Characterization of the glacier algal bloom proliferating on surface ice of Morteratsch glacier, Switzerland, was undertaken over a 3-day period in late August 2022 (21st August 2022–23rd August 2022). Surface ice samples were collected across three main sites (one site sampled per day) distributed from the terminus of the glacier at 2,100 m up to 2,300 m elevation above sea level (Figure 1), hereafter referred to as lower (46.418N, 9.933E; Figure 2D), middle (46.415N, 9.934E; Figure 2C), and upper (46.414N, 9.934E; Figure 2B) sampling locations. At each location, n = 10 randomly selected areas of supraglacial surface ice were sampled at the start (10:00–12:00), middle (14:00–16:00), and end (17:00–19:00) of the day to capture potential diurnal dynamism in the biogeochemistry associated with blooms, with n = 30 samples per lower/middle/upper sampling location. All sampling was undertaken during clear, sunny-sky conditions (Figure 2). For each individual ice sample a 20 × 20 × 2 cm depth patch of surface ice was sampled using an ice saw and metal scoop, pre-sterilised with 10% ethanol, directly into a sterile Whirl-Pak bag using standard sampling techniques (Williamson et al., 2018, 2020). Before collection of each ice sample, an image with a scale was taken, and the slope, aspect, elevation with respect to sea level, and GPS coordinates were recorded (directly adjacent to sampling patch to avoid contamination) using the Phyphox app on an Apple iPhone 14 (Staacks et al., 2018).

Ice samples were melted in the dark over 1–2 days at 4°C and homogenized prior to further processing. An initial 1.5 ml of each homogenized sample was placed into an individual 1.5 ml Eppendorf tube and fixed using Lugol's solution for subsequent cell counts and biovolume estimates at the University of Bristol, UK. A known volume of sample was then filtered through a pre-combusted (450°C for 5 h) 13-mm diameter GF/A filter (1.6 μm retention; Cytiva Whatman, Maidstone, UK), which was placed into an individual 1.5 ml Eppendorf tube and maintained at −20°C for subsequent cellular stoichiometry determination. Filtrates were collected into pre-acid-washed high density polyethylene (HDPE) bottles and maintained at −20°C for subsequent inorganic (ammonium, nitrate, nitrite, and phosphate) and organic (total organic carbon and nitrogen) aqueous geochemistry quantification. Samples were then driven to the University of Bristol, UK, and maintained within portable fridges at 4°C or freezers at −20°C.

To examine potential bottom-up control of glacier algal blooms at Morteratsch glacier, a nutrient spiking experiment was conducted ex situ with glacier algal communities sampled from surface ice. Fifteen samples were collected as above from lower Morteratsch glacier (46.418N, 9.933E) on 26th August 2022 and transported in the dark at 1°C within a portable fridge to Photon Systems Instruments, Czech Republic, where they were melted over 2 days as above. Approximately 7 days after sample collection, an incubation experiment was established with melted surface ice samples and their constituent glacier algal communities. Melted samples were initially pooled and then 50 ml distributed into n = 5 replicates across each of five treatments: control (no nutrient addition), +ammonium (+NH4+, 10 μmol L⁻¹), +nitrate (+NO3-, 10 μmol L⁻¹), +phosphate (+PO43−, 1 μmol L⁻¹), and +ALL (+NH4+ and +NO3- at 10 μmol L⁻¹ and +PO43− at 2 μmol L⁻¹). NH4+, NO3-, and PO43− spikes were added as NH4+-N, NO₃-N, and PO43−-P standards, respectively (Thermo Fisher Scientific, UK). Treatments were chosen to provide ~10 times the reported ambient inorganic nitrogen and phosphorus concentrations in supraglacial ice (Wadham et al., 2016; Smith et al., 2017) and maintain a 10:1 nitrogen:phosphorus ratio within our “+ALL” treatment (McCutcheon et al., 2021). Incubations were conducted within 50 ml Corning culture flasks with vented caps (Corning, New York, USA) and maintained at 4°C under 500 μmol photons m⁻² s⁻¹ of light on a 16:8 Light:Dark cycle provided by warm white fluorescent tubes with no UV provision. Destructive sampling of n = 5 replicate incubations per treatment proceeded at 0, 24, 72, and 120 h for the determination of cellular abundance/biovolume, algal photophysiology, and cellular stoichiometry as described below.

The cellular abundance (cells ml⁻¹) and average biovolume per species (μm³ cell⁻¹) were assessed for all samples collected in the field and used for incubation experiments. Cellular abundance was measured by counting cells on a modified Fuchs Rosenthal Hemocytometer (0.2 mm by 1/16 mm²; Hawksley, Lancing, United Kingdom) using a bright field Olympus BX41 microscope (Germany). Images of each sample were taken at 10 × and 40 × magnification with a MicroPublisher 6 CCD camera attachment (Teledyne Photometrics, USA), from which cells were counted and the width and radius measured for each sample using ImageJ software (Schneider et al., 2012). Final cellular biovolumes were calculated assuming species to be an average cylinder (Hillebrand et al., 1999). Power analysis was applied to determine the number of cells to measure to achieve a margin of error < 250 μm³ per species/sample (Jones et al., 2003). The total algal biovolume per sample (μm³ ml⁻¹) was calculated by multiplying species' cellular abundances per sample (cells ml⁻¹) by their corresponding mean biovolumes (μm³ cell⁻¹).

Rapid light response curves (RLCs; Perkins et al., 2006) were performed using Pulse Amplitude Modulation (PAM) fluorometry to characterize the photophysiological states of glacier algal communities as sampled from Morteratsch glacier and throughout nutrient spiking experiments. Measurements were conducted on 3 ml subsamples using a Walz Water-PAM fluorometer with attached red-light emitter/detector cuvette system and stirrer (Walz GmBH, Germany). Each sample was dark adapted for a minimum of 5 min prior to RLC measurement. RLCs consisted of nine sequential light steps of 20 s duration ranging in irradiance from 0 to 3,000 μmol photons m⁻² s⁻¹ of Photosynthetically Active Radiation (PAR). The maximum quantum efficiency (F_v/F_m) was calculated from minimum (F₀) and maximum (F_m) fluorescence yields measured in the dark-adapted state during the initial RLC step of 20 s darkness as Fv/Fm = Fm – Fo/Fm (Consalvey et al., 2005). Electron transport through photosystem II (PSII) was calculated across subsequent light steps in relative units (rETR) assuming an equal division of light between PSI and PSII as rETR = Y (PSII) × PAR × 0.5 (Consalvey et al., 2005). Analysis of all RLC data (rETR vs. PAR) followed Eilers and Peeters (1988) for calculation of the relative maximum electron transport rate (rETRmax).

Aqueous concentrations of NH4+, NO3+, NO2-, and PO43− were derived for all melted ice samples spectrophotometrically using a Gallery Plus Discrete Photometric Analyser (Thermo Fisher Scientific, UK). The limit of detection (LoD) for all nutrients was determined by the mean concentration plus three times the standard deviation of calibration blanks (n = 3). LoDs were 0.04 μmol L⁻¹ (NH4+), 0.02 μmol L⁻¹ (NO2-), 0.15 μmol L⁻¹ −(-NO3-), and 0.15 μmol L⁻¹ (PO43−). Precisions were ±2.0% (NH4+), ±1.0% (NO₂), ±1.3% (NO3-), and ±2.0% (PO43−) as determined by comparison with diluted 71.43 mmol L⁻¹ NH4+-N, NO2--N, and NO3--N and diluted 32.29 mmol L⁻¹ PO43−-P certified stock standards to a concentration of 3.6 μmol L⁻¹ (NH4+ and NO2-), 2.9 μmol L⁻¹ (NO3-), and 6.1 μmol L⁻¹ (PO43−) (Sigma TraceCERT^®).

Filtrate was also analyzed for total organic carbon (TOC) and total nitrogen (TN) concentrations via a TOC/TN Organic Carbon Analyser (Shimadzu, UK). Non-purgeable organic carbon (NPOC) was measured after the acidification of samples with 9N sulfuric acid and catalytic combustion at 720°C as CO₂. TN was measured as NO by chemiluminescence. The LoD was 18.3 μmol L⁻¹ (TOC) and 30 μmol L⁻¹ (TN), precision was ±2.0% as determined by comparison with diluted 41.7 mmol L⁻¹ TOC certified stock standards to a concentration of 8.3 mmol L⁻¹ and ±1.7% by comparison with diluted 14.3 mmol L⁻¹ TN certified stock standards to a concentration of 3.6 mmol L⁻¹ (Sigma TraceCERT^®).

Glacier algal cellular carbon (C) and nitrogen (N) contents were determined for n = 5 samples from Morteratsch glacier for each location and time point. For this, GF/A filters were freeze-dried for 24 h to remove water, re-weighed and wrapped in individual 16 mm tin disks prior to elemental analysis using a Vario PYRO cube^® (Elementar, Stockport, UK). The detection limits of elemental concentrations were 0.001% for both elements measured, and the coefficient of variation (CV) for C and N according to 12 replicates of an organic analytical standard (NC Soil Standard 338 40025, cert. 341506, C = 2.31%, N = 0.23%; ThermoFisher Scientific, Bremen, Germany) were 1.76% and 1.13%, respectively. The molar content of carbon and nitrogen per sample was derived from the total recorded carbon and nitrogen area as follows:

where Ar_x is the relative atomic mass of nutrient x (e.g., carbon), x [%] is the derived percentage of nutrient x present in a processed sample, w is the total weight of the processed sample. Data are presented as molar C:N ratios.

Analysis and plotting of data were completed using R v.4.2.1 (R Core Team, 2022). Each data series collected (cell abundance, biovolume, photophysiological, aqueous geochemistry, and stoichiometry) was tested for homogeneity of variance and normality of distribution. Statistical comparisons of photophysiology, aqueous geochemistry, and stoichiometry between sample locations and time points were investigated using one-way or two-way analysis of variance (ANOVA) with post-hoc Tukey Honest Significant Difference (HSD) analysis applied to all significant ANOVA results. ANOVA and Tukey HSD were also applied to comparisons between nutrient spiking treatments. To compare cell abundances and biovolumes between sampling locations and sampling times, Kruskal–Wallis rank sum tests (for non-parametric datasets) were employed. Linear regression analysis was employed to investigate potential relationships between cell abundance or biovolume and sample site slope. Aspect of the slope was treated as circular data (Von Mises distribution) and tested for correlation with cell abundance or biovolume using the circular correlation coefficient equivalent for Pearson's test for correlation (Fisher and Lee, 1983; Rao Jammalamadaka and Ramakrishna Sarma, 1988). Data were plotted using the ggplot2 package v.3.4.4.

Glacier algal communities were composed of both cosmopolitan glacier algal species at Morteratsch during 2022, with Ancylonema alaskana showing the greatest overall cellular abundance in all sampling locations as compared with A. nordenskiöldii (85% A. alaskana relative abundance vs. 15% A. nordenskiöldii across all data; Figure 3), though a more even split between species was apparent in our lowest sampling location based on total biovolume datasets (52% A. alaskana vs. 48% A. nordenskiöldii; Figure 4). These data are consistent with the study by Di Mauro et al. (2020) who reported an overall greater cellular abundance of A. alaskana relative to A. nordenskiöldii for the 2016 bloom in the same site; though contrasting findings based on DNA recovery. At present, we hold little-to-no knowledge on the factors that determine glacier algal community composition within or between sites and seasons; a key facet of bloom ecology yet to be understood. Indeed, it is likely that we do not yet have a full understanding of the true diversity of glacier algae across the cryosphere (Remias et al., 2023).

Glacier algal cellular abundance ranged from 1.78 × 10⁴ to 8.95 × 10⁵ cells ml⁻¹ during August 2022 on Morteratsch glacier; at the higher end of abundances reported across the cryosphere during large bloom years (Stibal et al., 2017; Williamson et al., 2018; Di Mauro et al., 2020) but not substantially greater than previous records. For example, Di Mauro et al. (2020) reported an average of 2.03 × 10⁴ cells ml⁻¹ from the same site in September 2016, during which melt may have been as little as a third of that in 2022 (World Glacier Monitoring Service, 2021, 2022). On the southwestern Greenland Ice Sheet (GrIS), known to harbor extensive glacier algal blooms, maximal abundances have been reported as 8.5 × 10⁴ cells ml⁻¹ (Stibal et al., 2017), 1.6 × 10⁴ cells ml⁻¹ (Williamson et al., 2018), and 3.0 × 10⁵ cells ml⁻¹ at the margin (Yallop et al., 2012); comparable to our observations in Morteratsch. This suggests a more complex relationship between melt and glacier algal abundance and/or the presence of other controls on overall bloom carrying capacity.

Spatiotemporal heterogeneity in glacier algal abundance recorded here indicated the potential melt-water-driven redistribution of (larger) glacier algal cells to down-glacier locations and presumptive export of biomass from the glacial system at Morteratsch; the latter a putative mechanism to set upper limits on bloom carrying capacity. At the diurnal scale, cellular abundance datasets recovered consistently higher abundances earlier in the day, when the volume of meltwater on the glacier surface was at its lowest (Figure 3). As melt increased throughout the day, corresponding estimates of abundance declined, reflecting dilution of assemblages by the higher melt fraction. The timing of sampling relative to the daily hydrograph is thus important for final estimates of abundance and should be considered in future sampling efforts.

Across our study locations, the abundance of A. nordenskiöldii was significantly higher in the lower region of the glacier (Figure 3), with a lower and more heterogenous abundance at mid and upper sites evidenced by a range of 0 cells ml⁻¹ to 8.0 × 10⁴ cells ml⁻¹ across samples. While the more abundant A. alaskana (cells ml⁻¹) was more evenly distributed across the glacier (Figure 3), when cellular biovolume (μm³ cell⁻¹) was incorporated into estimates, a clear spike in total biovolume (μm³ ml⁻¹) was apparent at our lower site, highlighting the presence of larger A. alaskana and A. nordenskiöldii cells in this region of the glacier (Figure 4). Interestingly, this heterogeneity in glacier algal distribution was independent of surface ice slope or aspect. Data thus reflected our previous findings (Williamson et al., 2018), whereby true spatial trends in GrIS glacier algal blooms at the km scale were reflected only by total biovolume datasets rather than estimates of cellular abundance alone.

The presence of more and larger cells at our lowest sampling location irrespective of slope or aspect could reflect preferential growth within this region of the glacier and/or preferential transport of larger cells down glacier by meltwater, serving to concentrate assemblages at the terminus, with subsequent export to downstream environments. At larger spatial scales, the duration since snow-line retreat is a known factor controlling the progression of blooms, such that areas of glacial ice uncovered the earliest (i.e., at lower altitudes) provide more time for glacier algal blooms to develop (Tedstone et al., 2017; Williamson et al., 2018). This dynamic has been noted across the ablation zone of the GrIS (Tedstone et al., 2017; Williamson et al., 2018, 2020) as well as for valley glaciers (Takeuchi, 2013). However, sampling sites targeted here covered a relatively small altitudinal range (~100 meters of elevation gain) during a record melt year with rapid snow line retreat (GLAMOS, 2022; Cremona et al., 2023), and it is unlikely that substantial differences in exposure duration were experienced across our transect during 2022. PAM analysis did not indicate differences in community photophysiological state across our sampling locations (Figure 5), e.g., Fv/Fm remained consistently high across all locations at 0.66 ± 0.12 (Figure 5B). Additionally, aqueous geochemistry data and nutrient spiking incubations (see below) gave no indication of preferential conditions across our sampling locations.

Alternatively, larger cells (i.e., the filaments of A. nordenskiöldii and larger cells of A. alaskana) may have been preferentially transported over the weathering crust, concentrating in the lowest glacier sampling site. At the macro-scale, a clear concentration of glacier algal biomass was apparent at the glacier terminus throughout our sampling campaign (Figure 2). We also observed obvious down-glacier cell transport within meltwater flowing over the weathering crust via microscopic observations, indicating potential down-glacier concentration of cells (data not shown). Preferential movement of larger cells contrasts a previous report by Irvine-Fynn et al. (2021), which suggested that smaller cells are more likely to be transported through the weathering crust of the GrIS. However, it is likely that hydrological controls of cell distribution and weathering crust properties on steep valley glaciers differ from those of the flatter “dark zone” of the ice sheet; hence, more information is required on the meltwater-driven redistribution of glacier algal cells across the diversity of surface ice environments and its role in regulating bloom expansion vs. driving loss of cells from the system.

Characterization of the bulk-phase aqueous geochemistry (Figure 6) once again demonstrated the paradox of highly abundant glacier algal communities thriving within oligotrophic surface ice (Williamson et al., 2018, 2020). Inorganic macro-nutrient concentrations ranged 0.22 – 4.36 μmol L⁻¹ of NH4+, 0.04 – 1.38 μmol L⁻¹ NO3⁻, 0.09 – 0.29 μmol L⁻¹ NO2-, and 0.22 – 0.85 μmol L⁻¹ PO43− (Figure 6), with a shift to dominance of organic speciation apparent by our sampling dates, i.e., 142.70 – 811.36 μmol L⁻¹ dissolved organic carbon (DOC) and 0.23 – 15.34 μmol L⁻¹ dissolved organic nitrogen (DON), reflecting known phase-shifts in the dominant geochemistry associated with blooms (Holland et al., 2019). Previously, Williamson et al. (2021) argued that glacier algae themselves may be adapted to low ambient nutrient concentrations through overall lower cellular N and P requirements. Indeed, we recorded similarly high C:N ratios here across sampling locations (average 12.28 ± 0.70 across all locations; Figure 7), with elevated C:N at our lowest sampling location corresponding to the presence of larger cells of both species; likely containing more carbon-rich phenolic pigmentation, rather than reflecting nutrient limitation at the terminus. To confirm this, algal communities sampled from our lower location were spiked with a series of nutrient additions (~10 times ambient NH4+, NO3-, and PO43− or a combination of all) and incubated for up to 120 h. Even after this 5-day duration, no clear indication of inorganic nutrient limitation was apparent for our sampled assemblages (Figure 8), with no difference across treatments in the maximum quantum yield in the dark-adapted state (Fv/Fm; Figure 8B) and minimal variance in rETRmax (Figure 8C).

These findings, taken together with previous studies (McCutcheon et al., 2021; Williamson et al., 2021), indicate that certainly at the point of sampling, glacier algae do not appear to be inorganic nutrient limited within surface ice environments of both mountain glaciers (here) or larger ice sheet systems (McCutcheon et al., 2021). This may reflect a true adaptation to low nutrient conditions (Williamson et al., 2021), efficient uptake and storage of those nutrients available within the system (Barcyté et al., 2019), and/or poor characterization of the available nutrients within the thin meltwater film that glacier algae reside. To date, researchers have typically employed the “bulk” method used here of ice sampling with subsequent melt and aqueous geochemistry characterization. This potentially dilutes the true nutrient available to glacier algae within their habitat, which may be enhanced due to ice solute exclusion and/or preferential elution of solute-rich fractions when glacier ice melts (Ginot et al., 2020; Holland et al., 2022). Thus, more work on true glacier algal nutrient requirements and nutrient availability within their habitat is needed to tackle the oligotrophic-bloom-paradox that is ubiquitous across the cryosphere.