Notice: Wiley Online Library will be unavailable on Saturday 27th February from 09:00-14:00 GMT / 04:00-09:00 EST / 17:00-22:00 SGT for essential maintenance. Apologies for the inconvenience.
•High-latitude ecosystems are important carbon accumulators, mainly as a result of low decomposition rates of litter and soil organic matter. We investigated whether global change impacts on litter decomposition rates are constrained by litter stoichiometry.
•Thereto, we investigated the interspecific natural variation in litter stoichiometric traits (LSTs) in high-latitude ecosystems, and compared it with climate change-induced LST variation measured in the Meeting of Litters (MOL) experiment. This experiment includes leaf litters originating from 33 circumpolar and high-altitude global change experiments. Two-year decomposition rates of litters from these experiments were measured earlier in two common litter beds in sub-Arctic Sweden.
•Response ratios of LSTs in plants of high-latitude ecosystems in the global change treatments showed a three-fold variation, and this was in the same range as the natural variation among species. However, response ratios of decomposition were about an order of magnitude lower than those of litter carbon/nitrogen ratios.
•This implies that litter stoichiometry does not constrain the response of plant litter decomposition to global change. We suggest that responsiveness is rather constrained by the less responsive traits of the Plant Economics Spectrum of litter decomposability, such as lignin and dry matter content and specific leaf area.
If you can't find a tool you're looking for, please click the link at the top of the page to "Go to old article view". Alternatively, view our Knowledge Base articles for additional help. Your feedback is important to us, so please let us know if you have comments or ideas for improvement.
The functioning of northern high-latitude and high-altitude ecosystems may be constrained by harsh environmental conditions, such as low temperatures, waterlogging, anoxic and acidic site conditions and low nutrient supply (Robinson, 2002). As a result, both productivity and soil organic matter breakdown are usually low, and most of these ecosystems are characterized by a low diversity of vascular plants (Wookey et al., 2009). However, they often harbour a wide diversity of cryptogams (Longton, 1997; Lang et al., 2012).
These Arctic and alpine ecosystems have been the focal point for global change research for decades for two important reasons. First, recent research has demonstrated powerful coupling between Arctic terrestrial ecosystems, surface energy budget and global atmospheric chemistry (Chapin et al., 2000a,b; Dorrepaal et al., 2009; Schuur et al., 2009), especially with regard to CO2 and CH4 concentrations and the subsequent feedback to the global climate system. The carbon (C) store function of these ecosystems is mainly a result of low C losses caused by litter decomposition in comparison with the amount of C fixed by net primary productivity (NPP). Second, both observational and modelling studies have shown that these ecosystems are subject to the most profound climatic changes on the planet (ACIA, 2005; IPCC, 2007, Callaghan et al., 2010). As many organisms in these ecosystems live close to their physiological limits, it is to be expected that they respond strongly to these environmental changes (Post et al., 2009; Wookey et al., 2009). A recent meta-analysis of the results of 61 circumpolar warming experiments showed that, although tundra vegetation exhibits strong regional variation in response to warming, the cumulative effects of warming over time on tundra vegetation – and associated ecosystem consequences – have the potential to be much greater than we have observed to date (Elmendorf et al., 2012). As Arctic and alpine ecosystems contain a wide spectrum of plant growth forms (cryptogams, deciduous and evergreen dwarf shrubs, graminoids and forbs), shifts in dominance among these growth forms represent potentially a substantial change in ecosystem structural characteristics and functional properties, such as productivity and litter decomposition.
The results of Elmendorf et al. (2012) raise the question: what factors will constrain the long-term response of Arctic and alpine ecosystems to global environmental change? There can be many other constraints, but here we concentrate on possible stoichiometric constraints on both productivity responses and the response of litter decomposition. The balance between these processes is an important determinant of ecosystem C balance and thus on the feedbacks between these ecosystems and the atmosphere. The stoichiometric constraints on these processes will most likely differ, as productivity responses are driven by the elemental constraints on the photosynthetic machinery, whereas the decomposition constraints are driven by the elemental constraints of the soil microorganisms involved. Indeed, earlier studies have shown that the nutrient limitation of litter decomposition in a given ecosystem cannot be predicted from the nutrient limitation of NPP in that system (Hobbie & Vitousek, 2000; Aerts et al., 2003).
It might be expected that, because of the harsh environmental conditions, plants in high-latitude ecosystems will show less variation in stoichiometric ratios than plants from other biomes. The reason for this might be that both nitrogen (N) and phosphorus (P) cycles are constrained to a similar extent by the environment, thereby leading to relatively low stoichiometric variation. However, a meta-analysis of foliar N/P mass ratios at a global scale has shown that, for northern biomes, the interspecific variation in green leaf stoichiometry does not deviate from the high variation observed in other biomes (Reich & Oleksyn, 2004; Kerkhoff et al., 2005). Another conspicuous result from these analyses was that foliar N/P ratios decrease with increasing latitude (or decreasing temperatures), so that plants from high-latitude ecosystems have lower N/P ratios than those from lower latitude biomes. An implication of the relatively high interspecific variation in foliar N/P ratios may be that the responsiveness of green leaf stoichiometry to global change is also high. Indeed, in a meta-analysis of published data, Güsewell (2004) found that phenotypic responses to external nutrient supply may cause up to a 50-fold variation in foliar N/P mass ratios. However, in this analysis, no high-latitude sites were included.
Stoichiometric constraints on productivity are often derived from foliar N/P ratios. This is largely based on an analysis of Koerselman & Meuleman (1996) of many fertilization experiments in wetlands in which foliar N/P mass ratios were a good diagnostic measure of the type of nutrient limitation (N vs P). They found that foliar N/P mass ratios < 14 indicate N-limited growth and ratios > 16 P-limited growth, whereas values between 14 and 16 are indicative of co-limitation of N and P. This concept has frequently been criticized, but it appears to be an easy method to estimate nutrient limitation, although the values of the critical N/P mass ratios appear to be less ‘fixed’ than suggested earlier (Güsewell, 2004; Soudzilovskaia et al., 2005). Previous fertilization studies have indeed shown that N- or P-limited growth in high-latitude tundra can be predicted very well from the foliar N/P ratios in the unfertilized control, for both Sphagnum mosses (Aerts et al., 1992) and vascular plants (Shaver & Chapin, 1995). This implies that stoichiometric constraints on the productivity responsiveness of plants to nutrient-related global changes can indeed occur when the N/P ratios in the plants are close to the critical N/P ratios as defined by Koerselman & Meuleman (1996), or have already exceeded them.
It is logical to assume that, in cold biomes, the responsiveness of litter decomposition to global change is constrained more by climatic factors than by litter traits. However, in a global meta-analysis (including data of all major biomes), Cornwell et al. (2008) showed that the variation in litter decomposition rates caused by litter traits was 18-fold, whereas the variation caused by climatic factors was six-fold. This shows the overriding importance of litter trait variation for litter decomposition. Unfortunately, the relatively simple and straightforward relationship found between productivity responses and foliar N/P ratios does not seem to hold for litter decomposition responses. The biochemical transformations in decomposing litter are performed by bacteria and fungi and, although the stoichiometric relations in microbial biomass are well constrained at a global scale (Cleveland & Liptzin, 2007), some global-scale meta-analyses have shown that litter decomposition rates cannot simply be predicted from stoichiometric ratios in plant litter, such as litter C/N ratios (Aerts, 1997; Cornwell et al., 2008). This is most likely because of the secondary compounds in plant litter and litter physical traits, which are very strong co-determinants of litter decomposition rates (Hättenschwiler & Vitousek, 2000; Shaw & Harte, 2001; Dorrepaal et al., 2005; Cornwell et al., 2008). These observations suggest that global change-driven changes in litter stoichiometry do not necessarily lead to proportional changes in litter decomposition rates, and that thereby the stoichiometric constraints on litter decomposition are less conspicuous than those on productivity. This is clearly a topic that urges further investigations, as the relationship between changes in stoichiometry on global change and changes in litter decomposition have not been investigated previously.
In this article, we explore whether and how plant stoichiometry constrains the responsiveness of the litter decomposition of plants in northern high-latitude and high-altitude ecosystems to global environmental change. Thereto, we investigate the following hypotheses: first, that cool and cold climate plant species show strong natural interspecific variation in litter stoichiometry (C/N, C/P, N/P) and stoichiometry responds strongly to global change; and, second, that the variation in stoichiometry caused by climate change is not very strongly related to decomposition responses.
Materials and Methods
Although leaf litter C/N ratios are frequently reported in high-latitude decomposition studies, fewer studies have assessed both green leaf and litter C/P and N/P ratios because of a strong underrepresentation of green leaf and leaf litter P analyses (cf. Yuan & Chen, 2009a). Therefore, we performed C, N and P analyses of green leaf and leaf litter in the leaf material collected in a sub-Arctic flora in the Torneträsk region of northern Sweden (at 68°N), an area characterized by a high diversity of landscape types and ecosystems (Karlsson & Callaghan, 1996). The species sampled comprised 76 plant species belonging to eight growth forms (22 forbs, 13 graminoids, 11 woody deciduous species, 10 evergreens, six N-fixers, seven hemi-parasites, five fern allies and two carnivorous plants) and three mycorrhizal types (ericoid, ectomycorrhizal and arbuscular mycorrhizal). As a result of this wide coverage of growth forms and the relatively species-poor character of most high-latitude ecosystems (Wookey et al., 2009), this dataset can be considered to be representative of high-latitude floras. Fresh, mature leaves of these species were collected at peak biomass (from 10 July to 9 August) and litter material was collected in September to coincide with natural leaf death. For the evergreens, dead leaves were collected throughout the snow-free period (early May to mid-October).
For the chemical analyses of both the green leaves and litter, only the petioles were used. Total C and N concentrations were determined by dry combustion of ground plant material with an elemental analyser (Perkin Elmer 2400 Series II). Total P concentrations were determined by digesting ground leaf material in 37% HCl : 65% HNO3 (1 : 4, v/v). P was measured colorimetrically at 880 nm after reaction with molybdenum blue.
In addition, an extensive set of litter material from a wide array of medium-term global change treatments is available from our Meeting of Litters (MOL) experiment. In this material, we determined C/N ratios and responses of leaf litter decomposition rates to these treatments (Cornelissen et al., 2007). The MOL experiment includes litter material from 33 circumpolar and high-altitude experiments from 18 sites in 10 countries. The litters originated from 44 plant species belonging to seven growth forms. As such, this is, to our knowledge, the most comprehensive and representative set of high-latitude and high-altitude litter material that has been subject to a wide variety of global change treatments. An overview of treatments and growth form representation is given in Table 1. The leaf litters were incubated for 2 yr in litter bags inside representative surface litter mixtures in outdoor litter beds in northern Sweden (68ºN), and all the experimental details are provided in Cornelissen et al. (2007). The advantage of this approach is that it is very suitable for large-scale comparative studies under realistic outdoor conditions. Moreover, it has been shown previously that a 2-yr incubation period is sufficiently long to establish a robust interspecific ranking of mass loss, where this ranking of mass loss is also robust to variation in incubation duration (Quested et al., 2003; Cornelissen et al., 2007). The two litter beds were at 340 and 980 m altitude, respectively, differing by 6°C in mean annual temperature and being within the range of climate warming predicted for the Arctic this century (ACIA, 2005; IPCC, 2007). Cornelissen et al. (2007) have reported the summarized results for litter mass loss rates, but stoichiometric changes and their relations with litter decomposition rates have not been reported so far, and are thus a major data source for this synthesis paper.
Table 1. Summary of data on the plant litter material incubated in the Meeting of Litters experiment performed in sub-Arctic Sweden
1Fertilizer experiments were run using different levels of nitrogen supply: 2.5, 5 or 10 kg N ha−1 yr−1.
Plus various combinations of treatments in most experiments
Representation of plant growth forms
Deciduous dwarf shrubs (11 spp.)
Evergreen dwarf shrubs (4 spp.)
Forbs (14 spp.)
Grasses (6 spp.)
Wood-rushes (Luzula 1)
Sedges (Carex 4 spp., Eriophorum 1, Kobresia 1)
Mosses (2 spp.)
Green leaf and leaf litter stoichiometric relations (expressed as mass ratios) are presented in box plots. As ratios are usually not normally distributed, box plots provide a concise overview of the ‘real’ distribution and are preferred to the reporting of means and standard errors only. Differences among stoichiometric ratios were determined with one-way ANOVAs after log transformation.
As a result of limitations in the amounts of material available for chemical analyses in the MOL experiment, for this experiment we only have data on litter C/N ratios. To determine the relation between litter stoichiometry and decomposition rates in the MOL dataset, we analysed the response ratios of the litter C/N ratio to all global change treatments, for both the single factor and multi-factor experiments. In all cases, response ratios were calculated with respect to the control treatment only, and were calculated as the log proportional change in the means of a treatment and control group (Hedges et al., 1999). Differences in response ratios for the C/N ratio were analysed by generating 95% confidence intervals by bias-corrected bootstrapping (Dixon, 1993). We analysed the litter decomposition rates for the single-factor global change treatments averaged over the two altitudes in the same way. Next, we compared the response ratios of litter decomposition rates with the concomitant response ratio of the litter C/N ratio in those treatments in a linear regression analysis. In this way, we could assess whether global change-induced changes in litter C/N stoichiometry scale with changes in litter decomposition rates.
Natural variation in green leaf and litter stoichiometric relations
Green leaf stoichiometric ratios showed considerable variation, especially for C/N ratios (Fig. 1), which had a coefficient of variation (CV) of 50.0%. The median green leaf C/N ratio was 21.6 and the median C/P ratio was 384 (CV = 33.5%). The variation in the ratios was mainly caused by variation in N and P concentrations (dry mass basis), as the C concentration showed relatively little variation (data not shown). The median green leaf N/P ratio was 12.5 and had the lowest CV (30.8%).
Similarly, the natural interspecific variation in litter stoichiometric ratios was also considerable (Fig. 2), but was slightly higher for litter C/N (CV = 54.8%) and C/P (CV = 42.8%) ratios than for green leaves. For the litter N/P ratio, the CV (30.3%) was equal to that of green leaves. All litter stoichiometric ratios were considerably higher (P < 0.001 in all cases) than for green leaves.
Global change-induced variation in litter C/N ratios and the relation with litter decomposition
All global change treatments induced significant changes (P < 0.05) in litter C/N ratios, except for the enhanced UV-B radiation experiments (although a small but significant (P < 0.05) reduction occurred in the two experiments that combined UV-B and warming), water addition experiments and experiments combining warming with defoliation (Fig. 3). Not surprisingly all treatments that included N fertilization (85 species–treatment combinations in total) resulted in significantly lower (P < 0.05) C/N ratios in the litter. At the other side of the response spectrum were the treatments that included warming (except the UV-B and warming combination): in these 114 species–treatment combinations, the litter C/N ratios were significantly (P < 0.05) increased. This was also the case for the treatments that included elevated CO2: in all 20 species–treatment combinations, litter C/N ratios were significantly (P < 0.05) increased compared with the control treatment by up to 40%. The effects of warming only on litter C/N ratios (with, on average, a 17% increase in C/N ratio) were similar and not significantly different from the effects of elevated CO2 only (with an average increase of 12%). Overall, there was a three-fold difference in C/N response between the treatments for which the deviation from a zero response ratio was largest. This is in the same range as the natural variation (with regard to the variation between the 10th and 90th percentiles) in litter C/N ratio among species (Fig. 2).
Despite the sometimes substantial responses of litter C/N ratios to fertilization, warming and elevated CO2, the responses of litter decomposition rates to these treatments were only significantly (P < 0.05) different from zero for the treatments for which the deviation from a zero response ratio was largest (Fig. 4), with defoliation, warming and fertilization leading to significantly increased decomposition rates (P < 0.05) and elevated CO2 to decreased decomposition rates (P < 0.05). As expected, the treatment that resulted in the strongest increase in litter C/N resulted in decreased decomposition rates, whereas the treatment that resulted in the strongest decrease in litter C/N resulted in increased decomposition rates. Across all treatments, the regression analysis pointed to a marginally significant (P = 0.06) relation between the response ratios of the litter C/N ratio and the litter decomposition rate. The most striking result, however, was that these response ratios of litter decomposition were about an order of magnitude lower than the response ratios of litter C/N. This implies that leaf litter decomposition rates are not very responsive to global change-induced variation in litter stoichiometry.
We found high interspecific variation in litter stoichiometric traits and, from the results of 33 circumpolar global change experiments included in the MOL experiment, it is clear that stoichiometric ratios in high-latitude and high-altitude plant species show high responsiveness to global change. However, there appears to be only a weak link between stoichiometric flexibility and litter decomposition rates: the response ratios of litter decomposition were about an order of magnitude lower than the litter C/N response ratios. These findings are discussed further below.
High interspecific variation and intraspecific flexibility in litter stoichiometric traits
In agreement with our first hypothesis, we found considerable interspecific variation in litter stoichiometric traits for the sub-Arctic flora investigated, especially for litter C/P and N/P ratios compared with litter C/N (Fig. 2). As litter C concentrations showed relatively little variation, this implies that litter P shows much more variability than litter N. This is even more so for green leaves. For green leaves, a similar pattern of P concentrations being more variable than N concentrations was reported by Güsewell (2004) for lower latitude plant species and by Ostertag (2010) for nutrient-limited Hawaiian forests.
Biological regulation of litter stoichiometric ratios depends on processes causing variation in green leaf stoichiometry and differences in resorption efficiency among C, N and P compounds during leaf senescence. Thus, the patterns seen in leaf litter are caused by ‘afterlife’ effects of green leaves (cf. Quested et al., 2003). We discuss these processes first. There are two major types of explanation for interspecific variation and intraspecific flexibility of green leaf stoichiometry at large spatial scales (Güsewell, 2004; Reich & Oleksyn, 2004; Kerkhoff et al., 2005), which can be summarized as the physiological hypothesis and the biogeochemical hypothesis (Reich & Oleksyn, 2004). The physiological hypothesis identifies internal physiological mechanisms as regulators of leaf stoichiometry in response to climatic variation. The biogeochemical hypothesis, however, postulates that leaf N and P typically reflect soil N and P availability for plants, which, in turn, is affected by climate, but also by geological history and by plant-specific uptake mechanisms. Given the fact that we sampled within one specific biome (in contrast with the global databases of Reich & Oleksyn (2004) and Kerkhoff et al. (2005)) with a relatively similar climate, it is most likely that the patterns in green leaf stoichiometry observed in this study reflect spatial variation in soil nutrient availability, in combination with species-specific uptake mechanisms, such as ectomycorrhizal uptake of organic N compounds, P uptake by arbuscular mycorrhizal species or symbiotic N fixation (Aerts & Chapin, 2000; Townsend et al., 2007). This contention is supported by an analysis of Freschet et al. (2011), who compared leaf trait distributions in plant assemblages from the community level (including high-latitude communities) to the global scale, and found that 50% of total global leaf trait variation occurs within communities. Thus, within a seemingly homogeneous environment (community scale), there are many co-occurring ‘trait solutions’ to the requirements of the local environment. This implies that trait variability, including stoichiometric trait variability, occurs at fine spatial scales.
Our results for green leaf N/P ratios are in line with the high-latitude values reported in the global meta-analysis of Kerkhoff et al. (2005). At the global scale, they found lower Arctic N/P ratios compared with lower latitude biomes. This global pattern is related to the lower temperatures and smaller plant sizes in high-latitude ecosystems (Elser et al., 2010) and the very low N deposition (Aerts et al., 2001) and generally P-depleted soils in the tropics (Reich & Oleksyn, 2004; Kerkhoff et al., 2005). However, not all tropical soils are P depleted and nutrient requirements may vary, even at the local scale (Townsend et al., 2007).
The conspicuously higher stoichiometric ratios in leaf litter compared with green leaves (cf. Figs 1, 2) is a result of differential resorption of C, N and P compounds during leaf senescence. Structural C is hardly resorbed, whereas, in the sub-Arctic flora studied, average nutrient resorption efficiencies are very high (N, 66%; P, 63%; Freschet et al., 2010). This is considerably higher than the global average N and P resorption efficiencies, which are c. 50% (Aerts, 1996; Killingbeck, 1996; Yuan & Chen, 2009b). The median stoichiometric ratios for leaf litter determined in this study compare well with the mean values for tundra ecosystems compiled from literature sources by Yuan & Chen (2009a). However, their dataset, with only five observations for N/P ratios, six for C/P ratios and 26 for C/N ratios, is considerably smaller than the present one. Thus, our dataset provides more robust data on litter stoichiometry in high-latitude ecosystems.
The ultimate effect of the high nutrient resorption from senescing leaves is that the stoichiometric ratios in leaf litter move far away from the required stoichiometric ratios for microbial growth that are, according to the data on microbial stoichiometry presented by Cleveland & Liptzin (2007), almost an order of magnitude lower than those in the litters. The overall result of this mismatch is a long period of microbial net nutrient immobilization in the litter which, together with the harsh abiotic conditions, further contributes to the low nutrient availability for plants in high-latitude ecosystems (Schmidt et al., 1999; Aerts et al., 2006; Rinnan et al., 2007).
In response to the wide array of global change treatments represented in the MOL experiment (Table 1), we found a three-fold variation in litter C/N ratio. This was in the same range as the natural interspecific variation in litter C/N in a sub-Arctic flora. This points to high intraspecific flexibility in litter stoichiometry in these high-latitude plant species. This overall pattern raises the question of whether this high intraspecific stoichiometric flexibility in response to global change is also reflected in changes in litter decomposition rates.
Climate change-induced variation in litter stoichiometry hardly affects decomposition
The high variation in litter C/N ratios caused by the various global change treatments in the MOL experiment was clearly not reflected in conspicuous changes in litter decomposition rates over a 2-yr period (Fig. 4). The response ratios of litter decomposition were about an order of magnitude lower than those of litter C/N. At first sight, this may seem strange, as it has often been found in interspecific comparisons that the litter C/N ratio is a good predictor of the litter decomposition rates (e.g. Cadish & Giller, 1996; Parton et al., 2007). However, more recently, it has been shown that litter identity (the full array of traits that is included in each species) is a much better predictor than simple stoichiometric traits of the interspecific variation in litter decomposition rates (Cornwell et al., 2008). The reason for the observed relation between the interspecific variation in stoichiometric ratios and litter decomposition is that these ratios are correlated with the full array of chemical and structural litter traits which, in combination, determine decomposability. Recently, Freschet et al. (2012) have shown that the full set of traits of living plants (which corresponds to ‘species identity’) reflect the ‘Plant Economics Spectrum (PES) of litter decomposability’. They demonstrated for a broad range of sub-Arctic species that the PES based on multiple traits of living plant organs has important ‘afterlife’ effects on litter decomposition, as most of the traits are conserved during the transition from living plant material to litter. Thus, they found that the decomposabilities of all organs (leaves, fine stems, coarse stems, fine roots, reproductive parts) were consistently controlled by the same structure-related traits (lignin, C and dry matter content), whereas nutrient-related traits (N, P, pH, phenols) had a more variable influence.
Our data show that, on global change, the connection between stoichiometric ratios and structural litter traits is decoupled. Although stoichiometric ratios are highly responsive to change, this is not the case for most of the other structural and chemical traits that are part of the PES of litter decomposition, such as lignin content, specific leaf area or phenol content (e.g. Hobbie & Vitousek, 2000; Aerts et al., 2003, 2006; Vivanco & Austin, 2008). These findings suggest that the relative constancy of the constitutive (secondary) litter chemistry and structural traits of plant species (‘species identity’ or PES) largely dampens the effects of experimentally or climate change-imposed changes in the more responsive litter chemistry traits, such as N and P concentrations, on litter decomposition.
From these results, we conclude that litter from plants of high-latitude ecosystems shows strong stoichometric flexibility in response to global change, but that the effect is only weakly reflected in changes in decomposition rates. This implies that litter stoichiometry does not constrain the response of plant litter decomposition to global change; we suggest that the response is constrained by the full PES of litter decomposability, which is not very responsive to global change.
Green leaf and litter material from the Torneträsk region was partly collected and analysed by Helen Quested. We further thank all the members of the Meeting of Litters Team for making litter from their global change experiments available to us for a previous paper: Terry Callaghan, Richard van Logtestijn, Juha Alatalo, Terry Chapin, Renato Gerdol, Jon Gudmundsson, Dylan Gwynn-Jones, Anne Hartley, David Hik, Annika Hofgaard, Inga Jonsdottir, Staffan Karlsson, Julia Klein, Jim Laundre, Borgthor Magnusson, Anders Michelsen, Ulf Molau, Vladimir Onipchenko, Helen Quested, Sylvi Sandvik, Inger Schmidt, Gus Shaver, Bjørn Solheim, Nadja Soudzilovskaia, Anna Stenström, Anne Tolvanen, Ørjan Totland, Naoya Wada, Jef Welker and Xinquan Zhao. Many thanks are also due to Richard van Logtestijn and Jurgen van Hal for performing most of the chemical analyses. Finally, we thank Terry Callaghan and the staff of the Abisko Scientific Research Station for facilitating the experimental part of this study.