Boreal feather mosses secrete chemical signals to gain nitrogen



  • The mechanistic basis of feather moss–cyanobacteria associations, a main driver of nitrogen (N) input into boreal forests, remains unknown. Here, we studied colonization by Nostoc sp. on two feather mosses that form these associations (Pleurozium schreberi and Hylocomium splendens) and two acrocarpous mosses that do not (Dicranum polysetum and Polytrichum commune). We also determined how N availability and moss reproductive stage affects colonization, and measured N transfer from cyanobacteria to mosses.
  • The ability of mosses to induce differentiation of cyanobacterial hormogonia, and of hormogonia to then colonize mosses and re-establish a functional symbiosis was determined through microcosm experiments, microscopy and acetylene reduction assays. Nitrogen transfer between cyanobacteria and Pleurozium schreberi was monitored by secondary ion mass spectrometry (SIMS).
  • All mosses induced hormogonia differentiation but only feather mosses were subsequently colonized. Colonization on Pleurozium schreberi was enhanced during the moss reproductive phase but impaired by elevated N. Transfer of N from cyanobacteria to their host moss was observed.
  • Our results reveal that feather mosses likely secrete species-specific chemo-attractants when N-limited, which guide cyanobacteria towards them and from which they gain N. We conclude that this signalling is regulated by N demands of mosses, and serves as a control of N input into boreal forests.


Pleurocarpous feather mosses have a wide global distribution and frequently dominate the ground layer in the boreal forest, which is the Earth's largest forested biome. They can reach a biomass of up to 3 t ha−1 and can account for up to a third of total forest primary productivity (Wardle et al., 2012). They also commonly host dinitrogen (N2)-fixing cyanobacteria that are epiphytic (i.e. living on the surface of moss leaves) and which serve as the main agent of biological nitrogen (N) input into mature forests (DeLuca et al., 2002). As such, there is emerging recognition of the important role of feather mosses in the functioning of boreal forests, for example in governing nutrient fluxes, carbon sequestration and the establishment and growth of vascular plants (Turetsky et al., 2012). However, despite the ecological importance of moss–cyanobacteria associations, little is known about the intrinsic factors regulating the capacity of mosses to host cyanobacteria, or the potential benefit that they may obtain from the association. An improved knowledge of the mechanisms involved will help us better understand the means by which N enters into extensive areas worldwide in which cryptogamic species are major contributors (Elbert et al., 2012).

In recent work, we have shown that strains of the N2-fixing cyanobacterial genus Nostoc were highly host specific, with different strains associating with each of the two dominant and most widespread boreal feather mosses, that is Pleurozium schreberi and Hylocomium splendens (Ininbergs et al., 2011). This result suggests that feather mosses are highly selective towards the cyanobacterial strains that they host. Consequently, these results raise the possibility that mosses may secrete compounds that serve as chemo-attractants for cyanobacteria, and that these cyanobacteria could in turn contribute to moss nutrition by providing mosses with fixed N, in the same manner as has been shown for vascular plant–cyanobacterial symbioses (Adams & Duggan, 2012).

First, we hypothesized that colonization of feather mosses by N2-fixing cyanobacteria occurs as a result of the secretion of specific chemical signals (a hormogonia inducing factor (HIF) and a chemo-attractant) that induce hormogonia formation (i.e. transient motile filaments serving as infection units) and guide them towards the host moss. Second, we hypothesized that the secretion of the chemo-attractant is maximized when the host moss is depleted in N. Third, we also hypothesized that the N2 fixed by the cyanobacteria is transferred to their host moss. Testing these hypotheses would inform on whether feather mosses secrete chemicals to attract cyanobacteria from which they then gain a nutritional benefit.

Materials and Methods

Sampling site, moss species and cyanobacterial strain

All mosses used in this study (Hylocomium splendens (Hedw.) W.P. Schimp, Pleurozium schreberi (Brid.) Mitt., Dicranum polysetum Sw. and Polytrichum commune Hedw.) were collected in northern Sweden (65°80′N; 19°10′E) in June 2010. Hylocomium splendens and P. schreberi were used as host species (Ininbergs et al., 2011), while D. polysetum and P. commune were used as nonhost species because no or few cyanobacteria occur on them (Scheirer & Dolan, 1983). A monoculture of Nostoc sp. isolated from H. splendens was used for all the experiments.

Hormogonia induction assessment

Gametophytes of all moss species were surface-sterilized to remove all epiphytic cyanobacteria using 0.15% sodium hypochlorite for 1 min followed by seven rinses with sterile water to remove any trace of sodium hypochlorite (Misaghi & Donndelinger, 1990). To verify that all cyanobacteria had been removed, the gametophytes were observed under an epifluorescence microscope equipped with a green excitation filter (510–560 nm). One millilitre of the Nostoc sp. culture (100 mg ml−1 in BG-110 medium; Rippka et al., 1979) was added to each well of two 12-well BD Falcon plates. For each moss species, four replicates were prepared, where each gametophyte (each c. 10 mg DW) was placed in a fitting cell culture insert (1 μm membrane pore size). Only the upper 2 cm green parts of the gametophyte were used as this is where most of the cyanobacterial N2-fixation activity occurs (Solheim et al., 2004). Each insert was placed in the cyanobacteria-filled wells of the plates. With this set-up, the gametophytes were in contact with the Nostoc culture but physically separated from it because the diameter of the Nostoc filaments is 1.5–2 μm, allowing no filaments to pass through the membrane. Nostoc cultures without bryophytes present in the cell inserts were used as negative controls. The plates were placed in a growth chamber for 72 h at 24°C under constant illumination of 18 μmol photons m−2 s−1. Hormogonia formation in the culture media was measured every 24 h using a hemocytometer. The pH of the culture media was determined before each counting.

Assessment of chemo-attraction

The ability of mosses to attract cyanobacteria from the culture medium was assessed as mentioned earlier, except that an 8 μm pore size membrane was used allowing the movements of hormogonia through the membrane, and that nine gametophyte replicates of each species were used. From these nine replicates, four were incubated in darkness at 24°C to exclude the possibility that cyanobacterial colonization was due to light attraction. Cyanobacterial colonization was quantified after 8 d as the proportion ratio of the total surface area occupied by cyanobacteria, using ImageJ software v.1.45s (Rasband, 1997–2012).

To assess whether chemo-attraction of cyanobacteria by mosses depends on the N demand of the host, we established two 1 m × 1 m field plots dominated by P. schreberi in northern Sweden (see earlier) in June 2007. One of the plots was fertilized by N, added as ammonium nitrate (NH4NO3) dissolved in water, at 5 kg N ha−1 yr−1, and the other was an unfertilized control amended with equivalent amounts of water. Nitrogen and water were distributed at seven application times from June to September between 2007 and 2010. This N addition rate is consistent with rates used in previous experiments in boreal forests to study the impairment of N2-fixation by N fertilization (Gundale et al., 2011). In September 2010, five P. schreberi gametophytes were randomly collected in each of three 0.2 × 0.2 m sub-plots. The gametophytes were exposed to cyanobacteria as described earlier, and cyanobacterial colonization was assessed after 8 d of incubation.

To test if chemo-attraction of cyanobacteria by P. schreberi is affected by the reproductive costs of producing sporophytes, we collected 24 P. schreberi gametophytes with well-developed sporophytes and 12 gametophytes without sporophytes from the field site in September 2012. For 12 of the 24 gametophytes with sporophytes, the sporophytes were manually removed to reduce the reproductive costs to the gametophytes. All gametophytes were arranged in 12 replicate blocks; each block consisted of all three gametophyte types (i.e. gametophytes with sporophytes, those with sporophytes removed, and gametophytes with no sporophytes). The mosses were left in a growth chamber for 10 d to acclimatize until chemo-attraction was assessed. All gametophytes were surface-sterilized before experimentation, and assessment of chemo-attraction and N2-fixation rates was performed 8 d after experimental set-up.

N2-fixation measurements

Nitrogen-fixation activity of the gametophytes used during the different experiments was measured using the acetylene reduction assay (ARA; Hardy et al., 1968). The gametophytes were placed individually in 16-ml glass tubes fitted with rubber septa. Ten per cent of the air volume was removed from the tube and replaced with acetylene. The tubes were incubated in a growth chamber at 24°C under constant illumination of 18 μmol photons m−2 s−1 for 4 h, and measurements of ethylene levels were performed as described by Ininbergs et al. (2011). A ratio of 3 moles of ethylene per mol of N was used to convert acetylene reduction rates to ng N2 fixed g moss−1 d−1 (DeLuca et al., 2002).

Nitrogen transfer between cyanobacteria and Pleurozium schreberi gametophytes

Cyanobacteria-hosting P. schreberi gametophytes were incubated with 15N2 gas to obtain 15N-enriched samples. Sixty gametophytes were placed in 12 serum bottles (five in each) sealed with rubber caps. Gametophytes in six of these bottles were exposed to 2% 15N2 gas while those in the other six were exposed to air; these were then incubated in a growth chamber under a light regime of 167 μmol photons m−2 s−1 for 12 h d−1 at 24°C. Half the bottles (i.e. three exposed to 15N2 gas and three exposed to air) were incubated for 1 wk, and the other half were incubated for 4 wk. The incubations were stopped by adding 2% freshly prepared paraformaldehyde solution to each bottle, and the bottles were further incubated for 16 h at 4°C. The gametophytes were washed twice with phosphate-buffered saline (pH 7.4) and stored at 4°C until further processed for secondary ion mass spectrometry (SIMS) analysis (for embedding and sectioning of samples, see Supporting Information Methods S1). This analysis is both a qualitative and quantitative technique that was used in this experiment to determine the 15N2 isotopic composition on the surface of moss cross-sections. Before SIMS analysis (for both 15N2-exposed and non 15N2-exposed gametophytes), leaf areas with cyanobacteria were identified by microscopy, and coated with 5 nm gold (Au) to provide a conducting surface. Scanning ion imaging was performed using a large geometry CAMECA ims1280 SIMS instrument (Cameca SAS, Gennevilliers, France). A 133Cs+ primary beam with 20 kV impact energy was critically focused to produce a < 1 μm analytical spot on the sample, with a beam current of c. 100 pA which was rastered over an area of 80 × 80 μm. Following the removal of the Au-coat from the area to be analysed by using a larger rastered primary beam, the secondary ion species of 12C14N and 13C15N were sequentially measured in an ion counting electron multiplier at a mass resolution (MM) of 5000. The dynamic transfer optical system was utilized to enable reconstruction of the spatial distribution of the measured ion species within the rastered sampling area. Each spot analysed consisted of 120 cycles through the masses 12C14N, 12C15N and 13C14N (1, 5 and 2 s integration time, respectively). Image processing of the analysed spots was performed using CAMECA Winimage software to generate accumulated ion maps of 15N and 14N. Calculation of the 15N/14N isotopic ratio of moss and cyanobacterial cells was performed by using high resolution spectra from the control (non 15N2-exposed) mosses and the 15N2-exposed mosses at mass peaks 26 and 27, respectively, to determine the major mass of 12C14N and 12C15N of the chosen area. The mean 15N/14N ratios of non 15N2-exposed samples were used to correct the 15N/14N values of 15N2-exposed samples from background noise. Due to the heterogeneity of resin-embedded samples, the total number of areas (n) analysed for 15N/14N ratio among moss and cyanobacterial tissues is not equal.

Statistical analysis

The ability of the bryophytes to induce hormogonia formation was tested in a linear mixed effects model using a Gaussian error, where the fixed effect was the moss treatment (i.e. each of the four moss species and the moss-free control) and the random effect was the time of exposure. The effect of time on hormogonia formation was tested by analysis of variance (ANOVA) for each time point, on Box-Cox transformed data. The effects of moss species, N amendment and sporophyte presence on cyanobacterial colonization and N2-fixation were each assessed by ANOVA, on Box-Cox transformed data. Tukey's HSD (honestly significant difference) post hoc tests were subsequently performed to investigate significant differences between moss species, and between treatments (i.e. N amendment and sporophyte presence) at = 0.05. The effect of 15N2 exposure on 15N2 enrichment in gametophytes was assessed by a nonparametric Wilcoxon test. All statistical analyses were carried out using R software v.2.13.1 (R Foundation for Statistical Computing, Vienna, Austria).


Hormogonia induction assessment

All four bryophyte species induced hormogonia differentiation in Nostoc sp. cultures after 24 h of incubation (Fig. 1). After 48 h, more than half of the cyanobacterial filaments in cultures had formed hormogonia when exposed to mosses, whereas only 9% of the filaments formed hormogonia in the absence of mosses. Depending on the moss species, hormogonia differentiation decreased or levelled out when cyanobacteria were exposed to mosses for longer than 48 h (Fig. 1). Measurements of the pH within the cell culture medium revealed a slight acidification of the medium over time regardless of the moss species (Fig. 1).

Figure 1.

Hormogonial formation of Nostoc sp. culture (percentage of total number of filaments) exposed to Pleurozium schreberi, Hylocomium splendens, Polytrichum commune and Dicranum polysetum over 72 h (mean ± SEM; data points accompanied by different letters are statistically significant at = 0.05 according to a linear mixed effects model combined to ANOVAs for each time point, followed by Tukey's test; = 4 each data point, df = 4 for each ANOVA), and pH variation in the cell culture inserts during the experiment (mean ± SEM; = 4 each data point). Nostoc sp. culture without moss was used as the control.

Assessment of chemo-attraction

After 8 d of incubation, H. splendens and P. schreberi shoots were abundantly colonized by Nostoc sp., whereas P. commune and D. polysetum remained uncolonized (Fig. 2a,d). Both H. splendens and P. schreberi were colonized equally well in dark and light conditions (data not shown). The cyanobacteria that colonized H. splendens and P. schreberi fixed N2 as confirmed by ARA (Fig. 2a). Moreover, P. schreberi shoots from N-amended plots were not colonized by Nostoc sp. (Fig. 2b,d). Additionally, we found that Nostoc sp. was more attracted towards sporophyte-bearing gametophytes and gametophytes from which sporophytes had been removed, than to gametophytes without sporophytes (Fig. 2c).

Figure 2.

Degree of cyanobacterial colonization (closed bars) and N2-fixation rates (open bars) of (a) Pleurozium schreberi, Hylocomium splendens, Polytrichum commune and Dicranum polysetum; (b) P. schreberi fertilized with N and without N addition; and (c) P. schreberi with and without sporophytes, and sporophytes removed. All data are collected after 8 d of incubation. For each panel, closed bars topped with different upper case letters indicate significant differences among treatments in cyanobacterial colonization, and open bars topped with different lower case letters indicate significant differences among treatments in N2-fixation (ANOVA followed by Tukey's HSD test, < 0.05). Statistical analyses performed on Box-Cox transformed data. Values are means ± SEM (= 5 a and b; = 12 for c). (d) Upper four photographs correspond to the experiment in (a) and show the degree of colonization of each of the four moss species by Nostoc sp. (in bright orange); lower two photographs correspond to the experiment in (b) and show the degree of colonization of N-amended and non N-amended P. schreberi by Nostoc sp. Degrees of freedom (df) for both N2-fixation and cyanobacterial colonization are: (a) df = 3; (b) df = 1; (c) df = 2.

Nitrogen transfer between cyanobacteria and P. schreberi gametophytes

The 15N2 fixed by the associated cyanobacteria was transferred to the host and accumulated within the moss tissue over time (Table 1, see Supporting Information Table S1 for values of δ15N‰ of 15N2-exposed and non 15N2-exposed cyanobacteria associated with P. schreberi). Even 1 wk after exposure, a significantly higher 15N/14N ratio was observed in the host tissue of 15N2-exposed gametophytes than in nonexposed gametophytes. This ratio corresponds to a 4.1% 15N enrichment when compared with natural abundance measured in nonexposed P. schreberi gametophyte tissues. After 4 wk of exposure, the enrichment in the moss tissue increased to 34.4% (Table 1). The cyanobacteria associated with P. schreberi also showed an enrichment of 15N when exposed to 15N2 gas (Table 1, Fig. 3). Contrary to the moss tissue, the enrichment of 15N in cyanobacterial cells did not significantly increase with time of exposure.

Table 1. 15N/14N isotopic ratios (mean ± SEM) of 15N2-exposed and control (non 15N2-exposed) Pleurozium schreberi gametophyte cells and their associated cyanobacteria
Exposure timeExposed to 15N2ControlP-Value
  1. Data are obtained from secondary ion mass spectrometry (SIMS) analyses 1 and 4 wk after initiation of 15N2-exposure.

  2. The number of spots (n) analysed with SIMS is given for each sample. P-Values indicate significant differences between cells exposed to 15N2 and control cells, according to a nonparametric Wilcoxon rank sum test.

  3. a

    Significant enrichment of 15N between P. schreberi cells after 1 and 4 wk of exposure (= 1.096e−05).

P. schreberi cells
1 wk

3.85e−03 ± 3.77e−03

(= 13)

3.65e−03 ± 3.02e−05

(= 13)

4 wk

4.97e−03 ± 2.16e−04 a

(= 65)

3.73e−03 ± 8.05e−05

(= 31)

Cyanobacterial cells
1 wk

7.76e−03 ± 3.07e−03

(= 8)

3.69e−03 ± 1.04e−04

(= 60)

4 wk

7.47e−03 ± 2.65e−03

(= 185)

3.73e−03 ± 8.05e−05

(= 21)

Figure 3.

Cross-sections of a Pleurozium schreberi gametophyte with cyanobacteria after 4 wk of 15N2 exposure. (a) Light microscopy image with a red box indicating the area subsequently analysed by secondary ion mass spectrometry (SIMS). (b) Close-up SIMS images of moss tissue (11–13, black arrows) and cyanobacterial cells (4–10 and 14, blue arrows) chosen for 15N/14N ratio measurements. The boxes (1–3, white arrows) to the left depict background areas. (c) Colour gradient showing the distribution of 15N-enrichment within moss and cyanobacterial cells.


Our aim was to shed light on the underlying mechanisms of N input into boreal ecosystems; more specifically, on which factors determine if a moss is a suitable host for cyanobacteria and what the fate is of N fixed by the associative cyanobacteria. Our results showed that all four mosses promoted hormogonia differentiation of Nostoc sp. after 24 h of incubation (Fig. 1), which supports our first hypothesis. Hormogonia are induced by a variety of environmental stimuli, including chemical signals (HIFs) released by hosts (Campbell & Meeks, 1989; Rasmussen et al., 1994). Therefore, we conclude that these mosses secrete a HIF in a similar manner to Gunnera or Anthoceros (Campbell & Meeks, 1989; Rasmussen et al., 1994). Moreover, the observed decrease or plateau after 72 h of incubation (Fig. 1) is in accordance with previous findings by Meeks & Elhai (2002), which showed that Nostoc hormogonia cease to move and revert to vegetative filaments after 48 h. Additionally, the acidification of the medium during the experiment did not appear to have any effect on hormogonia differentiation and suggests that the HIF secreted by each of the moss species is strong enough to override the possible negative effects of the pH on hormogonia differentiation (Hirose, 1987; Rasmussen et al., 1994).

Our results also showed that H. splendens and P. schreberi were the only mosses colonized by Nostoc sp. (Fig. 2a,d). This demonstrates that hormogonia are exclusively attracted towards these feather mosses due to the probable production of host-specific chemical compounds that would induce chemotactic responses in cyanobacteria, and therefore also provides support for our first hypothesis. Additionally, the cyanobacteria which colonized P. schreberi and H. splendens fixed N2 (Fig. 2a), suggesting that a functional symbiosis was formed for both of these mosses after colonization. While this study was not aimed at identifying the precise molecular structures underpinning why H. splendens and P. schreberi but not D. polysetum and P. commune become colonized by Nostoc sp., our results help to explain why D. polysetum and P. commune remain free of cyanobacteria in natural conditions. Thus, we conclude that although hormogonia differentiation is a crucial step in the cyanobacterial colonization of host plants, for successful colonization to occur the plant must also attract and guide the hormogonia towards suitable colonization sites (i.e. leaves).

In support of our second hypothesis, we found that P. schreberi from N-amended plots remained uncolonized by Nostoc sp., indicating that the moss ceases to secrete chemo-attractants under elevated N conditions (Fig. 2b,d). Furthermore, we found that the reproductive status of the moss positively impacted cyanobacterial colonization and plausible chemo-attractant secretion (Fig. 2c). This result also is consistent with sporophyte formation being a N-demanding process (Rydgren & Økland, 2002) and with sporophytes serving as N sinks (Renault et al., 1989). Moreover, the removal of sporophytes from the gametophytes probably induced changes in the source-sink relationship between these two components, creating a reduced N demand, and explains why gametophytes with sporophytes removed had intermediate colonization rates (Fig. 2c). Thus, P. schreberi is likely to allocate resources to the production of chemo-attractants only under N-limited conditions or when the moss has a greater need for N (e.g. during sexual reproduction). Source–sink relationships between N2-fixation and flowering have been widely reported for legumes (Sprent, 2009), and our results suggest that a similar phenomenon involving active production of chemo-attractants may apply for boreal forest feather mosses. Sporophyte development is seasonal (Longton & Greene, 1969) and these results may also help to explain the previously observed seasonality of feather mosses N2-fixation patterns (DeLuca et al., 2002).

Finally, a transfer of N was observed between cyanobacteria and P. schreberi when gametophytes were exposed to 15N2 gas (Fig. 3). Thus, feather mosses acquire N from their associated cyanobacteria, which eventually accumulates in moss tissues over time (Table 1), and therefore provides support for our third hypothesis. Although the levels of 15N within the cyanobacteria cells also increased throughout the experiment, no accumulation was observed with time of exposure (Table 1), most likely because the N fixed by the cyanobacteria is instead transferred to their host moss.

Moss–cyanobacteria associations are widespread in boreal forest, tundra and sub-Arctic ecosystems (Turetsky, 2003; Lindo & Gonzalez, 2010), and frequently represent the main form of biological N input into N-limited ecosystems. However, we have limited understanding of how N2-fixing cyanobacteria colonize mosses, and whether this fixed N becomes available to the moss and ultimately to the rest of the ecosystem (Gavazov et al., 2010). It has been suggested that cyanobacterial colonization of host mosses is a random phenomenon due to mechanical means, for example wind (Marshall & Chalmers, 1997). Our findings show that the initiation of associations between boreal feather mosses and hormogonia-forming cyanobacteria is principally directed by the mosses, and that the mosses regulate the degree of cyanobacterial colonization based on their need for N. Our findings further suggest that the ability of feather mosses to occupy N-limited ecosystems has evolved through chemical communication with cyanobacteria, and that cyanobacteria fulfil important functions for them through the provision of fixed N2. Ultimately, these findings reveal that the feather moss–cyanobacteria association operates as a functional symbiotic relationship driven by chemical signalling that supports the N economy of feather mosses, and refines our understanding of how feather mosses govern N input into boreal forests.


Funding was provided by the Carl Trygger Foundation for Scientific Research and Stiftelsen Oscar och Lili Lamms Minne. The authors are grateful to H. Ploug and S. Lindwall for technical assistance. The Nordsims facility is financed and operated under an agreement between the Nordic countries; this is Nordsims contribution no. 343.