. Large seed mass was most critical to explain seedling survival under dry conditions. It seems reasonable to think that the root-traits associated with high seed mass (high initial root biomass, high DS ratio) enhanced water acquisition and survival during drought.

Fast growth allocated to shoot biomass during wet conditions and root biomass under dry ones seems to explain the success of tree seedlings against fast-growing mosses. Fast seedling growth may be particularly important in systems where plant establishment is restricted to short windows of opportunity (Scheffer et al. 2008). Also in drylands, where tree seedling recruitment is restricted to wet events, fast growth is a critical trait to explain successful establishment (Holmgren et al. 2006; León et al. 2011).

At present, only two conifer species dominate the boreal bogs of the northern hemisphere: Pinus sylvestris in Europe, including Siberia, and P. mariana in North America and Canada. In our experiments, both species performed well under both dry and wet conditions. Although P. mariana was the most successful within its genus, the three Pinus species (P. sylvestris, P. banksiana and P. nigra) performed particularly well. Despite the evident importance of early seedling survival in recruitment success, it is of course not the only bottleneck that individuals face before they reach adult status. Consequently, extrapolating these early life phase responses under very controlled experimental conditions to predict range expansions of adult trees in natural bogs needs to be undertaken with great caution. It is likely that multiple interacting factors determine successful recruitment in the field, some taking effect beyond the young seedling stage covered in our experiment. Limited seed availability or dispersal in sparse tree cover landscapes may be one of the first limiting mechanisms preventing tree expansion (Pellerin & Lavoie 2003). Once seeds have germinated, the young seedlings need to reach a safe distance above the moss layer. In this phase, tree seedling survival is closely related to shoot growth in relation to moss growth rate. Several factors may further modify the balance between moss and seedling growth. Factors depressing moss growth such as drought or factors increasing shoot growth such as enhanced nutrient acquisition (Camill et al. 2010) or species-specific differences in shoot investments may facilitate seedling survival. In contrast, factors limiting shoot growth, such as foliage loss by fungal pathogens or herbivory (Camill et al. 2010), may increase the likelihood of seedlings being overgrown by moss. Once seedlings have reached a safe height above the moss surface, root competition with shallow-rooting herbs or shrubs may likely further take its toll (Sarkkola et al. 2005). Indeed, multiple recruitment limitation (i.e. source, dispersal, germination and seedling establishment) is often responsible for the speed of vegetation change in various ecosystems (Dullinger, Dirnbock & Grabherr 2004; Acácio et al. 2007).

Our study demonstrates that the fate of tree seedlings on peat moss depends strongly on traits that allow rapid growth just after germination in the early establishment phase.

Acknowledgements

We thank Jort Bosman for his help in tending the experiment and the State Forestry Service for allowing us to collect peatmoss in one of their bog reserves. The research was made possible by a grant from the Schure-Beijerinck Popping fund to JL and MH.

Data accessibility

Data of this paper have been deposited in the Dryad repository: http://dx.doi.org/10.5061/dryad.926nd. (Limpens et al. 2013)

Ancillary