Species composition

Diatoms grew in all of the culture material but grew more abundantly in the Bold’s Basal and BG-11 media, along with many species of green algae and cyanobacteria. Only the results for diatoms are presented in this study. Cultured material from each medium type was pooled for subsequent analysis. Diatom species from 27 genera were identified in the material sampled from the cryoconite holes and cirque glacier site (Table 1). The total number of genera ranged from 12 (Frøya Glacier) to 17 (cirque glacier). A number of genera were common to all sites, namely Achnanthidium, Diadesmis, Mayamaea, Navicula, Nitzschia and Psammothidium. Only two species of Luticola were identified: L. nivalis and L. ventricosa (Fig. 1). Three other distinct forms were identified, though they could not be reliably assigned to any species in the literature and could be new taxa for this genus (personal communication from Reference Van de Vijver, Mataloni, Stanish and SpauldingB. van de Vijver, 2010). This genus was restricted to the Greenland cirque glacier. Diadesmis contenta (Fig. 1) grew particularly well in the cultured samples from this site but was also found in the material cultured from the three cryoconite hole samples. By contrast, Psammothidium helveticum dominated the cultured material from Frøya Glacier and Austre Brøggerbreen. Three species of Gomphonema, G. parvulum, G. olivace-oides and G. angustatum, were recorded from Austre Brøggerbreen, and four species of Nitzschia were found: N. amphibia, N. paleaeformis, N. paleacea and N. palea var. debilis (Fig. 1). The latter species dominated the samples from Vestre Brøggerbreen.

Table 1. Genera (with codes recorded in parentheses) from cryoconite in Austre Brøggerbreen (C1) and Vestre Brøggerbreen (C2), Svalbard, and Frøya Glacier (C3) and cirque glacier (RT1), Greenland

Fig. 1. Selected diatom species isolated from Arctic cryoconite. 1 . Luticola ventricosa. 2. Luticola nivalis. 3. Hantzschia amphioxys. 4. Nitzschia amphibia. 5. Nitzschia palea var. debilis. 6. Pinnularia borealis. 7. Diadesmis contenta. 8. Muelleria cf. terrestris. 9. Planothidium frequentissimum. 10. Reimeria sinuata. 1 1 . Psammothidium marginulata. 12. Psammothidium subatomoides. 13. Caloneis molaris. 14. Stauroneis (?) sp. 1 . 15. Meridion circulare. 16. Mayamaea atomus var. permitis.

Spatial comparisons

To examine associations between samples, comparisons were made, at the generic richness level, between the diatom assemblages collected from cryoconite material and those identified from previously published work in polar environments, totalling 73 sites. The first split of the TWINSPAN analysis separated the Antarctic lake samples from the rest of the samples, based on the more frequent co-occurrence of Stauroforma and Psammothidium in the benthic diatom samples taken from Antarctic lakes and more frequent co-occurrence of Navicula, Staurosirella, Staurosira, Encyonema, Caloneis, Nitzschia and Achnanthidium at the other sites. In the second quadrat division, all the Svalbard lake samples and the Frøya Glacier sample formed a group (n=24; TWINSPAN group 1), though the glacier sample was borderline in terms of assignment. Genera preferring this group included Achnanthidium, Diatoma, Fragilaria, Cymbella, Encyonema, Tabellaria and Staurosirella. The Greenland cirque glacier diatom assemblage and other samples from cryoconite holes formed a second group (TWINSPAN group 2) together with the samples of diatom assemblages from terrestrial mosses in Svalbard and diatoms sampled from both submerged and emergent mosses in three shallow ponds on the Potter Peninsula. Genera preferring TWINSPAN group 2 included Luticola and Muelleria. The Antarctic lake samples formed two further groups. Samples assigned to TWINSPAN group 3 (n = 14) commonly supported Luticola and Stauroneis, while lakes in group 4 were more typically represented by Amphora, Craticula and Planothidium.

Spatial segregation of the sites is shown in the DCA samples plot (Fig. 2a). The eigenvalues of axes 1 and 2 were 0.551 and 0.152 respectively, and collectively explain 3 1% of the cumulative percentage variance in the species data. Symbols differentiate the four TWINSPAN groups on the DCA and indicate the clear separation of Antarctic lake sites (open diamonds and circles), to the right of the ordination, from the Arctic lakes to the left (closed diamonds), based on the composition of the diatom assemblages at the generic level. The samples from two of the cryoconite holes in Svalbard, the cirque glacier in Greenland and aquatic and terrestrial moss communities from both the Arctic (coded E1–3) and Antarctica (coded P1–3) occupy a central position on the ordination between the two sets of lake sites (Fig. 2).

Fig. 2. Detrended correspondence analysis (DCA) of diatom assemblages, analysed at the genus level, showing the first two axes of (a) sample plot and (b) species plot based on their diatom composition. Site codes for the Arctic are: S = Svalbard lakes; E = Edgeøya (Svalbard) terrestrial moss; C and RT1 = cryoconite. Site codes for the Antarctic are: P = Potter Peninusula ponds and L = lakes from Larsemann Hills and Rauer Islands. Symbols reflect four groupings from TWINSPAN analysis: closed diamonds are TWINSPAN group 1 ; closed circles are TWINSPAN group 2; open diamonds are TWINSPAN group 3; and open circles are TWINSPAN group 4. Taxa codes for diatoms found in cryoconite (see Table 1), with additional codes: Dent = Denticula, Crat = Craticula, Amp = Amphora, Cham = Chamaepinnularia, Staf = Stauroforma, Aul = Aulacoseira, Neid = Neidium, Pseu= Pseudostaurosira, Bra = Brachysira, Staa = Staurosira, Cycl = Cyclotella, Stel = Staurosirella, Cymb = Cymbella, Dipl = Diploneis, Tab = Tabellaria.

The DCA species biplot (Fig. 2b) displays the main variation in the diatom genera. The diatom genera Craticula, Stauroforma, Amphora, Diadesmis and Psammothidium situated more to the right of the ordination were commonly recorded in the benthic diatom samples in Antarctic lakes. Craticula, Amphora and Stauroforma were not recorded in any of the cryoconite samples analysed in the present investigation from the Arctic cryoconite. Some species common to all the cryoconite samples including Achnanthidium, Nitzschia and Navicula were positioned to the left of centre, along axis 1 , and were more commonly found co-occurring in Arctic assemblages. Along axis 2, a number of genera positioned towards the positive and negative end, including Eunotia, Brachysira, Surirella, Denticula, Neidium and Diploneis, were commonly recorded for some of the Svalbard lakes. These genera were notably absent from the Antarctic lake samples, though recorded as occasional or rare in the moss samples in Potter Peninsula. Surirella was recorded in Austre Brøggerbreen, and Eunotia and Frustulia were recorded in the cryoconite from Frøya Glacier, though no records of the other four genera were recorded in other cryoconite material sampled during this study.

Generic richness

The number of diatom cells in the raw cryoconite debris was relatively low, necessitating culturing to obtain a sufficient sample size. In studies of diatom species composition, the organic matter usually has to be removed to expose the frustules to obtain the required taxonomic resolution. One of the criticisms of these currently employed methods is that it is impossible to distinguish between living and dead cells. Empty frustules could therefore be included in cell counts. Between 12 and 15 live diatom genera were recorded in the cryoconite holes, and 17 genera were recorded for the cirque glacier. Although some genera were common to all samples, there were differences and, in total, 27 live genera were recorded. These numbers are comparable with mean local and regional numbers of diatom genera recorded from large-scale analyses of lakes in similar latitudes in the Northern Hemisphere (Reference VyvermanVyverman and others, 2010), which suggests that cryoconite may support a similar diatom richness to that of surrounding lacustrine environments. The number of live genera recorded was higher than has previously been recorded in cryoconite holes. Reference Mueller, Vincent, Pollard, Fritsen, Elster, Seckbach, Vincent and LhotskyMueller and others (2001) found many different diatom frustules in Canada Glacier, McMurdo Dry Valleys, Antarctica, though only species in the genus Muelleria were regularly recorded as living. In agreement with this study, a number of species of Luticola were noted, though their relative abundance, compared to other algae and cyanobacteria, was low. Seven species of diatom belonging to three genera, Achnanthes, Synedra and Gomphonema, were found in cryoconite holes on North Glacier, Mount Athabasca, Alberta, Canada (Reference Wharton and VinyardWharton and Vinyard, 1983). In other studies, no diatoms were found (Reference BroadyBroady, 1989; Reference Sawstrom, Mumford, Marshall, Hodson and Laybourn-ParrySawstrom and others, 2002). Pennate diatoms have previously been recorded in cryoconite from the glacial surfaces in Svalbard, but cyanobacteria were found to be numerically dominant compared to microalgae (Reference Stibal, Sabacka and KaštovskáStibal and others, 2006), a finding corroborated in other studies (e.g. Reference Mueller, Vincent, Pollard, Fritsen, Elster, Seckbach, Vincent and LhotskyMueller and others, 2001). Evidence obtained during this study indicates that cryoconite in Svalbard and Greenland can support a relatively high number of viable diatom genera.

Cell viability and sources of colonizing diatoms

The diatoms and other photosynthetic microbes in the cryoconite samples cultured in this study had previously been frozen (∼1–2 years), yet retained their viability. Freshwater pennate benthic diatoms are frequently recorded to form resting cells that can survive adverse conditions (Reference Sicko-GoadSicko-Goad, 1986). In particular, subaerial or aerophytic diatom species that commonly inhabit rock or soil can survive desiccation for long periods of time (Reference Round, Crawford and MannRound and others, 1990). A number of diatoms found in the cryoconite during this investigation including species of Diadesmis, Pinnularia, Hantzschia and Luticola are aerophytic. Aerophytic diatoms are less tolerant to desiccation than cyanobacteria (Reference Poulíčkova and HašlerPoulíčková and Hašler, 2007), and moisture content may be crucial to their survival (Reference Van de Vijver and BeyensVan de Vijver and Beyens, 1997). Some species within the genus Luticola (e.g. L. nivalis) are halophilous and can tolerate more saline conditions which may be experienced in areas of high evaporation. A number of representatives of this genus were recorded on the Greenland cirque glacier, where evaporative losses may be elevated, though they were not recorded elsewhere. Some of the diatom taxa recorded in the cryoconite are commonly recorded on mosses (e.g. Nitzschia palea var. debilis; Hantzschia amphioxys; Pinnularia borealis) or are truly epiphytic (e.g. Planothidium and Gomphonema species (Reference Vinocur and MaidanaVinocur and Maidana, 2010)), and cryoconite samples containing these shared a similar location on the ordination plot to those diatom assemblages sampled from aquatic, terrestrial or semi-terrestrial mosses from both the Arctic and Antarctica (Fig. 2a). Other motile, epipelic species in the genera Navicula, typically associated with lake or stream sediments, were also common in the cryoconite material. The presence of diatoms from a wide range of terrestrial, semi-terrestrial and aquatic sources indicates that cryoconites may act as a reservoir, accumulating a number of diatom species from a variety of local sources. Many of these diatom cells may be growing in conditions that fall outside their optima in terms of environmental preference, which would explain their relatively low abundance compared to other major autotrophic groups. However, their proven viability from the culturing techniques adopted in this study indicates that while diatoms may not be numerically dominant in cryoconite, the debris may contain a number of viable diatom cells. Given the tolerance of many of the diatom genera recorded in the cryoconite to desiccation and/or freezing, these cells could also serve to propagate other areas following ablation.

Comparisons between microbial assemblages in different types of supraglacial habitats indicate variation in the composition and relative abundance of species. Cyanobacteria-dominated cryoconite material and algae (mainly greens) were considered as ‘accessory’ organisms (Reference Kaštovská, Elster, Stibal and ŠantruckováKaštovská and others 2005). Chemical and physical properties were found to be overriding factors determining the fate of microbial cells deposited on the glacial surface (Reference Stibal, Sabacka and KaštovskáStibal and others, 2006). It is likely that only those diatom genera capable of tolerating low moisture content, or relatively higher salinity or conductivity (e.g. some species in the genera Pinnularia and Luticola), may proliferate in soil habitats including vegetated soils (Reference Van de Vijver and BeyensVan de Vijver and Beyens, 1997).

A number of the benthic diatom species recorded in this study (e.g. Achnanthidium minutissimum, Planothidium lanceolatum, Stauroneis anceps, Encyonema minutum, Nitzschia palea), commonly recorded in periphytic communities of lakes and rivers in European waters (Reference King, Clarke, Bennion, Kelly and YallopKing and others, 2006; Reference KellyKelly and others, 2008), are psychrotolerant rather than obligately cold-tolerant (Reference Seaburg and ParkerSeaburg and Parker, 1983), so they should be able to grow well in higher temperatures at lower altitudes and latitudes.

We recorded a number of ‘pioneer’ species in the cryoconite material belonging to three genera, Gomphonema spp., Achnanthidium spp. and Cocconeis spp., and these species typically represent the starting point in successional processes leading potentially to mature biofilms in lentic and lotic environments (Reference Yallop, Kelly, Crawford, Moss, Mann and PreisigYallop and Kelly, 2006). Diatom assemblages supporting the latter two of these taxa typify relatively ‘pristine’ waters indicating a ‘reference state’, defined by European legislation as sites relatively unimpacted by human activity (Water Framework Directive (WFD; EU, 2000). These pioneer cells are fast colonizers with rapid growth rates and adopt the R-selected life-history strategy (i.e. opportunistic species with high intrinsic growth rates) within a habitat matrix of disturbance frequency and nutrient resource supply (Reference Biggs, Stevenson and LoweBiggs and others, 1998; Reference Yallop, Kelly, Crawford, Moss, Mann and PreisigYallop and Kelly, 2006). Typically, in relatively undisturbed habitats, S-selected (i.e. stress-tolerant) species, including more green algae, and cyanobacteria as well as other diatoms, would join the pioneer species once the biomass increased and competition for resources became more intense. Many of these species may be nitrogen fixers (Reference Biggs, Stevenson and LoweBiggs and others, 1998). An isolated empty frustule of an S-selected diatom, Rhopalodia gibba, which contains nitrogen-fixing endosymbionts, was recorded in the material from cryoconite on Vestre Brøggerbreen (Yallop, unpublished data). The window of opportunity for further colonization of the cryoconite holes in the Arctic is limited to a period of weeks during the summer when the cryoconite holes are open. This period of time may be insufficient to allow for biofilms to reach a more mature stage. Repeated destabilization events and constant physical (e.g. flushing) and biological (e.g. grazing) stress may result in only the R-selected functional group surviving in the cryoconite holes. The unique habitat within the cryoconites may favour the growth of cyanobacteria and green algae, thereby accounting for the relatively low biomass of diatoms.

Origin of diatoms in cryoconite

The diatom assemblages from the cryoconite material bore many similarities, at the generic level, to the assemblages in some of the lakes from Svalbard, which could indicate they were a likely source of propagules for cryoconites. Reference Jones and BirksJones and Birks (2004) found that these lakes were split into three groups, based on the relative abundance of diatoms, at the species level, separated along gradients of pH, alkalinity, concentrations of cations, and conductivity. The diatom assemblages in the cryoconite material shared more similarities with their TWINSPAN group 2 sites which were sites of relatively lower pH, low Ca and lower conductivity. Limnological measurements of meltwaters of glaciers in the Arctic generally indicate acidic conditions of very low conductivity (Reference Remias, Holzinger and LützRemias and others, 2009). The lower pH of cryoconite holes may impose limitations on the growth of species with relatively higher pH optima. These variables have also been found to be important determinants in the spatial distribution of diatom assemblages in other polar habitats (Reference Antoniades, Douglas and SmolAntoniades and others, 2005; Reference Michelutti, Smol, Douglas and DouglasMichelutti and others, 2006). Differences between the diatom assemblages from the cryoconite samples in Greenland and Svalbard may be due to differences in pH and associated correlates of the meltwaters at these sites. The presence of acid-loving genera Eunotia and Frustulia in the Greenland samples, and their absence from the Svalbard samples, lends further support to this argument. The lack of similarity between the diatom assemblages of the cryoconite material and most of the Antarctic lakes is not surprising given the relatively high chloride concentrations of Antarctic lakes which favoured species like Craticula, not recorded in the cryoconite samples taken in this study.

Analysis of debris from the cryoconite on the surface of snow on a Himalayan glacier indicated the presence of some fragments of plant origin (Reference Takeuchi, Kohshima and SekoTakeuchi and others, 2001). Moss fragments could come from local aquatic and terrestrial sources and serve as vehicles for diatom transport to cryoconite holes. Additional propagules may arrive from geographically distant sources by aeolian transport. The observation that moss-derived diatom assemblages from Potter Peninsula bore a number of similarities to samples from the Arctic may be indicative of the taxonomic resolution of the study (i.e. generic rather than specific). Ongoing and recent advances in our understanding of the taxonomy and environmental preferences of some of the key taxa inhabiting these environments will help to address these questions (Reference Van, de Vijver and MataloniVan de Vijver and Mataloni, 2008; Reference Van de Vijver, Mataloni, Stanish and SpauldingVan de Vijver and others, 2010). The degree of endemism in Antarctic diatom samples is considered to be relatively high, which may be explained by their relative isolation compared with Arctic communities. At the generic level, we identified broadly distinct patterns of site groupings. A generic-level analysis may not work in every situation (Reference Chessman, Growns, Currey and Plunkett-ColeChessman and others, 1999), as some genera (e.g. Navicula) contain species with a wide range of ecological tolerance (Reference LoweLowe, 1974). However, using this higher level of taxonomy is still advantageous, given the degree of difficulty associated with identification of diatoms at the species level and the further complexities of cryptic and pseudo-cryptic species (Reference Mann, Evans, Brodie and LewisMann and Evans, 2007; Reference VanelslanderVanelslander and others, 2009). In situations where the genera are not particularly speciose, a generic level of analysis may be sufficient to pick up signals of environmental change, as found by Reference GrownsGrowns (1999), and this may be the case in polar environments. In accordance with Reference Bouchard, Gajewski and HamiltonBouchard and others (2004), our findings lend support to the use of these more ‘natural groupings’ to understand patterns of biogeography and environmental inference.

As glaciers retreat, the diatom cells residing in cryoconite have the potential to act as seeding agents for a variety of terrestrial and aquatic habitats in proglacial sites, which, given the large surface area covered by cryoconite holes, may present a greater contribution than had previously been realized.