Field site and sampling

The GrIS is the second largest body of ice on Earth. Its area is ∼1 700 000 km² (>10% of the world’s ice surface), and its ablation zone stretches up to 100 km inland, comprising all ‘glacier ecological zones’, i.e. large areas of barren ice, slush and wet snow (Reference HodsonHodson and others, 2010a). Samples of surface debris (cryoconite) were collected between 1 and 15 August 2009 in the southwest part of the GrIS, along a transect across the entire ablation zone of the ice sheet. The transect consisted of 11 points beginning at the terminus of Leverett Glacier and finishing ∼100km inland (Table 1; Fig. 1). Cryoconite samples were collected in sterile 0.5 L Whirl-Pak bags (Nasco, Fort Atkinson, WI, USA) with a sterile 50 mL syringe and frozen immediately after return to the field camp by Leverett Glacier. Samples remained frozen at –20°C prior to analysis, whereby they were slowly thawed in the laboratory. The coverage of surface debris was estimated at each sampling point using the quadrant method (Reference Stibal, Tranter, Benning and RehákStibal and others, 2008a).

Table 1.

Sampling points along the transect across the ablation zone of the GrIS and the presence of meltwater and surface debris (cryoconite) during sampling in August 2009. Distance is from the terminus of Leverett Glacier. +, present; –, absent

Fig. 1.

Total organic carbon and total nitrogen in the cryoconite debris samples collected along the transect.

Organic carbon analysis

The total carbon (TC) and nitrogen (TN) contents of the debris were determined using a EuroVector EA3000 Elemental Analyser (EuroVector, Milan, Italy). Inorganic carbon was measured on a Strohlein Coulomat 702 analyser (Ströhlein Instruments, Kaarst, Germany) adapted for this purpose. Both analysers were calibrated using certified standards. Detection limits were 10ppm for both elements, and the precision of determinations was 0.3%. The total OC content was calculated as the difference between TC and inorganic carbon content.

Carbohydrate concentrations in the debris samples were analysed by ion chromatography with pulsed amperometric detection. This followed an acid-extraction protocol, involving hydrolysis, neutralization and desalting, to convert any polysaccharides and sugar derivatives to lower molecular weight components (mono- and disaccharides) for ion chromatography determination (Reference Jensen, Holmer and ThamdrupJensen and others, 2005). Briefly, 250 mg of wet debris was weighed into 8mL glass screw-cap serum vials and fitted with Teflon cap liners. 5 mL of 1 M H₂SO₄ was added and the vial sealed. Samples were hydrolysed at 110°C for 24 hours. After cooling, they were shaken for 30 min and centrifuged. The supernatant was then neutralized to pH 4–5 by the addition of NaOH and an aliquot diluted for analysis and stored in sterile polypropylene tubes. Samples were stored at –20°C until analysis. Each extraction procedure was conducted in triplicate, and blanks run to assess possible contamination during the procedure. In addition, carbohydrate standards were subject to the same extraction procedure to determine the loss of mono- and disaccharides during the hydrolysis step. Losses ranged from 53% to 97%, with fructose and sucrose destroyed entirely, in line with losses observed elsewhere (Reference Borch and KirchmanBorch and Kirchman, 1997). Hence, the concentrations of carbohydrates present in sediment samples represent minimum estimates.

Chromatographic analyses were performed by an ICS-3000 dual-analysis Reagent-Free Ion Chromatography system (Dionex, Camberley, UK), equipped with Chromeleon 6.8 software. Carbohydrates were separated isocratically by a CarboPac PA20 column (3 × 150 mm), with in-line CarboPac PA20 guard column (3 × 30 mm), and detected by a Dionex ED40 electrochemical detector operating in a pulsed amperometric mode and using non-disposable electrodes. The mobile phase consisted of degassed 15mM NaOH flowing at 0.35 mLmin⁻¹. During initial runs, xylose and mannose tended to co-elute due to similar retention times. However, separation was deemed necessary for subsequent interpretations of the origin of the bulk carbohydrate component (e.g. xylose indicates decomposition of plant residues, whereas mannose suggests the accumulation of microbial metabolites). Separation was achieved during a 30 min run, using a degassed 3 mM NaOH mobile phase and 0.4 mL min⁻¹ flow rate. The retention times of the compounds of interest were determined by running mixed standard solutions, and concentrations of individual compounds were obtained by directly interpolating the peak area in respect to the corresponding regression equation, supplied by the standard calibration. Calibrations were linear, based on a minimum of five standards, prepared daily from aqueous serial dilutions of a 1 mgL⁻¹ intermediate stock standard (prepared weekly). Analytical blanks were run in parallel. Drift was determined by running a 100 μgL⁻¹ standard after every ten samples. The column was regenerated every 2–3 days by washing with 200 mM NaOH for 45 min, and re-equilibrating with 15 mM NaOH for a further 45 min, which removed carbonates and other contaminants from the column. Detection limits, calculated as 3x signal-to-noise ratio, were close to 0.5 μgL⁻¹ for all carbohydrates. Relative standard deviations for individual carbohydrate concentrations ranged from 0.9% to 45.7%

Chlorophyll extraction

Chlorophyll was extracted from 1 g of wet cryoconite debris in clean 15mL centrifuge tubes by adding 10mL 96% ethanol, shaking on a reciprocating shaker for 30 min at 200 rpm and leaving at 4°C in the dark for 24 hours. The tubes were then shaken again for 30 min at 200 rpm and centrifuged at 2600 rpm for 12 min. The extracts were analysed using a Shimadzu UV-Mini 1240 spectrophotometer (Shimadzu, Kyoto, Japan) at wavelengths of 665 and 750nm. Calibration was conducted using chlorophyll-a standards (1–10 000μgL⁻¹). Procedural blanks containing no cryoconite debris were also analysed.

Microbial abundance

10mg of freshly melted wet cryoconite debris and 1 mL of pre-sterilized deionized water was placed in a sterile Eppendorf tube and mixed by vortexing for 30s. In order to fix the sample, 0.2 μm filter-sterilized 37% formaldehyde (final concentration 1%) was added to the tube, followed by a brief vortexing. The suspension was then filtered onto a sterile 0.2 μm Anodisc filter (Whatman, Maidstone, UK). Dried filters were placed onto 100 μL drops of 2× SYBR Gold (Invitrogen, Eugene, OR, USA) for 15 min, dried and mounted on microscopic slides with the anti-fade agent Citifluor AF2 (Citifluor, Leicester, UK). More than 300 SYBR Gold stained cells were enumerated on each slide with an Olympus BX41 epifluorescence microscope (Olympus Optical, Tokyo, Japan) using the filter block U-N31001 (excitation 480 nm, emission 535 nm; Chroma Technology, Rockingham, VT, USA). Cells showing red chlorophyll autofluorescence were enumerated using the filter block U-MWU2 (excitation 330–385 nm, emission 420+; Olympus Optical, Tokyo, Japan). Two slides per sample were counted. A procedural blank with no cryoconite sediment was counted in parallel and the average blank counts were then subtracted from the experimental sample counts.

Distribution of cryoconite debris on the ice sheet

Close to the margin (point 0km), large amounts of debris were scattered in a thin layer covering much of the ice surface (Fig. 4a). This debris was different from that in cryoconite holes further up-glacier in being much coarser and lighter in colour. It is likely to have partly originated from pressure ridges (Fig. 4a), and so to have been of subglacial origin (Reference HodsonHodson and others, 2010a). More than 2 km inland, cryoconite debris was present on the surface (Fig. 4b), covering up to 10% of the surface, and increasing in its per cent areal coverage from 2 to 10km. From 10 to 50 km the amount of visible surface debris decreased again, with most covered by a layer of firn or ice (Fig. 4c and d). No cryoconite was found at 70 and 100 km inland. Since most microbial activity and organic matter is associated with the debris in glacier surface environments (Reference Sawstrom, Mumford, Marshall, Hodson and Laybourn-ParrySäwström and others, 2002; Reference HodsonHodson and others, 2007), this variation in debris distribution may be a key factor in determining the spatial distribution of productivity in the ecosystem.

Fig. 4.

Surface of the GrIS along the transect. (a) Terminus of Leverett Glacier with pressure ridges and supraglacial kames. (b) Open cryoconite holes and dispersed cryoconite debris at 8 km inland. (c) Ice-lidded cryoconite holes at 20 km inland. (d) At 50 km cryoconite debris is hidden below the ice surface.

Quantity and quality of organic carbon on the ice sheet

The TOC and TN contents found in the debris on the GrIS are within the range reported from smaller glaciers (Reference TakeuchiTakeuchi, 2002; Reference Stibal, Tranter, Benning and RehákStibal and others, 2008a). The C: N molar ratio of 8–9, determined in most samples, is typical of values found in soils (Reference Cleveland and LiptzinCleveland and Liptzin, 2007) and slightly lower than that found in cryoconite on other Arctic and alpine glaciers (Reference TakeuchiTakeuchi, 2002; Reference Stibal, Tranter, Telling and BenningStibal and others, 2008b), and suggests some biological control in the system. The C: N ratio in the sample from 4 km is very low due to the very high N content determined in the sample.

The fraction of carbohydrates within the debris OC pool is relatively small at most sites (1–16%), similar to the lower range of values found in ocean surface sediments (Reference Cowie, Hedges and CalvertCowie and others, 1992; Reference Kerhervé, Buscail, Gadel and ServeKerhervé and others, 2002). In samples from the 4–8km band, there is a significant increase in the contribution of carbohydrates (29–79%), suggesting an abundance of fresh biogenic organic matter (Reference YamaokaYamaoka, 1983; Reference Kerhervé, Buscail, Gadel and ServeKerhervé and others, 2002), for example, glucose and galactose, two hexoses well known to be labile and utilized by various heterotrophic organisms (Reference Hernes, Hedges, Peterson, Wakeham and LeeHernes and others, 1996). This suggests both efficient OC production by primary producers and the provision of a bioavailable OC substrate for heterotrophic metabolism.

Further insight into the OC source material and biochemical transformations can be derived from the distribution of carbohydrates. Compositional spectra dominated by galactose, rhamnose and fucose have been linked with macroalgae, macrophytobenthos and bacteria, respectively (Reference LiebezeitLiebezeit, 1988). Xylose is indicative of angiosperms (Reference Opsahl and BennerOpsahl and Benner, 1999) and a useful biomarker for plant residues within the sediment. Several ratios of carbohydrates have been employed to interpret OC data (Reference Biersmith and BennerBiersmith and Benner, 1998; Reference Khodse, Fernandes, Bhosle and SardessaiKhodse and others, 2008). We use the ratio of (galactose + mannose)/(arabinose + xylose) to provide an indication of the origin of the monosaccharides since arabinose and xylose are usually not synthesized by microorganisms and so originate mostly in plant materials (Reference Turchenek and OadesTurchenek and Oades, 1979). The determined ratios (>1) in our samples from all sites suggest an accumulation of microbial metabolites (i.e. autochthonous OC), with some input of allochthonous terrestrial carbohydrates from plant material. The highest (galactose + mannose)/(arabinose + xylose) ratio was found at the 10 km site (Fig. 2b), and the 35 and 50 km sites also had high ratios. This suggests that microbial production is dominant over allochthonous organic matter input higher up on the ice sheet.

The numbers of photoautotrophic cells were within the range found in cryoconite holes on small valley glaciers on Svalbard (Reference Sawstrom, Mumford, Marshall, Hodson and Laybourn-ParrySawstrom and others, 2002; Reference Stibal, Šabacka and KaštovskáStibal and others, 2006; Reference Stibal and TranterStibal and Tranter, 2007), although the concentrations of chlorophyll were lower than reported from Svalbard (Reference Stibal, Tranter, Benning and RehákStibal and others, 2008a). However, due to the very low and variable cell numbers and chlorophyll concentrations and the resulting relatively high measurement errors, it is difficult to discern a trend in the abundance of photoautotrophs in the debris along the transect. The heterogeneity in their distribution within the debris, consistent with observations from other glaciers, can be attributed to the tendency of the dominant filamentous cyanobacteria to form clusters (Reference Stibal and TranterStibal and Tranter, 2007).

The concurrent peaks in the carbohydrate proportion and the numbers of all microbial cells within the debris at ∼ 6 km inland mark a zone where high microbial growth and production coincides with potential in situ accumulation of organic matter. From this we can develop a concept of what controls the distribution of organic matter on the ice-sheet surface.

Controls of the distribution of organic matter content in debris on the ice sheet

Three key processes are likely to exert a control over the quantity and quality of OC in supraglacial debris on the GrIS. Two simple conceptual models (Fig. 5) present the possible effects of these processes on the OC content within the surface debris. These models are based on the following hypotheses and supported by the OC data shown in Figures 1 (total OC) and 3a (cell-contained OC).

Fig. 5.

Conceptual models of the OC distribution across the ablation zone of the GrIS. (a) Microbial abundance and the cell-contained OC (cf. Fig. 3a). (b) TOC (cf. Fig. 1). WZ–OC: wash-away zone; AZ–OC: accumulation zone.

First, the input of aeolian debris, containing microbial cells and other organic matter, to the glacier surface is dependent on the source and size of particles transported (Reference Bøggild, Brandt, Brown and WarrenBøggild and others, 2010). Adjacent deglaciated areas are the most likely source for the majority of cryoconite debris (Reference Stibal, Tranter, Telling and BenningStibal and others, 2008b; Reference Bøggild, Brandt, Brown and WarrenBøggild and others, 2010; Reference Xu, Simpson, Eyles and SimpsonXu and others, 2010). It is possible that loose organic matter fragments are transported further than mineral particles due to their lower relative density (Reference ZobeckZobeck, 1991), in which case the proportion of OC in the debris could increase inland (Fig. 5b). It is also expected that larger particles of plant residue will only be transported a short distance. Active microbial cells, on the other hand, are usually attached to mineral particles due to producing exopolymers (Reference HodsonHodson and others, 2010b), and so may be transported with mineral dust particles. In the latter case, the cell-contained OC within the debris would be constant along the transect (Fig. 5a). The role of debris that has melted out of old ice in this scheme is unclear; however, as its origin is likely far from the ice sheet and delivery by long-distance transport mechanisms (Reference Bøggild, Brandt, Brown and WarrenBøggild and others, 2010), no spatial variability of its OC content across the ice sheet would be expected.

Second, biological carbon transformation changes the concentration of OC in the system. Photoautotrophic microbes fix CO₂ from the atmosphere into organic matter, and thus add C to the system. On the other hand, all microbes respire, or ‘burn’ organic C into CO₂, back into the atmosphere and so remove C from the system. A wide range of in situ rates of photosynthesis and respiration and their ratios (P: R) have been determined on various glaciers to date (Reference HodsonHodson and others, 2007, Reference Hodson2010a,Reference Hodsonb; Reference Stibal, Tranter, Benning and RehákStibal and others, 2008a; Reference Anesio, Hodson, Fritz, Psenner and SattlerAnesio and others, 2009; Reference TellingTelling and others, 2010). This means that there is the potential for glacier surfaces to be net autotrophic (CO₂ sinks; P :R >1) or net heterotrophic (CO₂ sources; P: R < 1), and both cases may be of high regional importance (Reference Anesio, Hodson, Fritz, Psenner and SattlerAnesio and others, 2009; Reference HoodHood and others, 2009). However, it is important to realize that even a balanced P :R does not necessarily imply a constant quantity of the cell-contained OC in the system, since heterotrophic growth can also be sustained by the uptake of allochthonous OC (Reference Margesin, Zacke and SchinnerMargesin and others, 2002; Reference AnesioAnesio and others, 2010). As the presence of liquid water is likely to have a profound effect on all microbial processes on the ice sheet, lower-altitude areas closer to the ice-sheet margin are potentially more productive in terms of cell numbers and the cell-contained OC than more distant, and colder, inland areas.

Third, the wash-away effect of supraglacial meltwater may play an important role in the steep marginal zone of the ice sheet. Organic matter has been shown to be removed from cryoconite debris over time (Reference Stibal, Tranter, Benning and RehákStibal and others, 2008a), and the wash-away may result in a decreased content of OC in supraglacial debris in the marginal zones of ice sheets.

Figure 5a is a conceptual model of the three described key environmental effects on the abundance of microbial cells and the cell-contained OC within supraglacial debris on the GrIS. Viable cells are assumed to be transported with dust, so the input of the cell-contained OC is constant across the ablation zone in the model. The productivity component (change in the cell-contained OC in the system) of this model is dependent on the P: R ratio and the heterotrophic activity in the microbial community. We assume the simplest case of a balanced P: R, supported by data from Greenland (Reference HodsonHodson and others, 2010a) and Svalbard (Reference TellingTelling and others, 2010), and utilization of both autochthonous and allochthonous OC (Reference AnesioAnesio and others, 2010) in the model. The highest productivity is predicted to be related to the extent and duration of melting, so it is negatively correlated with distance from the edge of the ice sheet in the model. The wash-away effect occurs in the steepest part of the ice sheet, i.e. close to the edge. The combined effect of these processes (dust transport, productivity and wash-away effects) produces a maximum percentage of the cell-contained OC within supraglacial debris that is located at the border between the ‘carbon wash-away zone’, where rapidly flowing meltwater efficiently removes large amounts of carbon and transports it to subglacial and proglacial ecosystems, and the ‘carbon accumulation zone’, where the presence of liquid water and inputs of microbes and nutrients via dust transport decrease inland. This is supported by the microbial-abundance and carbohydrate data shown in Figure 3a.

Another scenario is suggested for the TOC content in the debris (Fig. 5b). Here the easier atmospheric transportability of loose organic matter compared to mineral dust is accounted for, and results in a proportional increase in the OC content of supraglacial debris with increased distance inland. The productivity component (change in the TOC in the system) of this model is only dependent on the P: R ratio in the microbial community, again assumed to be balanced (Reference HodsonHodson and others, 2010a,Reference Hodsonb), and not on secondary production that only recycles OC in the system. The wash-away effect of meltwater is identical to that in the previous model. The resulting percentage of the TOC within cryoconite debris then increases from the edge of the ice sheet inland, although the total amount of OC on the surface can actually decrease. This is supported by the TOC and TN contents shown in Figure 1 .

These models are very simple and only show the environmental effects on the percentage of OC in surface debris, rather than on the amount of OC per unit area, which is more important for a global picture of supraglacial productivity. It is also possible that the P : R ratios will be significantly different from balance at various sites. However, the models demonstrate how simple physical and biogeochemical processes may control C cycling in large ice-sheet ecosystems.