Cryoconite

1 Composition of granules

The materials comprising cryoconite can be divided into two main types: organic and inorganic. Organic matter (OM) includes living and dead microbes, their exudates, products of decomposition and allochthonous biotic and biogenic matter (e.g. Hodson et al., 2008; Langford et al., 2010; Takeuchi, 2002a, 2002b; Takeuchi et al., 2010; Wientjes et al. 2011). While there is clearly variability in OM abundance in cryoconite worldwide (Table 1) it invariably represents a key component of cryoconite granules. Spatial variability in OM has also been identified across individual glaciers (Langford et al., 2014; Stibal et al., 2010, 2012a). The quantity and quality of OM is known to influence aggregate formation in terrestrial soils, but this has rarely been examined in cryoconite (Langford et al. 2014).

OPEN IN VIEWER

Table 1.

Percentage fractions of organic matter (OM) in cryoconite samples (error is 1σ).

Location	Latitude (°N or S)	OM (%)	Reference
Tyndall Glacier, Patagonia	40° S	1.8 ± 0.4	Takeuchi et al. (2001c)
Rhikha Samba Glacier, Nepal	29° N	5.8 ± 0.9	Takeuchi et al. (2009)
AX010 Glacier, Nepal	28° N	2.0 ± 0.8	Takeuchi, unpublished
Yala Glacier, Nepal	28° N	7.8 ± 0.4	Takeuchi et al. (2001b)
Meikuang, Tibet	35° N	5.5	Takeuchi (2002a)
Gohza, Tibet	35° N	5	Takeuchi (2002a)
Xiao Dongkemadi, Tibet	33° N	13.2	Takeuchi (2002a)
Urumqi No.1, China	43° N	9.7 ± 1.6	Takeuchi and Li (2008)
Qiyi Glacier, China	39° N	8.6 ± 1.9	Takeuchi et al. (2005)
Akkem Glacier, Russia	49° N	4.8 ± 0.7	Takeuchi et al. (2006)
Gulkana Glacier, Alaska	63° N	8.2 ± 1.3	Takeuchi (2002b)
Worthington Glacier, Alaska	61° N	6.2 ± 0.6	Takeuchi et al. (2003)
Devon and Penny Ice Caps, Canadian Arctic	76° N (Devon) 67° N (Penny)	8.5 ± 6.7	Takeuchi et al. (2001a)
Qaanaaq Ice Cap, Greenland	77° N	5.1 ± 1.6	Takeuchi et al. (2014)
Nuna Ice Ramp, Greenland	76° N	13.9 ± 1.1	Gerdel and Drouet (1960)
Thule Ice Ramp, Greenland	76° N	18.3 ± 2.3	Gerdel and Drouet (1960)
Kronprinz Christian’s Land, Greenland	80°N	5	Bøggild et al. 2010
Blue ice near Kangerlussuaq, Greenland	67° N	7.1 ± 3.3	Hodson et al. (2010a)
Marginal ice near Kangerlussuaq, Greenland	67° N	1.9 ± 0.8	Hodson et al. (2010a)
Hansbreen, Werenskioldbreen, Nannbreen, and Austre Torellbreen, Svalbard	77° N	8.4 ± 5.0	Stibal et al. (2006)
Austre Brøggerbreen	79° N	9.7	Takeuchi (2002)
Taylor Valley, Antarctica	77° S	3	Foreman et al. 2007

In addition to biotic and biogenic OM, persistent organic pollutants (POPs) with carcinogenic and mutagenic potential (Hodson, 2014) can be incorporated into cryoconite. These molecules resist environmental degradation and might bioaccumulate in cryoconite. Whilst they likely have little impact on-site, they may influence downstream ecosystems. These pollutants have been linked to industrial emissions and are often deposited along with inorganic matter such as black carbon (BC) (Aamaas et al., 2011). BC, a product of incomplete combustion of fossil and biofuels, has attracted research attention because it is an extremely effective absorber of heat, both in the atmosphere and after deposition on ice and snow, possibly accelerating melt (Clarke and Noone, 1985; Hansen et al., 2000; Jacobsen, 2004; Xu et al., 2009a). The longevity of BC on ice surfaces is probably enhanced by entrainment into cryoconite, likely reducing the aggregate albedo (Bøggild, 2011; Xu et al., 2009b), although it is also possible that cryoconite microbes metabolise BC and reduce its potency.

Inorganic matter in cryoconite is dominated by mineral fragments, often dominated by phyllosilicate, tectosilicate and quartz (e.g. Edwards et al., 2011; Hodson et al., 2010a; Langford et al. 2010; Stibal et al., 2008a, 2008b); however, differences in source geology likely cause geographic variation in cryoconite mineralogy. Bullard (2013) pointed out that supraglacial dusts (that presumably form cryoconite granules) generally include materials from various local and distal sources, citing studies undertaken in Arctic Canada (Lawrence and Neff, 2009; Zdanowicz et al., 2000), Antarctica (Bory et al., 2010), the central (Drab et al., 2002; Prospero et al., 2012; Tegen and Rind, 2000) and south-west marginal zones of the Greenland Ice Sheet (Bøggild et al., 2010; Wientjes et al., 2011). This was recently corroborated by Nagatsuka et al. (2014) who used Sr and Nd isotopic ratios to describe contrasting origins for silicate minerals in cryoconite on a selection of central Asian glaciers. Fine dusts deposited from high atmospheric suspension are particularly important in the interior zones of large glaciers and ice sheets, while on smaller glaciers and ice-sheet margins a greater proportion of minerals are likely derived from local sources (Stibal et al., 2012a). While material can be transported from local sources (including exposed valley sides and moraines) to the ice surface by gusts of wind (Bøggild et al., 2010; Oerlemans et al., 2009), Porter et al. (2010) found unconsolidated morainic material to be reworked onto the surface of a Svalbard glacier by debris flows. The dominant sources of cryoconite minerals can also vary across glacier surfaces (Langford et al. 2011). Other possible sources include release from englacial storage by ablation (Atkins and Dunbar, 2009; MacDonnell and Fitzsimons, 2008; Wientjes et al., 2011), outcropping of basal tills (Stibal et al., 2012a) and deposition of micrometeorites (particularly noted on the Antarctic and Greenland ice sheets). Differences in source geology and depositional regimes likely explain geographical variations in the size of mineral fragments in cryoconite. For example, Takeuchi et al. (2010) found mineral fragments between 1.3 and 98 μm diameter in China, while Zarsky et al. (2013) found between 0.02 and 2000 µm in Svalbard.

The mineralogy of ice surface debris (including cryoconite) might influence spatial patterns of melt. Casey et al. (2012) found the mineralogy of supraglacial debris on Himalayan glaciers to correlate with surface temperatures and spectral reflectance. Since cryoconite mineralogy has been shown to influence its colour and reflectivity (Takeuchi et al., 2014; Tedesco et al. 2013) it may similarly influence ice surface albedo (Sugiyama et al., 2014). Mineralogy might also influence cryoconite microbes (Carson et al., 2009). For example, since cryoconite microbial communities are often phosphorus (P) limited (Mindl et al., 2007; Säwstrom et al., 2002; Stibal and Tranter, 2007), phosphates from rock debris might provide crucial nutrients. Tazaki et al. (1994) found silicate clays are entrained on snow algal cell surfaces. This is likely analogous to processes occurring in cryoconite, although direct evidence of mineralogical controls on cryoconite microbial activity has not yet been presented. Analysis of heavy metals has indicated the occurrence of mineral–microbe interactions (Nagatsuka et al. 2010) and Hodson et al. (2010a) highlighted the importance of cryoconite biota for extending rock-water interactions and catalysing chemical weathering. Dittrich and Lüttge (2008) suggested that microbes actively control water–solid interactions. Nevertheless, deep understanding of mineral–biota interactions in cryoconite is still lacking. Further inorganic material in cryoconite includes heavy metals (Singh et al., 2013) and radionuclides (Tieber et al., 2009), the accumulation and release of which may impact supraglacial and proglacial ecosystems.

The literature therefore indicates geographic variations in OM (further details in Section IV) and inorganic matter arising from different depositional and post-depositional processes. Quantifying the relative contributions of OM and inorganic matter sources and the impacts upon melt and microbial processes remain important outstanding research questions.

2 Shape and size of cryoconite granules

Cryoconite granules tend to be quasi-spherical, however their size and morphology is highly spatially variable (Figure 1, Table 2), likely reflecting the local balance between microbial growth, physical aggregation and erosion (Irvine-Fynn et al., 2010). Variations in granule size have been reported between glaciers; for example Zarsky et al. (2013) found larger granules to be more common on Aldegondabreen (Svalbard) than on other glaciers, possibly due to gentle slope gradients, fine material particles, nutrient input from debris avalanches and atmospheric deposition and high rates of microbial activity. Also on Aldegondabreen, Langford et al. (2014) identified an “edge effect” whereby granules were generally smaller towards the central areas of the glacier. This was attributed to biochemical and photophysical processes including TOC and carbohydrate concentration. The strong correlation between granule size and OM content (Langford et al., 2010, Stibal et al., 2010; Takeuchi et al., 2010) suggests that biota play a key role in determining the shape and size of cryoconite granules.

OPEN IN VIEWER

Figure 1.

(a) Discrete quasi-spherical cryoconite granules of 1–2.5 mm diameter sampled ∼2km inland of the ice margin on the Greenland ice sheet near Kangerlussuaq; (b) a complex of cryoconite granules ca. 10 mm diameter, sampled ∼34km inland of the ice margin on the Greenland ice sheet near Kangerlussuaq. Scale bar applies to both images.

OPEN IN VIEWER

Table 2.

Cryoconite granule sizes for a range of glaciers.

Location	Latitude	Granule Size (mm)	Reference
Yala Glacier, Nepal	28° N	0.39 ± 0.15	Takeuchi (2002a)
AX010, Nepal	28° N	0.29 ± 0.12	Takeuchi (2002a)
Rihka Samba, Nepal	29° N	0.54 ± 0.20	Takeuchi (2002a)
Meikuang, Tibet	35° N	0.55 ± 0.26	Takeuchi (2002a)
Gohza, Tibet	35° N	0.55 ± 0.28	Takeuchi (2002a)
Xiao Dongkemadi, Tibet	33° N	0.80 ± 0.35	Takeuchi (2002a)
Urumqi No.1, China	43° N	1.10 ± 0.39	Takeuchi et al. (2010)
Austre Brøggerbreen, Svalbard	79° N	0.49 ± 0.29	Takeuchi (2002a)
Penny Ice Cap, Canada	67° N	0.33 ± 0.11	Takeuchi (2002a)
Devon Ice Cap, Canada	76° N	0.33 ± 0.13	Takeuchi (2002a)
Longyearbreen, Svalbard	78° N	8.92 ± 6.93	Hodson et al. (2010)
Longyearbreen, Svalbard	78° N	8.59 ± 6.67	Irvine-Fynn et al. (2010)
Aldegondabreen, Svalbard	79° N	110 ± 35	Zarsky et al. (2013)
79 km transect along Leverett Glacier, Greenland	67° N	2.59 ±1.54	Cook (2012)

Growth and proliferation of autotrophic cyanobacteria at granule peripheries has often been suggested to be a driver of granule growth because their filamentous morphology enables them to entangle debris (Langford et al., 2010). Heterotrophic microbes might limit this growth by decomposing OM (Hodson et al., 2010a). However, decomposition of OM produces ‘sticky’ humic substances that may enhance granule cohesion (Langford et al., 2011; Takeuchi, 2002a; Takeuchi et al., 2001b). Granule size is therefore probably co-limited by the binding ability of filamentous autotrophs, the adhesion potential of other OM and rates of decomposition by heterotrophs (Takeuchi et al., 2010). The potential for cryoconite microbes to control granule size and morphology was demonstrated by Takeuchi et al. (2001b) who treated half a batch of cryoconite with biocide (CuSO₄) before leaving them on a glacier surface over an ablation season. Algal mats formed in untreated samples, whereas treated samples remained loose and disaggregated. The contrasting morphologies suggest that meltwater, wind or gravity driven movement of granules might influence their shapes. This is supported by the even distribution of cyanobacteria over granule surfaces (e.g. Hodson et al., 2010a; Takeuchi et al., 2001a, 2001b) despite only one side of settled grains being exposed to sunlight. Irvine-Fynn et al. (2011) showed continuous hydraulic redistribution (but very little net displacement) of cryoconite over a small plot on a Svalbard glacier that could be crucial for maintaining homogenous growth rates over their surfaces.

Cryoconite granules also exhibit distinctive internal microstructures (e.g. Langford et al., 2010; Takeuchi et al., 2001b, 2010) where granule interiors are dominated by mineral fragments and heterotrophic microbes, while granule surfaces are inhabited by photoautotrophs (e.g. Langford et al., 2010; Takeuchi, 2002a; Takeuchi et al., 2001b). This probably results from the extinction of light within the outer few mm of cryoconite granules. Indeed, Hodson et al. (2010b) showed both cyanobacterial photosynthesis and photosynthetically active radiation (PAR) receipt to be concentrated at granule surfaces. Thin section microscopy of Chinese cryoconite by Takeuchi et al. (2010) revealed more complex internal microstructures including concentric rings of OM that may have resulted from annual biomass accumulation or cycles of erosion and regrowth (Takeuchi et al., 2010). The number of rings per granule ranged from 2 to 7, with a mean of 3.5. Some granules contained sub-granules with their own concentric rings, suggesting amalgamation due to tight packing and rapid cyanobacterial growth. The amalgamation of individual granules might explain reports of large complex granules with diameters of up to 40 mm (Hodson et al., 2010a). Figure 2 shows four distinct classes of cryoconite microstructure proposed by Takeuchi et al. (2010).

OPEN IN VIEWER

Figure 2.

Cryoconite granule structures: (a) simple concentric ring structure; (b) granule containing a dark humic ring; (c) granule containing subgranules, each with individual concentric ring structures; (d) granule with a large central mineral grain; (e) an example of a granule with no specific internal structure. (Reprinted from Takeuchi et al. (2010), Annals of Glaciology with permission from the International Glaciology Society.)

In general, larger and more stable grains exist where cyanobacteria are more abundant (Takeuchi et al., 2010). This is because granule growth is driven by the growth and proliferation of microbes, entanglement of debris in cyanobacterial biomass and cementation by cohesive humic and extracellular polymeric substances (Hodson et al., 2010b; Langford et al., 2010, 2014; Takeuchi et al., 2001b, 2010). However, the precise biotic and abiotic processes controlling granule growth and erosion are still poorly understood, and we have little knowledge of their spatio-temporal variations. These are important knowledge gaps since larger granules probably have longer residence times, and therefore more prolonged albedo-reducing effect on ice surfaces (Bøggild et al., 2010; Hodson et al., 2010a; Irvine-Fynn et al., 2011) while also providing more microhabitats for varied microbial communities (Langford et al., 2010; Zarsky et al., 2013). This suggests granule size could have important glaciological and ecological implications, as well as potentially being an indicator of local biogeochemical processes (Langford et al., 2014).

1 Formation and evolution of cryoconite holes

Cryoconite exists in many several supraglacial environs (Figure 3) of which cryoconite holes are by far the most common and best understood. Cryoconite holes are quasi-cylindrical depressions on ice surfaces that commonly have depths and diameters ranging from millimetres to tens of centimetres. Cylindrical holes are common; however, complex shapes also form and likely reflect interactions between regional and local topography, aspect, ice type, hydrology and sediment dynamics (Cook et al., 2010; McIntyre, 1984).

OPEN IN VIEWER

Figure 3.

A range of cryoconite habitats. (a) Wide shot of ice surface at ‘S6’, 38 km inland of the ice margin, Greenland ice sheet. Cryoconite holes are visible in the foreground, stream cryoconite and cryoconite mantle is visible in the midground. (b) Aerial view of cryoconite holes at the same location. (c) Loose cryoconite granules, fine dusts and algae dispersed upon the ice surface (T. Irvine-Fynn for scale).

Research to date has almost invariably assumed circular hole planforms, presumably to provide a simplified model system for studying the fundamental mechanisms of hole development. It is now well established that cryoconite holes form due to accelerated melting of ice beneath accumulations of cryoconite (Agassiz, 1846; Gribbon, 1979; McIntyre, 1984; Nordenskiöld, 1875; Phillip, 1912; Wharton et al., 1985). This occurs because cryoconite has low albedo relative to the surrounding ice surface, meaning it efficiently absorbs solar radiation. Gribbon (1979) proposed a conceptual model (see equation (1), Table 3) showing the solar radiation received by cryoconite sediment diminishing as it sinks in ice. There is, therefore, a critical depth where the melt beneath cryoconite is equal to that of the surrounding ice surface. This is referred to as the ‘equilibrium depth’. Equilibrium depths vary over space and time according to environmental conditions (particularly the balance between radiative and turbulent heat fluxes in the local ice surface energy budget), the albedo of cryoconite and the extinction coefficient of light (k) defined by Beer’s law. Beer’s law describes the exponential decay of solar radiation (I) as a function of distance (z) through ice (Oke, 1987)

I z * = I 0 * − k z

1

OPEN IN VIEWER

Table 3.

Equations relating to the thermodynamics of cryoconite hole evolution.

	Equation	Definitions	Reference
Equation (1)	(1 − α) *S e−kz + fGa + M = 0	α = albedo; S = incident radiation; k = extinction coefficient for sunlight; z = depth; M = downward melt; Ga = undefined heat flux	Gribbon (1979)
Equation (2)	ρiLv = Q 1 − Qc	ρi = ice density; L = latent heat of fusion for water; v = absolute sediment velocity; Q1 = shortwave energy absorbed by the sediment; Qc = defined in equation (3)	Jepsen et al. (2010)
Equation (3)	Qc = −K λ Tice(z, t)	K = thermal conduction coefficient; λ = thermal conductivity of ice	Jepsen et al. (2010)
Equation (4)	Q1 = T (1 − R) (1 + S)   Io(t) exp(−kz)	T = fraction of downwelling shortwave radiation transmitted into the ice through a semi-opaque surface layer; R = all-wave albedo of the sediment; S = ratio of upwelling to downwelling shortwave flux at the sediment; Io(t) = downwelling shortwave flux above the ice surface; k = all-wave absorption coefficient for ice.	Jepsen et al. (2010)
Equation (5)	Qv = aT (1 − R)   (1+ S) Io (t)   exp(−kz)	Qv = thermal energy directed to the hole floor	Cook (2012)
Equation (6)	QL = (1 − a) T (1 − R)   (1 + S) Io(t)   exp (−kz)	QL = thermal energy directed to the hole walls	Cook (2012)

Coupling between equilibrium depth and environmental conditions suggests cryoconite hole depth could be a crude indicator of synoptic energy balance (McIntyre, 1984). McIntyre (1984) used polythene screens to shade holes and observed dramatic changes to hole shapes, confirming the role of direct solar irradiance as a primary control on hole morphology. Microbial activity had previously been suggested to contribute significantly to the heat balance at the hole floor (Gerdel and Drouet, 1960) but this was discredited by the work of McIntyre (1984), Fogg (1988), Gribbon (1979) and Wharton et al. (1985). Recent work by Hollesen et al. (2015) has shown that bioheat in organic rich permafrost can accelerate ice-melt, but this is unlikely to occur in cryoconite unless it is extremely rich in organic matter or present in thick deposits. There is evidence that the melt water in cryoconite holes might influence hole evolution by acting as a radiation filter, sink of latent heat and medium for convection and advection of heat away from hole floors (McIntyre, 1984). Sinking dark cryoconite beneath a layer of reflective water may also raise the albedo of the ice surface, altering the local energy balance regime (McIntyre, 1984). Hydraulic processes might therefore influence hole morphology, but this has not yet been examined in detail.

Podgorny and Grenfell (1996) provided an analytical model of radiation absorption by cryoconite that demonstrated the importance of solar radiation and sediment albedo; however, it dealt specifically with cryoconite holes developing on the floors of melt pools on sea ice and is probably not directly applicable to holes on terrestrial ice. Similarly, a model for shortwave radiation-driven cryoconite hole formation was produced by Jepsen et al. (2010) and supported by field and laboratory experiments; however, this model considers closed holes in perennial lake ice and omitted atmospheric exchanges of sensible and latent heat, limiting its application to cryoconite holes on Antarctic lakes. No solvable model for cryoconite hole depth evolution on Arctic ice currently exists.

Most studies of cryoconite hole evolution have assumed that heat fluxes are exclusively vertical, although we now know that is not the case. Cook et al. (2010) proposed a process of horizontal evolution in response to sediment supply. Horizontal heat fluxes through thickened layers of cryoconite sediment promote melting of hole walls, causing the hole to expand at the base and the holes walls to overhang. Cryoconite can then fall into the newly created space at the hole periphery and the sediment layer thins. A new equilibrium state is attained when cryoconite is spread into a layer just one grain thick (single grain layer, ‘SGL’) and the widening at the hole floor has propagated upwards due to surface ablation, producing a hole with straight walls and a wide aperture. The precise mechanism of granule redistribution across the expanded hole floor remains uncertain, but recent observations (Cook, unpublished data) suggest the primary mechanism is probably gravity-driven sliding of cryoconite down sloped hole floors due to uneven solar irradiance, while redistribution driven by air bubbles escaping from the melting hole floor may contribute (Cook et al., 2010) in areas where cryoconite granules are small.

Cryoconite holes can therefore evolve in three dimensions in response to changes in the local ice surface energy budget and sediment supply. They tend towards equilibrium morphologies characterised by maximal areal coverage and exposure of granules to solar irradiance, which promotes photosynthesis (Cook et al., 2010). These equilibrium states are common in cryoconite holes in slow moving, low gradient ice that is free from topographic shading during periods of shortwave radiation dominated surface energy balance (Cook et al., 2010; 2012; Hodson et al., 2010a), although in steep, hummocky ice and in areas where rates of sediment delivery are particularly high or during periods of turbulent heat flux-dominated surface energy balance, thicker sediment layers exist. This implies that although the process of sediment layer equilibration might be ubiquitous in cryoconite holes, the attainment of SGL can sometimes be inhibited by synoptic conditions. The fundamental processes of vertical and lateral equilibration of cryoconite holes can be summarised in a conceptual model (Figure 4) wherein thermal energy absorbed by cryoconite debris is directed to the hole floor (Q_V) or transferred laterally for melting hole walls (Q_L) in varying proportions according to the thickness of the debris layer (equations (5) and (6), Table 3).

OPEN IN VIEWER

Figure 4.

Flow diagram depicting the vertical and horizontal development of cryoconite holes. The term ‘I*’ refers to solar radiation and ‘SGL’ stands for single grain layer. Ice is depicted using light grey shading and water is depicted using diagonal hatching.

There are several specific challenges that must be overcome before a solvable model of cryoconite hole evolution can be developed. Firstly, the mechanism of sediment redistribution on hole floors must be better constrained. Second, a much deeper understanding of the thermodynamic processes operating in cryoconite holes, including in the water column, is required. Furthermore, equations (5) and (6) (Table 3) relate thermal fluxes to ice–debris contact area, but this assumes constant layer thickness and will therefore only be applicable to holes in morphological equilibrium, which are only common on stable, slow moving ice in the interior zones of large glacier and ice sheets. Elsewhere, shallow holes contain thick sediment layers and granules can be dispersed directly upon the ice surface (Hodson et al., 2007, 2008; Irvine-Fynn et al., 2011; Stibal et al., 2008) and understanding these may require a revised approach. Current models are limited to well-developed cryoconite holes and there is a lack of research into hole initiation which likely relies upon complex surface debris dynamics, microscale hydraulics and spatial heterogeneities in surface melt rates (Lancaster, 2002). Finally, these models assume an uninterrupted pathway towards morphological equilibrium, whereas periods of turbulent flux dominated ice surface energy balance can reduce the depths of cryoconite holes, sometimes causing them to melt out and redistribute sediment onto the ice surface (Hodson et al., 2007). Similarly, natural cryoconite systems can be disturbed by rainfall events. Only during extended periods of meteorological stability characterised by high incident radiation receipt can cryoconite holes be expected to evolve as described by the models in Table 3.

2 Cryoconite holes and synoptic energy balance

The complex interplay between synoptic energy balance conditions and cryoconite hole evolution was studied by Hodson et al. (2007) who showed that frequent disruption and emptying of Svalbard cryoconite holes occurred in areas of high melt rate. This was due to hydraulic mobilisation of sediment from shallower holes. Hodson et al. (2010b) and Irvine-Fynn et al. (2011) both used time lapse imagery to show that during sunny periods when radiative fluxes dominate ice surface energy balance, cryoconite holes deepen as predicted by Gribbon (1979). This causes the albedo of the ice surface to increase locally since dark sediment sinks further beneath melt water which, while still dark has greater reflectivity than cryoconite. In contrast, periods of turbulent flux dominated energy balance cause cryoconite holes to shallow, moving dark cryoconite closer to the ice surface and reducing the local ice surface albedo. Tracking of individual cryoconite granules suggested continuous redistribution of cryoconite sediment, usually in random directions and resulting in little net displacement, but occasionally melt water flow caused rapid movement of cryoconite between holes (Irvine-Fynn et al., 2010). Hodson et al. (2010a) further suggested frequent hydraulic redistribution of granules between cryoconite habitats and glacier zones. Patterns of melt-in and melt-out of cryoconite influence the microtopography and therefore roughness of the ice surface, with implications for surface albedo (Cutler and Munro, 1996; Warren et al., 1998) and turbulent heat fluxes (Munro and Davies, 1977). These observations imply that cryoconite hole morphology is not only an indicator but may also be a driver of energy balance across melting ice surfaces. In general, periods of prolonged radiative-flux dominated ice surface energy balance are associated with deeper, more stable cryoconite holes.

3 Feedbacks between cryoconite granules and cryoconite holes

The current model of cryoconite granule formation has been inferred from the relationships between OM and granule morphology, along with observations of granule microstructures and biogeochemistry. Langford et al. (2010) suggested the following mechanism:

Blooms of filamentous and unicellular phototrophs (particularly cyanobacteria) on ice, snow and slush surfaces act as nets for wind-blown debris (Figure 5(a)). During snow and slush melt, debris is strained through these nets, forming biofilms that are deposited on ice surfaces.

OPEN IN VIEWER

Figure 5.

Scanning electron microscope images of cryoconite from 38 km inland on the Greenland Ice Sheet (near Kangerlussuaq) showing (a) a ‘net’ of cyanobacterial filaments binding mineral fragments; (b) EPS mineral interactions along the sheath of a cyanobacterial filament.

After the granules have settled, cryoconite holes can begin to form. Hole formation is primarily driven by abiotic processes of radiative and turbulent heat fluxes; however, microbial activity alters the albedo of cryoconite granules (Takeuchi, 2002a; Takeuchi et al., 2014; Tedesco et al., 2013). Since albedo controls energy absorption and therefore thermal fluxes through cryoconite granules, accelerated melt could be described as microbially mediated. Once granules reside on cryoconite hole floors, further redistribution is unlikely due to the low competence of hydrologic flow through the weathering crust, unless heterogeneous irradiance causes hole floors to slope (Cook et al., 2015). Hydrological monitoring by Cook et al. (2015) also suggests that even if granules are entrained into flowing meltwater, granules are unlikely to be evacuated onto the ice surface. Integrating this information with the well-known association between stability and extended periods of net carbon fixation on Arctic ice surfaces (Anesio et al., 2009; Cook et al., 2012; Hodson et al., 2007) and the tendency for cryoconite holes to widen in favour of autotrophy (Cook et al., 2010) suggests that photosynthesis-driven granule growth creates and maintains favourable conditions for further photosynthesis on Arctic ice surfaces. This in turn drives further granule growth and stabilisation (Langford et al., 2010, 2014). This feedback mechanism is not currently well-understood, but it does point towards the high ecological significance of cryoconite autotrophs, possibly representing keystone taxa (Paine, 1969) or autogenic ecosystem engineers (Jones et al., 1994).

In summary, cryoconite granule formation is ultimately driven by biological processes and results in the formation of cryoconite holes, which further promote biological activity. However, this explanation is a great oversimplification that cannot explain the complex associations between the changeable abiotic environment, biological processes and hole morphology. Complex coupling exists between cryoconite surface coverage, albedo, microtopography and melt processes. These processes may have implications for the process of glacier and ice sheet wastage and require further investigation. What is abundantly clear is that cryoconite holes should be viewed as diverse, dynamic, responsive entities on ice surfaces.

1 Microbes in cryoconite

Cryoconite is an important microbial habitat and a major component of supraglacial ecosystems (Anesio and Laybourn-Parry, 2011). Although recognition of this is commonly credited to Hodson et al. (2008), Kohshima (1984a) identified a complex microbial community in ice and snow around cryoconite holes twenty-four years earlier. Kohshima’s subsequent work in the Himalaya, Tien Shan and Andes (Kohshima, 1984b, 1985, 1987a, 1987b, 1989) described abundant and diverse microbiota and meiofauna and linked them with accelerated ablation via ice surface albedo reduction. Takeuchi et al. (1998) later explicitly linked algal growth on a Himalayan glacier to summer mass balance. Kohshima and Takeuchi’s work during this period provided a crucial basis for developing our understanding of microbially-mediated glacier wastage and recognition of supraglacial microbial ecology; Kohshima et al. (1993: 1) stated: “Himalayan glaciers are never abiotic environments. They are simple and well closed ecosystems; housing various microbes, insects and copepods”, cementing a new paradigm of glacier ecology that recognised the diversity and significance of life on ice.

In the late twentieth century, Japanese scientists made advances in the microbiology of the supraglacial environment, while European and American scientists focussed specifically upon cryoconite. Wharton et al. (1981) broke a 20 year hiatus in cryoconite ecology research, identifying cryoconite holes as microbial niches. Wharton et al. (1985) further described them as discrete ecosystems with “distinct boundaries, energy flow and nutrient cycling”, establishing an ecological context with an inherent reductionism that persists to the present day. Vincent (1988) further suggested that cryoconite ecosystems were more complex than other glacial habitats. Despite this work, an assumption of lifelessness on ice surfaces permeated cryosphere literature until at least the late 1980s with writers such as Pyne (1986) stating that there was “no terrestrial ecosystem… in Antarctica”. It was a further twenty-five years before Anesio and Laybourn-Parry (2011) proposed the cryosphere to be one of the earth’s major biomes, with cryoconite holes representing important sites of concentrated microbial activity and biodiversity. Globally, between 10²¹ and 10²⁶ cells have been estimated to be contained within the porous near surface ice that provides the substrate for cryoconite hole formation (Irvine-Fynn and Edwards, 2014).

Within cryoconite holes, biota is found both in cryoconite granules and the overlying meltwater (e.g. Säwstrom et al., 2002), with by far the greatest microbial abundance in the granules (Anesio et al., 2009; Mieczen et al., 2013). Anesio et al. (2009) found the concentration of bacteria in cryoconite meltwater to be ∼300× less than in debris. Nevertheless, microbes in melt water may represent important contributors to the supraglacial ecosystem. Stratification of ciliates in the water column in cryoconite holes was identified by Mieczen et al. (2013) in accordance with temperature and nutrient concentration gradients. Sixteen phyla of ciliate were identified in total with algiverous and mixotrophic species inhabiting the surface layers and bacterivores occupying the deeper water adjacent to bacteria-rich cryoconite sediment. Suspension of microbial cells in melt water flowing through porous near-surface ice called the ‘weathering crust’ (Muller and Keeler, 1969) may provide an important mechanism of mixing and redistributing microbes in the supraglacial zone. This porous ice layer likely permits admixtures of soluble resources to be homogenised and translocated through the surface ice (Irvine-Fynn and Edwards, 2014; Irvine-Fynn et al., 2012). Punctuating this porous ice layer are cryoconite holes, which likely act as longer-term storage units for biomass (Cook et al., 2015; Hodson et al., 2008) and sites of enhanced biogeochemical cycling, probably utilising resources delivered by hydrologic fluxes through the weathering crust.

Cryoconite granules represent by far the most biodiverse supraglacial microbial habitat, harbouring communities of several trophic levels. Food webs in cryoconite are underpinned by photoautotrophy: primary producers use mainly solar energy to fix atmospheric CO₂ into OM, providing substrate for heterotrophs. Of the primary producers, cyanobacteria often dominate both the biomass and the C fixation in cryoconite holes (Christner et al., 2003; Hodson et al., 2010; Mueller et al., 2001; Porazinska et al., 2004; Stibal and Tranter, 2007; Stibal et al., 2006; Takeuchi et al., 2001b; Zarsky et al., 2013) while simultaneously promoting the growth of granules by entangling minerals and additional OM. Heterotrophs metabolise OM both fixed by primary producers in situ (autochthonous) and delivered from elsewhere (allochthonous). Heterotrophic communities in cryoconite are often diverse and usually dominated by a wide range of bacteria. This has been demonstrated using culture-dependent studies (e.g. Margesin et al. 2002) and clone libraries (e.g. Cameron et al., 2012; Edwards et al., 2011). In Svalbard, at least seven phyla of heterotrophic bacteria were identified in clone libraries (Edwards et al., 2011), with their abundance varying between different glaciers. Sampling of Arctic, Antarctic and Alpine cryoconite by Anesio et al. (2010) showed bacterial abundances ranging from 0.05 × 10⁹ cells g⁻¹ in Antarctica to 1.40 × 10⁹ cells g⁻¹ in Svalbard. Rates of bacterial carbon production were not correlated to abundance and varied greatly across the glaciers. In addition to bacteria, higher heterotrophs also inhabit cryoconite holes, surviving by grazing upon smaller organisms. These include tardigrades, rotifers, copepods, ice worms and midge larvae (e.g. DeSmet and Van Rompu, 1994; Zawierucha et al. 2013). Several taxa of tardigrada and rotifera in Spitsbergen cryoconite holes were characterised by DeSmet (1988; 1990; 1993), DeSmet et al. (1988) and DeSmet and Van Rompu (1994). Ciliates usually represent the most complex protozoa in cryoconite holes and are crucial for nutrient recycling through metabolism of primary producers (Mieczen et al., 2013). To date, only a few studies have isolated yeasts and fungi in cryoconite (Edwards et al., 2013b; Margesin and Fell, 2008; Singh and Singh, 2012) and only Edwards et al. (2013b) evaluated their spatial variability, finding distinct communities on adjacent glaciers and suggesting ice surfaces could represent reservoirs of fungal diversity.

Viruses are prominent in cryoconite holes (Anesio et al., 2007; Bellas and Anesio, 2013; Bellas et al., 2013; Hodson et al., 2008; Säwstrom et al., 2007) and could be considered ‘predators’ exerting top-down controls on bacterial populations since they infect and cause the death of bacteria through viral lysis. Viral lysis is crucial for bacterial mortality and thereby influences the recycling and export of carbon and nutrients in cryoconite holes. Viral abundances of about 0.6 × 10⁶ ml⁻¹ and 20 × 10⁶ ml⁻¹ in water and sediment have been identified (Anesio et al., 2007) in Svalbard. The frequency of virus-infected cells in cryoconite communities (13%) was shown by Säwstrom et al. (2007) to exceed temperate freshwater ecosystems (2%). Most recently, Bellas et al. (2015) analysed viral genomes from Arctic cryoconite metaviromes revealed genomic signatures of unusual life strategies which are thought necessary for longer-term interaction with their hosts. Viral shunts represent a crucial mechanism by which DOC is recycled or made labile in cryoconite holes, thus affecting the flow of resources to higher trophic levels and truncating the cryoconite food web (Laybourn-Parry et al. 2001).

Similarities in community structures in polar cryoconite have been illustrated using 16 S rRNA gene clone libraries (Cameron et al., 2012), implying a degree of cosmopolitanism in cryoconite microbial assemblages. However, the abundance of particular species has been shown to vary locally (Cameron et al., 2012), suggesting adaptation to environmental conditions at the community level. Stibal et al. (2008) used epifluorescence microscopy to determine the abundance of phototrophs (0.25 × 10³ to 8.0 × 10³ cells mg⁻¹) and heterotrophs (10 × 10³ to 50 × 10³ cells mg⁻¹) in cryoconite from Werenskioldbreen (Svalbard), finding much more biodiversity in heterotrophic communities than the cyanobacteria-dominated autotrophic communities (further described by Stibal et al. 2006; Stibal and Tranter, 2007). At lower latitudes and on some small polar glaciers, heterotrophs have been suggested to be more abundant than primary producers, and may also be supported by allochthonous OC. For example, Edwards et al. (2013c) found genes associated with proteobacteria and bacteriodetes to be more abundant than those associated with cyanobacteria in a metagenome of cryoconite taken from Rotmoosferner (Austrian Alps). Examining community structures is useful for understanding microbial dynamics on glaciers, but has also been linked to cryoconite granule morphology, perhaps indicative of biotic-abiotic feedbacks between microhabitat structure and microbial ecology. For example, Takeuchi et al. (2001c) described poorly aggregated, fine-grained cryoconite in Patagonia where both autochthonous and allochthonous OM were relatively scarce, limiting granule building efficacy. In contrast, cryoconite in the Tien Shan had more abundant biota and higher OM fractions and consequently formed larger, more stable granules.

While cryoconite hole biota can be categorised into distinct trophic levels, functionality is often shared by various organisms throughout the community. For example, granule growth results from the production of cyanobacterial biomass, but also through humification of OM and production of EPS by heterotrophic bacteria, fungi and yeasts. Biogeochemistry is mediated by several organisms including nitrogen fixing bacteria and those involved in ammonification and nitrogenation, including ammonia-oxidising Archaea (Zarsky et al., 2013). Biological darkening of cryoconite granules (and the wider ice surface) is carried out by organisms including algae and bacteria whose activity results in the accumulation of dark humic material and photo-protective pigments (Quesada et al., 1999; Takeuchi, 2002a; Takeuchi et al., 2010, 2014; Tedesco et al., 2013).

To date, microbiological and molecular studies on glacier surfaces have been limited to snapshots, and we await insight into temporal shifts and environmental responses in cryoconite at the community level. However, molecular analyses are becoming increasingly affordable and accessible and are consequently being employed more frequently. It is becoming increasingly clear that close examination of ecosystem structure and function will be crucial for understanding biotic–abiotic processes on glacier surfaces. For a more comprehensive review of the microbes inhabiting cryoconite holes, we direct the reader towards Kaczmarek et al. (2015).

2 The origins of cryoconite biota

The origins of cryoconite biota are currently unclear, with robust evidence in favour of any dominant source or mode of delivery currently lacking. However, several hypotheses have emerged from the literature, mostly linking cryoconite biota to the aeolian biome. Swan (1992) proposed the existence of an aeolian biome based upon observations of microalgae, bacteria and spiders on high altitude and polar ice. Active microbes have since been confirmed to survive in suspension in the atmosphere, some of which are eventually deposited on ice surfaces (Pearce et al., 2009; Sattler et al., 2001; Swan, 1992). There, cryo-tolerant species survive, especially under favourable conditions in cryoconite holes, whereas others perish due to environmental and competitive stresses (Anesio and Laybourn-Parry, 2011; Pearce et al., 2009). Cryoconite biota may therefore be delivered to glacier surfaces directly from the atmosphere via both wet and dry deposition. This may be from local or distant sources. Alternatively, biota may be deposited onto fresh snow packs or precipitated out of atmospheric suspension in snow fall (probably providing nuclei for ice crystallisation; Edwards et al., 2014a) and either redistributed hydrologically following snow melt or incorporated into glacier ice in the accumulation zone. Microbes interred englacially may be stored for 10²–10⁵ years, some being preferentially partitioned into water veins (Mader, 1992) before being melted out in the ablation zone. Biota in basal ice and sediment may be extruded onto the ice surface by thrust faulting or via pressure ridges and subsequently blown upglacier. There may also be biota blown or simply dropped onto glaciers from local valley sides.

Suspension of microbes in the high atmosphere may pre-select microbial communities for survival on glacier surfaces. Evidence suggests long term survival in the atmosphere requires similar adaptation to survival on ice surfaces (tolerance to UV, cold and desiccation; Pearce et al., 2009) and viable communities may therefore be shaped to some degree prior to deposition. However, there are commonalities in species abundance between cryoconite holes, soils, freshwater, marine, frozen lake and activated sludge habitats (Edwards et al., 2011) that indicate wide cosmopolitanism of many cryoconite microbes. Furthermore, post-depositional structuring of communities has been suggested to be dynamic over short time scales, responding rapidly to local biotic and abiotic stresses (Edwards et al., 2014b; Priscu and Christner, 2004). There is most likely a fine balance between cosmopolitanism and endemism that reflects complex associations of sources and modes of delivery of biota, post-depositional ecological dynamics and local environmental regimes that collectively shape cryoconite microbial communities. Amplicon pyrosequencing of cryoconite bacterial 16 S rRNA genes from six Arctic and Alpine glaciers and the southwestern margin of the Greenland Ice Sheet (Edwards et al., 2014b) reveals that cosmopolitan generalist taxa (cf. Barberan et al. 2012) are predominant in cryoconite bacterial communities (Figure 6). This is consistent with the notion that priority effects may be important in colonization of nascent cryoconite granules by bacteria, first indicated by Edwards et al. (2013c) as a result of phylotype abundance distribution models supporting deterministic community assembly as a result of several stages of succession. To an extent, environmental variability may buffer against priority effects (cf. Tucker & Fukami, 2014), promoting species richness and evenness as insurance against fluctuating environmental conditions. Indeed, varying levels of bacterial community evenness correlated with ecosystem function measurements in Svalbard cryoconite (Edwards et al., 2011). By these means, perturbations at the ice surface may trigger reconfiguration of cryoconite associated microbiota. Further support that cryoconite granules are formed as a result of microbial succession is provided by the striking negative correlation in the relative abundance of taxa assigned to r-selected, early-colonizing Betaproteobacteria versus typically K-selected, late-colonizing Alphaproteobacteria (Figure 6; Edwards et al., 2014b). It is clear that the processes of microbial succession in supraglacial habitats such as cryoconite not only require, but merit further study as currently, our understanding of these processes is basic yet the implications for the development of biodiverse communities as a result of glacier melting are considerable. In short, the question of whether microbial succession interacts with glacier atrophy prior to the development of the glacier forefield is raised.

OPEN IN VIEWER

Figure 6.

Population trends in Arctic and alpine cryoconite bacterial communities revealed by 16S rRNA gene amplicon pyrosequencing (Edwards et al., 2014b). (a) A negative correlation between the relative abundance of 97% id OTUs assigned by ribosomal database project taxonomy to Alphaproteobacteria and Betaproteobacteria classes is clearly apparent (Pearson’s r = −0.88, p ≤ 0.0001). (b) Dominance of ‘generalist’ OTUs sensu Barberan et al. (2012). Bubble size is proportional to cumulative relative abundance of each Operational Taxonomic Unit. Raw data are available at EBI-SRA (PRJEB5067-ERP004426).

3 Cryoconite biogeochemistry

Biogeochemical cycling in cryoconite holes has been intensely researched over the past decade due to the potential for storage, transformation and export of nutrients. Cryoconite ecosystems have been found to be co-limited by temperature and nutrient stress (Säwstrom et al., 2002), meaning knowledge of hole biogeochemistry is crucial for understanding cryoconite hole microbiology. Research has concentrated upon three main elements: carbon (C), nitrogen (N) and phosphorus (P), with a particular focus upon C. These nutrients will be studied individually here.

Net ecosystem production

Net ecosystem production (NEP) describes the balance between autotrophy (fixation of CO₂ into organic molecules) and heterotrophy (metabolism of organic molecules back into atmospheric CO₂):

NEP = PP − R

7

where PP = primary production and R = community respiration (Hodson et al., 2010).

This has been a particular focus for glacier microbiologists because NEP determines whether a community represents a C sink or a C source (e.g. Hodson et al., 2010b; Stibal et al., 2008a), and also whether there is an overall increase in dark OM on a glacier surface. In cryoconite holes autotrophy is opposed by heterotrophy, perhaps providing biotic control on granule morphology and glacier ablation (e.g. Hodson et al., 2008). NEP both drives and indicates community structure and function and hole biogeochemistry. For example, C fixation increases the pH of cryoconite hole melt water (Stibal and Tranter, 2007; Stibal et al. 2010), particularly important in Antarctic systems where cryoconite holes are decoupled from surface and atmospheric exchanges by thick ice lids. Several crucial abiotic controls upon NEP have been identified, including sediment arrangement (Cook et al., 2010; Telling et al., 2012a), PAR, N and P availability, solar angle and hydrologic regime (e.g. Hodson et al., 2007, 2010b; Irvine-Fynn et al., 2011; Mindl et al., 2007; Stibal et al., 2008b, 2012b; Telling et al., 2012a). Cook et al. (2010) and Telling et al. (2012a) suggested a threshold thickness of 3 mm above which systems are net heterotrophic, below which net autotrophic, and at which NEP is balanced. Telling et al. (2012a) found sediment thickness to explain more than half the variation in NEP in Arctic cryoconite due to the increased surface area of cryoconite exposed to PAR. Cook et al.’s (2010) mechanism of lateral equilibration suggests that given low gradient, slow moving ice and moderate sediment delivery, cryoconite holes evolve towards robust net autotrophy. Where ice surface are steeper or faster moving thick, net heterotrophic layers are more likely to form. Therefore, although rates of microbial activity in cryoconite holes are broadly similar to aquatic ecosystems (e.g. Anesio et al., 2009) NEP varies considerably over both space and time (Table 4). This has been further corroborated by various Greenland (Cook et al., 2012; Stibal et al., 2010, 2012b; Yallop et al., 2012) and Svalbard (Langford et al., 2014) transect studies which have found spatial patterns of NEP controlled primarily by granule size limits imposed by disaggregation and removal of biomass by melt water, while nutrient and PAR availability are also important. Under seasonal snow, photosynthesis probably ceases and systems become net heterotrophic; however, this remains untested. Outside of cryoconite holes, thick, net heterotrophic accumulations of cryoconite can develop under distinct redox and hydrological conditions (Hodson et al., 2008).

OPEN IN VIEWER

Table 4.

Previously reported rates of PP, R and NEP in debris in natural cryoconite holes. The methods used to derive these data are shown: DO₂ refers to changes in dissolved oxygen concentration, ¹⁴C refers to incorporation of traceable radioisotopes of C and ΔTDIC refers to changes in dissolved inorganic C concentration. ‘n.d.’ refers to values that were not reported. A and H refer to net autotrophy and net heterotrophy respectively.

Location	Lat.	NEP	PP	R	Bacterial production	Unit	A or H	Method	Source
Midtre Lovenbréen, Svalbard	79° N	n.d.	0.63 ± 0.03 (26 July) 24.6 ± 0.75 (5 Aug) 31.2 ± 2.10 (11 Aug) 157 ± 4 (22 Aug)	n.d.	n.d.	µg C L-1 h-1	n.d.	14C	Säwstrom et al. (2002)
Werenskioldbreen, Svalbard	79° N	n.d.	3.4 ± 2.2	0.44 ± 0.17	n.d.	µg L-1 h-1	n.d.	ΔTDIC	Stibal and Tranter (2007)
Midtre Lovenbréen, Svalbard	79° N	n.d.	n.d.	1.17 ± 0.18	0.04 ± 0.02	µg C g-1 h-1	A	DO2	Hodson et al. (2007)
Werenskioldbreen (Svalbard)	79° N	n.d.	4.3	n.d.	n.d.	µg C L-1 yr-1	n.d	14C	Stibal et al. (2008a)
Midtre Lovenbréen (Svalbard)	79° N	n.d.	353 ± 248	28.2 ± 4.37	39.7 ± 17.9	µg C g-1 d-1 (ng C g-1 h-1 for bacterial prod.)	A	14C (3H Leucine for bacterial prod.)	Anesio et al. (2009) (Anesio et al. (2010) for bacterial prod.)
Austre Brøggerbreen (Svalbard)	79° N	14.6	48.0 ± 35.9	15.3 ± 5.02	8.62 ± 6.41	µg C g-1 d-1 (ng C g-1 h-1 for bacterial prod.)	A	14C (3H Leucine for bacterial prod.)	Anesio et al. (2009) (Anesio et al. (2010) for bacterial prod.)
Vestre Brøggerbreen (Svalbard)	79° N	n.d.	208 +/- 106	34.3 ± 2.18	n.d.	µg C g-1 d-1	A	14C	Anesio et al. (2009) (Anesio et al. (2010) for bacterial prod.)
Stubacher Sonnblickees (Austria)	47° N	n.d.	147 ± 78.3	42.1 ± 7.91	0.13 ± 0.14	µg C g-1 d-1 (ng C g-1 h-1 for bacterial prod.)	A	14C (3H Leucine for bacterial prod.)	Anesio et al. (2009) (Anesio et al. (2010) for bacterial prod.)
Froya Glacier (Svalbard)	74° N	n.d.	115 ± 56.3	n.d.	n.d.	µg C g-1 d-1 (ng C g-1 h-1 for bacterial prod.)	n.d.	14C	Anesio et al. (2009) (Anesio et al. (2010) for bacterial prod.)
Patriot Hills (Antarctica)	80° S	n.d.	n.d.	n.d.	11.2 ± 4.11	ng C g-1 h-1	n.d.	3H Leucine
McMurdo Dry Valleys (Antarctica)	77° S	n.d.	n.d.	n.d.	23.4 ± 11.8	ng C g-1 h-1	n.d.	3H Leucine
Longyearbreen (Svalbard)	78° N	-2.03 ± 6.41	17.2 ± 9.7	19.2 ± 5.5	n.d.	µg C g-1 d-1	H	ΔTDIC	Hodson et al. (2010b)
Vestfold Hills (Antarctica)	67° S	n.d.	2.1 ± 1.5	1.86 ± 1.51	n.d.	µg C g-1 d-1	A	ΔTDIC	Hodson et al. (2010b)
Greenland Ice Sheet (near Kangerlussuaq)	67° N	n.d.	18.7 ± 10.1	20.9 ± 8.2	n.d.	µg C g-1 d-1	H	ΔTDIC	Hodson et al. (2010b)
Greenland ice sheet (nr Kangerlussuaq)	67° N	-0.14 (average blue ice areas) 0.01 ± 0.01 (at ice margin)	1.56 (average blue ice areas) 0.01 (at ice margin)	1.74 (average blue ice areas) 0.01 ± 0.01 (at ice margin)	n.d.	µM C g-1 d-1	A (blue ice areas) H (at ice margin)	ΔTDIC	Hodson et al. (2010a)
Canada Glacier, Antarctica	77° S	n.d.	∼ 1.4**	∼ 2.2**	n.d.	µg C g-1 d-1	H	DO2	Bagshaw et al. (2011)
Austre Brøggerbreen, Vestre Brøggerbreen, Midtre Lovenbréen, Svalbard	79° N	-0.12 ± 4.1	18.7 ± 10.3	18.7 ± 9.1	n.d.	µg C g-1 d-1	H	ΔTDIC	Telling et al. (2012a)
Greenland ice sheet (nr Kangerlussuaq)*	67° N	6.11	24.5	18.4	n.d.	µg C g-1 d-1	A	ΔTDIC	Stibal et al. (2012b)
Antarctic Blue Ice (mean values from 11 sites)	68 ° S	0.23	2.22	1.99	5.73 ± 6.00	µg C g-1 d-1 (ng C g-1 d-1 for bacterial prod.)	A	ΔTDIC	Hodson et al. (2013)
Rotmoosferner (Austria)	46° N	n.d.	3.71	86.6	n.d.	µg C g-1 h-1	H	14C	Edwards et al. (2013c)
Canada Glacier (Antarctica)	77° S	15.3 ± 11.7	n.d.	n.d.	2.4 ± 1.6	ng C g-1 h-1		3 H Leucine (bact. Prod.) 14C (PP)	Telling et al. (2014)
Greenland Ice Sheet (nr Kangerlussuaq) Cryoconite in holes	67° N	1.0 ± 0.33	0.76 ± 0.47	0.24 ± 0.11	n.d.	mg C L-1 d-1		DO2	Chandler et al. (2015)
Greenland Ice Sheet (nr Kangerlussuaq) Dispersed Cryoconite	67° N	0.64 ± 0.31	0.24 ± 0.17	0.4 ± 0.2	n.d.	mg C L-1 d-1		DO2	Chandler et al. (2015)

*Stibal et al. (2012b) reported values for NEP, PP and R at nine sites along a 79 km transect. The averages of all sites is presented here; however, there was significant spatial variability and the reader is directed to the original study.

**Bagshaw et al (2011) showed some temporal variability in PP and R values in Antarctic cryoconite – results here are representative of rates following a period of stabilisation in full sunlight.

NEP in cryoconite holes is impacted by bacterial activity (Foreman et al., 2007; Hodson et al., 2007; Mindl et al., 2007) predominantly via heterotrophic C oxidation, much of the energy from which drives bacterial growth and production. Anesio et al. (2010) examined bacterial production in Antarctic, Arctic and Alpine cryoconite, finding rates between 0.13 ng C g⁻¹ h⁻¹ (Stubacher Sonnblickees, Austria) and 39.7 ng C g⁻¹ h⁻¹ (Midtre Lovenbreen, Svalbard). The same study found bacterial doubling times to usually be <5 days (where >60 days was suggested to signify negligible contribution to cryoconite biogeochemistry) apart from in Antarctic melt water and radioactive cryoconite on Stubacher Sonnblickees (Austria). However, bacterial production only represents transformation of OC into biomass, omitting growth and respiration (Anesio et al., 2010; Hodson et al., 2007). To characterise bacterial C fluxes, bacterial growth efficiency (BGE) and respiration rates are also required (Hodson et al., 2007). Anesio et al. (2010) estimated bacteria in Arctic cryoconite to use only 0.4–2.4% of available OC (1.2–7% in Antarctica), suggesting autochthonous production alone was more than sufficient to sustain bacterial production. Low rates of production were therefore unlikely to be due to OC limitation. Temperature, P, viral lysis, and grazing of bacteria by ciliates were identified as much more likely limiting factors, although on Stubacher Sonnblickees, rates were probably also limited by radioactivity (∼140,000 Bq) persisting from the 1984 Chernobyl disaster and bomb tests in the 1950s and 1960s (Tieber et al., 2009).

Recognising the need for standardised reporting of NEP, PP and R, Telling et al. (2010) evaluated various measurement techniques. Measuring changes in dissolved inorganic C (ΔTDIC) consistently outperformed radiolabel incorporation and dissolved O₂ methods. A standard procedure was proposed by Hodson et al. (2010b) and Telling et al. (2010) whereby ΔTDIC incubations last for whole days and results normalised for dry mass. Rates of PP, R and NEP previously reported in the literature are shown in Table 4.

1 Directing future research

Glaciology has progressed from an abiotic to a biotic paradigm; however much of the modern literature assumes the supraglacial biome to be simple, quasi-static and exclusively comprising organisms contained in discrete cryoconite holes and algal blooms. Regarding cryoconite, there remains a tendency to ignore the diversity of hole shapes and sizes present on natural ice surfaces and instead discuss (and sample) symmetrical, cylindrical holes with flat floors and even sediment arrangements. Since hole morphology impacts the photic conditions at the hole floor (Cook et al., 2010) and cryoconite microbes are sensitive to changes in their environments (Edwards et al., 2014b) it follows that hole morphology probably influences cryoconite microbial ecology and therefore deserves research attention. Furthermore, if cryoconite holes undergo morphological evolution and migration then their contribution to supraglacial melt might be continuous. It has also recently been recognised that cryoconite holes are not spatially isolated entities occasionally interlinked by ephemeral, discrete hydrological flowpaths, but rather represent dynamic sites of organic and inorganic matter storage and high microbial activity punctuating a spatially expansive aquatic ecosystem within the porous near-surface ice layer known as the weathering crust (Irvine-Fynn et al., 2012; Muller and Keeler, 1969). This weathering crust ecosystem is then widely connected to other glacial and extraglacial environments (Figure 7). Since the top of the weathering crust represents the interface between ice and atmosphere and is exposed to incoming solar radiation, it is the primary site of radiative and sensible heat exchange and its properties therefore strongly influence supraglacial melt rates. Key variables in the surface energy balance include surface roughness and albedo, both of which are influenced by microbial activity, yet biology has not been incorporated into any predictive models of glacier melt. To do so will require better knowledge of the interactions between microbes and the ice matrix they inhabit. To improve our understanding of ice-microbe interactions, microbial ecology and biogeochemistry in the supraglacial zone, cryoconite and other microbial habitats must be studied in concert with the evolution of ice surface roughness, topography and hydrology. We propose a framework for integrating these biological and glaciological processes and introduce the term ‘biocryomorphology’ in reference to ice-microbe interactions.

OPEN IN VIEWER

Figure 7.

Three diagrams that show carbon fluxes through supraglacial systems. Black arrows denote flux of C. (a) Carbon cycling within an individual cryoconite hole; (b) Connectivity between cryoconite holes and other glacial and extraglacial environments; (c) Interconnectivity between cryoconite holes via hydrologic fluxes through the weathering crust. We gratefully acknowledge Antony Smith (DGES, Aberystwyth University) for his assistance with this figure.