Mineral phosphorus drives glacier algal blooms on the Greenland Ice Sheet

The Greenland Ice Sheet (GrIS) has experienced increasing surface melt and runoff, coincident with the emergence of a low-albedo Dark Zone along its western margin. Albedo is modulated by ice surface structure and light-absorbing particulates including glacier algae, black carbon, and mineral dust. Glacier ice algae produce pigments that reduce albedo and have been estimated to directly account for 9–13% of surface melting in the Dark Zone, with additional indirect effects. However, the interannual variability, intensity, and spatial extent of algal blooms remain insufficiently constrained. This study investigates the controls on glacier algal growth, testing the hypothesis that phosphorus is a limiting nutrient and that mineral dust provides the nutrient source driving bloom development and associated albedo reduction.

Prior work has established that algal pigment production lowers ice albedo and accelerates melt; algae can cover large fractions of the Dark Zone and their abundance and physiology influence albedo. Light-absorbing particulates include black carbon and mineral dust, but their direct roles in albedo vary with composition and grain size. Nutrient availability has been implicated in cryospheric microbial dynamics; snow algae have been linked to mineral-derived nutrients and respond to fertilizer additions, and correlations between mineral dust loading and algal abundance have been observed. Dissolved and particulate nutrient pools (N, P) exist in supraglacial environments, with reports of available inorganic and organic nitrogen. The geochemical provenance of dust in west Greenland suggests local sources, and dust properties influence transport and albedo impacts.

• Study area and sampling: Surface snow and ice were sampled along a southwest GrIS ablation zone transect at Sites 1–5 during summer 2016 and 2017; local rocks were collected near Russell Glacier (Site 6) in 2018. Habitats sampled included clean ice (CI), high algal biomass ice (Hbio), Hbio snow, dispersed cryoconite (DCC) ice, cryoconite holes (CCH), a floating biofilm, and supraglacial stream water.
• Nutrient addition experiment: Conducted at Site 4b (2017). Hbio ice with 8.0 ± 2.1 × 10^3 cells mL^-1 was melted in the dark (24 h), then incubated 120 h in 30 mL flasks (quadruplicates) under five treatments: control; +NH4+ (10 µM); +NO3− (10 µM); +PO4^3− (10 µM); +ALL (NH4+ 10 µM, NO3− 10 µM, PO4^3− 2 µM; N:P=10:1). Photophysiology measured at 24, 72, 120 h via WaterPAM rapid light curves after 20 min dark adaptation to derive Fv/Fm and rETRmax; statistics by two-way ANOVA (treatment, time) and Tukey HSD. Cell counts assessed microscopically.
• Phosphorus speciation: Modified SEDEX sequential extraction (Steps I, III, IV, V) quantified exchangeable (Pexch), mineral (Pmin), and organic (Porg) P in particulates from Hbio ice/snow, DCC ice, and CCH.
• Fluid chemistry: Filtered (0.22 µm) melted samples analyzed for major cations by ICP-MS; phosphorus by segmented flow injection analysis or LWCC spectrophotometry; anions by ion chromatography. Replicates had <10% SD.
• Carbon and nitrogen: Elemental analyzer quantified total/organic carbon (TC/TOC) and total N (TN) on 0.7 µm-filtered particulates; TOC measured after in situ decalcification.
• Mineralogy: Powder XRD (Bruker D8, Cu Kα) with Rietveld refinement (Topas) determined mineral phases; semi-quantitative due to shallow mounts and hand grinding. Phases grouped as quartz, plagioclase, K-feldspar, pyroxenes, amphiboles, micas; hydroxylapatite assessed.
• Microbial community: DNA extracted from 26 samples; amplicon sequencing of 16S, 18S, ITS2 on MiSeq. Processing with QIIME2 (dada2; taxonomic assignment via Greengenes, SILVA, UNITE), rarefaction thresholds (16S: 5500; 18S: 15000; ITS2: 15000), NMDS (Bray-Curtis) and ANOSIM in R to assess clustering by site/habitat; barplots of relative abundances.
• SEM: Samples fixed, dehydrated, HMDS-dried, sputter-coated; imaged on Hitachi SEM to visualize microbe–mineral associations and exopolymer matrices.
• REE provenance: Particulates ashed, dissolved (HF/HNO3; HCl), analyzed by HR-ICP-MS; compared to local rock REE and literature sources; Eu/Eu* anomaly used to infer provenance.
• Particle size: Organic removal by H2O2, disaggregation, washing; particle size measured by CPS disc centrifuge.
• LAP mass loading: Field filtration through 5 µm and 0.2 µm pre-weighed filters; mass load per melted volume determined; biomass fraction approximated from TOC+TN, remainder assigned to mineral dust. All sampling metadata and counts in Supplementary Tables.

• Phosphorus limitation: After 120 h, phosphate additions (+PO4^3− and +ALL) significantly increased photophysiological performance (higher Fv/Fm and rETRmax) relative to control, +NH4+, and +NO3−, which did not differ from each other. Two-way ANOVA: for Fv/Fm—treatments F4,30=64.8, P=0.000699; time (24 vs 120 h) F4,30=28.75, P=0.00000839. For rETRmax—treatments F4,30=16.71, P=0.000000264; time F4,30=35.07, P=0.00000174. No significant differences at 24 or 72 h; response manifested at 120 h, consistent with luxury P storage and slow algal doubling time (~5.5 ± 1.7 days).
• Nitrogen was not limiting: Nitrate or ammonium additions did not improve photophysiology relative to control; dissolved inorganic N was low/below detection in some samples, though prior reports indicate available inorganic (~1 µM) and organic N (~14 µM) in Hbio ice.
• Particulate P pools: In Hbio ice particulates, Porg accounted for up to 86% of solid-phase P; Pexch <31%; Pmin ~17% comprised the remainder. Pmin positively correlated with TOC (r=0.80), TN (r=0.83), and Porg (r=0.94). As Pmin increased, particulate C:N:Porg ratios decreased toward Redfield (normalized C:N:P vs Pmin r=0.94). Measured TOC:N:Porg in Site 4 Hbio particulates ranged 690:48:1 to 2615:196:1.
• Cell abundance from C and N: TOC and TN translate to glacier algal abundances of ~1.2–5.6 × 10^4 cells mL^-1, consistent with regional counts; <1% of TOC attributable to bacteria.
• Mineralogy and P source: Dust dominated by plagioclase (41–54 wt%) and quartz (18–30 wt%), with amphiboles (4–14 wt%), pyroxenes (<10 wt%), K-feldspars (3–12 wt%), micas (1–6 wt%), and kaolinite (<3 wt%). Hydroxylapatite was detected at trace levels (<1.1 wt%), providing the likely Pmin source; highest hydroxylapatite abundance observed in the algal biofilm (1.1 wt%).
• Enhanced weathering and cations: Closer to the margin (Sites 4–5), ferromagnesian phases were depleted and meltwaters had elevated dissolved cations (Na+, Mg2+, Al3+, K+, Ca2+, Fe2+). Example at Site 4a: dissolved Fe was 69 ± 20 µg L^-1 in Hbio ice vs 36 ± 5 µg L^-1 in DCC ice and 16 ± 13 µg L^-1 in clean ice. Dissolved P was 3–4× higher in Hbio and DCC ice than in clean ice or stream water at Site 4a.
• Microbial communities: Community composition clustered by site across bacteria, fungi, and algae (ANOSIM significant). Algal assemblages were dominated by Ancylonema nordenskioeldii and Mesotaenium sp. (66–99% of algal reads). SEM showed microbial exopolymer facilitating adhesion to and trapping of mineral grains, supporting bioweathering and nutrient retention.
• Dust provenance and transport: REE patterns of particulates were homogeneous with a positive Eu anomaly, matching local sources and excluding major distal inputs (Asian/African dust with negative Eu anomaly). Grain size analysis showed 99% of dust particles <20 µm, consistent with atmospheric transport.
• Particulate loading and albedo: Hbio ice contained >30× more particulates than clean ice (394 ± 194 vs 19 ± 6 µg mL^-1). Mineral dust comprised 94.2 ± 0.5 wt% of particulate dry mass in Hbio and DCC ice. Radiative transfer modeling and in situ spectra indicate mineral dust exerts negligible first-order albedo reduction compared to pigmented glacier algae; instead, mineral dust exerts a second-order control by supplying limiting P that modulates bloom development and thus albedo.

The study directly links mineral-derived phosphorus to glacier algal bloom development on the GrIS Dark Zone. Nutrient addition experiments demonstrate phosphorus, not nitrogen, limits algal photophysiological performance. Sequential extractions and mineralogy reveal hydroxylapatite in mineral dust as the key P source, with biotic weathering by microbial communities transferring P from the mineral to organic pools. Elevated dissolved cations and reduced ferromagnesian phases near the margin suggest intensified weathering where blooms are more prolific. Microbial exopolymers facilitate dust retention, promoting local recycling and nutrient availability. REE signatures confirm dust is locally sourced and of a size readily transported atmospherically. Although mineral dust has limited direct radiative impact in this region due to its largely felsic composition and scattering properties, it indirectly controls albedo by fueling algal blooms. These findings refine the understanding of biogeochemical controls on ice darkening and melt and suggest that areas with greater P-bearing dust availability and exposure are predisposed to more intense blooms and melt, establishing a positive feedback: melting releases dust, heterotrophs mobilize nutrients, algae proliferate, and albedo declines further.

This work quantitatively connects mineral phosphorus in dust—specifically hydroxylapatite—to glacier algal bloom development on the Greenland Ice Sheet, establishing phosphorus as the limiting nutrient and mineral dust as a second-order control on albedo and melt via nutrient supply. The microbial community enhances mineral bioweathering and retains nutrients at the ice surface, reinforcing a positive feedback between algal growth and melting. Dust is dominantly local and fine-grained, enabling widespread delivery. Incorporating mineral nutrient availability into predictive models of GrIS albedo and melt will improve projections of biological darkening and sea-level contributions. Future research should constrain bloom initiation timing relative to snowpack retreat, quantify spatial variability of P-bearing dust and its mobilization, resolve the potential optical effects of mineral–algal aggregates (e.g., lensing), and expand geographically comprehensive datasets to parameterize nutrient–albedo feedbacks across the ice sheet.

• Spatial and temporal scope: Sampling was limited to selected sites in southwest GrIS during parts of the 2016 and 2017 melt seasons; interannual and broader spatial variability may not be fully captured.
• Mineralogical quantification: XRD refinements were semi-quantitative due to shallow mounts and hand grinding, introducing uncertainties in phase proportions, especially for minor phases like hydroxylapatite.
• Experimental timeframe: The 120 h incubation captured photophysiological responses but not biomass increases, given slow algal doubling times; longer-term responses remain unmeasured.
• Generalizability of radiative effects: The negligible direct albedo effect of mineral dust pertains to the studied dust composition and grain-size distributions; other regions with different dust properties may experience stronger mineral-driven albedo reductions.
• Unresolved processes: Potential optical interactions (e.g., lensing by mineral grains) and detailed mechanisms of nutrient recycling and retention were suggested but not quantified. Factors controlling bloom timing, extent, and intensity remain knowledge gaps.