Dynamics of organic matter in algal blooms on the Greenland ice sheet

Algal diversity, DOC concentrations and OM molecular characteristics

18S rRNA marker gene sequencing confirmed that the snow-algae Chloromonas and the glacier ice-algae Ancylonema^7,17,22 dominated the eukaryotic community composition in each habitat (90% and 65% respectively, Fig. 1). Snow samples contained solely snow-algae taxa, whereas ice samples still contained snow-algae remnants (12%, Fig. 1) from snow-melt, as previously reported⁴. The two habitats differed not only in algae taxonomy, but also in algae life cycle stage, with snow-algae normally considered to be close to the resting stage and ice-algae still actively growing and dividing²³.

Compositional differences in all OM (both DOM and POM and both light and dark samples) in the purple glacier ice-algae (GIA) and red snow-algae (RSA) dominated habitats based on Bray Curtis dissimilarity. (a) Non-metric multidimensional scaling (NMDS) of ice-algae (purple labels at left) and red snow-algae (red labels at right) samples, with the vertical and horizontal axes (NMDS1 and NMDS2) explaining the OM molecular variability (grey x); the legend in the upper left details the information in the sample labels: type of incubation (dark, D and light, L) and time (T in days) for DOM and POM samples, and type of extracts for POM (NaOH and hot water). Plotted at left and right of the NMDS plot in (a), is the relative abundance of algal species in the initial samples (T0_Ice and T0_Snow) based on 18S rRNA sequencing analyses (see further information in methods). (b) and (c) van Krevelen diagrams with molecular formulae plotted according to their H/C and O/C ratios using NMDS1 loadings ≤ 0.45 for ice-algae (at left) and ≥ 0.45 for snow-algae (at right); displayed in each diagram is the overall contribution (in %) of saturated, unsaturated aliphatics, highly unsaturated, aromatics and condensed aromatics (Tables S4 and S5) as well as the elemental composition of the formulae (CHO, CHON, CHOS, CHOP) all relative to the total number of formulae in each habitat signal (4,078 in GIA and 4,749 in the RSA). Note: due to insufficient particulate material in the red snow at intermediate time steps, POM could only be analyzed in the initial and final time point sample.

Over the 24-day incubations, dissolved organic carbon (DOC) concentrations (0.7 ± 0.05 in snow and 1.2 ± 0.1 mg-C l^-1 in ice) increased ca. 2- to 6-fold (Fig. S3). Water-soluble (POM-HW) and water-insoluble extracts (POM-NaOH, Fig. S1) also differed with water solubilizing 2-fold less DOC from snow compared to ice (1184 ± 56 vs. 2128 ± 216 mg-C l^-1), while water-insoluble DOC concentrations were similar (885 ± 57 vs. 1090 ± 187 mg-C l^-1, Fig. S3). In the POM-extracts of snow, DOC concentrations remained similar over time, while in the ice they decreased 1 to 2.5-fold (Fig. S3). The different algal genera also produced compositionally different OM (Figs. 1 and S4, Tables S1, S2 and S3). Snow-algae had significantly more phosphorus and sulfur compounds compared to ice-algae (Figs. 1b-c and S5, Tables S4 and S5). Up to 1% of the phosphorus- and sulfur-containing compounds in the initial snow-algae POM were also present in the initial ice-algae POM, likely due to the snow-algae present in the initial ice-algae sample (Table S1).

Nitrogen-containing compounds made up 30–38% of the total formulae in both habitats, but in snow, they were 2-fold higher in DOM, whereas in ice, they were 13-fold higher in POM (Figs. 1b–c and S4). Differences in abundance of phosphorus- and nitrogen-containing compounds between these habitats likely reflect algae metabolic activity²⁴ and nutrient availability²⁵. In snow, nutrients are from dry/wet deposition, while on glacial ice, nutrients are from snow/ice melting and mineral dissolution^1,25. Like other algae²⁶, snow- and ice-algae may use diverse strategies for phosphorus uptake. These adaptations, along with the difference in nutrients, and algae life cycle stages, are likely responsible for the higher abundance of phosphorus compounds in snow compared to ice-algae OM. In algae, sulfur is essential for metabolism and environmental adaptations²⁷ and in our snow, sulfur compounds likely reflect Chloromonas response to oxidative stress from high solar radiation²⁸. In contrary, glacier ice-algae, with their darker purple pigments²⁹, are likely better adapted to light stress and may not actively produce sulfur compounds.

Unsaturated aliphatics (H/C ratios > 1.5), which are linked to labile algae-derived OM³⁰, were 2.5-fold more prevalent in snow than ice. In ice, almost all unsaturated aliphatics were POM-derived (Figs. 1b–c, S4 and S6, Tables S4 and S5). In such a bloom, POM is mainly algal (Supplementary note 1), as confirmed by protein analyses on same sample types³¹. Thus some labile unsaturated aliphatics may include water-soluble lipids and proteins from algae^6,8,32. However, the advantage for snow-algae of releasing unsaturated aliphatics when they transition to the cyst life stage is unclear (Fig. S4d). Highly unsaturated, aromatics and condensed aromatics (H/C ratios < 1.5), were especially higher in the ice-algae samples. This aligns with the DOM composition reported from a similar glacier ice-algal habitat¹² but is distinct from snow-algae habitat (Figs. 1 and S6, Tables S4 and S5).

Condensed aromatics were significantly more abundant in ice-algae (13%) compared to snow-algae OM (2%) and in snow, almost all condensed aromatics were in DOM (Figs. 1, S4 and S6). The origin of condensed aromatics on glaciers is unclear and could be allochthonous¹⁹. If formed from incomplete combustion of biomass or fossil fuel burning, they represent particulate wind-delivered black carbon (BC) known to reduce snow albedo³³. However, if all our condensed aromatics would be BC or dissolved BC^15,19, it is unclear why BC would be more abundant in ice- than snow-algae blooms. Alternatively, our condensed aromatics could be algae-produced compounds. Indeed many aromatics and condensed aromatics in ice (66% and 72%, Table S4) and snow (95% and 96%, Table S5) could belong to the extended molecular category of oxy-aromatic phytochemicals, considered abundant in plants³⁴. Some of these compounds might include phenolic to polyphenolic metabolites produced by algae³⁵, falling within the mass range of those in our samples (100–800 Da, Table S4 and S5).

Among phenolic metabolites, sugar-containing tannins and flavonoids, can assist in UV protection, pigmentation, cell-signaling and antioxidant activity^35,36. In glacier ice-algae, the abundant purpurogallin phenolic pigments aid in photoprotection against excessive solar radiation and cellular heat generation^18,29. In snow-algae, aromatics were mainly present in DOM and had lower O/C ratios and higher sulfur contents (Figs. 1c and S4d, Tables S4 and S5). Since glacier ice-algae phenolic pigments were previously documented only in higher plants, the presence of sulfur-free and sulfur-containing aromatics in ice- and snow-algae, respectively, may also be related to their taxonomy (ice-algae are Zygnematophytes and snow-algae are Chlorophytes). Aromatics, with their conjugated carbon double bonds, absorb light, crucially influencing the light-absorption and optical properties of chromophoric DOM (CDOM)²¹, with CDOM in these blooms closely linked to algae biomass and pigments³⁷. We demonstrate that these blooms produced distinctive OM patterns, containing aromatic compounds, particularly abundant in glacier ice-algae habitats (Figs. 1 and S4), that will invariably influence the GrIS color. Their impact on GrIS darkening depends not only on their light-absorbing properties, but also on their potential accumulation over the summer, influenced by their production and degradation. To constrain OM production and degradation, we assessed molecular variations over time under light and dark conditions (Fig. 1b–c, Tables S4 and S5).

Release of OM from particles in light

The light exposure induced DOM production or release from POM can occur abiotically or biotically through photosynthesis. Photoproduced algal DOM may dominate over heterotrophy and progressively increase, especially during the early experimental period when photosynthesis was not DIC limited. We evaluated biotic and abiotic changes by following the release of water-soluble and water-insoluble compounds from POM (Fig. S1), as evidenced by the shared formulae between POM and DOM found exclusively in light incubated samples. We show that up to 18% and 36% of the initial ice- and snow-POM formulae were transferred to their DOM pools (Figs. 2 and S7, Tables S4 and S5). Despite constituting less than half of the initial POM, these formulae accounted for > 50% and 70% of the DOM-T0 composition (Figs. 2 and S7).

Time resolved influence of solar radiation on the transfer of water-soluble OM from POM to DOM in the glacier ice-algae (left panel) and red snow-algae (right panel) experiments. van Krevelen diagrams with molecular formulae according to their H/C and O/C ratios for (a) and (b) water-soluble ice-algae OM at T0 (empty) and over time (filled red symbols) only under light conditions; (c) and (d) ice-algae DOM at T0 (purple) and DOM shared with water-soluble ice-algae OM over time (dark tone symbols); (e) and (f) water-soluble snow-algae OM at T0 (empty) and over time (filled red symbols) only under light conditions and (g) and (h) snow-algae DOM at T0 (red) and DOM shared with water-soluble snow-algae OM over time (dark tone symbols). Symbol shapes indicate organic compounds without (CHO) or with heteroatoms (CHON, CHOS, and CHOP). The contribution of formulae over time, indicated by the increasing dark color intensity in each panel, is expressed as percent relative to the total number of formulae in POM-T0 for glacier ice-algae (total 2,945 formulae) and red snow-algae experiments (total 3,066 formulae); for details of POM-T0 for each experiment see Tables S4 and S5. For ease of viewing DOM-T0 symbols in ice and snow are displayed in purple and red one size bigger and those uncover by dark symbols represent formulae in DOM-T0 that were not shared with the initial POM. Note: due to insufficient particulate material in the red snow at intermediate time steps, POM could only be analyzed in the initial and final time point sample.

Autotrophically produced DOM was particularly evident during the first nine days of the experiments, with photoproduced formulae comprising ~ 50% or more than those in DOM-T0 (Fig. S8). Such a DOM photoproduction has been also documented in Antarctic clean snow³⁸. In both algal experiments, the autotrophically produced formulae were 29–41% of those released from POM (Fig. S9, Tables S4 and S5), suggesting that light alone could induce additional processes that refill the DOM pool, potentially explaining more than half of its composition (Fig. S9). Over time, water-soluble formulae decreased, along with those transferred to DOM in the ice samples (Fig. 2a–d), while water-insoluble displayed more variability (Fig. S7a-d). This trend was mirrored in the DOC from DOM and POM extracts (Fig. S3). Such a DOC increase could indicate abiotic leaching, a process inferred for cryoconite particulates³⁹. This may be more common, yet not well studied POM-DOM transfer mechanism in algal-rich ice/snow environments. In snow, water-soluble and -insoluble formulae decreased with time, alongside with those transferred (Figs. 2e–h and S7e-h). However, unlike in ice, these changes were not reflected in the DOC (Fig. S3).

In the ice experiments, the transferred OM was related to unsaturated aliphatics and oxygen-rich highly unsaturated (O/C > 0.5) and aromatics (Fig. 2c–d, Table 1). Despite the fact that ice-algae POM had aliphatics and highly unsaturated nitrogen-containing compounds, < 4% of those in POM-T0 were transferred or were rapidly used up and thus almost absent in DOM (Fig. 2b,d). In snow, transferred OM contained nitrogen and phosphorus, and was slightly oxygen-poorer, and related to highly unsaturated and unsaturated aliphatics (Figs. 2g–h and S7g-h, Table 1). Overall, photoproduction diversified the DOM-T0 composition in both habitats, increasing the number of formulae without heteroatoms (i.e., CHO) in the ice vs. nitrogen and sulfur compounds in the snow. Some of these sulfur compounds were aromatic and present only in the snow-DOM likely due to autotrophic photoproduction (Figs. 2g–h and S8c-d). Our findings indicate that bloom-derived DOM composition depends on autotrophic metabolic production and the overall POM composition. While POM is algae dominated (Supplementary Note1), it could also contain a minor OM fraction from other microorganisms or atmospheric deposition, that can affect the DOM pool by releasing OM during light exposure, contributing to the DOC released from glaciers^14,40. The role of algal blooms is however, relevant since their DOC concentrations can be over one-fold higher (Fig. S3) than in ice/snow surfaces without blooms (usually < 0.5 mg-C l^-1)^15,19,20. The bloom-derived DOM will have a greater impact on GrIS surface darkening, as during the diurnal freeze-thaw cycles this OM will become ice-locked. Although part of the ice sheet surface DOM is exported^9,14, the daily freeze-in, the slow DOM circulation on the weathering crust¹² and the presence of “sticky” exopolymeric substances in these habitats⁴¹, likely maintain the DOM composition and prevent its degradation thus fostering its accumulation over time. We found higher DOC and aromaticity in the ice vs. the snow samples. However, aromatics in DOM are easily photodegraded, which supplies labile DOM for heterotrophic microorganisms⁴² and thus likely remove aromatics at the end of the summer.

Alteration of DOM

DOM changes in the dark samples revealed that ~ 50% (ice) and 35% (snow) of the DOM-T0 formulae, were heterotrophically degraded (Figs. 3a–b and e–f), aligning with the high bioavailability and rapid turnover of glacier DOM^14,43. This heterotrophically degraded DOM was more diverse compared to the photodegraded DOM that represented a far smaller fraction (6–16% were solely degraded in light-incubated samples; Fig. S10). This contrasts with ~ 70% of the DOM formulae photodegraded in bacterial-dominated clean snow³⁸. Our samples were exposed to light radiations high enough to breakdown photo-reactive DOM⁴⁴, although the experimental bottles filtered out the highest energy of UV-B (see Methods), and thus the magnitude of the photochemical effects are more conservative. Our results suggest that our microalgae-DOM was less photosensitive (Fig. S10) likely due to the algae producing fresher, less aromatic DOM³⁰. Alternatively, glacier microalgae, which are physiologically adapted to strong solar radiation²⁹, may produce photoresistant DOM as reflected especially in our ice experiments (Fig. S10). In these, highly unsaturated and unsaturated aliphatics without heteroatoms were heterotrophically degraded, while in snow experiments, compounds with nitrogen, sulfur and phosphorus compounds were also degraded (Fig. 3a–b and f ,Tables 1, S4 and S5). These degraded nitrogen- and phosphorus-containing compounds were mainly oxygen-poor unsaturated aliphatics and highly unsaturated, while sulfur compounds included oxygen-rich unsaturated aliphatics and aromatics (Fig. 3f). Furthermore, ice samples had ~ 4 times more aromatics resistant to bio- and photo-degradation, while snow had ~ 3 times more highly bioavailable, unsaturated aliphatics (Figs. 1, 3 and S10a, Table 1). Degradation of hydrogen-rich aliphatics by heterotrophic processes (Fig. 3a and e–f), aligns with their bioavailability^30,43 and potential utilization as a carbon source, being incorporated by heterotrophic microorganisms or transformed into new compounds.

Heterotrophic molecular signals in DOM of glacier ice-algae (GIA) and red snow-algae (RSA) dominated habitats at dark conditions. van Krevelen diagrams with molecular formulae according to their H/C and O/C ratios for (a) and (b) initial DOM (DOM-T0, purple) and progressively degraded or (c) and (d) produced formulae over time due to heterotrophy in glacier ice-algae experiments (dark tones); (e) and (f) initial DOM (DOM-T0, dark red) and progressively degraded or (g) and (h) produced formulae over time due to heterotrophy in red snow-algae experiments (dark tones). Heterotrophic degradation was related to a consistent decrease in the mass peak intensity of molecular formulae until they either reached zero by the specified time or they decreased consistently without reaching zero at the end of the experiment. Heterotrophic production of molecular formulae was related to a consistent increase in mass peak intensity until they reached a maximum by the specified time. Symbol shapes indicate organic compounds without (CHO) or with heteroatoms (CHON, CHOS, and CHOP). Both degradation and production (dark tones) expressed gradual change over time in percentage relative to the total number of formulae in DOM-T0 in glacier ice-algae or red snow-algae (for details see Tables S4 and S5). For ease of viewing symbols for the DOM-T0 in ice and snow habitats are displayed in purple and dark red one size bigger than those produced or degraded. Plotted at left and right of figures (a) and (f) is the relative abundance of microbial species in the initial samples (T0_Ice and T0_Snow) based on 16S rRNA sequencing analyses (see further information in methods).

After 22/24 days in darkness, heterotrophic metabolism resulted in a 74% (ice) and 24% (snow) increase in DOM-T0 formulae (Fig. 3c–d and g–h). In the ice, this increase was mainly in aromatics without heteroatoms related to oxy-aromatic phytochemicals, while in the snow this included nitrogen- and sulfur-containing compounds, partly related to oxy-aromatic phytochemicals (Fig. 3c–d and g–h, Tables 1, S4 and S5). Aromatic phytochemicals, have been related to cell signaling and stress resilience in microalgae³⁶. Thus, their production in the blooms is expected and probably triggered by the extreme glacier conditions⁴. Accumulation of aromatic phytochemicals in the ice experiments indicates that they are not available for heterotrophic degradation (Fig. 3c), unlike the rapidly degraded sulfur-containing phytochemicals in the snow (Fig. 3f). Heterotrophic alteration removed labile DOM in both algal experiments. This degradation is expected to increase the abundance of refractory carboxyl-rich alicyclic compounds (CRAM)⁴⁵. However, CRAM increased mainly in the snow, while in the ice some were heterotrophically degraded (Fig. 3a), suggesting microbial degradation of more refractory DOM in the ice habitat. Snow-algae blooms are earlier in the season and glacier ice-algae blooms peak later⁴. Our findings indicate that snow-algae provide bioavailable nitrogen-, sulfur- and phosphorus-bearing compounds that are quickly cycled by heterotrophs after snow melts. Although our experimental setup does not allow us to determine the origin of these compounds, the high abundance in our snow-algae photoproduced DOM (Fig. S8d) indicates that autotrophic bacteria or algae produce some of these bioavailable compounds (Fig. 3f). The 16S rRNA sequencing data (Fig. 3) confirmed a higher relative abundance of bacteria found in freshwater systems under the influence of ice melting and involved in iron cycling⁴⁶ (ca. 23% of Ferruginibacter) and the ITS2 rRNA data (Fig. S11) revealed the abundance of cold adapted fungi (Cryolevonia and Phenoliferia) in the snow-algae habitat, suggesting a better adaptation to degrade more bioavailable snow-algae DOM. The ice-algae habitat was dominated by chemoautotrophs and ice-adapted genera (ca. 40% Hymenobacter and 22% Parafrigobacterium), which were present in snow at lower proportions (with ca. 26 and 12%, respectively). The high relative abundance of these microorganisms on the ice habitat suggest that they may be more versatile to degrade less bioavailable ice-algae DOM. Fungi such as Microbotryomycetes present higher relative abundance in the snow (Fig. S11). Their high relative abundance in our and other supraglacial habitats⁴⁷ indicates that they are important OM decomposers⁴⁸ and key players in algae-DOM alteration. Bacteria can also channel OM from primary production and in our samples sulfur oxidizers like Acidophilum⁴⁹ may be responsible for the particular low abundance of sulfur compounds in the ice-algae habitat (Fig. 3). Moreover, arysulfatases make sulfate available to Chlamydomonas during periods of sulfate deficiency²⁷. These enzymes may also occur in snow-algae potentially explaining sulfur degradation in the snow experiments (Fig. 3f), yet whether these change with environmental factors and bloom dynamics remains unknown. Cyanobacteria, which are dominant in cryoconite hole habitats, can also contribute to the overall surface ice OM pool when they become dispersed on the ice surfaces. However, in our initial glacier ice experimental sample cyanobacteria were present at below 0.7% in relative abundance at the genus level (included in “others” in Fig. 3), while in the snow sample they were fully absent. The most abundant cyanobacteria in our initial ice sample were the genus Phormidesmis (0.58%) and Pseudanabaena (0.06%).

Refractory OM, represented by formulae always present and not degraded with time, exhibited higher aromaticity in ice-algae compared to snow-algae habitats (Fig. S12). In ice, refractory OM also included unsaturated aliphatic and highly unsaturated primarily without heteroatoms, while in snow, they contained nitrogen and phosphorus (Fig. S12, Tables 1, S4 and S5). These distinctive snow- and ice-algae habitat signals can be either ice-locked during winter or released into streams in summer (Fig. 4). Aromatics associated to oxy-aromatic phytochemicals (43%) or BC (12%), were unique to ice habitats (Table 1 and Fig. S12), indicating that they play a bigger role in albedo reduction than in snow-algae habitats. This was, not only due to the role of glacial ice-algae phenolics and BC as light absorbers^18,50 but due to their resistance to degradation and thus preferential accumulation. However, a detailed assessment of the role of pigmented glacier ice-algae blooms and BC on GrIS albedo reduction and to disentangle if these molecular signals are from atmospheric deposited BC, glacier ice-algae aromatic phytochemicals or both is still to be studied.

Conceptual diagram of the influence of light and heterotrophic processes on the OM composition of glacier ice- and snow-algae dominated habitats. As melting progresses, solar radiation refills > 50% of the DOM pool composition by stimulating autotrophic photoproduction and likely abiotic release of OM from POM, characterized in the red snow-algae habitat by higher diversity of heteroatoms compared to glacier ice-algae habitats. The red snow-algae habitat provides N-, S-, P-bearing compounds of unsaturated aliphatics, highly unsaturated and aromatics that are largely degraded by heterotrophic microorganisms that keep these compounds in low abundance after snow melts and bare-ice becomes predominant. Heterotrophic degradation removes 35 and 50% of the DOM found in red snow-algae and glacier ice-algae blooms, while solar radiation only 16 and 6%, respectively. The OM composition was strongly linked to the algae-dominating (glacier ice-algae or snow-algae) and influenced its degradation, resulting in an accumulation of resistant light-absorbing aromatics in glacier ice algae habitats (related to oxy-aromatic phytochemicals, black carbon or both) that can lead to more GrIS darkening (Table 1, Fig. S12).