•This study investigates the influence of vegetation composition on carbon (C) sequestration in a moss-dominated ecosystem in the Arctic.
•A 13C labelling study in an arctic wet meadow was used to trace assimilate into C pools of differing recalcitrance within grasses and mosses and to determine the retention of C by these plant groups.
•Moss retained 70% of assimilated 13C over the month following labelling, which represented half the growing season. By contrast, the vascular plants, comprising mostly grasses, retained only 40%. The mechanism underlying this was that moss allocated 80% of the 13C to recalcitrant C pools, a much higher proportion than in grasses (56%).
•This method enabled elucidation of a plant trait that will influence decomposition and hence persistence of assimilated C in the ecosystem. We predict that moss-dominated vegetation will retain sequestered C more strongly than a grass-dominated community. Given the strong environmental drivers that are causing a shift from moss to grass dominance, this is likely to result in a reduction in future ecosystem C sink strength.
The importance of soil carbon (C) in the global C budget is well established, with northern ecosystems in particular representing significant soil C stocks (e.g. Davidson & Janssens, 2006). Soil C stocks are a function of inputs and outputs of C, both of which are strongly controlled by abiotic factors, particularly climate and hydrology; soil C typically accumulates where cold or wet conditions limit decomposition. However, the importance of biotic factors is increasingly being recognized, with the role of plant species composition being key to both C assimilation and loss within ecosystems (Dorrepaal, 2007). In addition to abiotic controls, the chemical composition of plant litter, which differs among plant growth forms, is an important control on decomposition rates (e.g. Hobbie et al., 2000; Dorrepaal, 2007). Indeed, our ability to predict future ecosystem C dynamics depends on greater understanding of plant species traits and their relationship to both C assimilation and litter decomposability (Cornwell et al., 2008; De Deyn et al., 2008). Rather little is known about the roles of different plant species in these central components of ecosystem C sequestration (Fenner et al., 2007), but these roles will be of key importance in the face of the many environmental drivers that are causing change in species composition. Here we compare the assimilation of C into pools of different decomposability by two plant growth forms, grasses and mosses.
One of the key characteristics of northern ecosystems, and also of temperate peatlands which are similarly important C sinks, is a high abundance of moss. Where mosses are present at high cover within arctic vegetation, they are responsible for a significant proportion of ecosystem C uptake. In both the high Arctic (Svalbard) and the sub-Arctic (Sweden), Douma et al. (2007) found that, in vegetation with 100% moss cover beneath the vascular canopy, mosses accounted for an average of 60% of ecosystem C uptake. Similarly, moss biomass was shown to be a good predictor of gross ecosystem productivity across a range of habitats on Svalbard (Sjögersten et al., 2006). These studies were undertaken in summer, with maximum vascular canopy; mosses may dominate C assimilation even more strongly at the beginning and end of the growing season, before vascular plant shoot growth and after their senescence. In addition to contributing significantly to total assimilation of C, the moss layer is key to its long-term sequestration as it limits decomposition both indirectly, by reducing soil temperature and thus microbial activity (Gornall et al., 2007), and directly, by producing litter of poor substrate quality (Hobbie et al., 2000).
Arctic ecosystems, and indeed temperate peatlands in which mosses are typically dominant, are subject to strong environmental drivers which have the potential to alter the relative abundance of coexisting mosses and vascular plants. For example, herbivory is recognized to have a major influence on the state of the vegetation in the Arctic (Van der Wal, 2006). Graminoids tend to be promoted because of their ability to recover from repeated defoliation and because their growth is stimulated by nutrients from excreta, and by the warming of the soil that results from grazing and trampling of the moss layer. Moss productivity is simultaneously reduced by such grazing and trampling and by increased shading by grasses, and so the vegetation shifts from moss to graminoid dominance. This shift in vegetation composition is not only seen in response to herbivory. Meta-analyses of climate change experiments in the Arctic have shown grasses to have a significantly positive response to warming and cryptogams to have no response (Dormann & Woodin, 2002) or a negative response (Walker et al., 2006). Similarly, nutrient enrichment results in a generally positive response of graminoids, and a concomitant inhibition of lower plants (Van der Wal et al., 2003; Van Wijk et al., 2004). In all these scenarios the increase in the relative abundance of grasses could result in increased gross ecosystem productivity. At the same time, however, moss productivity is reduced as a result of both the direct impact of the environmental driver and increased shading by the graminoid canopy (e.g. Van Wijk et al., 2004; Van der Wal et al., 2005), and so the assimilation of C into recalcitrant moss litter will be reduced. Thus different drivers have similar impacts on arctic vegetation composition, with potentially important consequences for ecosystem C sequestration.
This study investigates the traits of these two coexisting plant growth forms, mosses and grasses, which influence the C dynamics of the ecosystem. Specifically, we employ a field 13CO2 tracer experiment in a tundra wet meadow to test the hypotheses that, firstly, mosses allocate a greater proportion of assimilated C to recalcitrant pools than vascular plants (predominantly grasses), and, secondly, the retention of assimilated C over the growing season is greater in mosses than in vascular plants. The results contribute to a mechanistic understanding of key C sequestration traits and can be applied to scenarios of vegetation change to predict the consequences for ecosystem C cycling.
Materials and Methods
The experiment was carried out on Svalbard, Norway in Adventdalen (78°10.2′ N, 16°06.5′ E). The valley has a mean annual temperature of −6.7°C; the January mean temperature is −15.3°C and the July mean temperature is 5.8°C (1961–2000 average; Svalbard airport weather station; http://www.met.no). Mean annual precipitation is 190 mm (1988–2000 average). The area is underlain by permafrost which starts to thaw in June to a maximum depth of c. 50–70 cm in August and the growing season lasts from mid-June until mid-August. The experimental site is a wet meadow habitat in which the soil is water-saturated throughout the growing season, resulting in anoxia below c. 10–20 cm. There is an organic horizon of variable depth composed largely of poorly decomposed moss, below which the mineral soil consists of silt and stones. The vascular plant community of the meadow is dominated by the grasses Dupontia spp. R. Br and Calamagrostis stricta (Timm.) Koeler, with horsetail Equisetum arvense L. also being abundant. There is also 100% moss cover, dominated by Calliergon richardsonii (Mitt.) Kindb., with c. 10 cm depth of live moss shoots.
Experimental set-up and in situ13CO2 labelling
In order to investigate the fate of C in arctic tundra, we used in situ13CO2 pulse labelling of vegetation components in the wet Dupontia meadow. Five blocks were established within an area of c. 100 m × 50 m, each block being separated by at least 10 m. In order to determine the relative importance of mosses vs vascular plants in terms of C sequestration, paired plots (each 30 cm × 30 cm) were established within each block in July 2005. One plot of each pair was for labelling of moss and one for labelling of vascular plants. To take account of within-site heterogeneity, the paired plots were repeated within each block, at least 2.5 m apart, and values from these two replicates were averaged, giving one value per block in the analysis (n = 5). At least 24 h before 13CO2 labelling the plots were trenched to sever underground connections and base collars were inserted just below the water table position (c. 2 cm deep) to provide a good seal. At the same time, in one of the paired plots all aboveground vascular plant material was clipped to ensure that only mosses could assimilate the released tracer C. In the other plot mosses were covered with 1–2 cm depth of small black plastic beads, thus allowing tracer C assimilation by the vascular plants only. The black beads were removed after labelling.
13C-labelled CO2 was delivered to each plot in a 30 cm × 30 cm Perspex chamber attached to the plastic collar which was already inserted into the soil. To keep headspace to a minimum, the height of the chambers on moss and vascular plant- labelled plots was 5 and 10 cm, respectively. Labelling took place between 11 and 20 July. Lamps were hung above the plots to ensure light-saturating conditions during labelling. Using a pump system connected to an absorber column, the CO2 concentration in the chamber was reduced to c. 100 ppm. At this point the CO2-absorber column was removed from the system and a second loop including a jar containing 0.2 g of 98%13C-enriched sodium bicarbonate was connected to the chamber pump set-up. Using a syringe, 20 ml of 0.1 M HCl was added to the jar through a septum. The resultant evolved 13CO2 was pumped to the chamber where it mixed with the remaining headspace CO2. Air was circulated through the closed chamber and jar system for the following 20 h to maximize uptake. Before removal of the chamber, three samples were taken of the final headspace gas. These were analysed for 13CO2 (Delta PlusXP isotope ratio mass spectrometer interfaced to a Gas Bench II and PAL auto-sampler; Thermo Finnigan, Bremen, Germany) to enable calculation of total 13C assimilation in each plot.
Between 7 and 11 August 2005, c. 1 wk before the onset of vegetation senescence, plots were harvested for above and belowground biomass. In total, an area of 20 cm × 30 cm was harvested. From within this area, five cores of 5 cm diameter were taken down to the frozen soil. The cores from plots in which moss had been labelled were each sampled for live green moss; those from vascular plant-labelled plots were sampled for live aboveground vascular plant tissue, rhizomes and roots. Each vegetation fraction was pooled among cores within a plot for analysis. The remaining turf area from each plot was harvested for estimation of live moss or vascular plant aboveground biomass, as appropriate. Unlabelled moss and vascular plant (above and belowground) material was collected away from labelled plots to determine the natural abundance of 13C.
Seasonal fate of carbon All plant material was ground in a ball mill. Subsamples of 1 mg were analysed for percentage C and 13C enrichment using an elemental analyser linked to a mass spectrometer (Thermo Finnigan Delta Plus Advantage IRMS interfaced via a Conflo III to a FlashEA 1112 Elemental Analyser; Thermo Finnigan). Total 13C present in the vegetation within a plot was calculated from biomass and 13C enrichment values, and was then expressed as a percentage of overall plot uptake, providing an estimate of 13C recovery.
Lability of assimilated carbon Subsamples of ground material were used in a proximate C analysis (Ryan et al., 1990; Loya et al., 2004). This allows plant material to be spit into four fractions: nonpolar extractives (fats, waxes and oils), polar extractives (carbohydrates, soluble phenolics and amino acids), acid-soluble fraction (carbohydrates and cellulose), and acid-insoluble fraction (lignin and lignin-like compounds). This process involves analysis of the initial material and then three sequential extractions, after each of which residual material is analysed and the amounts of C and 13C removed by that extraction are calculated by subtraction.
To extract nonpolar extractives, 1 g of sample was incubated in 100 ml of methylene chloride within a sonicating water bath for 45 min. Then 0.5 g of the residue was added to 75 ml of deionized water and heated to 100°C for 3 h to extract polar extractives. To extract acid-soluble fractions, 0.2 g of the new residue was added to 3 ml of 72% H2SO4 and placed in a water bath at 30°C for 30 min, after which 84 ml of deionized water was added to the acid and the solution incubated for 1 h in a heating block set at 100°C. The remaining material constituted the acid-insoluble fraction. After each step the residue was filtered out of solution using ashed GF/F filters and dried, and a subsample analysed for percentage C and 13C enrichment using an elemental analyser attached to a mass spectrometer.
Biomass estimates from the two types of 13C-labelled plots (moss-labelled and vascular plant-labelled) indicated similar amounts of live vascular plant and moss tissue present (442 and 456 g m−2, respectively; Fig. 1a). The majority of vascular plant biomass (89%) was located belowground in roots and rhizomes (Fig. 1a). Analysis of total tissue C content revealed a similar pattern (data not shown). Moss contained a higher proportion of recalcitrant C (82% of C in acid-soluble and acid-insoluble fractions) than vascular plants as a whole (69% recalcitrant).
The uptake of the 13C tracer during labelling was satisfactory; on average 94.1% (range 77.6–100%) of added 13C in the vascular plant plots and 98.5% (range 93.7–100%) in the moss plots was taken up. One month after labelling, all four plant pools were enriched in 13C, with the highest concentrations in live aboveground vascular plant and moss tissues (Fig. 1b).
Total recovery of the 13C tracer, expressed as a percentage of plot-specific total uptake, was substantially greater for mosses (on average 70.4%) than for vascular plants (40.3%; Fig. 1c). Within vascular plants, most of the 13C tracer (61%) was detected in belowground components. Thus, although live aboveground plant biomass was most enriched in 13C (Fig. 1b), its small contribution to total biomass (Fig. 1a) meant that most of the 13C was held belowground, particularly in the rhizomes, which were noticeably more enriched than the roots.
Subsequent analysis of the plant material revealed that vascular plants and mosses also differed in the way new C was allocated to fractions of different recalcitrance (Fig. 2a). In mosses, 80% of all 13C recovered was in fractions of relatively high recalcitrance (acid-soluble and acid-insoluble), whereas in vascular plants these fractions accounted for 56% of 13C. Comparison of the relative distribution of 13C within each of the different plant parts revealed that vascular plant roots had a profile similar to that of mosses, with 75% of all 13C present in the two fractions of relatively high recalcitrance (Fig. 2b). By contrast, this was only 47% in rhizomes and 57% in aboveground vascular plant tissue. The starkest pattern was revealed for the most recalcitrant fraction, with only 1.6% of the 13C tracer present in acid-insoluble compounds in aboveground plant tissue, whereas this was 22, 29 and 38% for rhizomes, roots and mosses, respectively. Inspection of 13C tracer recovery data shows that only small amounts of label were found in recalcitrant fractions in any of the vascular plant components compared with the large amount detected in mosses (Fig. 2c).
This experiment has clearly demonstrated the importance of moss in ecosystem C assimilation and, importantly, in the sequestration of C in recalcitrant forms which are strongly retained within the system. Moss within the tundra meadow assimilated as much 13CO2 as did the vascular plants, predominantly grasses, confirming the significant contribution of mosses to the gross productivity of arctic vegetation (e.g. Sjögersten et al., 2006; Douma et al., 2007). Subsequent retention of assimilated 13C over 4 wk (half the growing season) by the moss was remarkably high, at 70%, and much greater than the 40% retention by the vascular plants. Interestingly, this is very similar to the reported 64% retention of assimilated 13C over 23 d by Sphagnum in a temperate peatland (Fenner et al., 2004), demonstrating the relevance of our results to other moss-rich ecosystems. The greater persistence of C within the moss than within the vascular plants is a clear reflection of the greater allocation of assimilates to more recalcitrant C pools by the moss. That mosses decompose more slowly than vascular plants, and thus constitute an important contribution to C storage in highly organic soils, is well known (e.g. Hobbie et al., 2000); our data elucidate a key mechanism by which this occurs.
A month after labelling, 13C enrichment was greatest in the photosynthetic material of the vascular plants (grasses), followed by live moss; but the proportion held in recalcitrant forms, largely in structural material, was very much greater in the moss. This provides a mechanistic explanation for similar differences in 13C retention between plant growth forms observed in experiments in temperate peatlands which did not determine the form in which the 13C was sequestered. Ward et al. (2009) observed bryophytes to retain all assimilated 13C over 6 d, whereas the turnover of 13C in graminoid leaves was 19% per day. Similarly, Sphagnum showed no turnover of assimilated 13C over 48 h, whilst two graminoid species had c. 45% turnover (Fenner et al., 2007). The authors’ suggestion that this may be explained at least in part by greater incorporation of 13C into structural material in the moss is strongly supported by our field experimental data.
Within the vascular plants, enrichment decreased from the shoots, through the rhizome to the roots, although, because of their greater biomass, the majority of the 13C recovered was in the belowground parts of the plants. The rhizomes contained a similar proportion of 13C in relatively labile forms to the shoots (47 and 57%, respectively), presumably in C storage pools which can be remobilized to meet growth demands, particularly in spring. The roots, which do not have the same C storage function as the rhizomes, held three-quarters of their 13C in recalcitrant form, representing assimilate that has been allocated to root growth. Such greater recalcitrance of C in roots than in shoots has been noted in other studies (see De Deyn et al. (2008), although Loya et al. (2004) found no difference in the distribution of C and of 14C label between C fractions in leaf and root litter of a sedge). This high proportion of recalcitrant 13C in the roots is similar to that in the moss, yet, because of their much lower enrichment and biomass, the roots hold an order of magnitude smaller total pool of 13C assimilate in recalcitrant forms than the moss.
The fact that moss retained a greater proportion of assimilated 13C than the vascular plants over a month, and sequestered a greater proportion of it into more recalcitrant forms, indicates that vegetation composition will affect ecosystem C dynamics. In fact, the relative sequestration of 13C into the pool of C most resistant to degradation, lignin and related compounds, was almost five times greater in the moss than in the vascular plants, partly reflecting the fact that hardly any 13C at all was incorporated into such compounds within the grass leaves. The chemical composition of live plant material, both vascular and moss, will be reflected in the litter produced by the plant and in turn influence the turnover of the C within the soil, with lignin-like compounds in particular having a long residence time (De Deyn et al., 2008). In a long-term incubation of four tundra soils, that containing the highest proportion of C in the recalcitrant (acid-soluble and acid-insoluble) pools lost least CO2 and total C (Shaver et al., 2006). Thus, the relative abundance of mosses within tundra vegetation, and indeed other plant communities, will strongly influence the persistence of C within an ecosystem.
Given this demonstrated importance of the growth form composition of the plant community, a change in vegetation composition is likely to have important consequences for ecosystem C sequestration. A shift from moss to grass dominance in arctic vegetation is predicted in response to several different drivers, including herbivory (Van der Wal, 2006), nutrient enrichment (Van Wijk et al., 2004) and climate change (Dormann & Woodin 2002;Walker et al., 2006). In all these circumstances, we would predict a consequent reduction in long-term sequestration of C. This is borne out by recent peatland studies. High nutrient applications to an ombrotrophic bog resulted in a loss of mosses and an increase in vascular plant biomass, the result of which was an overall decrease in net C assimilation (Bubier et al., 2007). Furthermore, a shift in vegetation from Sphagnum to graminoid cover caused by elevated atmospheric CO2 concentration resulted in decreased retention of assimilated C (Fenner et al., 2007).
An understanding of plant species traits influencing decomposability has been identified as being crucial to the prediction of future ecosystem C dynamics (Cornwell et al., 2008). In this study, tracing of 13C assimilate not only into the component plant functional groups within a tundra meadow, but also into the C pools of differing recalcitrance within those plant groups, has proved a very effective tool for increasing such understanding. Moss has been demonstrated to allocate a much higher proportion of assimilated C to recalcitrant pools than grasses do, and thus to retain C more strongly within the ecosystem over a growing season and potentially in the longer term. This finding is pertinent to all ecosystems with high moss cover, particularly in the context of the many current drivers that are likely to promote a shift from moss to graminoid dominance.
This work was funded by NERC (NE/C514866). We are also grateful for extensive support from the Macaulay Land Use Research Institute, Aberdeen, UNIS, Svalbard, and members of the FRAGILE project working at the same field site.