Dust from the dark region in the western ablation zone of the Greenland ice sheet

Dust from the dark region in the western ablation zone of the Greenland ice sheet I. G. M. Wientjes, R. S. W. Van de Wal, G. J. Reichart, A. Sluijs, and J. Oerlemans Institute for Marine and Atmospheric research Utrecht (IMAU), Utrecht University, Princetonplein 5, 3854 CC, Utrecht, The Netherlands Department of Earth Sciences, Utrecht University, Budapestlaan 4, 3584 CD Utrecht, The Netherlands Alfred Wegener Institute for Polar and Marine Research, Am Handelshafen 12, 27570, Bremerhaven, Germany Biomarine Sciences, Institute of Environmental Biology, Utrecht University, Laboratory of Palaeobotany and Palynology, Budapestlaan 4, 3584 CD Utrecht, The Netherlands


Introduction
The western ablation zone of the Greenland ice sheet contains a region, which is darker than the surrounding ice ( Fig. 1). This dark region is situated some tens of kilometres from the margin and stretches from 65 • N to 70 • N. The dark appearance implies lower albedos and therefore enhanced melt rates. Unravelling the underlying 5 processes that force the low albedo is critical to determine the existence of this region. More knowledge about the dark region will help to predict how this region will develop in the future, and what its influence will be on the mass balance of Greenland. Accumulation of melt water at the surface could cause the observed dark region, since water might drain slowly due to the cold ice and the small surface slope (Van de Wal and Oerlemans, 1994;Knap and Oerlemans, 1996;Greuell, 2000). However, Wientjes and Oerlemans (2010) compared the spectral signature of the dark region to that of the surrounding brighter ice, showing a pattern that is typical for ice containing more dust. This dust can have two main sources; recent enhanced deposition of dust or accumulation through outcropping of old dust layers. Because the dark region appears 15 every year during the summer season, fixed at the same position at some distance from the margin, the dust unlikely represents only recent dry or wet deposition. In addition, satellite images reveal a wavy pattern in the dark ice, which is typical for the outcropping of tilted, stratified ice layers (Fig. 2). Therefore, the dust was probably initially deposited higher on the ice sheet, travelled through the ice sheet towards the 20 margin and accumulated in the ablation zone (Wientjes and Oerlemans, 2010). In the ablation zone, a horizontal surface ice profile perpendicular to the margin of the ice sheet represents a time line; the closer to the margin, the older the ice (e.g. Reeh et al., 1987). Dust causing the dark region may thus originate from a period when more dust settled in the accumulation zone of the ice sheet. It might have been deposited during 25 cold periods, due to dryer and windier conditions (e.g. Fuhrer et al., 1999;Ruth et al., 2003). Alternatively, the dust could be volcanic material that was deposited on the ice (e.g. Clausen et al., 1997;Mortensen et al., 2005). Reeh et al. (2002) (2005) found bands of dark ice near Jakobshavn, closer to the margin, containing dust from a colder glacial period. A similar dark region has been found in North-East Greenland, probably caused by dust that was deposited high on the ice sheet during the Pleistocene (Bøggild et al., 1996. Dust on the ice is often called cryoconite, as suggested by Nordenskjöld (1875).
When this dust accumulates on the ice surface, it absorbs sunlight and melts in the ice forming cryoconite holes. The holes deepen until the melt rate at the bottom of these holes equals the ablation rate at the ice surface (Gribbon, 1979). They are typically a few centimetres in diameter, and several centimetres to decimetres deep, water filled and with the cryoconite lying on the bottom (e.g. Gajda, 1958). Because the cryoconite 10 holes contain nutrients released from the dust minerals, water and sunlight, they form an ideal habitat for microorganisms (Wharton et al., 1985). Cryoconite can significantly lower the albedo of the ice surface in the visible part of the spectrum (Takeuchi et al., 2001;Takeuchi, 2009;Bøggild et al., 2010). The locally formed organic matter of the cryoconite is one of the main factors for this albedo lowering. Takeuchi (2002) 15 found that the amount of dark coloured organic substances, possibly the residue of bacterial decomposition of organic matter, determines the optical characteristics of the cryoconite. In addition, the aggregation of dark particles and organic debris by extracellular polymeric substance affects the light absorbency of the cryoconite (Hodson et al., 2010). 20 To elucidate the origin and characteristics of the surface dust in the dark region, cryoconite from the dark ice as well as from the brighter ice was collected and analysed. We sampled at various locations along the K-transect, a series of surface mass balance and weather station sites in western Greenland (Van de Wal et al., 2005). The sites are situated on a line perpendicular to the ice sheet margin around 67 • N, and cover Introduction

Samples
Cryoconite was collected from six sites along the K-transect (Table 1) on 2 and 3 September 2009. The number of samples was restricted by logistical constraints and surface conditions like snow or frozen holes. Table 1 shows the number of samples 5 collected. At S8 we only managed to take small sized samples, which did not yield enough material for the full suite of analyses. At some locations the cryoconite holes were deeper than expected; at S6 they were even about 50 cm deep. Although the holes were variable in depth among the locations, they had similar depths at individual sites, consistent with previous findings (e.g. Gribbon, 1979). At S4, S5 and S6 large patches or streams of dust were present on the surface (Fig. 4). We collected dust from the holes as well as from these dust patches and streams. In between sampling, tools were rinsed with distilled water to prevent cross contamination among the samples. Analyses revealed no significant differences between dust from surface patches and streams, or dust from cryoconite holes.

Geochemical analyses and microscopy
The cryoconite was studied with an optical microscope as well as a scanning electron microscope (SEM, XL30FEG, FEI). Together with SEM, Energy-Dispersive X-ray spectroscopy (EDX) was performed, to qualitatively determine the bulk mineralogical composition of the individual observed grains. For these measurements, material was 20 mounted on a stub, dried and coated with carbon. The bulk mineralogical composition of the whole samples was qualitatively analysed by X-Ray Diffraction (XRD), using a powder diffractometer (AXS D8 Advance, Bruker) with Co-Kα radiation. The different minerals were identified based on their spectral characteristics. Total Organic Carbon (TOC) and Nitrogen (N) concentrations were measured using Introduction analyses by repeated rinsing with 1 M HCl (4 h, 12 h) followed by washing with demineralized water. Concentrations measured on the decalcified residue were back corrected to their original values using the determined weight loss. Because samples from S7 were too small for this procedure, they were decalcified by acidification with 25% HCl in situ within silver sample cups as described in Nieuwenhuize et al. (1994). Based on 5 duplicate analyses and including in-house standards the relative standard deviations for TOC and N were both better than 2% of the recorded weight percentage of TOC. Total dust samples were dissolved in an acid mixture of 2.5 ml mix acid (HClO 4 :HNO 3 = 3:2) and 2.5 ml HF, and subsequently evaporated until gel was formed before being taken up again in 25 ml 4.5% HNO 3 suprapur acid. Major elements were 10 quantified using Inductively Coupled Plasma Atomic Emission Spectroscopy (ICP-AES, Varian), minor and trace elements using Inductively Coupled Plasma Mass Spectroscopy (ICP-MS, X-Series II, Thermo Scientific). The elemental concentrations were calibrated against standard solutions. Duplicates were used to calculate relative standard deviations. The precision for all elements shown in this paper is better than 2.5% 15 relatively, with exception of Antimony, Arsenic and Mercury, which have a precision of 7.8%, 8.7% and 12.7% respectively.

Grain shape and composition
Visual inspection using a light microscope and SEM analyses reveal that the overall 20 grain shapes at the various sites are similar. Most grains have rather angular shapes and are characterized by sharp edges (Fig. 5a). The fragile, acute angles are typical for volcanic material, but have also been observed in glacial grains (Mahaney, 2002). However, the high amounts of these angular-faceted, sharp-edged grains are found in the dark region as well as in the reference ice, where precipitation rates are low and TCD 4,2010 Dust from the ablation zone in West Greenland volcanic source. The other possibility for volcanic material to reach the ablation zone is settling in the accumulation zone of the ice sheet during periods of volcanic activity, followed by transport through the ice, and subsequent accumulation through prolonged melting in the ablation zone. This outcropping volcanic material would, hence, be enriched in volcanic grains. Still, also grains transported together with the volcanic dust, 5 but originating from different sources should accumulate here. The overwhelming presence of acute edged grain particles, and the lack of rounded grains suggest that these grains became angular and sharp-edged due to glacial transport itself, independent of their origin. Due to the grain tot grain contact in the ice, the grains may suffer abrasion. Abrasion features on some of the grains (see Fig. 5b) confirm that the grains indeed 10 have been transported through the ice. In theory, the material could also be entrained from the underlying bedrock. This seems unlikely, however, as the higher sites (S7 and S8) are located close to the equilibrium line, and particle trajectories through the ice towards these sites do not get close to the bottom. We therefore conclude that at least part of the dust comes from higher parts of the ice sheet, outcropping in the ablation 15 zone.
The chemical composition of individual grains was assessed by EDX, whereas the bulk dust mineralogy was determined using XRD. Although the relative abundance of the various minerals might differ between the sites, all the samples analysed contained largely the same minerals. The EDX showed grains, each consisting of a single min-20 eralogy, mainly quartz, feldspars and plagioclases. A dominant contribution of these 3 minerals to the bulk dust composition is in line with the results of the XRD analyses. The samples mainly consist of quartz and albite with some anorthite. Within these dominant components minor amounts of phyllosilicate were detected as well.
The cryoconite at S4 seems coarser relative to the other sites. Visual interpretation 25 with SEM revealed that material from this site contains larger grains than observed in the other samples. In addition, the EDX analyses indicate that the dust from S4 consists of quartz particles with some feldspars and plagioclases, whereas at other locations, the relative abundances of plagioclases and feldspars seem higher with respect to the quartz concentration. S4 is located close to the margin, allowing for a possible influx of some wind-blown material from the nearby tundra. Possibly even some material entrained from the bedrock has been added to the other grains at this site. The high relative abundance of feldspars and plagioclases relative to the quartz in the samples could be due to a lesser degree of chemical weathering, or a more felsic nature 5 of the source rock for the dust. For this reason, sources that have undergone significant weathering, such as old deserts, can be excluded as the origin of the cryoconite, because they mainly consist of quartz. Maggi (1997) investigated changes in mineralogy of atmospheric dust in the GRIP ice core spanning the time interval between the Eemian and the last glacial maximum, 10 and found two distinct assemblages with different characteristics during cold and warm episodes. One consists mainly of quartz, illite, chlorite, micas, and feldspars, related to cold mid-high-latitude source areas; and the other consist mainly of kaolinite and Fe (hydr)oxides, related to warm and humid low-latitude source areas. As our samples does not contain kaolinite and as we did not detect any Fe (hydr)oxides-coated grains 15 using EDX, our samples compare most favourably with the first assemblage of Maggi (1997), excluding low-latitude areas as a source for the cryoconite. Svensson et al. (2000) investigated twelve glacial dust samples from the GRIP ice core, and indicated eastern Asia being the main source for these samples originating from the Younger Dryas, the Bølling, the Last Glacial Maximum and cold and mild glacial periods back 20 to 44 kyr B.P. He found higher kaolinite/chlorite ratios and lower feldspar content in the dust from cold glacial periods relative to warmer periods. This is in contrast with the findings by Maggi (1997). However, as the mineralogical composition of our samples differs from the mineralogical composition of all samples from Svensson et al. (2000), eastern Asia does not qualify as the main source for the dust in the dark region. This is Summarizing, part of the dust consists of outcropping material that initially settled higher on the ice sheet during dustier periods, although common source areas for glacial dust, like Asian desserts, can be excluded as provenance for this dust.

Elemental composition
The samples contain similar concentrations of most elements. However, the samples 5 from S6 and S7 contain less inorganic material, because they were diluted by higher amounts of organic matter, as we will explain in Sect. 3.3. To overcome misinterpretation, we corrected for the amount of organic matter in the samples. We have measured the TOC and N concentrations of the samples and we assume that the total organic matter can be described by the mass of CH 2 O and N. We multiplied the relative TOC 10 concentrations for each site by 2.5, based on the relative molecular weights ratio of C:CH 2 O is 12:30, and add the relative concentration of N. The percentage of organic matter obtained in this way, was used to correct all the results shown in this paper.
When plotting elements against each other, samples seem to cluster into two groups with S4 and S5 in one cluster, and S6 and S7 in the other cluster. This suggests com-15 positional differences between samples from the dark region and from the reference ice, as S6 and S7 are both located in the dark region, whereas S4 and S5 are located in the brighter ice close to the margin. Figure 6 shows this clustering when Tin (Sn) is plotted against Aluminium (Al). The subdivision in two groups is also observed for other elements. The results for SHR are different from both the two clusters, but as 20 there is only one single sample from this site, it is hard to assign a third cluster and draw firm conclusions.
For each element we divided the average concentration for samples from the dark region by the average concentration for samples from the reference ice. Figure 7 shows the results for part of the elements measured with either ICP-AES or ICP-MS. Values 25 above 1 indicate that these elements are relatively enriched in the dark region with respect to the reference ice. Especially Caesium (Cs) is much more abundant in the dark region, but also heavy metals such as Lead (Pb) (Hg) have larger concentrations in the dark region. As these are all elements that are relatively enriched in the atmosphere by anthropogenic pollution, a recent source (after industrial revolution) for these elements seems likely. S6 and S7 (the sites in the dark region) are located some tens of kilometres away from the equilibrium line. It is unlikely that dust has travelled this distance through the ice sheet within a short period of about 5 100 years. Therefore these elements do apparently not originate from the part of the dust that crops out from the ice. This implies recent deposition, which does not show up near S4 and S5, probably because most snow is blown away immediately at these sites, and melt rates are higher with respect to the dark region. In addition, these elements might accumulate due to the uptake by microorganisms, which are much 10 more abundant in the dark region (Sect. 3.3). Figure 8 shows the rare earth elements (REE) normalized to the averaged upper continental crust (Taylor and McLennan, 1995). The results for each location are averaged. For most of the elements, this value lies close to 1. That means that the average composition of the collected surface dust resembles the average composition of the 15 upper earth crust, as well as the average composition of volcanic material, as most of the upper earth crust originates from volcanic material. However, Icelandic volcanoes show a different pattern and can therefore be excluded as possible origin of this dust. For comparison, the REE pattern from tephra of the February 2000 eruption of Hekla (Moune et al., 2006) is added in Fig. 8. In addition, the REE patterns for the different 20 sampling sites resemble each other closely. REE patterns of dust samples show often more variation, for example glacial dust from ice cores in Antarctica (Basile et al., 1997). As all sites are located at different distance from the margin, the outcropping material originates from different periods. If dust from a single event, like a large volcanic eruption, would have settled in the accumulation zone of the Greenland ice sheet 25 and is outcropping now at one of these sites, it should alter the REE pattern for that specific site. As this is not the case, we exclude a single volcanic eruption as possible provenance for the dust. As the REE patterns for the different sites do not show large variations, they seem to have a common source area, that provide dust to the Greenland ice sheet at all these different periods. This makes a regional source most plausible. We conclude that the dust from different sites has mainly the same provenance; although dust from the dark region is enriched with elements from the current atmosphere by higher accumulation rates. The outcropping dust seems to have a regional 5 origin.

Microorganisms
Visual inspection with a light microscope using normal light revealed that the cryoconite also contains microorganisms (Fig. 9). No organisms are present at Site S4 and only small granules containing organic matter were detected in samples from S5. However, 10 samples from all other sites contain long yellow-brownish filamentous strings that represent cyanobacteria (Takeuchi, personal communication). These bacteria are present, not only as loose strings in the samples (Fig. 9a), but are also aggregated in granules, together with mineral particles and organic matter. At S6 these cryoconite granules appeared very dark-coloured using the transmitted light microscope and are therefore 15 large and thick-walled (Fig. 9b), whereas the cryoconite granules from SHR, S7 and S8 were covered with yellow or brown strings of cyanobacteria (Fig. 9c). In addition, we noted abundant occurrence of green algae at the higher altitude sites (Fig. 9d).
The TOC measurements on dry cryoconite confirmed our microscope observations that organic matter was abundantly present at sites from the dark region. Figure 10   20 shows the averaged percentage of total organic carbon (TOC) and nitrogen (N) for the different sites. The ratio C/N ranges between 11.3 and 13.5 for all the samples from SHR, S6 and S7. Because the TOC and N values are low for the samples from S4 and S5, their C/N ratio is determined by the slope of the trend line through the plotted values of C against N. This is done to overcome misinterpretation as result of 25 small nitrogen leakage during the measurements. The C/N ratio is around 12.2 for the samples from these sites. Takeuchi (2002) found C/N ratios between 9 and 14 (mostly around 11) for different cryoconites of nine glaciers from the Himalaya, Tibet, and the 2567 Introduction Arctic. In addition, Hodson et al. (2010) found a C/N ratio of 13 ± 1 for cryoconite from Longyearbreen, a valley glacier in Svalbard. Therefore, our findings seem to compare well with these literature values.
Concentrations of TOC and N seem to increase with elevation. All the samples from S4 have a TOC content <0.2 weight percent (%) and all the samples from S5 have a 5 TOC content <0.8% (the lowest value is 0.36%). In contrast, the samples from S6 and S7 contain between 4.5% and 5.5% TOC. From SHR we have only one measurement, indicating a TOC content of 2.6%. The total organic carbon in the dark region (S6 and S7) is rather high. Takeuchi (2002) found values of 0.54-4.37% TOC for the cryoconites from nine different glaciers, with 4.37% for Austre Brøggerbreen in Spitsbergen, Nor-10 way and respectively 2.22% and 2.06% for Penny Ice Cap, Greenschield Glacier in Baffin Island and Devon Ice Cap, Sverdrup Glacier in Devon Island, both in Nunavut, Canada. Bøggild et al. (2010) found cryoconite from northeast Greenland containing 5% organic matter, and Takeuchi (2009) found 2.9% to 8.8% organic matter in the surface dust from ice from the Gulkana Glacier, Alaska Range, USA. If we assume that 15 organic matter can be described by the mass of CH 2 O, the amount of organic matter corresponding with 5% TOC is 12.5%, calculated by using relative molecular weights. Together with the measured N, this becomes 13%. This value is considerably higher than the values found by Bøggild et al. (2010) and Takeuchi (2009). However, as organic matter might not only contain CH 2 O and N, this might be an underestimation, 20 with real values of organic matter being even higher.
It is not obvious why the biological activity increases with altitude. The amounts of useful elements provided by the cryoconite are equal in the dark region with respect to the reference ice. Phosphorus, which is an essential nutrient for the cyanobacteria, is even higher at S4 compared to the other sites. Also the pH of the ice, that might 25 affect the biological activity, is found to be the same at all sites, between 4.6 and 5.2, both in the cryoconite holes as well as in the surrounding supraglacial streams. These values are slightly lower than values for surface ice from the Gulkana Glacier in Alaska, which are between 4.9 and 5.7 (Takeuchi, 2001). At the higher elevations, the ice is Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | also longer covered by snow, which is more likely to reduce algal growing. However, global radiation strongly increases with altitude at the Greenland ice sheet, because of a decreasing cloudiness and a more transparent atmosphere (Konzelmann et al., 1994). This effect might be an important factor for the biological activity. Stibal et al. (2006) suggest that microorganisms prefer fine sediment in the cry-5 oconite. As the cryoconite at S4 contains coarser material relative to the other sites, this might also contribute to the absence of organisms at S4. Takeuchi (2001) found that the abundance of algae increase with altitude at Gulkana Glacier, Alaska, caused by the outwash of the algae at lower sites where more melt occurred, as the microorganisms on this glacier mainly consist of unicellular green algae that do not form cry-10 oconite granules. However, if aggregations formed with cyanobacteria, organic matter and mineral particles, these granules are more resistant against flushing by melt water Takeuchi (2001). In our case these aggregations formed, and it is very unlikely that they completely washed out at S4, were we found no microorganisms but considerable amounts of dust. We also found green algae, which were only abundant at the higher 15 elevations. In contrast to the cyanobacteria, these green algae could be flushed away from the glacier at the lower sites, due to the higher melt rates. Despite the fact that it is uncertain what causes the occurrence of cryoconite granules at higher altitudes, it is well known that they have a high light absorbency (e.g. Takeuchi et al., 2001). As there are abundant of these granules in the dark region, we 20 can conclude that biological activity will certainly contribute to the surface darkening process.

Conclusions and discussion
In order to find out more about the characteristics of cryoconite from the dark region, surface material was collected and analyzed from locations inside the dark region, 25 and from locations of brighter ice closer to the ice margin. Using transmitted light microscopy as well as electron microscopy, we observed triangular-faceted, sharp-edged grains, which are either volcanic or glacial grains, both indicating glacial transport. This implies that at least part of the dust is outcropping from the ice, in the dark region as well as in the reference area. Geochemical analyses revealed that the mineralogical and elemental composition of the material is mostly the same for both areas. However, enrichment with elements from anthropogenic sources in the dark region suggests an 5 influence from recent atmospheric material by precipitation.
In addition, abundant green algae and cyanobacteria, forming cryoconite granules, were found in the dark region. The relative amount of microorganisms in this region is much higher compared to the reference ice. It is not quite clear what causes these differences, although global radiation, sediment texture and meltwater supply might 10 play a role. The TOC values found for the dark region are high compared to literature values. As the organic matter in the cryoconite is well known to have a high light absorbency, the dark region is not only caused by outcropping dust, but biological activity also significantly contributes to the darkening of the surface.
The REE patterns for the different sites reveal a source with a chemical composition 15 similar to the upper continental crust. Therefore, the material does not seem to contain meteorites, which are found incidentally on the Greenland ice sheet in relatively high amounts (Maurette et al., 1987). Besides, meteorites were not revealed by the microscopy at all, and seem therefore not present in the dust samples. The outcropping dust could be volcanic material, but as the dark region is several tens of km wide, 20 a series of several eruptions are more likely to have caused the dark region than one single big eruption. However, nearby Icelandic volcanoes provide a different REE pattern and are therefore excluded as possible source. Comparison of the mineralogical composition of the dust with literature values shows that the samples do not contain material from Asian deserts, which is often found in ice cores during glacial periods. the investigated dust has a local source. Therefore, the dark region is probably caused by higher amounts of local dust, during periods when more dust settles in the accumulation zone of the ice sheet. Together with this dust, microorganisms could have settled on the ice sheet, outcropping at present in the dark region and multiplying rapidly.
To find out more about the origin of outcropping dust, ice cores from several meters 5 need to be analysed, at some depth below the surface to avoid influence from recent dry or wet deposition. Such cores will also provide quantitative information about the amounts of outcropping dust. Also grain size sorting might reveal information about the origin of the dust. For example, Bøggild et al. (2010) found clay size grains in cryoconite from northeast Greenland, revealing dust from Pleistocene origin. Analyses Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | in the Alaska Range), Hydrol. Process., 15, 3447-3459, 2001. 2568, 2569 Takeuchi, N.: Optical characteristics of cryoconite (surface dust) on glaciers: the relationship between light absorbency and the property of organic matter contained in the cryoconite, Ann. Glaciol., 34, 409-414, 2002. 2560, 2567, 2568 Temporal and spatial variations in spectral reflectance and characteristics of 5 surface dust on Gulkana Glacier, Alaska Range, J. Glaciol., 55, 701-709, 2009Glaciol., 55, 701-709, . 2560Glaciol., 55, 701-709, , 2568 Takeuchi, N., Kohshima, S., and Seko, K.: Structure, formation, and darkening process of albedo reducing material (cryoconite) on a Himalayan glacier: a granular algal mat growing on the glacier, Arct. Antarct. Alp. Res., 33, 115-122, 2001. 2560, 2569 Taylor, S. R. and McLennan, S. M.: The geochemical evolution of the continental crust, Rev.