Moss species effects on peatland carbon cycling after fire


  • Kate H. Orwin,

    Corresponding author
    1. Soil and Ecosystem Ecology Laboratory, Lancaster Environment Centre, Lancaster University, Lancaster LA1 4YQ, UK
      Correspondence author. E-mail:
    Search for more papers by this author
  • Nicholas J. Ostle

    1. Plant and Soil Biogeochemistry Group, Centre for Ecology and Hydrology, Bailrigg, Lancaster LA1 4AP, UK
    Search for more papers by this author

Correspondence author. E-mail:


1. Peatlands are a significant store of global carbon (C) and may be particularly sensitive to climate change. Understanding the drivers of C cycling in peatlands is therefore important for predicting feedbacks to climate change. One such driver may be the identity of component plant species.

2. Moss species contribute significantly to peatland vegetation, yet we understand comparatively little of their role in short-term C cycling and whether that role is consistent regardless of environmental context.

3. We examined the impact of three dominant, co-occurring moss species (Hypnum jutlandicum, Sphagnum capillifolium and Plagiothecium undulatum) on C cycling across a long-term fire frequency experiment, where plots have been burnt every 10 years, every 20 years or have not been burnt since 1954. We measured species effects on in situ net ecosystem exchange (NEE) and decomposition environments, and moss species litter decomposition rates in a laboratory experiment. The fire experiment also provided an ideal opportunity to test whether moss species effects were consistent regardless of context.

4. Moss-dominated patches were on average net sources of CO2 over the main growing period, with Sphagnum-dominated patches showing the lowest NEE. The presence of mosses reduced peat temperature, but this was insufficient to cause differences in the decomposition rate of a standard substrate. Sphagnum and Hypnum litter decomposed more slowly (17 and 16% mass loss, respectively) than Plagiothecium litter (64% mass loss) over a 42-week incubation period.

5. Fire frequency treatments had few effects on measures of C cycling. Moss effects were generally consistent regardless of fire frequency treatment, but Plagiothecium and Sphagnum litter collected from the no-fire plots had higher rates of decomposition than litter collected from the plots burnt more frequently.

6. In summary, postfire, both inter- and intraspecific differences in mosses had significant effects on short-term C cycling. Consequently, changes in both moss species composition and variation within species may need to be taken into account to accurately predict C cycling in peatland ecosystems subject to fire.


Peatlands are a globally important store of soil carbon (C) (Gorham 1991; Dise 2009), and as they occur primarily in the high latitudes, they are likely to be one of the more sensitive ecosystems to climate change (IPCC 2007). These two factors combined mean that it is particularly important to understand what drives C cycling and thus C storage in these systems. Although the major driver of high C storage in peatlands is low temperature and water-logged conditions (Baird et al. 2009), it is increasingly recognized that the composition of peatland plant species can also significantly influence ecosystem C cycling (De Deyn, Cornelissen & Bardgett 2008; Ward et al. 2009). A major component of peatland vegetation is formed by mosses, which can contribute a high proportion of total plant biomass and NPP (Ward et al. 2007; Lang et al. 2009; Lindo & Gonzalez 2010) and often have a high species richness (Limpens et al. 2008) that encompasses a wide variety of traits (Skre & Oechel 1981; Cornelissen et al. 2007; Lang et al. 2009). Despite the potential for a significant contribution of moss species to C cycling we have, compared with vascular plants, a poor understanding of their functional role in ecosystems (Cornelissen et al. 2007). This is partly because many studies to date have focused on single genera (e.g. Malmer et al. 2003; Gunnarsson 2005) or treat mosses as a single functional group (e.g. Verville et al. 1998; Ward et al. 2009), although several more recent studies have begun to compare comprehensive ranges of species (Marschall & Proctor 2004; Turetsky et al. 2008; Lang et al. 2009). Most studies have also examined the impact of moss species on single C processes (e.g. decomposition rates or photosynthetic rates), limiting the depth of our understanding of the role of mosses in overall ecosystem C cycling.

The influence of moss species on peatland ecosystem C cycling is likely to occur through multiple direct and indirect processes. Moss species directly affect gross and net peatland ecosystem CO2 fluxes as a result of their photosynthesis and respiration and through the quality and quantity of their litter inputs (Lang et al. 2009). They also have a wide range of indirect effects on C cycling as they influence soil temperature, hydrology and chemistry (Turetsky 2003; Cornelissen et al. 2007; Gornall et al. 2007; Lindo & Gonzalez 2010), which together regulate the activity of soil decomposers and C loss through microbial respiration. The the exact effect of a given plant species on C cycling can vary depending on growth stage or other environmental contexts such as soil moisture, disturbance regimes and climate (Skre & Oechel 1981; Sveinbjornsson & Oechel 1983; Jackson et al. 2011). Gaining further understanding of both the impact of moss species on C cycling and whether these effects are consistent regardless of environmental context is essential to fully understand their role in C cycling and what impact changes in their composition may have on feedbacks to climate change.

Here, we use a unique, long-term peatland fire frequency experiment (>50 years) at Moor House National Nature Reserve in the UK to examine the effects of three dominant moss species [i.e. Sphagnum capillifolium (Ehrh.) Hedw., Plagiothecium undulatum Hedw. and Hypnum jutlandicum Holmen & Warncke; Fig. 1] on ecosystem C cycling. This fire frequency experiment provides an ideal opportunity to test in detail how environmental context affects the impact of species on C cycling: as fire frequency decreases from every 10 years to every 20 years and finally to not being burnt for >50 years, the fungal/bacterial ratio of the microbial community increases, vegetation cover increases and vascular plant composition changes from Eriophorum dominated to Calluna dominated (Hobbs 1984; S.E. Ward, N.J. Ostle, S. Oakley, H. Quirk, A. Stott, P.A. Henrys, W.A. Scott & R.D. Bardgett, unpublished data). These changes are likely to affect underlying belowground respiration, soil temperature, light availability and the ability of mosses to intercept nutrients in rainfall (Malmer et al. 2003), which are in turn likely to influence moss species effects on C cycling. We aimed to test the hypotheses that (i) coexisting moss species have different effects on ecosystem C cycling, as assessed by measurements of net ecosystem exchange (NEE), moss litter decomposition rates and the effect of mosses on the decomposition environment, and (ii) that moss species effects on ecosystem C cycling are dependent on environmental context, in this instance the effects of fire frequency.

Figure 1.

 Images of the three moss species used in this study: (a) Hypnum jutlandicum, (b) Sphagnum capillifolium and (c) Plagiothecium undulatum. Photographs by K.H. Orwin.

Materials and methods

Study site

The study site was located on acidic ombrogenous upland blanket peat at Moor House National Nature Reserve, in the North Pennines of England (54° 65′N, 2° 45′W). It has a mean monthly temperature of 10 °C and an average annual rainfall of 1900 mm. The Hard Hill long-term fire frequency experiment was the focus of this work. Vegetation in all plots was burnt when the experiment was set up in 1954, followed by fire every 10 years, every 20 years or no fire events since 1954. Plots are 30 × 15 m and are organized in a randomized block design across a 1-km2 area, with four replicates of each fire frequency treatment. At the time of this study (2008–2009), the 10-year fire plots were last burnt in winter 2007 (i.e. 1–2 years ago) and the 20-year fire plots in 1996 (i.e. 12–13 years ago). The no-fire plots had not been burnt for 54–55 years. Controlled burning removes the litter layer and most of the vegetation; some Sphagnum hummocks survive. Plots are colonized in the first 2–3 years by mosses and graminoids (primarily Eriophorum vaginatum), followed by Calluna vulgaris. At the time of measurement, the 10-year fire plots had a low vegetative biomass and were dominated primarily by mosses with some graminoids. Vegetative biomass on the 20-year fire plots and the no-fire plots was considerably higher with close to 100% cover. The 20-year fire plots contained a relatively even mix of mosses, Eriophorum and C. vulgaris, whereas the no-fire plots were dominated by mosses and C. vulgaris. Both Hypnum and Plagiothecium increase in abundance as the time since the last fire increases, whereas Sphagnum maintains a consistent presence across the fire gradient (Hobbs 1984). Hypnum and Sphagnum contribute more to above-ground vegetation mass than Plagiothecium. Owing to the nature of the experiment, measured responses were expected to be influenced by both fire frequency and the time elapsed since the last fire. For simplicity’s sake, we use the term ‘fire frequency’ to indicate both covarying effects.

To test whether fire frequency altered the impact of mosses on C cycling, we focused on three moss species (S. capillifolium, P. undulatum and H. jutlandicum; hereafter referred to by genus name only; Fig. 1). These species were chosen because they are commonly found in moorland, peatlands, forests and ericaceous heathland throughout Europe and often grow in the same vicinity as each other. They were the dominant moss species at our field site and occurred in all plots across the fire gradient. This allowed us to separate out species and environmental context effects. They are all tolerant of extreme temperature ranges, high moisture levels and desiccation (Smith 2004). Sphagnum has an upright growth form and produces lawns or hummocks (Fig. 1). It maintains waterlogged conditions via water-holding hyaline cells and can acidify the local area by sequestering available cations (Turetsky et al. 2008). Its tissues are rich in phenolics that suppress peat enzyme activity and therefore reduce decomposition rates (e.g. Freeman, Ostle & Kang 2001). Sphagnum litter is generally nutrient poor with samples from our study site containing an average of 10 mg nitrogen per gram of tissue. Plagiothecium and Hypnum are commonly known as feather or fern mosses. Plagiothecium has sparsely branched, flattened shoots that grow in surface mats and through the peatland litter layers (Fig. 1). Hypnum’s shoots branch to a greater extent and can form relatively deep mats. It is also often found throughout the litter layer (Watson 1981; Fig. 1). Hypnum litter collected at our study site had a nitrogen content similar to Sphagnum, whereas Plagiothecium contained twice as much, that is, an average of 9 and 18 mg N per gram, respectively.

C cycling measurements

To test whether these three dominant moss species varied in their effects on short-term C cycling and determine whether effects interacted with fire frequency, we measured three aspects of C cycling: in situ CO2 fluxes (i.e. gross respiration and NEE), moss species’ impact on the decomposition environment beneath them (via in situ litter decomposition rates) and how variation in the quality of moss litter affected decomposition rates in a controlled laboratory-based study. Combined, these three measurements covered the main mechanisms by which mosses are likely to affect short-term peatland C cycling.

  • 1In situ C cycling: To examine whether moss species identity and fire frequency interact to affect C loss and gain from the study system, we identified two patches dominated by each moss species in each plot, along with two bare (no vegetation) patches. One patch of each species in each plot was used to measure gas fluxes and the other patch to measure the impact of moss species on the decomposition environment. For gas fluxes, a plastic base ring (110 mm external diameter) was fitted to the peat surface in each patch in 2008, and then left for 1 year to allow vegetation to recover. Flux measurements were made monthly during peak biomass production, in May, June and July 2009. We measured gas fluxes using a portable IRGA (EGM-4) attached to a SRC-1 soil respiration chamber (PP Systems, Herts, UK). Photosynthetically active radiation at the level of the moss canopy and soil temperature at 5 cm peat depth were measured at the same time. To measure net ecosystem CO2 exchange (NEE), the respiration chamber was fitted with a 155-mm-transparent extension hood (total volume = 3764 cm3), which allowed 90% of ambient PAR to reach the vegetation. Gross ecosystem (i.e. plant + soil) respiration fluxes were measured using a darkened chamber of the same volume. The difference between the two measurements was used as an estimate of moss photosynthesis rates (see Ward et al. 2007). Flux measurements were made during closures of <3 min, with one measurement per patch per sampling day. Preliminary analysis indicated that this length of time was sufficient to accurately measure CO2 fluxes. During this time, over 100 individual CO2 measurements are made and integrated by the instrument, and tests gave an R2 for CO2 concentrations vs. time of >0·95. The short period of closure time used here also ensured that environmental conditions within the chamber were not altered significantly during measurement by avoiding the effects of chamber warming.

To measure how moss species identity and fire frequency interact to affect longer-term C loss as a result of differences in the decomposition environment, a litter bag (5·5 × 5·5 cm, 1·2-mm mesh) containing 0·5 g of air-dried E. vaginatum litter (i.e. a standard homogenized litter) was placed in the second set of patches in the field, that is, under each moss species and on bare ground in each plot. Litter bags were placed just beneath the litter layer in all patches, on the FH layer surface. A subsample of the air-dried litter was used to determine initial moisture content and subsequently used to calculate initial litter bag oven-dry weights. Litter bags were buried in November 2008 and collected 1 year later. Decomposed litter was removed from the litter bags, cleaned and oven-dried at 60 °C for 48 h, and the % mass loss calculated.

  • 2Ex situ decomposition experiments: To assess how moss species affect peatland C cycling through the quality of their litter, and how this interacts with fire frequency, we conducted a controlled laboratory experiment composed of a factorial manipulation of the origin of the moss litter and the origin of the peat used as an inoculum, where origin refers to each of the fire treatments (i.e. burnt every 10 years, every 20 years or not burnt for >50 years). Sphagnum litter was collected from all plots, but because of the low abundance of Plagiothecium on the plots burnt every 10 years and of Hypnum on the plots burnt every 10 and 20 years, we were only able to collect sufficient litter from the lower fire frequency plots for these species. Litter was defined as nonphotosynthetic material that was not obviously decayed for Hypnum. For Sphagnum, we defined litter as material between 2 and 5 cm from the top of the capitulum (Aerts et al. 2001; Turetsky et al. 2008). Identifying litter for Plagiothecium was more difficult; we used the lower sections of stems that were not obviously decayed. Peat was collected from each plot to 10 cm depth to use as an inoculum. The assay was conducted in Petri dishes, using the method described in Wardle et al. (1998). Each dish contained 4·25 g dry weight of peat and 0·3 g of air-dried moss litter separated by a layer of 1·2-mm mesh. A subsample of air-dried litter of each moss sample was oven-dried to determine initial oven-dry mass. The moisture content of each peat sample was maintained at field moisture as differences in water table may be one way in which fire treatments may indirectly affect decomposition rates. Petri dishes were sealed and incubated at 14 °C for 42 weeks. Moisture loss was checked every 3 months and adjusted back to initial values with distilled H2O. We maintained the structure imposed by the randomized block design of the field experiment by ensuring that litter and peat samples collected from within the same block were kept together. For example, litter collected from each of the fire frequency treatments in block A was decomposed on peat samples collected from block A. Following decomposition, litter was cleaned, oven-dried at 60 °C for 48 h and mass loss calculated. We also analysed ground initial litter and decomposed samples for C and N on an elemental CN analyser (Leco Truspec CN, St. Joseph, Michigan, USA).

Statistical analysis

The impact of moss species identity and fire frequency on in situ measures of C cycling were analysed using a split-plot anova, with block as a random effect, fire frequency as the main-plot effect and moss identity as the sub-plot effect. The laboratory decomposition study was analysed using a normal anova, with block as a random effect and peat and moss origin as main effects. Data were checked to ensure they met the assumptions of anova and log-transformed if they did not.


In situ C flux measurements

The fire frequency treatments had little effect on ecosystem CO2 fluxes, with the only significant effect being higher levels of NEE and lower estimated photosynthesis rates in the 20-year fire plots than in the 10-year fire plots in May (Fig. 2). Most variation in gas fluxes was caused by the identity of the moss species (Fig. 2). Patches dominated by Sphagnum, Hypnum or Plagiothecium across the fire frequency gradient tended to be, on average, net sources rather than net sinks of CO2. However, Sphagnum patches showed significantly lower NEE compared with the other moss species and bare peat across all three measurement dates. Patches dominated by Plagiothecium and Hypnum both had similar effects on NEE, but had a lower net flux compared with the bare soil in some months only. The significant species effect on NEE appeared to be caused by a combination of patches dominated by Sphagnum showing lower gross respiration rates in June (P < 0·05) and July (< 0·1), and faster rates of photosynthesis in May (P < 0·05) and July (< 0·1) than the other species. Moss effects on ecosystem CO2 fluxes were independent of fire frequency treatments (no significant interaction terms, Fig. 2).

Figure 2.

 Effect of moss species identity and fire frequency on short-term gas fluxes from blanket peat at Moor House National Nature Reserve, on three occasions during the growing season in 2009. Graphs show means and standard errors. Effects of each factor were tested using a split-plot anova, with fire frequency (Freq) as the whole-plot factor and moss species identity (Moss) as the sub-plot factor. Groups of bars with the same capital letter above them do not differ significantly in response to fire frequency. Bars with the same lower case letter above them are not significantly different to each other, when tested for the effect of moss identity. Where lower case letters are present within a panel, they represent the effect of moss identity across all fire frequencies at that sampling date. Post hoc tests were performed using least significant difference (LSD) tests. NS not significant, +P < 0·1, *P < 0·05 and ***P < 0·001.

Across all measurement dates, the presence of mosses tended to reduce soil temperature compared with the bare soil (Table 1). Fire affected soil temperature as well, with warmer temperatures found in the 10-year fire plots compared with the 20-year and no-fire plots at all measurement dates (Table 1). Moss species did not appear to inhabit areas with different levels of photosynthetically active radiation (P > 0·05 for all measurement dates, data not presented), but in June mosses in the 10-year fire plots received significantly more PAR (mean = 429 μmol m−2 s−1) than those in the other fire treatments (20-year fire treatment mean = 102 μmol m−2 s−1; no-fire treatment mean = 103 μmol m−2 s−1; F = 1·04, P = 0·0410). Despite the measured changes in soil temperature, neither moss species identity nor fire treatment had a significant effect on in situ mass loss of the standard Eriophorum litter substrate (Fire frequency: F = 0·17, P = 0·85; Moss species identity: F = 0·50, P = 0·68; interaction term: F = 0·42, P = 0·86). The litter lost an average 33·99% of its mass over the 1-year period.

Table 1.   Effect of moss species and fire treatment on soil temperature at 5 cm depth
Temperature (°C) inMayJuneJuly
  1. Data were analysed using a split-plot anova and are presented as means with standard errors in brackets. Means followed by the same letter within each column are not significantly different to each other (least-significant difference test). There were no significant interactions between moss species identity and fire treatment at < 0·1.

Moss species effects
 Bare peat6·8 (0·3) a13·2 (0·4) a10·0 (0·2) a
 Hypnum6·4 (0·1) b12·7 (0·4) ab9·7 (0·2) b
 Plagiothecium6·7 (0·2) ab12·5 (0·4) bc9·8 (0·2) b
 Sphagnum6·4 (0·1) b12·2 (0·3) c9·8 (0·2) b
 F statistic2·956·303·73
 P value0·06080·00220·0236
Fire treatment effects
 10-year fire7·1 (0·2) a14·0 (0·2) a10·4 (0·2) a
 20-year fire6·3 (0·1) b12·4 (0·1) b9·6 (0·1) b
 No-fire6·4 (0·1) b11·6 (0·2) c9·5 (0·1) b
 F statistic6·2726·8519·42
 P value0·05850·0010·0024

Ex situ decomposition experiment

We first tested for main effects of moss species identity on C and mass loss rates and initial litter C:N ratios using the results of litter collected from and decomposed on peat from the no-fire treatments. Moss species identity significantly affected mass and C loss rates, with Plagiothecium showing a much higher loss rate than both Hypnum and Sphagnum (Table 2). This was reflected to some extent in initial litter C:N ratios, with Plagiothecium showing a significantly lower C:N ratio than Hypnum, but no difference compared with Sphagnum (Table 2).

Table 2.   Effect of moss species identity on initial litter C:N ratios, and mass and C loss during a 42-week laboratory incubation at 14 °C. Data are derived from moss litter collected and decomposed on peat from the no-fire control plots
Moss speciesInitial C:N ratio% Mass loss% C loss
  1. Data are means with standard errors in brackets. Means followed by the same letter within each column are not significantly different to each other (least significant difference test).

Hypnum53·02 (6·50) a16·47 (0·48) b27·15 (8·18) b
Plagiothecium27·54 (1·79) b63·65 (4·46) a66·20 (5·28) a
Sphagnum44·46 (5·26) ab16·98 (1·53) b23·21 (2·92) b
F statistic5·993·5579·13
P value0·0370·0010·000

Comparison of effects of fire treatments (peat) and moss origin) on litter mass and C loss rates showed few effects of peat origin, with the only significant effect being a lower C loss rate of Sphagnum litter on peat collected from the 10-year fire plots compared with that collected from the 20-year fire and no-fire plots (Fig. 3). In contrast, moss litter origin showed a strong effect where it could be tested, with both Sphagnum and Plagiothecium litter collected from the no-fire plots showing a higher mass and C loss rate than litter collected from the plots subjected to regular fire (for C loss see Fig. 3; for mass loss see Appendix Table S1 in Supporting Information). This pattern was not reflected in the initial C:N ratio of the litter, as this variable was unaffected by litter origin within species (Plagiothecium: F = 1·3 = 0·34; Sphagnum: F = 2·9, P = 0·13; data not presented). The effects of moss origin did not interact significantly with peat origin (Fig. 3).

Figure 3.

 Effect of the origin of moss litter (Moss) and the origin of the peat (Peat) it was decomposed on, on % C lost over 42 weeks of a laboratory incubation at 14 °C, where origin refers to fire treatment. Data are not available for litter of Hypnum from the 10-year and 20-year fire plots or of Plagiothecium from the 10-year fire plots because of low litter availability on these plots. Graphs show means and standard errors. Data were analysed separately for each species using anova. Groups of bars with the same capital letter above them do not differ in their response to the origin of the moss litter, whereas bars with the same lower case letter above them do not differ in their response to the origin of the peat. Where lower case letters are present within a panel, they represent the effect of peat origin across all moss litter origin levels, for that species. Post hoc tests were performed using least significant difference (LSD). NS not significant, *P < 0·05, **< 0·01 and ***P < 0·001.


The importance of moss species identity for ecosystem C cycling

Our results confirmed the hypothesis that coexisting moss species have different effects on ecosystem C cycling and showed that these effects were expressed more strongly for some aspects of C cycling than for others. Although none of the tested species appeared to alter the peat environment underneath them sufficiently to change the decomposition rate of a standard litter substrate, Sphagnum-dominated patches were consistently able to reduce NEE to a greater extent than the other two moss species during the growing season, and Plagiothecium litter showed a much higher C loss rate than either Hypnum or Sphagnum.

The greater effect of Sphagnum on NEE than the other species appeared to be due to both slightly higher photosynthetic rates and lower gross respiration rates. The lower respiration rate of Sphagnum-dominated patches may be because of low plant respiration rates or because Sphagnum plants suppressed soil microbial respiration by producing recalcitrant litter (Fig. 3, Lang et al. 2009), acidifying the environment (Turetsky 2003; Baird et al. 2009), producing anti-microbial compounds (Freeman, Ostle & Kang 2001; Turetsky 2003) or by reducing soil temperature (Eckstein 2000) to a greater extent than the other species. However, given that we detected no difference in mass loss of the standard litter substrate decomposed under each moss species, Sphagnum metabolic C-use efficiency and its relatively low litter quality (Table 2, Lang et al. 2009) are probably the more important mechanisms in this instance.

The lack of an effect of the different moss species on decomposition rates of a standard substrate in situ contrast with two other studies that used reciprocal transplant methods to show that substrate mass loss can be influenced by the moss species under which it was decomposed (Belyea 1996; Fenton, Bergeron & Pare 2010). However, Belyea (1996) examined moss species that typify different hydrological regimes (hummocks, lawns and hollows), whereas the three species examined here and the placement of the litter bags on the FH layer meant that the hydrology experienced would have been relatively similar. Fenton, Bergeron & Pare (2010) compared species that have similar or the same habitat preferences to those used here (Pleurozium schreberi (a feather moss with similar growth habits to Hypnum) and S. capillifolium) and found a significant effect on decomposition rates occurred for only one of the four litter types tested, suggesting an overall weak effect. Combined, these results suggest that hydrology is a stronger driver of decomposition rates than indirect effects of mosses on other aspects of the decomposition environment.

The comparatively low C loss rate of Sphagnum and Hypnum litter (16–17%) observed in the controlled laboratory decomposition experiment confirms the finding of other studies (Aerts, Verhoeven & Whigham 1999; Lang et al. 2009) that moss species tend to have low decomposition rates. However, Plagiothecium litter had a much higher C loss rate (66%) over the 42-week incubation period, suggesting that at least some moss species do not fit this general trend. These results suggest that only some mosses will contribute to C sequestration through producing low litter quality and that treating mosses as a single functional group that has similar effects on ecosystem C processes may not be appropriate.

Although moss species effects on C cycling did vary, their effect on NEE were rarely sufficient to result in patches being a net sink for CO2, and with the exception of Sphagnum-dominated patches, did not always result in lower NEE than the bare soil. This suggests that their photosynthetic rate and effects on microclimate were insufficient to offset increased peat and plant respiration rates during the warmer months. Given that other studies where both vascular and non-vascular plants are included in NEE measurements tend to show that peatlands are strong net sinks during the growing season (e.g. Ward et al. 2007), this implies that the contribution of nonvascular plants to C sequestration is relatively minor. However, it has also been suggested that moss species may be photosynthetically active over a greater proportion of the year than vascular plant species (Campioli et al. 2009; Street et al. 2012) and that C turnover in mosses may be much slower than that of vascular plants (Woodin et al. 2009). Combined with the low C loss rate of Sphagnum and Hypnum litter, this suggests that their actual contribution to C sequestration on a year-on-year basis of at least some species may be greater than that indicated by their effect on NEE in the present study.

Fire frequency and its interaction with moss species effects

Fire frequency effects on C cycling were relatively rare, with the 20-year fire plots showing a higher NEE and lower photosynthetic rate than the 10-year fire plots in May only and with a higher decomposition rate seen in peat collected from the 20-year and no-fire control plots for Sphagnum litter only. It is possible that the higher soil temperature found in the 10-year fire plots in May is indicative of a slightly higher air temperature in these plots (probably due to the higher amount of bare ground and consequent lower albedo effect), which may have stimulated photosynthesis to occur earlier than in 20-year fire plots. Species’ growth rates may also be higher immediately after a disturbance resulting in higher rates of photosynthesis in the 10-year burn plots, which were last burnt only 2 years before measurements were made. The higher NEE recorded in 20-year burn plots in May compared with the 10-year burn plots may also have been because these plots had reached a later stage of succession and so had a higher vascular plant biomass and higher abundance of the sedge E. vaginatum. This may have resulted in greater peat CO2 fluxes via root respiration and metabolism of root exudates (Crow & Wieder 2005), which moss photosynthesis may not have been able to offset. The trend in NEE from the bare plots across the fire frequency treatments lends some support to this hypothesis (Fig. 2).

The higher decomposition rate of Sphagnum litter in the 20-year and no-fire control plots may have been because of the increased fungal/bacterial ratios in the peat from these plots (Ward et al., unpublished data), as fungi are thought to play a greater role than bacteria in decomposing complex substrates (de Boer et al. 2005; Romani et al. 2006). There is also some indication that nutrient availability may be higher in the no-fire plots (Ward et al. 2007), potentially stimulating decomposition of Sphagnum litter. It is interesting to note that there was no effect of peat origin on Hypnum litter decomposition, even though it decomposed at a similar rate to Sphagnum and had a similar N content. This implies that decomposition of these two litters was limited by different factors and that complex and synergistic interactions among the decomposer community, litter quality and peat chemistry may determine decomposition rates in this system.

Overall, the fact that fire frequency effects on C cycling were only significant infrequently, whereas moss species effects were nearly always significant suggests that moss species identity is a more important driver of variation in short-term C cycling than the effects of vegetation burning per se (not withstanding losses of CO2 to the atmosphere as a direct consequence of fire). This is consistent with Fenton, Bergeron & Pare (2010) who also found stronger effects of species identity on moss litter decomposition rates compared with disturbance regime (in that instance forest harvest types) and adds further weight to the concept that vegetation composition and traits are important drivers of C cycling (De Deyn, Cornelissen & Bardgett 2008).

Previous studies and our measurements have shown changes in microbial and vegetation community composition (Hobbs 1984; Ward et al., unpublished data), as well as some differences in abiotic environmental conditions (Ward et al. 2007; Table 1) across the fire gradient. These differences in environmental context were insufficient to alter the relative impact of moss species identity on short-term C fluxes or in situ decomposition rates. However, the C loss rate of both Sphagnum and Plagiothecium litter collected from the no-fire plots was significantly higher than that found for litter collected from the other fire treatments. Although moss litter C:N ratios did not change significantly depending on which fire treatment it was collected from, it is possible that the concentration of other compounds that are known to affect decomposition rates (e.g. phenolics (Turetsky 2003)) change as mosses get older or as their competitive interactions with vascular plants change (Malmer et al. 2003). It is also possible that moss chemistry and subsequent decomposition rate is directly affected by fire where mosses survive. Further research is required to determine the exact mechanism behind this result.


Our results indicate that moss species have different effects on short-term C cycling in peatland and that, post-disturbance, these effects can be as or more important a driver of variation in C cycling as larger-scale environmental effects caused by the disturbance itself. Moss species effects on C cycling appeared to be primarily mediated by differences in species C-use efficiency and litter quality as opposed to indirect effects on the microclimate around them. Our results also suggest that variation in environmental context can result in sufficient plasticity in moss species traits to significantly alter their effects on C cycling processes. Such inter- and intraspecific variation needs to be taken into account to accurately predict how changes in species composition in response to global change factors are likely to affect peatland C cycling.


We would like to thank Chris Dove, Jane Mangler, Simon Oakley, Dan Shepherd, Joey Talbot, Sue Ward and Mike Whitfield for technical assistance. We thank Natural England for use of the field site.