The role of mosses in carbon uptake and partitioning in arctic vegetation


Author for correspondence:

Lorna E. Street

Tel: +44 114 222 7990



  • The Arctic is already experiencing changes in plant community composition, so understanding the contribution of different vegetation components to carbon (C) cycling is essential in order to accurately quantify ecosystem C balance. Mosses contribute substantially to biomass, but their impact on carbon use efficiency (CUE) – the proportion of gross primary productivity (GPP) incorporated into growth – and aboveground versus belowground C partitioning is poorly known.
  • We used 13C pulse-labelling to trace assimilated C in mosses (Sphagnum sect. Acutifolia and Pleurozium schreberi) and in dwarf shrub–P. schreberi vegetation in sub-Arctic Finland. Based on 13C pools and fluxes, we quantified the contribution of mosses to GPP, CUE and partitioning.
  • Mosses incorporated 20 ± 9% of total ecosystem GPP into biomass. CUE of Sphagnum was 68–71%, that of P. schreberi was 62–81% and that of dwarf shrub–P. schreberi vegetation was 58–74%. Incorporation of C belowground was 10 ± 2% of GPP, while vascular plants alone incorporated 15 ± 4% of their fixed C belowground.
  • We have demonstrated that mosses strongly influence C uptake and retention in Arctic dwarf shrub vegetation. They increase CUE, and the fraction of GPP partitioned aboveground. Arctic C models must include mosses to accurately represent ecosystem C dynamics.


There is growing evidence that climate change is driving shifts in the productivity and composition of Arctic vegetation. Remotely sensed data indicate ‘greening’ of the land-surface (Myneni et al., 1997; Verbyla, 2008) and ground-based observations suggest that the range and abundance of tundra shrubs are increasing (Tape et al., 2006; Forbes et al., 2010). Increases in plant productivity are expected in response to rising air temperatures, which lead to earlier snow melt and a longer growing season (Groendahl et al., 2007), as well as greater availability of nutrients as a result of accelerated decomposition and deepening of the active layer (Keuper et al., 2012; Natali et al., 2012). Experimental warming studies have confirmed plant growth responses to increases in temperature (Elmendorf et al., 2012).

Enhanced plant productivity may partially offset increases in respiratory CO2 release which result from warmer soils and melting permafrost (Schuur et al., 2009), though greater plant growth may also lead to net soil carbon loss (Hartley et al., 2012). Terrestrial carbon (C) stocks in the permafrost region are large, with c. 1600 Pg contained in soils (Tarnocai et al., 2009) and 60–70 Pg in vegetation (McGuire et al., 2009). Loss of this C from the terrestrial biosphere to the atmosphere could therefore result in an accelerating feedback on climate warming. Modelling studies indicate that the Arctic was a small sink for CO2 at the end of the last century (McGuire et al., 2000), but the future C balance of the Arctic will be determined by how much long-term C storage results from the greening trend.

A key difficulty in understanding the consequences of vegetation change for terrestrial C balance is determining how gross primary productivity (GPP) relates to C storage. This requires quantification of (1) the fraction of GPP that is used for growth versus that re-released to the atmosphere through autotrophic respiration (RA), also known as the carbon use efficiency (CUE); (2) the partitioning of assimilated C between plant pools, and (3) the longevity of those pools. In this paper we use the term ‘incorporation’ to describe the conversion of recent photosynthate (sugars) into structural or storage compounds. We follow the definitions of Litton et al. (2007) and refer to ‘partitioning’ as the flux of C to a particular component as a fraction of GPP and ‘carbon allocation’ as a general term describing partitioning or incorporation or both.

It has often been assumed that CUE is well constrained at c. 50% (Gifford, 1994, 2003; Waring et al., 1998; Williams et al., 2005) and, while 50% may represent a central tendency, large differences in CUE have been reported between ecosystems (Zhang et al., 2009), between forest types (DeLucia et al., 2007), between species (Albrizio & Steduto, 2003) and between stages of plant development (Van Iersel, 2003). Litton et al. (2007), for instance, report a large range of 47–71% CUE in forest ecosystems. Furthermore, controls over the partitioning of C between plant pools, particularly the fluxes of C to aboveground versus belowground tissues, remain a critical uncertainty in global C models (Williams et al., 2005; Litton & Giardina, 2008; Chapin et al., 2009). Again, constant fractions are often assumed (Ise et al., 2010) even though observations suggest that C partitioning varies with forest type, stand age and resource supply (Gower et al., 2001; Litton et al., 2007). There is limited understanding of the factors governing CUE and partitioning in Arctic ecosystems, in particular the role that mosses play, despite their significant contribution to biomass. This therefore currently restricts our ability to interpret the consequences of vegetation change for Arctic C balance.

Mosses are generally expected to respond negatively to increasing temperatures (van Wijk et al., 2004; Walker et al., 2006); however, ambient conditions are important in dictating the direction of response, which can also be species-specific (Lang et al., 2012). The contribution of mosses to ecosystem-scale C dynamics in the Arctic has recently begun to receive greater attention (Douma et al., 2007; Cornelissen et al. 2007 Huemmrich et al., 2010; Turetsky et al., 2012). Research suggests that C partitioning in mosses is variable; in the high Arctic, Woodin et al. (2009) found 70% retention of 13C tracer in photosynthetic tissues after 4–6 wk in Calliergon richardsonii, as a result of high rates of incorporation into recalcitrant pools. By contrast, 20% of recently assimilated C in Polytrichum piliferum, a relatively productive species, is retained in photosynthetic tissues; 80% being respired or translocated (Street et al., 2011).

The objective of this study was to determine the fate of assimilated C in sub-Arctic plant communities and to quantify, for the first time, the contribution of mosses to ecosystem CUE and above/belowground C partitioning. We used 13C pulse-labelling to track recently fixed C in Sphagnum sect. Acutifolia and Pleurozium schreberi-dominated moss patches and in dwarf shrub vegetation with a P. schreberi understory. To quantify short-term C turnover, we measured the 13C content of tissues and respired CO2 over 19 d. To quantify the longevity of incorporated C, we measured recovery of 13C in plant tissues after 1 yr. We asked: what is the CUE of Sphagnum and P. schreberi, and a mixed P. schreberi–dwarf shrub community? What percentage of GPP is allocated to P. schreberi versus that allocated to above- and belowground vascular pools? And, what proportion of incorporated C is recoverable in plant biomass after 1 yr? We hypothesized that CUE for P. schreberi would be greater than for Sphagnum, as P. schreberi is metabolically inactive during dry periods, which may act to reduce respiratory C demand. We expected that P. schreberi would account for a significant component of GPP in the dwarf shrub vegetation, and would therefore act to increase community-level CUE. We also hypothesized that recovery of incorporated C after 1 yr in mosses would be intermediate between that of evergreen shrub species, which have long-lived leaves and woody stems, and that of herbaceous vascular species, which turn over large amounts of leaf and stem material annually.

Materials and Methods

Site description

Our field site was located c. 50 km south of Kevo Sub-Arctic Research Station (, in northern Finland (69°29′N, 27°14′E, elevation 250 m). The site lies in the sub-Arctic forest–tundra transition just north of the pine forest tree-line. The mean annual temperature at Kevo is −2°C (−16°C in January; 13°C in July). Our plots were situated in an area typical of the transition between wetland and birch forest, two dominant land cover types in the region. The vegetation was characterized by dwarf shrubs such as Betula nana, L. Empetrum nigrum L. and Vaccinium vitis-idaea L. and feather mosses (mostly Pleurozium schreberi (Willd. ex Brid.) Mitt.) growing on raised hummocks, with wetter Sphagnum-dominated patches between.

Overview of experimental set-up

We pulse-labelled three community types with 13CO2: Sphagnum sect. Acutifolia (Russow) Schimp. (hereafter referred to as Sphagnum) with vascular canopy removed (S1–4); P. schreberi with canopy removed (P1–4); and P. schreberi with an intact canopy of E. nigrum, V. vitis-idaea and Rubus chamaemorus L. (V1–4). We removed the vascular plants in order to compare moss respiratory 13CO2 fluxes without including the vascular component. We included an intact shrub canopy to assess the contribution of P. schreberi to community C dynamics, and to prevent long-term effects of the lack of canopy on C turnover in P. schreberi, which dries readily. Vascular cover in the Sphagnum plots was < 20%, and we assumed that the effect of canopy removal on C respired would be small. The first 19 d (t1 to t19) includes the key period of short-term C turnover during which photosynthate is allocated to growth or re-respired (Street et al., 2011). During this period we measured the 13C content of ecosystem respiration (RE), and modelled total 13CO2 flux, comparing respiratory losses to tissue 13C retention. After 19 d, most 13C remaining would be incorporated and in a slower turnover pool (Street et al., 2011); to assess longer term C turnover, we sampled plant tissues after 1 yr. A list of abbreviations and symbols is given in Table 1.

Table 1. List of abbreviations and symbols
GPPGross primary productivityμmol m−2 s−1
CUECarbon use efficiency%
R A Autotrophic respirationμmol m−2 s−1
R E Ecosystem respirationμmol m−2 s−1
R M Moss respirationμmol m−2 s−1
t 0…407 Sampling time point 0…407 d after labellingd
u Moss relative water content% dry mass
R 0 Respiration at 0°C (fitted parameter)μmol m−2 s−1
βFitted parameter describing respiration response to temperature°C−1
A t Pulse-derived, normalized 13C content of CO2 or plant tissuesatom%
a Pulse-derived, normalized 13C content of CO2 or plant tissues at t0atom%
k Decay constant of pulse-derived, normalized atom% of gas or tissuesd−1
c 0 Asymptote of decay curve for pulse-derived, normalized atom% of gas or tissuesatom%
S1–4Labelled Sphagnum plots
P1–4Labelled Pleurozium schreberi plots
V1–4Labelled P. schreberi–dwarf shrub plots
c cap Sphagnum capitulum
c sub Sphagnum subcapitulum
c 3 Sphagnum stem section 2–3 cm below ccap
c 4 Sphagnum stem section 3–4 cm below ccap
f y1 P. schreberi frond (0–2 cm)
f y2 P. schreberi frond (2–4 cm)
f SM P. schreberi frond (senescing material below 4 cm)

13C pulse-labelling

Twelve plots (three community types; four replicates) were pulse-labelled with 98 atom% 13CO2 using 0.55 × 0.55 × 0.20 m clear acrylic chambers, from 13:00 to 18:00 h on 7 July 2008. Labelling followed the methods of Street et al. (2011); one half of each chamber was darkened in order to quantify the nonbiological back-diffusion of 13C from the surface (Subke et al., 2009). The mean CO2 concentration in the chamber headspace during labelling was 456 ppm (see Supporting Information Table S1 for individual plot values). Twenty-four hours before labelling we inserted 0.2-m-diameter ‘13C return collars’ to a depth of 0.15 m on the ‘dark’ and ‘light’ sides of the labelled plots; we used these collars to collect respired CO2 for 13C analysis. To avoid severing roots and stems in the plots with an intact canopy (V1–4), we did not insert the collars; instead, we created a seal using plumber's putty (Heinemeyer et al., 2011). Before labelling, all plots were watered to ensure mosses were fully hydrated, and further water was applied to the P. schreberi (P1–4) plots by removing the chamber for c. 10 s mid-labelling.


Environmental conditions

Moss temperature at 2 cm depth was monitored every 5 min using thermocouples connected to CR1000 loggers (Campbell Scientific, Logan, UT, USA), and surface wetness using HOBO leaf wetness sensors with HOBO Micro-Station loggers (Onset Inc., Bourne, MA, USA). Precipitation data are from a nearby meteorological station (c. 200 m from the site).

Ecosystem respiration

RE fluxes on the ‘light’ and ‘dark’ treatment 13C return collars were measured using an LI8100 Soil CO2 flux system (Li-Cor, Lincoln, NE, USA) on six occasions from 6 to 10 July. Most measurements were made between 13:00 and 17:00 h; measurements on the 10 July were also made at 07:00 h. The temperature range over which respiration measurements were taken was 6–22°C. Chambers remained closed for a period of 2 min; we estimated fluxes based on the linear regression obtained > 30 s after closure.

13C content of ecosystem respiration

We measured the 13C atom% of RE from the 13C return collars at discrete time intervals of 0 (immediately after labelling), 0.3, 0.5, 1, 1.5, 2, 3, 4, 6, 10, 11, 19 and 42 d after labelling (hereafter abbreviated to t0, t0.3 etc.). Measurements at t10 included half the replicates as a lightening storm interrupted sampling. The collar headspace was closed using PVC lids of identical diameter to the inserted collars. We sealed the lids with 4-cm-wide rubber seals, then immediately took an 18-ml gas sample through a septum in the lid, followed by two further samples 30 min later. Each gas sample was transferred to an evacuated glass Exetainer (Labco Ltd, High Wycombe, UK). Within 1 month, all gas samples were analysed for CO2 concentration and δ13C using continuous-flow isotope ratio mass spectrometry at the University of York, UK (CF-IRMS; SIRAS Series2; Micromass, UK, Pro-Vac Services, Crewe, UK).

Isotope calculations

All calculations to derive isotopic mixing ratios were based on absolute isotopic concentrations of 12CO2 and 13CO2 in gas samples, determined by mass spectrometry. Mixing ratios are expressed as atom% 13C, that is,

display math(Eqn 1)

(13C, the isotopic fraction of 13C in CO2 expressed as atom%; [13C] and [12C], the absolute concentrations of 13CO2 and 12CO2, respectively, in gas samples.) The concentrations of soil-derived CO2 in the isotope flux chambers 30 min after lid closure were calculated as:

display math(Eqn 2)

([Cresp], the CO2 concentration derived from soil- and vegetation fluxes; [Csample], the CO2 concentration of samples obtained after 30 min of chamber closure; [Cair], the CO2 concentrations of samples taken immediately after closing chambers, assumed to represent ambient atmospheric CO2.) The 13C : 12C isotopic mixing ratio of respiration-derived CO2 (13Cresp) was calculated using a two-source mixing model:

display math(Eqn 3)

(13Cresp and 13Cair, the 13C : 12C isotopic mixing ratios of soil-derived CO2 and ambient air, respectively.)

13C content of moss and vascular plant tissues

We sampled moss tissues 0, 0.5, 1, 2, 3, 4, 6, 10, 19, 42 and 407 d after labelling (t42 was 18 August 2008 and t 407 was 18 August 2009). For Sphagnum we removed several stems and separated the capitula (ccap) from the next 1 cm of stem, the subcapitula (csub). For P. schreberi we separated the tip of the frond which included the current year's growth (fy1) (approximately the top 2 cm) from the previous years’ growth (fy2) (the next 2–4 cm), although fy2 will have included small amounts of new growth (Benscoter & Vitt, 2007). At t407 we sampled deeper 1-cm sections along the Sphagnum stem, 2–3 cm from the surface (c3) and 3–4 cm from the surface (c4), and brown P. schreberi senescing material (fSM) below the green layer which was beginning to degrade. Leaves of E. nigrum (= 4 for all dates), V. vitis-idaea ( 3 for all dates except t2 (= 2), t6 (= 1), and t10 (= 2)) and R. chamaemorus ( 3 except for t10 (= 2)) were also sampled. At t4 and t42 we sampled stems, coarse roots (> 1 mm) and fine roots (< 1 mm) for E. nigrum (= 4), V. vitis-idaea (= 2) and R. chamaemorus (= 3) from the labelled plots outside of the 13C return collars (to a maximum of c. 10 cm, to minimize damage to the plots). At t407 we harvested entire turfs to a maximum depth of c. 15 cm from the V1–4 13C return collars and sampled leaves, stems, coarse roots and fine roots for 13C content and biomass.

Tissue samples were immediately sealed in plastic bags, then placed on ice, before being transported back to the laboratory and frozen at −18°C within 4 h. Samples were dried at 70°C for 3 d before being milled for isotope analysis (Mixer Mill MM200; Retsch, Haan, Germany). The relative water content (u;% dry mass) of every labelled moss sample was measured gravimetrically.

Vegetation biomass

To quantify moss biomass, we took 4 × 4 cm cores from all plots at t42 and t407. At t407 we harvested vascular plants from the return collars and separated leaves, stems and coarse roots to measure biomass. As large a sample of fine-root material as possible was collected for each species for 13C analysis (at least 15 g) by removing fine roots which were still attached to the plant. The total fine-root biomass within the return collars was estimated from a calibration relationship between vascular leaf area index (LAI) and fine-root biomass obtained at the same site (Sloan, 2011). We proportioned fine-root biomass between species based on relative contributions to aboveground biomass.

Pleurozium schreberi respiration

Pleurozium schreberi turfs (0.15 × 0.1 m) were cut using a knife, and placed into aluminium trays. The turfs were then watered until fully waterlogged, with excess water being allowed to drain from the trays. As the mosses dried we measured respiration at repeated intervals; every day during the first week, then once every 2 d. CO2 fluxes were measured using an LI6400 photosynthesis system (Li-Cor) connected to a 0.2 × 0.2 × 0.13 m clear acrylic chamber following the methods of Douma et al. (2007), with the exception that the turfs were placed onto an acrylic sheet before measurement. We darkened the chamber using an opaque plastic sheet. The weight of the turfs was recorded at each measurement to calculate water content (u) relative to oven-dry mass. In between measurements the turfs were stored outside under an open transparent shelter to ensure that light intensity and humidity were similar to field conditions.


Pleurozium schreberi respiration

Our measurements of RE on the 13C return collars were made under ambient conditions between 6 and 10 July when the mosses were dry. On wetting, mosses could contribute further respiratory CO2 flux, leading to a potential underestimation of total RE and thus 13CO2 flux. In order to quantify the contribution of P. schreberi RA to total RE in the P. schreberi (P1–4) and dwarf shrub plots (V1–4) we modelled moss respiration (RM) based on the moisture responses of the excised turfs. The P. schreberi respiration model was based on temperature and moisture responses where:

display math(Eqn 4)

And where:

display math(Eqn 5)

R0 is respiration at 0°C (μmol m−2 s−1), β is a fitted parameter (°C−1) and T is air temperature (°C). Wr is an adjustment factor for tissue moisture content (u; % dry mass), and wa, wb and wc are fitted parameters controlling the response of respiration to u.

Ecosystem respiration

We fitted exponential RE–temperature response curves for each plot using RE measurements (LI8100 flux system) from the 13C return collars:

display math(Eqn 6)

(R0, the respiration at 0°C (μmol m−2 s−1); β, a fitted parameter (°C−1); T, air temperature (°C).) The fitted RE–temperature response curves were then used to extrapolate total RE over the 19-d period on the basis of air temperature. We added modelled P. schreberi respiration to our RE estimates for P1–4 and V1–4. We used surface wetness as a proxy variable for tissue moisture (u) for P. schreberi, assuming that when the wetness probe recorded 100%, u was 600% (of dry weight), the maximum value recorded (data not shown), and when the surface wetness probe recorded 0%, u was 20%, a typical value for feather mosses in humid air (Dilks & Proctor, 1979). We did not explicitly include the effect of water content on Sphagnum as water contents were stable over the 19 d following labelling (Fig. 1b).

Figure 1.

(a) Air temperature (°C) and rainfall (mm), (b) Sphagnum relative water content (u) (% dry mass) based on destructive measurement and (c) Pleurozium schreberi relative water content based on destructive measurements (closed circles), and estimated from a surface wetness probe (dashed line), for 19 d following 13C labelling. Error bars are ± 1 SE.

13C concentrations

All values for plant tissue and CO2 13C concentrations are given in units of atom% excess above natural abundance (NA), that is, atom% pulse-derived 13C. We normalized 13C atom% excess across plots to a labelling concentration of 10 atom%, based on the linear regressions between labelling concentration and initial enrichment, in order to compare 13C label recovery across plots. The excess 13C contents for vascular stems, coarse roots and fine roots were assumed to be zero at t0; we linearly interpolated atom% values between t0, t4 and t42.

Respiratory 13CO2 flux and biomass 13C pool sizes

To calculate 13CO2 respiratory flux we multiplied modelled RE (RE + RM for plots P1–4 and V1–4) by 13CO2 atom%, linearly interpolated over time. We also calculated 13CO2 fluxes based on independent estimates of CO2 flux from syringe-sampling of the 13C collar head space. The 13C content (pulse-derived and normalized by labelling concentration) of each biomass pool was calculated by multiplying 13C atom% by C content (% by mass) and biomass (g). Biomasses for vascular leaves, stems and coarse roots were based on harvests taken at t407. Biomasses for mosses were based on harvests taken at t42 and t407. We expressed 13CO2 flux and biomass 13C as percentages of total 13C assimilated by assuming that total photosynthetic 13C uptake was equal to the amount of 13C present in tissues at the point of greatest enrichment after labelling.

Statistical analysis

We fitted exponential decay models to compare short-term (t0 to t19) 13C atom% dynamics in moss and vascular plant leaves, and in respired CO2 for each plot type. For the tissue data we fitted a model which includes a nonzero asymptote (co) (Subke et al., 2012):

display math(Eqn 7)

(At, pulse-derived, normalized 13C atom% of tissues or CO2; a, pulse-derived, normalized 13C atom% of tissues or CO2 at t0; k, a fitted decay constant.) For respired CO2 we subtracted the ‘dark’ 13C atom% then fitted exponential decay models with c0 set to 0, as by t19 13C in respired CO2 was negligible.

To test the significance of temporal trends in 13C we compared fitted models using Eqn 7 to a constant model (At = c0) using maximum likelihood ratio tests within the ‘nlme’ (Pinheiro et al., 2012) package for R 2.15.0 (R Development Core Team, 2012). We included variance functions to account for heteroscedasticity in the data where appropriate. We used one-way ANOVA with post hoc comparisons between groups to compare 13C recovery over 1 yr (Tukey's HSD) (R Development Core Team, 2012). To quantify CUE and C allocation between moss and vascular plant pools, mean values are provided with standard errors (= 4).


Environmental conditions

Air temperature varied between 2.2 and 21.3°C between t0 and t19 (Fig. 1a). Photosynthetic photon flux density (PPFD) exceeded 1000 μmol m−2 s−1 on cloud-free days; the lowest daily values of PPFD were between 2 and 35 μmol m−2 s−1 (data not shown). No precipitation was recorded between the end of labelling and t6. From t6 onwards the longest period without recordable (> 0.5 mm) rainfall was 2 d; moisture events were recorded on the surface wetness probe daily (Fig. 1c). Not all moisture events that registered on the surface wetness probe corresponded to measurable rainfall (Fig. 1a,c); for example, dew might evaporate before contributing to the rainfall signal from a tipping bucket gauge. The maximum measured u of Sphagnum tissues between t0 and t19 was 1150% (at t19), and the minimum recorded was 965% (at t3). For P. schreberi, the maximum recorded u was 474% (at t10), and the minimum was 5.4% (at t2) (Fig. 1b,c).

Respiratory CO2 fluxes

Based on the fitted RE–temperature response curves, the average rates of RE at 20°C were 1.9 ± 0.3 μmol m−2 s−1 for the dwarf shrub plots, 1.7 ± 0.2 μmol m−2 s−1 for Sphagnum, and 1.2 ± 0.2 μmol m−2 s−1 for Pschreberi (Fig. 2). Mean fitted Q10 values (the factor by which respiration increases) between 10 and 20°C were 1.7 ± 0.09 for Sphagnum, 1.5 ± 0.1 for P. schreberi and 2.5 ± 0.7 for E. nigrum. Based on the empirical model of RM for excised turfs, P. schreberi (Fig. 3b) respiration at 20°C was 0.8 μmol m−2 s−1 when = 400% and zero when < 15% (Fig. 3a). The R2 of modelled versus measured RM for P. schreberi was 0.734 (Fig. 3c).

Figure 2.

Measured ecosystem respiration (RE) for (a) Sphagnum, (b) Pleurozium schreberi and (c) P. schreberi–dwarf shrub plots. Different symbols represent different plots, and lines are the fitted exponential temperature response curves for each plot.

Figure 3.

Response of Pleurozium schreberi respiration (RM) to (a) tissue relative water content (u) (all data) and (b) temperature (when 200% < < 400%), and (c) modelled versus measured P. schreberi respiration (all data; slope = 0.99; intercept = 0.002 μmol m−2 s−1).

13C in respired CO2

RE was enriched in 13C immediately following labelling for both ‘light’ and ‘dark’ treatments for all vegetation types. For Sphagnum and P. schreberi (no canopy), ‘dark’ (i.e. abiotic, diffusive 13C flux) 13CO2 enrichment was > 50% of total 13CO2 enrichment at t0.3. The enrichment of the abiotic CO2 flux decreased to < 1% of the initial value 1.7 d after labelling for Sphagnum, 21 h after labelling for P. schreberi and 4.8 h after labelling for the dwarf shrub plots (Fig. 4d–f). The biological 13C signal also decreased over time for all plot types (Fig. 4d–f), but for P. schreberi, there was an increase in 13C enrichment between t6 and t11, which was also indicated by additional (but incomplete) CO2 sampling at t10. There were significant differences between vegetation types in the fitted 13C decay curve parameters (Table 2). 13C enrichment at t0 was greatest for the dwarf shrub plots (= 1.37 atom%) and lowest for P. schreberi plots (no canopy) (a = 0.06 atom%). The 13C decay rate for the dwarf shrub plots (= 1.16 d−1) was also greater than for Sphagnum (= 0.24 d−1) and P. schreberi (= 0.15 d−1) (Table 2). Problems with leakage for the dwarf shrub plots (where the collars could not be inserted into the ground), however, meant that much of the data had to be excluded, and no data were available for t19.

Table 2. Fitted coefficients describing the change in 13C atom% (normalized and in excess of natural abundance) of plant tissues and soil respiration over time (Eqn 7)
Tissue type a k c 0 R 2
Sphagnum capitulum0.016 (0.013–0.018)
Pleurozium schreberi 0–2 cm0.053 (0.049–0.057)
Empetrum nigrum leaf0.39 (−0.45 to 1.24)0.10 (0.02–0.19)0.08 (0.01–0.14)0.15
Rubus chamaemorus leaf0.49 (0.25–0.74)1.2 (−0.19 to 2.6)0.20 (0.09–0.31)0.35
Vaccinium vitis-idaea leaf0.13 (0.078–0.18)0.17 (0.001–0.35)0.03 (−0.025 to 0.078)0.52
Plot type a k R 2
  1. For Sphagnum and P. schreberi tissue, there was no significant time trend, so we provide an average value.

Sphagnum 0.26 (0.22–0.31)0.24 (0.19–0.28)0.67
P. schreberi (no canopy) 0.057 (0.035–0.080)0.15 (0.087–0.21)0.20
P. schreberi–dwarf shrub1.37 (1.22–1.51)1.16 (0.94–1.38)0.95
Figure 4.

(a–c) Top panels: normalized enriched (pulse-derived) 13C content (atom%) of tissues for (a) Sphagnum capitulum and subcapitulum, (b) Pleurozium schreberi 0–2 cm (fy1) and 2–4 cm (fy2) of frond with vascular canopy removed and (c) Empetrum nigrum leaves and P. schreberi 0–2 cm (fy1) and 2–4 cm (fy2) of frond with canopy intact. Black symbols are for E. nigrum, grey symbols are for the dark treatment, and white symbols are for the light treatment. (d–f) Normalized enriched (pulse-derived) 13C content of ecosystem respiration for (d) Sphagnum, (e) P. schreberi and (f) P. schreberi–dwarf shrub plots. Grey symbols are for dark treatment and open symbols are for light treatments. ^ indicates = 2 because sampling was interrupted. Error bars are ± 1 SE.

There was close agreement between 13CO2 flux based on RE–temperature response curves and the collar-based estimates of 13CO2 flux for all plots (Fig. 5a–c). 13CO2 flux for P. schreberi increased between t5 and t11 (Fig. 5b).

Figure 5.

(a–c) 13C flux, based on modelled ecosystem respiration (RE) from RE–temperature response curves and based on measured RE from head space sampling on the 13C return collars for (a) Sphagnum, (b) Pleurozium schreberi and (c) P. schreberi–dwarf shrub plots. (d–f) Per cent of total label uptake remaining in tissues, based on tissue isotope content (symbols) and modelled loss of 13C in ecosystem respiration (solid line) for (d) Sphagnum, (e) P. schreberi without vascular canopy and (f) P. schreberi with vascular canopy. In (f), closed symbols show total ecosystem retention, and open symbols show moss retention as a percentage of total ecosystem 13C uptake. Error bars are ± 1 SE.

13C in plant tissues

Initial enrichment of photosynthetic tissues was linearly related to labelling chamber 13C atom% (e.g. R2 = 0.94 for Sphagnum (ccap), R2 = 0.96 for P. schreberi (fy1) and R2 = 0.74 for E. nigrum leaves) and enrichment of tissues on the ‘dark’ sides of the plots was negligible (9.2 ± 1.7 × 10−4 atom%) (Fig. 4a–c). The time-point of greatest enrichment was t0.5 for P. schreberi (no canopy) and E. nigrum leaves, t1 for Sphagnum and t6 for P. schreberi within the dwarf shrub community. One Sphagnum plot (S4) had a low chamber labelling concentration and a high signal to noise ratio.

We found no statistically significant trend in moss tissue 13C content over the first 19 d, as the data for Sphagnum (both including and excluding S4) and P. schreberi (P1–4) did not provide support for the exponential decay model over the constant model. The apparent trend in 13C enrichment over time for Sphagnum was negative, but for P. schreberi there was no consistent negative trend (Fig. 4a,b). 13C enrichment of Sphagnum tissues was significantly greater than for P. schreberi; c0 for P. schreberi was 0.016 atom% and for Sphagnum was 0.053 atom% (Table 2). Exponential decay models were supported over the constant model for all vascular plant tissues, indicating significant negative trends in 13C content over time. Rubus chamaemorus had the highest initial and final enrichment, reflected by fitted values of a and c0 which were significantly higher than for E. nigrum and V. vitis-idaea, and more than ten times the fitted values of c0 for P. schreberi (Table 2). Rubus chamaemorus also had the highest fitted value of k, although confidence intervals for k overlapped across all species (Table 2).

Carbon use efficiency

Based on the loss of 13C from moss tissues (excluding S4), Sphagnum CUE was 68 ± 5% by t19. The estimate of CUE based on modelled 13CO2 flux was in close agreement at 71 ± 3% (Fig. 5d). For P. schreberi, CUE based on the loss of 13C from tissues (62 ± 13%) was less than the estimate of CUE based on the amount of 13C respired (81 ± 6%) by t19, that is, c. 20% of the initial 13C uptake was unaccounted for by the modelled 13CO2 flux (Fig. 5e). CUE for the dwarf shrub community based on tissue 13C content was 58 ± 10%, but there was also a discrepancy between the loss of 13C calculated from tissues and the modelled 13CO2 respiratory flux (74 ± 6%) (Fig. 5f).

Carbon allocation between moss and vascular plant pools

The biomass of current year's growth of P. schreberi fronds (fy1 = 157 ± 16 g m−2) in the canopy removal plots was comparable to the biomass of Sphagnum capitula (ccap = 114.2 ± 6.8 g m−2). In the dwarf shrub–P.schreberi heath, fy1 biomass was slightly lower (103 ± 31 g m−2), as was estimated cover (88 ± 13%) (Table 3). In the dwarf shrub plots, total ‘green’ moss biomass (fy1 + fy2) was 179 ± 30 g m−2 compared with vascular leaf biomass of 89 ± 30 g m−2. Total vascular (leaf, stem and root) biomass was 693 ± 135 g m−2 compared with total P. schreberi biomass (including senescing material) of 225 ± 35 g m−2 (Tables 3, 4). Empetrum nigrum contributed the largest fraction of the vascular biomass pool with 138 g m−2 of aboveground (leaf plus stem) biomass, compared with 65 g m−2 for V. vitis-idaea and 17 g m−2 for R. chamaemorus. On the basis of a general calibration relationship between leaf area and fine-root biomass (Sloan, 2011), we estimated fine-root biomass to be c. 57% of total plant biomass (Table 4).

Table 3. Peak growing season moss biomass and per cent cover (plots S1–4, P1–4 and V1–4), for August 2008 and 2009
SpeciesTissueBiomass (g m−2) (± 1 SE)Cover (% ± SE)
Sphagnum Capitulum (0–1 cm)114.2 ± 6.8128.5 ± 21.4100 ± 0
Subcapitulum (1–2 cm)112.0 ± 4.2194.5 ± 22.6
2–3 cm146.5 ± 19.3
3–4 cm156.5 ± 23.6
Total626 ± 74.2
Pleurozium schreberi (P1–4, no canopy)Current year's growth157 ± 165.8 ± 0.6100 ± 0
Previous year's growth217 ± 30342 ± 86
Senescing material109 ± 31
Total (excl. SM)375 ± 34348 ± 86
P. schreberi (V1–4, with canopy)Current year's growth103 ± 3170 ± 1688 ± 13
Previous year's growth120 ± 34109 ± 23
Senescing material46 ± 7
Total (excl. SM)223 ± 65179 ± 30
Table 4. Vascular plant biomass and cover (plots V1–4), for August 2009
TissueSpecies cover (%) (± 1 SE)
Empetrum nigrum Vaccinium vitis-idaea Rubus chamaemorus Total
All45 ± 1914 ± 115 ± 465 ± 23
 Biomass (g m−2) (± 1 SE)


(attached, senesced)

36 ± 12(36 ± 15)41 ± 15–12 ± 6–89 ± 30 (125 ± 43)
Stem102 ± 3424 ± 48.4 ± 5.5135 ± 40
Coarse roots34 ± 1124 ± 53.4 ± 1.461 ± 17
Fine roots226 ± 56114 ± 934 ± 16373 ± 44
Total (incl. attached, senesced)433 ± 114203 ± 2957 ± 27693 ± 135

In the dwarf shrub community, total pulse-derived 13C uptake by Pschreberi was 27.9 ± 9.2 mg 13C m−2 compared with 120.8 ± 54.1 mg 13C m−2 by vascular plants; that is, P. schreberi accounted for 25 ± 12% of total 13C uptake, over the labelling period (Table 5). After 19 d, 21.1 ± 6.7 mg 13C m−2 remained in the moss layer, while 43.7 ± 17.3 mg 13C m−2 was recovered in aboveground biomass, and 12.5 ± 1.9 mg 13C m−2 was recovered in belowground biomass (Table 5). After 19 d, therefore, 19.5 ± 8.9% of total gross 13C uptake was found in P. schreberi tissues, 28.4 ± 6.0% in aboveground vascular biomass and 10.3 ± 2.5% in belowground vascular biomass (Fig. 6).

Table 5. 13C in plant tissues 0, 19, 42 and 407 d after labelling (mg 13C m−2 ± 1 SE)
Species t 0 t 19 t 42 t 407
  1. n.d., not determined.

Capitulum33.3 ± 2.328.3 ± 2.117.7 ± 2.710.1 ± 2.6
Subcapitulum9.9 ± 2.46.7 ± 0.3510.8 ± 2.920.9 ± 3.6
2–3 cm stemn.d.n.d.n.d.1.1 ± 0.3
3–4 cm stemn.d.n.d.n.d.0.28 ± 0.1
Total43.2 ± 4.135.0 ± 2.128.5 ± 3.831.1 ± 5.8
Pleurozium schreberi
Current year11.3 ± 1.610.5 ± 1.61.33 ± 1.10.76 ± 0.19
Previous year6.7 ± 2.85.4 ± 0.861.5 ± 0.465.7 ± 2.3
Senescing materialn.d.n.d.n.d.0.76 ± 0.19
Total17.7 ± 3.615.9 ± 2.62.8 ± 1.46.5 ± 2.4
Empetrum nigrum
Leaves (green)31.6 ± 11.117.8 ± 6.212.1 ± 5.43.8 ± 1.6
Leaves (senesced, attached)n.d.n.d.n.d.1.04 ± 0.5
Stemn.d.2.1 ± 0.72.5 ± 1.18.3 ± 5.2
Coarse rootn.d.0.3 ± 0.20.4 ± 0.30.9 ± 0.7
Fine rootn.d.7.7 ± 1.515.3 ± 4.510.9 ± 6.9
Total E. nigrum31.6 ± 11.127.8 ± 7.830.5 ± 9.523.8 ± 13.5
Vaccinium vitis-idaea
Leaves36.4 ± 12.810.7 ± 3.93.0 ± 1.12.6 ± 0.9
Stemn.d.0.95 ± 0.160.03 ± 0.230.75 ± 0.15
Coarse rootn.d.1.04 ± 0.280.84 ± 0.260.29 ± 0.11
Fine rootn.d.3.09 ± 0.300.69 ± 0.441.09 ± 0.42
Total V. vitis-idaea36.4 ± 12.815.8 ± 4.44.6 ± 1.44.7 ± 1.4
Rubus chamaemorus
Leaves51.8 ± 32.710.2 ± 8.63.9 ± 3.10.05 ± 0.04
Stemn.d.1.62 ± 0.702.1 ± 1.70.24 ± 0.13
Rootn.d.0.36 ± 0.130.20 ± 0.170.03 ± 0.02
Total R. chamaemorus51.8 ± 32.712.5 ± 9.26.2 ± 5.00.32 ± 0.17
P. schreberi
Current year19.4 ± 6.415.1 ± 4.79.9 ± 2.28.2 ± 0.23
Previous year8.5 ± 2.86.0 ± 2.05.3 ± 1.12.4 ± 0.76
Senescing materialn.d.n.d.n.d.0.38 ± 0.14
Total P. schreberi27.9 ± 9.221.1 ± 6.715.2 ± 3.010.5 ± 3.3
Total vascular120.8 ± 54.156.3 ± 18.441.7 ± 14.429.1 ± 13.3
Total148.8 ± 51.377.6 ± 16.857.0 ± 13.640.1 ± 17.0
Figure 6.

13C recovery in leaves, stems and roots (coarse + fine) of vascular plants (Empetrum nigrum, Vaccinium vitis-idaea and Rubus chamaemorus) and moss (Pleurozium schreberi) for plots V1–4, 4 and 19 d after labelling. Recovery is expressed as a percentage of gross primary productivity (GPP), where GPP is defined at the total pulse-derived 13C enrichment at t0. Error bars are ± 1 SE.

13C recovery after 1 yr

Total 13C recovery after 1 yr varied between c. 90% (Sphagnum) and 10% (R. chamaemorus) (Fig. 7). Differences between species and pools in the total recovery of 13C between days t19 and t407 were not statistically significant (> 0.5), except for Sphagnum ccap vs csub (= 0.035). The 13C content of the capitulum was 60% lower after 1 yr, while the subcapitulum 13C content increased > 300%. Four per cent of the total 13C present in Sphagnum tissues at day t19 was recovered below the subcapitulum region at t407 (Table 5, Fig. 7). For P. schreberi in the dwarf shrub plots c. 60% of the 13C in tissues at t19 was recovered after 1 yr; recovery within individual tissues, that is, fy1 and fy2, was similar at c. 67% and c. 60%, respectively. After 1 yr, senescing tissues of P. schreberi (fSM) contained c. 2% of the total 13C present at t19 (Table 5, Fig. 7).

Figure 7.

Proportion of 13C in tissues at day 19 (t19) recovered 1 yr later (t407) for (a) Empetrum nigrum, (b) Vaccinium vitis-idaea, (c) Rubus chamaemorus, (d) Sphagnum and (e) Pleurozium schreberi, and (f) total recovered for all species on same axes for comparison. Pools along x-axes for vascular plants are leaf (Lf), stem (St), coarse roots (Crt) and fine roots (Frt). For mosses, pools include capitulum (ccap), subcapitulum (csub), same-year growth (fy1) and previous year's growth (fy2). Error bars are ± 1 SE.

For E. nigrum, total 13C recovery between t19 and t407 was 78% (Fig. 7). On average, the amount of 13C in E. nigrum stems and coarse roots increased between t19 and t407, although there was large variation between plots and the increase was not statistically significant. Over the same period, the amount of 13C in E. nigrum leaves and fine roots decreased. The amount of 13C in attached senesced leaves after 1 yr (t407) was c. 10% of the total amount present in green leaves the previous year (Fig. 7). In V. vitis-idaea, total 13C recovery was 40%, with the highest recovery in stem tissue (c. 80%). We were unable to recover significant amounts of 13C for R. chamaemorus, although there was an increase in the amount of 13C in R. chamaemorus stems between t19 and t407 and very high 13C atom% in coarse roots and rhizomes in one of the plots.


Moss CUE

Using independent approaches based on (1) loss of 13C from labelled plant tissues and (2) modelled respiratory return of 13CO2, we estimated Sphagnum CUE to be 68–71%, and P. schreberi (without canopy) CUE to be 62–81%. These results are slightly higher than the conservative 40–60% range assumed for vascular plants, but the value for Sphagnum is consistent with Woodin et al. (2009), who measured 70% 13C label retention in mosses in high Arctic systems. The Woodin et al. study considered a longer time period (1 month vs 19 d), but considering the minimal change in Sphagnum 13C after t19 (13C recovery between t19 and t407 was c. 90%) these values are approximately comparable.

For several days after labelling, the moisture content of P. schreberi fell below the minimum required for respiration; our chamber CO2 flux measurements showed that respiration ceased when tissue water content was < 15%. The rapid decrease in 13CO2 from the P. schreberi plots therefore must be attributable, at least in part, to a shut-down of metabolism as the mosses dried. After heavy rain there was a measurable increase in the respiratory return of 13CO2. The implication is that short-term turnover of C in desiccation-tolerant mosses such as P. schreberi cannot be modelled as simple decay functions, as is commonly done for plant C pools (Carbone & Trumbore, 2007; Carbone et al., 2007).

The calculations of CUE are based on the assumption that C fixation can be estimated from the time-point of highest tissue enrichment, but this did not occur immediately after labelling. This delay could have been caused by: uptake of 13C after the labelling period ended – the CO2 data show that there was diffusive return of 13C after labelling which may have prolonged label uptake; and loss of labile substrates (either respiration or leakage/exudation from cells) between sampling and freezing, although we attempted to minimize this by immediately placing samples on ice, and freezing within 4 h. The close agreement between 13C return based on CO2 and tissue data in Sphagnum, however, suggests that any loss between sampling and freezing was small.

For P. schreberi and the dwarf shrub–P. schreberi community there was disagreement between CO2 and tissue-based estimates of label return which equated to c. 20% of label uptake. It should be noted that the tissue-based estimates relate only to plant roots and do not include mycorrhizal hyphae and other microorganisms associated with roots, which are likely to have incorporated a small but not insignificant amount of the label (De Deyn et al., 2011). This missing sink may have contributed to the discrepancy. Leakage from the respiration chambers in the dwarf shrub plots could also have led to an underestimate of RE.

In the case of P. schreberi, where collars were inserted to 15 cm and leakage was therefore less likely, a possible cause for the discrepancy between CUE based on tissues and that based on respired CO2 was underestimation of 13CO2 flux. For instance, we probably underestimated the time period during which the mosses were moist; the surface wetness probe would have dried more quickly than moss tissues, and feather mosses are also known to absorb some water from underlying organic layers (Carleton & Dunham, 2003). 13CO2 that was dissolved in extracellular water would have been lost on drying, although, if this was a significant loss term, we would expect to see it for Sphagnum also. We also concentrated 13CO2 sampling efforts over the first 5 d after labelling in anticipation of the greatest change in 13CO2 flux during this period (Street et al., 2011). This meant low temporal resolution c. day 10. It is possible, therefore, that the pulse of 13CO2 that resulted from re-wetting of the moss layer was underestimated.

Moss contribution to CUE and aboveground:belowground C partitioning

We estimated CUE for the dwarf shrub community to be 58–74% over 19 d. If we suppose that the moss layer was absent, but in every other respect the canopy structure remained the same, total 13C label uptake by vascular plants would have been 120.8 mg 13C m−2 (Table 5). After 19 d, 56.3 mg 13C m−2 remained in above- and belowground vascular tissues (Table 5). In the absence of a moss layer, therefore, the CUE of an otherwise identical vascular plant canopy would have been c. 47%. We conclude that under the particular conditions of this study the presence of mosses increased the CUE of the vegetation. In the dwarf shrub plots, P. schreberi accounted for c. 25% of the total initial uptake of 13CO2 (Table 5). C fixed by mosses cannot be directly partitioned belowground because mosses do not have roots. For the plant community as a whole, including the moss layer, we estimated that c. 10% of total label uptake (GPP) was incorporated into belowground plant tissues after 19 d (Fig. 6). For an otherwise identical vascular canopy growing over bare soil, this figure would have been c. 15%: a 1.5-fold difference.

Turnover of structural C pools

Over a period of 1 yr, the majority of moss photosynthate incorporated as growth was retained within the tissue in which it was fixed, in the case of Sphagnum moving downwards in the profile. Only 3% was recovered in Sphagnum below the subcapitulum, indicating that the shoots grew by a maximum of c. 1 cm over the year. On average we recovered 64% of incorporated C in P. schreberi (including senescent moss tissue) over 1 yr. Loss of incorporated C was mostly from nonsenescent tissues and could have been caused by leaching of dissolved organic carbon after re-wetting (Wilson & Coxson, 1999).

Contrary to expectations, average recovery of 13C was lower for E. nigrum (67%) than Sphagnum (80%). The data suggest senescence of at least 14% of green leaves per year, and indicate some re-allocation of C to stem and coarse roots between growing seasons. This is in agreement with results obtained by Subke et al. (2012), who showed consistent increases of pulse-derived label in root biomass of E. nigrum 1 yr after a 13C pulse in shrub communities in northern Sweden. As expected, 13C recovery for R. chamaemorus was low as this species replaces all aboveground tissues annually. High enrichment in R. chamaemorus stems and coarse roots in one plot indicated remobilization of C, but this was impossible to reliably quantify because of limited replication and the possibility for lateral or downward growth of rhizomes out of the sampled area between years.

Significance of mosses at the landscape scale

We labelled plots with 100% Sphagnum cover and 88–100% P. schreberi cover. It is beyond the scope of this study to provide an up-scaling analysis, but it is possible to provide landscape context. Aerial photography indicates that, within 3.6 × 1.2 km surrounding the location of the field experiment, c. 47% of the land surface was birch forest; c. 15% was mire-forest transition and c. 35% graminoid lawn and lichen communities (T. C. Hill, pers. comm.). Within the ‘mire-forest transition’ in which we located our study, Sphagnum spp. were dominant within 10 m of the mire, with average cover > 50% (Street, 2011). Feathermoss cover (mostly P. schreberi) was c. 10% and total moss cover was estimated to be c. 50%. Average vascular plant cover in the same area was 45% (Street, 2011) compared with 65% in labelled plots.

Differences in C turnover between Sphagnum and P. schreberi are related to species’ moisture regulation strategy; Sphagnum moisture content did not fluctuate significantly over the duration of the study, despite variable rainfall. C turnover in P. schreberi, however, was driven by large fluctuations in tissue moisture content. Both landscape position and moss canopy traits will dictate moisture dynamics (Elumeeva et al., 2011), so a necessary first step in incorporating mosses into landscape C models will be a realistic representation of their water and energy balance. The challenge for landscape modelling is to represent the spatio-temporal heterogeneity in moisture conditions, and its links to (micro)topography. Modelling needs to focus initially on simulating the critical processes in wetter and drier components of the landscape, at fine scales (< 1 m). Geostatistical analyses of hydrological and moss species pattern can then provide insights into up-scaling strategies that avoid bias and minimize error.


We found that: CUE is 68–71% for Sphagnum and 62–81% for P. schreberi; P. schreberi can account for 25% of GPP in a dwarf shrub community; and 64–80% of moss photosynthate incorporated as growth at mid-growing season remains in tissues after 1 yr, whereas for vascular plants this range is 20–67%. We have shown that a moss understory acts to increase ecosystem CUE and increases the fraction of assimilated C which is partitioned to aboveground versus belowground plant pools. Our results highlight the need to include mosses within vegetation C models for realistic representation of the C cycle in high-latitude ecosystems.


We thank Carole Lowther for help with fieldwork, and the staff at Kevo Research Station. We also thank H. Cornelissen and three anonymous reviewers for their helpful comments on the text. This work was funded by a NERC PhD studentship to L.E.S. and NERC-funded ABACUS project, part of IPY. J-A.S. acknowledges funding by NERC grant NE/E004512/1. G.K.P. and V.S. acknowledge NE/D005884/1.