Cryoconite mass assessment

Mass loading of cryoconite and other surface debris was estimated using two short transect studies in the northeast and southwest margins of the ice sheet (at KPCL and KG, respectively; Fig. 1). The former transect has already been described by Reference Bøggild, Brandt, Brown and WarrenBøggild and others (2010) and is of the order of 5 km, extending from the equilibrium line to the glacier margin. This includes marginal ice with basally derived debris, Holocene blue ice with deep well-developed cryoconite holes, and then Pleistocene ice characterized by shallow cryoconite holes and dispersed cryoconite particles on the ice surface. The second transect (at KG) is just 2 km from the margin in an area where the equilibrium line is more than 100 km inland. This includes a ∼700m marginal zone characterized by basally derived debris and ∼1300m of Holocene blue ice with well-developed cryoconite holes. The mass loading of cryoconite debris on both these ice facies at KG was therefore assessed by first estimating the area of surface debris cover and then establishing the mass per unit area of cover. The former was achieved by processing 42 quadrat images of the ice surface in the marginal zone and 205 quadrat images from the blue ice zone, using ImageJ software (for complete details see Reference Irvine-Fynn, Bridge and HodsonIrvine-Fynn and others, 2010). The mass of debris per unit area of cover was then derived after processing images of 12 cryoconite holes of different size in the same way and establishing the dry mass within them after sampling their entire contents and oven-drying them at 40°C. For cryoconite and other debris lying upon the ice surface, 18 processed images of 5 cm thick surface ice samples removed using an ice axe were used instead.

Three bulk debris samples from the marginal ice zone and the Holocene blue ice zone at KG and from stakes 1, 7 and 11 at KPCL in the study of Reference Bøggild, Brandt, Brown and WarrenBøggild and others (2010) were also disaggregated using sodium hexametaphosphate dispersing agent and subject to particle-size analysis using a Horiba laser diffraction sizer under standard manufacturer’s operating conditions.

Biological characterization

Organic-matter content was analysed from the marginal and blue ice zones of the KG transect by standard loss-on-ignition at 450°C. Bulk microbial biomass of two large samples collected from multiple cryoconite holes at both the KG blue ice zone and the KPCL Holocene ice was also undertaken. Our exact methodology is reported by Reference Langford, Hodson, Banwart and BøggildLangford and others (2010), who also examined these samples. Briefly, 10mg cryoconite debris was suspended in 1 mL phosphate-buffered saline and incubated with DNA-specific fluorescent stain (DAPI), filtered onto a black 0.2 μm filter and mounted on a glass slide. Analysis was undertaken under fluorescent light using a Zeiss Avioplan 2 epifluorescence microscope with a DAPI filter cube. Fifteen random fields of view were imaged and enumerated at 0.5 μm Z intervals. The autofluorescence of the photosynthesizing microorganisms was distinguished using an FTIC (fluorescein isothiocyanate) broad filter (Reference Langford, Hodson, Banwart and BøggildLangford and others, 2010).

Light and dark incubations

Batch incubations of debris from the blue ice zone and marginal ice zone were employed in order to establish whether the microorganisms detected using the microscopy constituted an active community. We followed the exact methodology reported by Reference HodsonHodson and others (2010) and Reference TellingTelling and others (2010). Unfortunately, this could only be conducted at KG. Briefly, both the light and dark incubations were conducted on the ice, allowing direct estimation of net ecosystem production (NEP) and respiration (R), respectively. Primary production (PP) was then calculated by difference:

Two sets of estimates were established for each batch of incubations. The first was based on total dissolved inorganic carbon (TDIC):

The second was based on dissolved O2:

The respiratory and photosynthesis quotients were assumed to be unity, so that 1 μM O₂ consumption was assumed to result in 1 μM TDIC during respiration (and vice versa during photosynthesis). All results were then expressed as a carbon flux (μgCg⁻¹ d⁻¹) standardized to the dry mass (g) of the debris incubated. No significant mineral weathering contributions to TDIC and O₂ changes were detected in the ion chemistry of the incubations. Light incubations were conducted in three 200 mL glass flasks, and dark incubations were conducted in three foil-wrapped bottles. Each batch was incubated for 1–3 days in large cryoconite holes during the interval 22 August-3 September 2008 (i.e. day of year (DOY) 235–247). An additional batch of seven dark incubations was also undertaken at 2°C over the course of 4days to establish longer-term respiration rates. When all the incubations ended, 30 mL of sample was abstracted for TDIC analysis using the headspace method, and dissolved O₂ was assessed immediately using a Hach luminescence dissolved-oxygen meter. Full analytical details are given by Reference HodsonHodson and others (2010). Hourly irradiance at the ice surface was recorded with a Skye pyranometer and a Campbell Scientific 510 data logger.

Hole evolution study

In addition to the incubations, a spatial sampling programme of dissolved O ₂ in cryoconite holes was undertaken. This included the monitoring of two adjacent circular holes of near-identical area (70.1 and 73.1 cm²) and initial depth (17 and 16 cm, respectively). The dissolved oxygen, depth and water content of these holes was then followed over a 10 day period after removing a l l the cryoconite from one of the holes (hereafter the ‘blank hole’). Differences between the holes were then used to assess NEP i n the cryoconite community that was left behind in the other so-called ‘occupied hole’.

Mass distribution

No cryoconite holes were found in the marginal ice zone at KG, where coarser particles with no aggregate structure and low organic-matter content dominated (Table 1). A far greater proportion of the ice surface was debris-covered here (average 42%) than in the blue ice zone, where well-developed holes typically 15–60cm deep (average 24cm) covered about 3.2% of the ice surface. Here the heterogeneity of cryoconite cover was far greater than in the marginal ice zone and values ranged from 0 to 39% of the ice surface in a similar manner to the blue ice at Thule Ramp and KPCL (Reference Gerdel and DrouetGerdel and Drouet, 1960; Reference Bøggild, Brandt, Brown and WarrenBøggild and others, 2010). The mass loading within the cryoconite holes (∼0.21 gcm⁻²) was also very similar to the upper values reported by Reference Bøggild, Brandt, Brown and WarrenBøggild and others (2010) at KPCL (see Table 1), yet was greater than other parts of this transect and several Svalbard glaciers (Reference Cook, Hodson, Telling, Anesio, Irvine-Fynn and BellasCook and others, 2010). Reasons for this are unclear, but might reflect the thickness (not measured) and density of the debris. For example, Table 1 shows variable organic-matter content across the ice sheet, being greatest in the Thule area. Cryoconite densities might therefore be lowest here as a result of the reduced mineral contribution to the overall cryoconite volume. Table 1 also shows that the marginal ice of the KG transect was far coarser than any of the other sites upon the blue ice, where a greater proportion of particles less than 63 p.m was apparent. The marginal ice surface debris therefore appears to be derived from basal ice and moraine, rather than cryoconite derived from dust.

Table 1.

Cryoconite hole characteristics for Kronprins Christians Land (KPCL), Kangerlussuaq (KG) and Thule Ramp. Values in parentheses are the number of observations. ‘Mass per unit hole area’ refers to the dry mass of entire cryoconite holes of different areas (see Methods)

Biology

Bacteria, cyanobacteria and algae were identified in samples from both sites. Bacterial and photoautotroph abundance was greatest in cryoconite from KPCL, which also showed a proportionally greater content of unicellular photoautotrophs than cryoconite from KG. Microorganism abundance was greatest in samples with a high organic-matter content and percentage of fine-sized constituent particles (silt and clay), in agreement with a more complete analysis of these and other cryoconite samples reported by Reference Langford, Hodson, Banwart and BøggildLangford and others (2010). The high standard deviations on the photoautotrophic counts represent the difficulty in disaggregating clusters of photoautotrophs, so the cell counts in Table 1 are indicative only. They do, however, confirm the presence of the significant microbial biomass suggested by pioneering work elsewhere in Greenland (Reference SteinböckSteinbock, 1936; Reference Gerdel and DrouetGerdel and Drouet, 1960).

Hole evolution study

Cold conditions occurred during this study, causing the formation of an ice lid over both the blank and occupied holes. Despite this, absorption of radiation by the occupied hole led to significant deepening relative to the blank hole (Fig. 2a). In fact, regression analysis suggests that the rate of deepening in the occupied hole was 1.32 cm d⁻¹, almost ten times that observed in the blank hole (0.14 cm d ^{- 1} , although in this case the regression model was not significant at p=0.05). Owing to the cold weather, measurement of dissolved O₂ every afternoon (at ∼1500h) first required breaking the ice lid (which would soon form again and persist between measurements). It was therefore decided to fully equilibrate the two holes to the external atmosphere after O2 measurement by gentle stirring. Since a new ice lid would form after each daily measurement, no headspace formed above the water, so differences between the aqueous O2 mass of the two holes (ΔO₂ in mg) were the result of differential rates of hole wall melting (ΔO₂ ^wall ) and NEP Therefore:

and

where 12.13 is the dissolved O2 of glacier ice at the site (in mgL⁻¹) and (V^cryo—V^blank) is the daily increase in the occupied hole volume relative to the blank holes. NEPO², in mg O2, can be converted to μgCg⁻¹ d⁻¹ by assuming a respiratory quotient of 1 and standardizing to the dry mass of cryoconite in the hole (8.9g). Daily NEPO² and changes in ΔO₂ and ΔO₂ ^wall are depicted in Figure 2b. In spite of significant O₂ supply during hole deepening, the O₂ concentration was always lowest in the occupied hole, indicating that ecosystem activity was characterized by net respiration. The average NEPO² was –3.28 ± 1.8 μMCg⁻¹ d⁻¹ and can be compared with the incubation data presented below.

Fig. 2.

Evolution of the blank and occupied cryoconite holes: (a) depths and O₂ difference; (b) cumulative dissolved O₂ difference and O₂ supply from hole deepening, along with daily net ecosystem production (negative values indicate net respiration).

Incubations

Table 2 and Figure 3 show the results of the light and dark incubations. Also included are respiration values for 4day dark incubations at 2°C. The data clearly show that while biological activity within the blue ice cryoconite was significant (respiration rates 1.03–2.6 μMCg⁻¹ d⁻¹), very little occurred in the marginal debris (<0.02 μMCg^{- 1} d⁻¹). A strong relationship was also apparent between changes in dissolved O₂ and TDIC, with a slope of –1.025. This is marginally in excess of - 1 , as a result of the slightly greater rates of O2 consumption relative to TDIC production during the 4 day dark incubations. Without these data, the slope of Figure 3 becomes –1.002 and therefore supportive of respiratory and photosynthesis quotients of unity.

Table 2.

Respiration (R^O2 , R^TDIC), net ecosystem production (NEP^O2 , NEP^TDIC) and photosynthesis (PP^O2 , PP^TDIC) according to dissolved O₂ and TDIC changes, respectively. BI denotes data for the blue ice zone and M for the marginal ice zone. OH is the occupied hole, RS is the 4 day dark incubation at 2°C, IR is the average incident radiation throughout the incubations and d the duration in days. The end dates (in 2008) of the incubations are shown in parentheses, and negative NEP values show an excess of respiration over photosynthesis

Fig. 3.

Incubation results showing the values for respiration (R) and photosynthesis (PP) deduced from changes in dissolved O₂ (x-axis) and TDIC (y-axis). Legend: ‘glacier’ denotes the in situ incubations conducted at both the blue ice zone and the marginal zone (the latter plotting at the origin of the graph); ‘incubator’ denotes the 4 day dark incubation study.

NEP estimates according to the light bottle incubations were close to zero and therefore subject to large errors (not shown). The implication is that rates of photosynthesis were close to, but typically less than, the rates of respiration occurring in the incubations. Further, the average NEP^O2 for the hole evolution study (–3.28μM g⁻¹ d⁻¹) was three times greater than the highest NEP^O2 estimate from the blue-ice light incubations. It is unclear whether the bottle incubations or the hole evolution study are subject to the greatest uncertainty. However, a robust observation is that estimates of NEP^O2 from both the incubations and the hole evolution study were negative and therefore show net respiration during the field campaign.

Biological production and net carbon balance

The carbon transfers typical of the blue ice zone in Table 2 clearly indicate biological production associated with cryoonite debris upon the GIS. However, the coarser marginal debris was notable for its negligible PP and R according to our methods. The implication therefore is that sediment source and environmental factors such as the flushing rate by meltwater (e.g. Reference Stibal, Lis, Lawson, Mak, Wadham and AnesioStibal and others, 2010) are important for supporting biological activity upon the ice surface.

Results from the hole evolution study show how the onset of freezing air temperatures during this campaign ensured that radiative warming of the cryoconite debris was the most significant source of heat for melting near the ice surface. As a result, quite significant deepening of the cryoconite holes occurred beneath their ice lids, an effect that formed the basis of Reference GribbonGribbon’s (1979) work on cryoconite thermodynamics. Here we wish to show that these effects are also important in the context of biological processes because the deepening liberates new nutrients, oxygen and TDIC for biological production from the glacier ice wall (cf. Reference TranterTranter and others, 2004) and reduces the light incident upon the photosynthesizing microorganisms in the cryoconite. For example, Figure 2a shows how the depth of the occupied hole increased by 60% over 10 days, causing a depth difference between the two holes that was strongly correlated to cumulative solar radiation receipt at the ice surface (p=1 × 10⁻⁶, n = 9). Further, while Figure 2b shows that O₂ supply from the ice melt was the dominant term in the O₂ balance of the occupied hole, respiration still maintained a significantly lower dissolved O2 concentration compared with the blank hole (Fig. 2a). Therefore, lower light penetration down to cryoconite within rapidly deepening ice-lidded cryoconite holes is most likely to have been responsible for maintaining net respiration throughout this experiment. We believe that this explains why the correlation between photosynthesis and irradiance in our incubations was weak (p=0.08, n = 5). It is also very probable that our study has not captured photosynthesis rates at their maximum, since these are most likely promoted by higher air temperatures during the main melt season, when shallower cryoconite holes are likely to be open and subjected to greater levels of irradiance. For example, photosynthesis rates presented in Reference Anesio, Hodson, Fritz, Psenner and SattlerAnesio and others (2009) are up to an order of magnitude greater and include a value of 9.6 μMg⁻¹ d⁻¹ from Frøya Glacier near Zackenberg (Fig. 1). Therefore, the use of our incubation measurements for the estimation of carbon fluxes at the ice-sheet scale is likely to be more uncertain for photosynthesis than for respiration.

A basic model of biologically mediated carbon transfers on the GIS

Abundant cryoconite in the northeast, northwest and southwest quarters of the ice sheet, coupled with our new evidence of biological activity during the summer, clearly suggest that ice-sheet carbon fluxes deserve attention, not least because Reference Anesio, Hodson, Fritz, Psenner and SattlerAnesio and others (2009) have proposed a global impact of cryoconite NEP equivalent to 65 Gg Ca⁻¹. In order to assess the GIS contribution to this potential flux for the first time, we modified a basic carbon budget model developed by Reference HodsonHodson and others (2007) so that the potential carbon fluxes associated with respiration (CR^snow and CR^ice) and photosynthesis (CP^ice) could be estimated thus:

CR^snow represents respiration by cryoconite beneath saturated snow or slush, and CRi^ce represents respiration when cryoconite is exposed by snow retreat and photosynthesis is possible (CPi^ce). A _bio represents the area of the ice sheet covered in wet, biologically active cryoconite debris, i is the proportion of this zone that is no longer snow-covered, M_d is the mass per unit area of cryoconite (gcm^{- 2} ) and c is the percentage cover of cryoconite within Ab _io. R^snow and Ri^ce are respiration rates before and after removal of snow cover, respectively, whilst PP^ice is the rate of photosynthesis after snow-cover removal (all in μMCg⁻¹ d⁻¹). The term d_m represents the time-step of the calculations in days, and k is a constant of 1200 for unit conversion.

There are major uncertainties in almost all the parameters defined above, so we restricted our calculations to the basics by using fixed values of R^snow, R^ice and PP^ice and a monthly time-step. The maximum extent of Ab _io was assumed to be equivalent to the average end-of-summer bare ice zone over the interval 2001–08. The basal ice zone was ignored on account of the surface area being trivial. The growth of A _bio and its bare ice proportion, i, were achieved by driving the runoff-retention model of Reference Janssens and HuybrechtsJanssens and Huybrechts (2000) at a 5 km × 5 km spatial resolution using downscaled European Centre for Medium-Range Weather Forecasts (ECMWF) meteorological (re)analysis data and empirically derived ice-sheet-surface lapse rates. Full details of this process are given by Reference Hanna, Huybrechts, Janssens, Cappelen, Steffen and StephensHanna and others (2005). The typical end-of-summer expanses of A _bio(1-i) (slush) and A _bio (i) (bare ice) are shown in Figure 1.

The fixed values of Md (0.2gcm^{- 2} ) , c (3%), R^snow, R^ice (both 1.74μMCg⁻¹d⁻¹) and PP^ice (1.56μMCg⁻¹ d⁻¹) used initially were average values for the blue zone in Tables 1 and 2. The product of Md and c implies an impurity loading of 66gm⁻² , which can be compared to the range 16–1400 gm⁻² derived for the KPCL transect by Reference Bøggild, Brandt, Brown and WarrenBøggild and others (2010) and the 25gm⁻² estimate used to represent all glacier ice ablation areas outside Antarctica by Reference Anesio, Hodson, Fritz, Psenner and SattlerAnesio and others (2009). It is appreciated that fixed values are unlikely and that R^snow and Ri^ce might not be the same. However, there are simply no data as yet to improve these estimates and we only seek indicative flux estimates in this study.

The model described above resulted in values for CR^snow, CPi^ce and CRi^ce that are shown in Table 3. The resultant NEP fluxes are also shown. In addition, results for a second scenario are included that try to account for the clear evidence that PP is low in our dataset. For this scenario, PP^ice was 14.7 μMCg⁻¹ d⁻¹, the average rate of PP deduced from non-lidded cryoconite holes for a selection of Svalbard, Greenland and Austrian glaciers by Reference Anesio, Hodson, Fritz, Psenner and SattlerAnesio and others (2009). This was used to yield a high fixation rate across the GIS (^MaxCPi^ce). Table 3 shows that respiration fluxes are approximately ten times greater from bare ice surfaces than from slush-buried surfaces, largely because Ab _io rapidly loses its snow cover during the summer and we have been conservative when defining its extent (202 000 km² or 12% of the GIS). Total respiration fluxes of the order of 25 GgCa⁻¹ result, representing much of the glacial respiration flux of 35 GgCa⁻¹ proposed by Reference Anesio, Hodson, Fritz, Psenner and SattlerAnesio and others (2009) for all glaciers outside Antarctica. However, the outcome is less clear for carbon fixation by photosynthesis, being ∼20GgCa⁻¹ according to our incubation data and 180 GgCa⁻¹ using average values from Reference Anesio, Hodson, Fritz, Psenner and SattlerAnesio and others (2009). The impact of this uncertainty upon net ecosystem demand for carbon is very significant, with estimates ranging from a net respiration flux of –5.3 GgCa⁻¹ to a net carbon sink of 152 GgCa⁻¹. Therefore, significant carbon fluxes (at the 10¹–10²GgCa⁻¹ range) are likely to be associated with biological production upon the GIS, but since we have no evidence to suggest that the rates presented by Reference Anesio, Hodson, Fritz, Psenner and SattlerAnesio and others (2009) are representative, we tentatively suggest that the lower values are the most realistic. More research on spatio-temporal changes in cryoconite ecosystem activity across the ice sheet are clearly required to test this assertion, especially for evaluating the rates of photosynthesis.

Table 3.

Annual carbon fluxes (inGgCa⁻¹) for respiration in snow and bare ice habitats (CR^snow and CR^ice) and photosynthesis upon bare ice (CP^ice and ^MaxCP^ice). NEP I is (CP^ice – CR^ice – CR^snow) and NEP II is (^MaxCP^ice – CR^ice – CR^snow)