SEARCH

SEARCH BY CITATION

Keywords:

  • boreal forest;
  • cyanobacteria;
  • feather mosses;
  • nifH diversity;
  • nitrogen fixation

Summary

  1. Top of page
  2. Summary
  3. Introduction
  4. Materials and Methods
  5. Results
  6. Discussion
  7. Acknowledgements
  8. References
  • Recent studies have revealed that nitrogen fixation by cyanobacteria living in association with feather mosses is a major input of nitrogen to boreal forests. We characterized the community composition and diversity of cyanobacterial nifH phylotypes associated with each of two feather moss species (Pleurozium schreberi and Hylocomium splendens) on each of 30 lake islands varying in ecosystem properties in northern Sweden.
  • Nitrogen fixation was measured using acetylene reduction, and nifH sequences were amplified using general and cyanobacterial selective primers, separated and analyzed using density gradient gel electrophoresis (DGGE) or cloning, and further sequenced for phylogenetic analyses.
  • Analyses of DGGE fingerprinting patterns revealed two host-specific clusters (one for each moss species), and sequence analysis showed five clusters of nifH phylotypes originating from heterocystous cyanobacteria. For H. splendens only, N2 fixation was related to both nifH composition and diversity among islands.
  • We demonstrated that the cyanobacterial communities associated with feather mosses show a high degree of host specificity. However, phylotype composition and diversity, and nitrogen fixation, did not differ among groups of islands that varied greatly in their availability of resources. These results suggest that moss species identity, but not extrinsic environmental conditions, serves as the primary determinant of nitrogen-fixing cyanobacterial communities that inhabit mosses.

Introduction

  1. Top of page
  2. Summary
  3. Introduction
  4. Materials and Methods
  5. Results
  6. Discussion
  7. Acknowledgements
  8. References

Diazotrophic, or nitrogen-fixing, bacteria represent the largest input of nonanthropogenic nitrogen (N) into most of the Earth’s terrestrial and aquatic ecosystems (Vitousek et al., 1997; Canfield et al., 2010). There is also increasing recognition that several cyanobacterial species are important contributors to N input in many terrestrial and aquatic ecosystems world-wide (Rai et al., 2002; Stal & Zehr, 2008). These cyanobacteria can function either as free-living bacteria or in associations with vascular plants, fungi, diatoms, marine sponges, corals and bryophytes (Rai et al., 2002; Lesser et al., 2004; Usher et al., 2004). Over the past decade, evidence has accumulated that N2 fixation by cyanobacteria living in association with feather mosses is a major source of N input to boreal forest ecosystems (DeLuca et al., 2002; Markham, 2009), particularly those of late-successional status (Zackrisson et al., 2004; Lagerström et al., 2007). Nitrogen-fixing cyanobacteria have been shown to be hosted by several species of feather mosses, including Hylocomium splendens, Pleurozium schreberi, Ptilium crista-castrensis and Sphagnum caprifolium (DeLuca et al., 2002; Solheim & Zielke, 2002; Houle et al., 2006; Lagerström et al., 2007; Markham, 2009). Nitrogen fixation by cyanobacterial associations with feather mosses may be of considerable significance to the global N cycle, given that boreal forests occupy 11% of the Earth’s terrestrial surface (Bonan & Shugart, 1989) and store more organic matter than any other terrestrial biome (Anderson, 1991).

Little is known about the composition and diversity of the cyanobacterial communities that are associated with feather mosses, or how the rate of N2 fixation through this association is dependent upon cyanobacterial community properties. The cyanobacteria in these associations are known to belong to the heterocystous genera Nostoc, Stigonema and Calothrix, but have so far only been identified morphologically or by using culture-based molecular techniques at a coarse level (Gentili et al., 2005; Houle et al., 2006; Zackrisson et al., 2009). The genetic variability and composition of these cyanobacterial communities, and how N2-fixation rates within the feather moss environment are influenced by cyanobacterial community characteristics, remain entirely unexplored. Genetic inventories in other systems, such as marine microbial mats and stromatolites, have revealed highly diverse diazotrophic communities composed of both cyanobacteria and heterotrophic bacteria (Steppe et al., 2001; Omoregie et al., 2004; Yannarell et al., 2006; Bauer et al., 2008; Severin & Stal, 2010). Similarly, diverse bacterial communities have also been observed in association with seagrasses (Uku et al., 2007; Hamisi, 2010). Based on these previous studies, there are strong grounds for predicting that cyanobacterial communities associated with feather mosses may also have a level of diversity that is well in excess of what has been revealed to date using morphological and culture-based techniques (Gentili et al., 2005).

Our study focused on characterizing the diversity of the nifH gene in N2-fixing cyanobacterial communities associated with each of two feather moss species (P. schreberi and H. splendens) on each of 30 contrasting forested lake islands in northern Sweden. These islands have previously been shown to vary greatly in soil fertility, decomposer processes and plant productivity (Wardle et al., 1997, 2003), and in the amounts of N2 fixation by cyanobacteria on both moss species (Lagerström et al., 2007; Gundale et al., 2009). We chose to focus on the nifH gene because of its key role in the N2-fixation process through coding for a structural component (Fe-protein) of the nitrogenase enzyme, and because of its widespread use as a molecular marker for diazotrophs. Specifically we sought to test two hypotheses. First, we hypothesized that both diversity (i.e. richness) and community composition (i.e. types and relative abundances) of nifH genes will be responsive to both island size and moss species, because factors that may affect the occurrence of N2-fixing cyanobacteria vary both across islands (i.e. nutrient availability) and across moss species (in terms of the microhabitat that they provide for cyanobacteria). Secondly, we hypothesized that the composition and diversity of nifH genes would serve as a powerful predictor of N2 fixation across the 30 islands. In testing these hypotheses, our ultimate goal was to develop our understanding of genetic diversity among boreal forests ecosystems for a key group of microorganisms that perform a unique role in the functioning of these forests through governing ecosystem N inputs.

Materials and Methods

  1. Top of page
  2. Summary
  3. Introduction
  4. Materials and Methods
  5. Results
  6. Discussion
  7. Acknowledgements
  8. References

Sampling sites and moss sampling

The study system consisted of 30 forested islands in two adjacent Swedish lakes situated in the northern boreal zone, Lake Hornavan and Lake Uddjaure (65°55′N to 66°09′N and 17°43′E to 17°55′E). All islands were formed following the retreat of land ice c. 9000 yr ago. The mean annual precipitation is 750 mm and the mean temperature is +13°C in July and −14°C in January. The only major extrinsic factor that varies among islands is the history of lightning-ignited wildfire, with larger islands having burned more frequently than smaller islands because of their larger area for intercepting lightning (Wardle et al., 1997, 2003). Previous studies on these islands have shown that with decreasing size and increasing time since fire, they increasingly enter a state of ‘ecosystem retrogression’ (Peltzer et al., 2010) in which there is a reduction in soil fertility, plant biomass and ecosystem productivity (Wardle et al., 1997, 2003, 2004). The islands were divided into three size classes with 10 islands per class, that is, large (> 1.0 ha), medium (0.1–1.0 ha) and small (< 0.1 ha), in accordance with previous studies on these islands (Wardle et al., 2003; Wardle & Zackrisson, 2005). The mean time since the last major fire on these islands is 585, 2180 and 3250 yr on the large, medium and small islands, respectively. Vegetation composition varies greatly across these size classes; the tree vegetation is dominated by Pinus sylvestris, Betula pubescens and Picea abies on large, medium and small islands, respectively, while the dwarf shrub layer is dominated by Vaccinium myrtillus, Vaccinium vitis-idaea and Empetrum hermaphroditum on large, medium and small islands, respectively (Wardle et al., 1997, 2003). The size classes also differ greatly in soil fertility, with the lowest nutrient availability on the small islands (Wardle et al., 1997; Lagerström et al., 2009). The feather mosses Pleurozium schreberi (Bridel) Mitten and Hylocomium splendens (Hedwig) Schimper are common on all islands, and previous work has shown that the associative N2 fixation for both moss species varies greatly among islands as a result of local environmental conditions (Lagerström et al., 2007; Gundale et al., 2010).

Mosses were collected in June 2009, which corresponds to the first main seasonal peak for N2 fixation associated with feather mosses (DeLuca et al., 2002). For each island, we collected 30 shoots of P. schreberi and 15 shoots of H. splendens (which are generally much larger than P. schreberi shoots) in a grid pattern from a 20 × 20 m sampling area as described by Lagerström et al. (2007). In this grid we identified 10 sampling points for collecting P. schreberi shoots and five points for collecting H. splendens shoots, and in the vicinity of each point we randomly selected three individual shoots. Sampling points located < 2 m from overstory trees or in dense shrub vegetation were not used. The sampling area was always located at similar distances from the shore regardless of island size to prevent edge and microclimatic effects from confounding the results (Wardle et al., 1997, 2003). Only the upper part (i.e. the first 5 cm) of each shoot was used. After weighing, these shoots were placed into six 22-ml glass vials with 2 ml of deionized water, that is, three vials containing 10 P. schreberi shoots and three containing five H. splendens shoots (Gundale et al., 2010). Thus, the estimated weights for our samples were similar regardless of the moss species (i.e. mean ± SE of 0.60 ± 0.05 mg weight basis). Before N2-fixation analysis, all vials were placed unclosed and upside-down into the moss layer at a natural forest site near the laboratory in Umeå, and allowed to acclimatize under natural conditions for 72 h.

Nitrogen-fixation measurements

Nitrogen-fixation rates were estimated using the acetylene reduction assay (ARA) (Hardy et al., 1968), as described by DeLuca et al. (2002). Before the incubation, mosses were misted with deionized water and each tube was fitted with a septum. Once sealed, 10% of the total headspace of each vial was evacuated and replaced with acetylene. Mosses were then incubated at the same natural forest site in Umeå (see sampling sites and moss sampling), and samples were analyzed for total ethylene production after 24 h, on a Perkin Elmer Clarus 500 GC, with Turbomatrix 40 headspace injector (Perkin Elmer, Waltham, MA, USA). Two sets of three vials were also included as negative controls; one set with the mosses and no acetylene to assess the endogenous moss ethylene production, and the other set of vials with 2 ml of deionized water and acetylene to assess the concentration of ethylene in the acetylene used; ethylene concentrations in both controls were confirmed to be negligible. All samples were then placed at −80°C directly after the N2-fixation measurement, before extraction of DNA. Ethylene values were converted to micromoles N per day per gram of moss on a fresh weight basis (following drying at room temperature) using the ideal gas law. The values obtained were converted to micrograms of N2 fixed per day per gram of moss using a ratio of 3 mol of ethylene reduced per mol of N. This ratio has previously been confirmed by 15N labeling to be appropriate for both P. schreberi and H. splendens (DeLuca et al., 2002; Zackrisson et al., 2004; Lagerström et al., 2007).

Culturing and morphological identification of cyanobacteria

Thirty random shoots of both moss species were placed on agar (0.9%) plates prepared with BG110 medium (Rippka et al., 1979). The plates were incubated for 1 month in a growth chamber at 24°C with a constant light regime of 18 μmol photons m−2 s−1. Single colonies were transferred to BG110 medium and used as internal standards in the density gradient gel electrophoresis (DGGE) analysis. Isolated strains were observed under an Olympus BX 60 F5 epifluorescence microscope (Olympus Optical Co., Ltd, Tokyo, Japan) with a green filter (510–560 nm), and pictures were recorded with an Olympus DP 50 digital camera (Olympus Optical Co. Ltd).

DNA extraction

For each sample (5 (for H. splendens) or 10 (for P. schreberi) moss shoots from each of the three acetylene reduction assay incubations were pooled together), DNA extraction was initiated by freezing the sample in liquid N2 and grinding moss shoots carefully until a fine and homogeneous powder was obtained, of which 50 mg was mixed with 500 μl of extraction buffer (3.1 mM potassium ethyl xanthogenate, 40 mM ammonium acetate, 5 mM Tris-HCl, pH 7.4, and 0.81 mM EDTA, modified from Tillett & Neilan, 2000). The mixture was then subjected to three cycles of mechanical cell lysis treatment (3 × speed 5 for 30 s with 1 min cooling on ice between cycles; Bio 101 Thermo Electron Corporation FastPrep FP 120 homogenizer; San Jose, CA USA). To assist cell lysis, acid-washed glass beads were added to each sample (0.4 g of 212–300 μm beads and 0.4 g of 425–600 μm beads; Sigma-Aldrich Co., St. Louis, MO, USA). Samples were incubated at 65°C for 2 h with 50 μl of 10% SDS, vortexed for 5 s and placed on ice for 30 min. Each sample was purified twice by mixing with 500 μl of phenol:chloroform:isoamyl alcohol and once with 500 μl of chloroform:isoamyl alcohol, vortexed for 5 s and centrifuged for 10 min at 16 000 g between each treatment. After each purification treatment, the supernatant was transferred to a new, sterile 1.5-ml microtube. DNA was precipitated with 500 μl of sterile absolute isopropanol and 50 μl of sterile ammonium acetate (4 M) at −20°C for 30 min. The precipitated DNA was centrifugated at 16 000 g for 20 min. The DNA pellet was washed with 70% ethanol, air-dried for 15–20 min and re-suspended in 50 μl of sterile TE buffer (1 mM EDTA and 10 mM Tris-HCl, pH 8.0). Samples were kept at −20°C until further analysis. In addition, DNA from Nostoc punctiforme PCC73102 and Anabaena PCC7120 was used as internal controls, in order to ensure the primer affinity to multiple nifH copies within their genomes. As no nifH sequence from the genus Stigonema is available in public databases, DNA extraction was also performed on Stigonema ocellatum NIES2131, as described above.

Polymerase chain reaction (PCR) amplification and cloning

PCRs were performed using an Eppendorf Mastercycler EPgradient S thermal cycler (Eppendorf, Hamburg, Germany) and Phire hot start II DNA polymerase (Finnzymes OY, Espoo, Finland). nifH genes were amplified using the cyanobacterial selective CNF/CNR primer set (Olson et al., 1998), with an additional GC clamp at the 5′ end (Díez et al., 2007). Reaction mixture and conditions were set up according to the manufacturer’s instructions with 32 amplification cycles and an annealing temperature of 55°C. PCR amplifications were performed three times and products were pooled to reduce PCR biases. After DGGE analysis, eluted DNA from excised bands was re-amplified with the same primers and conditions described above for nifH gene amplification, and subjected to a three-cycle PCR amplification to obtain heteroduplex-free samples (Thompson et al., 2002). Eluted DNA of excised bands from the second DGGE analysis was re-amplified using the same primers (without the GC clamp) and conditions as above for nifH gene amplification. This was followed by purification of the PCR products using the QIAquick PCR purification kit (Qiagen GmbH, Germany). Purities and concentrations were assessed using a spectrophotometer (NanoDrop Technologies, Inc., Wilmington, DE, USA). Purified samples were sequenced using the CNF primer (GATC Biotech, Konstanz, Germany).

As a complement to the DGGE analysis, one sample of H. splendens was selected for nifH gene cloning. We chose a sample from a representative small island (S1) which we knew had a high N2-fixation rate, and chose a sample of H. splendens because this species characteristically hosts a higher diversity of phylotypes. General nifH primers PolF/PolR (Poly et al., 2001) were used to amplify a fragment overlapping the CNF/CNR fragment. PCR was performed as previously described (Bauer et al., 2008), but with an annealing temperature of 55°C and 30 amplification cycles. Cloning was performed using a commercial kit (TOPO TA cloning; Invitrogen, USA) according to the manufacturer’s instructions. Plasmids were extracted from transformed colonies using a commercial kit (QIAprep spin miniprep kit; Qiagen, Germany). Eight plasmids (concentration 50–100 ng μl−1) were sequenced using the M13 reverse primer (Eurofins, Ebersberg, Germany).

Denaturing gradient gel electrophoresis

In order to separate partially amplified nifH genes of different sequences, DGGE analyses were carried out as described by Bauer et al. (2008) with a denaturing gradient of 35–45%. To enable comparison of the gels, a ladder made of four nifH gene fragments covering the whole dispersion of the expected bands was loaded onto each gel and used as markers on the left-hand, middle and right-hand lanes. The reproducibility of the DGGE patterns was confirmed by performing DGGE two or three times. A second DGGE analysis was performed with re-amplified DNA of the excised bands from the previous DGGE analysis under the same conditions.

Fingerprinting analysis

To assess the similarity between moss-associated cyanobacterial communities of the different islands based on the presence/absence of nifH phylotypes, we used Fingerprinting II 3.0 (Bio-Rad Laboratories, USA). After normalizing and aligning all DGGE gels obtained according to the ladder, prominent and sharp bands were manually assigned in the program. A binary matrix, representing the presence/absence of bands in each sample, was then calculated by performing band matching with a position tolerance and an optimization of 1% each. Finally, to cluster the different band patterns obtained previously, a dendrogram based on the binary band matching table was constructed using Dice’s similarity coefficient (Dice, 1945) and the unweighted pair group method using arithmetic averages (UPGMA). The feature ‘standardized characters’ was enabled to permit all characters to have an equal influence on the similarity. A database was also created where each sample (DGGE fingerprinting pattern) is represented by an entry and will be used as a comparison tool in future work.

Sequence analysis

Sequences retrieved from DGGE and cloning were subjected to BLASTN 2.2.24 + searches (Altschul et al., 1997). Close matches as well as representatives from other representative orders and phyla were included in further analysis. In addition to GenBank sequences, additional nifH sequences were downloaded from complete cyanobacterial genomes available from Integrated Microbial Genomes (IMG; http://img.jgi.doe.gov/). All nifH copies from genome-sequenced heterocystous cyanobacteria were included in an attempt to identify nifH gene copies in our samples. In addition to GenBank and IMG sequences, reference sequences from Stigonema ocellatum NIES2131, isolated cultures from the mosses (Nostoc sp. 1 and Cylindrospermum sp.) and reference strains (Nostoc punctiforme PCC73102 and Anabaena PCC7120) were included in the phylogenetic analysis to ascertain we were not retrieving artificial products. Nucleotide sequences were aligned in geneious, using the geneious algorithm, and a phylogenetic tree was inferred using FastTree (Price et al., 2010). Sequences generated in this study are deposited in GenBank (accession numbers JF827722JF827766).

Data analysis

To summarize the nifH gene compositional data in a smaller number of variables, principal components analysis (PCA) was used to derive the primary and secondary ordination axes (i.e. those axes that maximize variation among the samples in terms of community composition) across the 60 samples (2 moss species × 30 islands) so that the significance of effects of island size and moss species could be directly evaluated and compared. The two principal axes, PC1 and PC2, explained 26.0 and 20.2% of the total variation, respectively, in the entire data set. Split-plot ANOVA was used to assess the effects of island size, moss species and their interaction on each response variable (N2 fixation, number of bands (i.e. band diversity), and PC1 and PC2), with island size as the main plot factor and moss species as the subplot factor. Pearson’s correlation coefficient (r) was then used to determine the relationship for each moss species between N2 fixation and the number of bands and ordination scores for the band data, and thus whether N2 fixation could be predicted from nifH gene diversity and composition. All data were transformed as necessary to satisfy the assumptions of parametric data analysis.

Results

  1. Top of page
  2. Summary
  3. Introduction
  4. Materials and Methods
  5. Results
  6. Discussion
  7. Acknowledgements
  8. References

Morphological identification of cyanobacteria

Two genera of cyanobacteria were successfully isolated and identified by light microscopy from both P. schreberi and H. splendens; Nostoc and Cylindrospermum (Fig. 1a–d). Three different morphotypes were observed for Nostoc that varied noticeably in cell shape and size. A third genus, Stigonema sp. (Fig. 1e), was observed on both moss species but was uncultivable. In addition, all filamentous cyanobacteria observed contained heterocysts.

image

Figure 1. Cyanobacterial species on the mosses Pleurozium schreberi and Hylocomium splendens observed under ultraviolet fluorescence with a green filter: Nostoc sp. 1 (a), Nostoc sp. 2 (b), Nostoc sp. 3 (c), Stigonema sp. (d) and Cylindrospermum sp. (e). Bars, 25 μm.

Download figure to PowerPoint

Fingerprinting (DGGE) analysis

The binary band matching table (expressing the presence or absence of bands in the different samples) revealed a total of 35 distinct bands (Fig. 2). Thirty-three partial nifH sequences were retrieved after excision and PCR re-amplification of the predominant DGGE bands. There were large variations in the number of phylotypes found for each moss sample, with the number of bands ranging from one to 14. Out of the 35 unique bands obtained, only 12 (or 34%) were shared by both mosses (Fig. 2). Six phylotypes were specific to P. shreberi and 17 to H. splendens. Thus, phylotype diversity was greater for H. splendens (mean number of phylotypes = 7.8) than for P. schreberi (mean number of phylotypes = 5.7) across all island size classes (Fig. 3). We could not detect any pattern in the distribution of phylotypes with regard to island size, and the majority of phylotypes (21 out of 35) were detected in all island size classes. Moss species, but not island size, had a statistically significant effect on the number of phylotypes present in the moss samples, as well as both primary and secondary ordination score values obtained by PCA (Table 1).

image

Figure 2. Binary band matching table and dendrogram of the cyanobacterial community present in the moss samples generated from density gradient gel electrophoresis (DGGE) profiles. Band positions are indicated as a percentage of the total gel length. Support values are indicated for each branch. Sample IDs represent the different islands (S, small; M, medium; L, large; 1–10 of each).

Download figure to PowerPoint

image

Figure 3. Average number of nifH phylotypes for Pleurozium schreberi and Hylocomium splendens on islands of three different size classes (small, gray columns; medium, black columns; large, white columns; = 10 of each). Error bars, ± SE (= 10).

Download figure to PowerPoint

Table 1.   Results from split-plot ANOVA testing for the effects of moss species and island size on nitrogen (N) fixation, number of bands, and the first and second principal component axis scores (PC1 and PC2) for the bands
Response variableMoss species (M)Island size (I)M × I
  1. Data are F-values with P-values in brackets.

  2. Degrees of freedom: M, 1,27; I, 2,18; M × I, 2,27.

  3. *Analysis on rank-transformed data.

N fixation1.2 (0.294)0.3 (0.723)0.1 (0.939)
Number of bands9.6 (0.004)0.1 (0.907)0.0 (0.960)
PC1*173.0 (< 0.001)1.4 (0.269)1.8 (0.191)
PC2*34.0 (< 0.001)0.3 (0.740)0.2 (0.852)

Based on the binary band matching table, a dendrogram was generated (Fig. 2). It showed that the nifH fingerprinting patterns grouped in two distinct clusters: one ‘Hylocomium cluster’ and one ‘Pleurozium cluster’. A fingerprinting pattern from one P. schreberi sample did not cluster with any other sample and remained as a distinct branch, as a result of possessing only a single phylotype. This may have been caused by the omission of weak bands in the corresponding DGGE profile during the calculation of the binary band matching table. With a few exceptions, the branches’ supporting values were ≥ 70%, which is the minimum value for well-supported branching.

Sequence analysis

The sequence analysis revealed five distinct clusters with nifH sequences generated in this study (Fig. 4). All clusters contained sequences originating from heterocystous cyanobacteria, including the true-branching Stigonematales. One cluster included the culture strain of Stigonema and contained nifH sequences from our samples that were identified both by DGGE and by cloning. According to the pattern analysis, this cluster contained some of the most commonly occurring phylotypes from both moss species (i.e. bands D, J, 13, 14 and 6 in Fig. 2). All the other clusters contained nifH sequences that were most closely related to different Nostoc species. One cluster (nifH2) contained sequences that we know to be comprised of nifH copies (i.e. nifH2 from various cyanobacterial genera), on the basis of sequence similarity with nifH copies located outside the nif operon in complete genomes of heterocystous cyanobacteria. One cluster contained only one, abundant phylotype (band P) that represents a heterocystous cyanobacterium which is most closely related to the symbiont of the water fern Azolla but whose genus is still inconclusive (Baker et al., 2003). The use of general nifH primers and cloning did not result in any cluster that was not also retrieved using cyanobacterial selective primers and DGGE, suggesting that no putative diazotrophs other than cyanobacteria are abundant in this association.

image

Figure 4. Approximately-maximum-likelihood phylogenetic tree from alignments of partial nifH genes amplified directly from Hylocomium splendens and Pleurozium schreberi samples (color coded), reference strains from isolated cultures and culture strains, and sequences from public databases (National Center for Biotechnology Information (NCBI) and Integrated Microbial Genomes (IMG)). Presence on the three different island sizes (small, squares; medium, circles; large, stars) is indicated using symbols. The tree was inferred using FastTree.

Download figure to PowerPoint

N2 fixation and relationship with phylotype community properties

Nitrogen-fixation rates were highly variable within the two moss species and among islands, and ranged from c. 0.3 to 34.4 μg N g FW moss−1 d−1 for P. schreberi and from 0.5 to 18.4 μg N g FW moss−1 d−1 for H. splendens (Fig. 5). All samples exhibited nitrogenase activity, confirming that the moss–cyanobacteria association was not only present (as shown by the molecular data) but also actively fixing N2 in all samples from all islands. Neither island size nor moss species had a statistically significant effect on N2-fixation rates (Table 1) and could therefore not explain the large variations in fixation rates between samples. The N2-fixation rates for H. splendens exhibited a statistically significant negative correlation with both the number of phylotypes and the primary ordination score (representing phylotype community structure) (Fig. 5). By contrast, N2-fixation rates for P. schreberi exhibited a marginally nonsignificant relationship with the number of phylotypes and no relationship with the primary ordination score (Fig. 5).

image

Figure 5. Relationship of nitrogen-fixation rates for Pleurozium schreberi and Hylocomium splendens with the number of bands (i.e. phylotypes) obtained by density gradient gel electrophoresis (DGGE) (a, b), and primary ordination scores (c, d) summarizing the DGGE band data (PC1) for each of the 30 islands (each island represented by a separate point).

Download figure to PowerPoint

Discussion

  1. Top of page
  2. Summary
  3. Introduction
  4. Materials and Methods
  5. Results
  6. Discussion
  7. Acknowledgements
  8. References

We observed clear differences in nifH gene diversity and composition between the two moss species through both fingerprinting and sequence analysis, consistent with our first hypothesis, that is that diversity and community composition of nifH genes would be responsive to moss species. We also found that two clusters of nifH sequences contained nifH phylotypes exclusively present on H. splendens. Further, only about one-third of phylotypes (excluding sequences retrieved by cloning and DGGE bands that were not successfully sequenced) were found on both moss species. This shows that moss species that coexist across large environmental gradients can harbor different diazotrophic communities and that the composition of N2-fixing cyanobacteria on P. schreberi and H. splendens is largely host specific. The underlying reasons why the two moss species host different cyanobacterial communities remain to be investigated, but are likely to involve the closeness of the association that they form with cyanobacteria, and differences in the physical or chemical microenvironments that they provide for the cyanobacterial communities. Host specificity has previously been suggested for the Azolla–cyanobacterial symbiosis as a result of symbiotic coevolution (Papaefthimiou et al., 2008), but to our knowledge there are no comparable previous reports of host specificity in associations between mosses and cyanobacteria.

Despite the strong effects of moss species on phylotype community, there were no differences in phylotype composition or diversity among the island size classes for either species, in contrast to the predictions of our first hypothesis. This is despite the large differences in several environmental factors between island size classes, notably soil fertility, light availability and soil moisture (Wardle et al., 1997), all of which are known to influence moss performance (Lagerström et al., 2007; Lindo & Gonzalez, 2010) and the N2-fixation process (Cleveland et al., 1999). Our work suggests that those factors that drive N2-fixation rates may therefore not be the same as those that influence cyanobacterial diversity and community composition. Instead, host species identity (and therefore, presumably, the microenvironment that the host provides) appears to be a much stronger driver of nifH gene diversity and composition than large-scale differences in environmental factors. This is consistent with the findings of Opelt et al. (2007), who showed that Sphagnum-associated eubacterial communities are specific to individual moss species, irrespective of the sampling site.

Our second hypothesis was that the composition and diversity of nifH genes would serve as predictors of N2 fixation across the 30 islands. The data showed a significant relationship between the number of nifH phylotypes and N2-fixation rates only for one of the species (H. splendens), and therefore only partially supported this hypothesis. While many experimental studies have reported positive relationships between organism diversity and ecosystem process rates (Balvanera et al., 2006; Cardinale et al., 2006), several studies have reported weak or negative relationships between diversity and processes across natural environmental gradients (Wardle et al., 1997; Grime, 1998; Grace et al., 2007). These relationships emerge when those taxa that dominate the process are also highly competitive and exclude other species, for example through competitive exclusion (Grime, 1974; Grace, 1998; Creed et al., 2009). Our results are consistent with such a mechanism; depending on the island from which it was sourced, H. splendens supported either a small number of phylotypes that were highly capable of N2 fixation, or a more diverse assemblage consisting mostly of phylotypes that had a poor capacity for N fixation. However, further quantitative analyses are required for a more complete picture of how the presence of diazotrophs is related to variations in N2 fixation.

While our results provide evidence of high nifH gene diversity and of differences in diversity and composition among co-occurring feather mosses, we emphasize that some caveats have to be considered in their interpretation. It is highly likely that a proportion of the nifH phylotypes we are reporting represent gene copies rather than true genotypes. The nifH2 cluster contained only sequences that were present in H. splendens samples, just as in Nostoc cluster II, suggesting that these sequences together represent one or more strains that have multiple nifH copies within their genomes. Genome analysis of Nostoc punctiforme revealed three different copies of the nifH gene (Meeks et al., 2001), and if gene duplication also occurs in feather moss-associated cyanobacteria, then our use of phylotype number as a measure of the diversity of putative N fixers will have resulted in an overestimate. However, and in contrast to studies that have used conventional prokaryotic molecular markers (such as the 16S rRNA gene) which are not specific enough to distinguish between closely related strains in symbiotic systems (Papaefthimiou et al., 2008), our study enables high resolution of the genetic variation of nifH phylotypes/copies of natural diazotrophic communities. We also note that our study has investigated a much larger number of samples from contrasting environments than is characteristic for this type of work. Investigation of multiple samples in this way is an essential step for evaluating the variability that exists among contrasting natural ecosystems in the genetic characteristics of associative cyanobacterial communities.

Consistent with earlier studies on feather mosses (Gentili et al., 2005), we were able to confirm through microscopy the presence of heterocystous cyanobacteria known to fix N2, such as the genera Nostoc and Stigonema. We were also able to obtain cultures of Cylindrospermum sp., an N2-fixing cyanobacterium not previously reported as being associated with mosses. However, the lack of sample sequences that clustered with the isolated Cylindrospermum suggests that this strain was not dominant in the mosses, at least at the time of sampling. Further, although Stigonema appears to be commonly associated with feather mosses, there is a dearth of available nifH sequences from this genus, especially relative to some other N2-fixing genera such as Anabaena and Nostoc, for which complete genomes are known and which we were able to use as a reference for nifH gene copies in our study. For this reason, we sequenced the nifH gene of Stigonema from a culture strain of freshwater origin available from a culture collection. When we included that reference sequence in our phylogenetic analysis, we could clearly distinguish a cluster of Stigonema sequences from our samples; we were, however, unable to discriminate between different strains and/or nifH gene copies of Stigonema. Further, our reference strain showed only one nifH fragment corresponding to the right sequence length, suggesting that Stigonema has only one conventional nifH gene. However, strains of Nostoc and Anabaena have been reported to contain different numbers of nifH genes (Thiel et al., 1995; Meeks et al., 2001), and on the basis of only one culture strain, we cannot exclude the possibility of multiple nifH copies in all Stigonema species. The Stigonema found in these moss associations is phylogenetically located among the Nostoc clusters and not close to other members of the Stigonematales, which is consistent with previously reported polyphyletic origins of the Stigonematales (Gugger & Hoffmann, 2004).

In conclusion, in this study we were able to demonstrate that the diazotrophic communities associated with two boreal feather moss species show a high degree of host specificity. By contrast, there were no detectable differences among island size classes in the diversity or composition of nifH genes despite large differences among them in the availability of nutrients, soil moisture and light. Certain phylotypes were rarely found or absent on most islands, and most phylotypes were found on a large range of islands but typically only associated with only one of the moss species. These findings show that the identity of the host species rather than extrinsic environmental factors is the primary direct determinant of the cyanobacterial community, and that environmental factors should influence this community primarily indirectly through affecting moss species composition. Our study also addresses the link between the community-level characteristics of a functionally important group of microorganisms and a key ecosystem function that it regulates. We found that, for H. splendens, but not P. schreberi, N2 fixation was significantly associated with nifH gene diversity and composition, which is indicative of a linkage between cyanobacterial community characteristics and N2-fixation ability, at least for some feather moss species. However, controlled experiments are needed to fully elucidate the possible role of diazotrophic diversity in maintaining ecosystem properties such as N input. Such studies would also need to examine the level of gene expression of the community, including the specific features of phylotypes that determine their capacity to fix N2. Nevertheless, our study suggests that cyanobacterial communities and N2-fixation ability can both vary greatly among boreal forest ecosystems, that moss species composition is a powerful determinant of cyanobacterial community composition, and that factors that affect the relative performances of different moss species in boreal forests will have important consequences for the N2-fixing cyanobacterial community.

Acknowledgements

  1. Top of page
  2. Summary
  3. Introduction
  4. Materials and Methods
  5. Results
  6. Discussion
  7. Acknowledgements
  8. References

We would like to acknowledge the Foundation in Memory of Oscar and Lili Lamm and the Carl Tryggers Foundation for Scientific Research for financial support. We would also like to thank Morgan Karlsson, SLU, Umeå, for assistance during field work and Helena Gustafsson, SLU, Umeå for assistance in the laboratory.

References

  1. Top of page
  2. Summary
  3. Introduction
  4. Materials and Methods
  5. Results
  6. Discussion
  7. Acknowledgements
  8. References