Data and samples were collected over several summer field-campaigns 2008–2013 due to the remote location and complicated logistics for diver-based work; most in situ studies were conducted in August 2012 and 2013 on the same specific tufa column in Ikka Fjord, SW Greenland (61°11.528′N 48°01′662′W). The Ikka Fjord is long and narrow with a maximum depth of 30 m oriented from NE-SW in a glacial valley bordered by steep, ~500 m high Precambrian mountains mainly composed of carbonatite, syenite, and gneiss. Tufa columns are only found in the shallow inner part of the fjord, “Ikka Bund,” that is protected by a geological land barrier forming a narrow entrance to the inner fjord (Figure 1); this barrier renders column formation unaffected by icebergs entering the outer fjord. A tidal range of ~3 m provides a large exchange of water, where seawater enters the fjord and partly mixes with the freshwater from the submarine springs and the many surrounding terrestrial streams leading into the inner fjord. All underwater measurements were performed by certified commercial scuba divers investigating a single ikaite column during ~25 dives with additional dives for sampling and documentation. The investigated column was branching off from a bigger column that belongs to a range of columns on a ridge protruding from one of the most massive ikaite formations “The Atoll,” oriented toward NNE (Figures 1, 2).

The upper photic part of the column had its apex situated at a depth of ~7 m at mean sea level. The column apex exhibited persistent and rapid formation of fresh ikaite crystals. A continuous flow of meteoric water seeping from its porous ikaite matrix created visible clouding from spring water mixing with the surrounding seawater, and rapid re-crystallization in cuts from harvested samples was observed. A small flat platform shaped the column apex, measuring ~45 × 25 cm, with three minor, 10–15 cm wide and ~5 cm high protruding freshly deposited ikaite formations (Figure 2). After completion of the in-situ measurements, the top of the investigated column was cut off with a saw and cross sections showed the presence of distinct green and yellowish colored zones of variable thickness in the ikaite matrix (Figure 3).

Images of the ikaite crystal matrix and associated microbes (Figures 3B,C) were obtained from the freshly harvested ikaite column top by a digital SLR camera (EOS 7D digital SLR camera, CANON Europe Ltd., Middlesex, UK) fitted on the camera port of an epifluorescence microscope with 5x and 20x objectives (Axiostar Plus FL microscope with Plan-Apochromat objectives, Carl Zeiss GmbH, Germany).

Light measurements of photon scalar irradiance (PAR; 400–700 nm) at the ikaite surface were performed in-situ with a spherical micro quantum sensor (US-SQS/A, Heinz Walz GmbH, Effeltrich, Germany) connected to an underwater microsensor meter (Underwater Meter System, Unisense A/S, Århus, Denmark). Light measurements were done in parallel with, and in close proximity to, the site of O₂ and pH profile measurements (see below). Photon irradiance (PAR) at the water surface was monitored by a universal light meter (ULM-500, Heinz Walz GmbH, Effeltrich, Germany).

Temperature loggers (iButton®–DS1922L, Maxim Integrated, San Jose, CA, USA) were placed on the ikaite column at three different depths in August 2013 recording temperature at an interval of 5 min for 7 days. The temperature loggers were placed at the top platform of the column of interest, at ~7 m mean sea level, and 1.5 m above and below this depth, Tidal data of the neighboring fjord Kangilinnguit (Grønnedal), were obtained from the Danish Meteorological Institute (www.dmi.dk) for the same period as temperature recording was performed.

The chemical conditions in the ikaite matrix were measured by detailed mapping of O₂ concentrations and pH in the outermost layers of the column apex. Profiles of O₂ and pH were obtained in situ by the use of underwater pressure-compensated Clark-type O₂ needle electrodes (OX-N, Unisense A/S; Revsbech, 1989) and pressure-compensated pH glass needle electrodes with a separate silver/silver-chloride open-ended reference macroelectrode (pH-N, Unisense A/S). The electrodes were connected to a four-channel underwater microsensor meter (Underwater Meter System, Unisense A/S, Århus, Denmark). The O₂ electrodes were linearly calibrated at in situ water temperature from signal readings in aerated seawater and in seawater made anoxic by addition of sodium dithionite. The pH electrodes were calibrated at in situ water temperature by a two-point calibration in standard pH buffers (pH 7 and 10).

The needle sensor measurements were carried out handheld by a diver, who could carefully insert the needle sensors into the porous ikaite matrix in steps of 5 mm, following markings on the needle tips (Figure 3). Profiling was done from the column surface to a depth of ~15–25 mm inside the column. Due to the fragility of the pH-electrodes, most pH measurements only reached a maximum profile depth of 15 mm, as the sensor glass tip was frequently damaged when attempting to go deeper. No repetitive measurements were done in the same sensor “hole,” and all individual measurements were thus done in separate positions on the apex of the ikaite column.

Concentration profiles of O₂ were measured at varying orientations on the tufa column relative to the incident sun light, to map any differences in O₂ in relation to changes in the prevailing incident light conditions. The measured concentration profiles thus represent O₂ levels at different light environments from the surface of the column to a depth of 20–30 mm within the matrix. All measuring points were sorted according to their orientation on the column, yielding profiles at different sun angles relative to the column apex, i.e., east, south, west, north, and exposed (top). The definition of N, S, E, and W was done by the divers compass. Measurements within an orientation zone were obtained in a line going down the column with a few cm between measurements on the side of the column, whereas top measurements were obtained more randomly on the top of the column. For each profile measurement, the microsensor was inserted normal to the ikaite surface. The measured O₂ profiles were further separated according to time of day; when the sun was at its highest around noon, and in darkness 1–3 h after sunset, and between 2 years (August 2012 and 2013).

Diel in-situ dynamics of O₂ concentration and photon scalar irradiance were measured during two 24 h cycles; one diel cycle with 100% cloud cover (completely overcast), and one diel cycle with 0% cloud cover (blue sky). During these measurements, the tip of the O₂ needle sensor was fixed at a depth of ~15 mm inside the ikaite matrix, yielding a relatively high O₂ reading at the beginning of a cycle. The photon scalar irradiance sensor was positioned on the ikaite surface in close proximity to the insertion point of the O₂ electrode. About 4–5% of the incident irradiance remained at a depth of 15 mm, as calculated from the ikaite surface light measurements obtained in this study and light attenuation coefficients of ikaite measured in the laboratory (Trampe et al., unpublished data).

The effective quantum yield of photochemical energy conversion via photosystem (PS)II activity in the ikaite matrix was assessed in-situ under natural light conditions by a submersible, diver-operated pulse-amplitude modulated (PAM) fluorometer (Diving-PAM, Heinz Walz GmbH, Effeltrich, Germany; Schreiber et al., 1997; Kühl et al., 2001) in August 2013. The fluorometer was equipped with a fiber-optic cable (Ø 5.5 mm; DIVING-F, Heinz Walz GmbH, Effeltrich, Germany) that was directed toward the ikaite crystal matrix harboring surface-associated phototrophs for monitoring chlorophyll fluorescence yields with a weak non-actinic measuring light. From measurements of the fluorescence yield before and during a strong saturating light pulse, the effective quantum yield of PSII (ϕ_PSII), equivalent to the fraction of energy that is photochemically converted in PSII, was determined as (Genty et al., 1989): ϕPSII=(Fm′−F′)/Fm′ where F′ is the fluorescence yield measured under ambient light conditions, and Fm′ is the maximal fluorescence yield measured during the saturation pulse. Subsequently, the relative photosynthetic electron transport rate (rETR) was calculated as: rETR=ϕPSII × E(PAR)×0.5 where E(PAR) is the incident photon irradiance of ambient photosynthetically active radiation (400–700 nm) and the constant 0.5 is from the assumption that the photon energy absorbed is equally divided between PSII and PSI. The ETR rates are relative since we did not measure the absorption cross-section of PSII.

It was possible to automate measurements of rETR under a series of increasing levels of actinic irradiance as provided via the fiber-optic light cable. With a short (10 s) exposure to each irradiance level before a measurement pulse, so-called rapid light curves (RLC) of PSII activity vs. photon irradiance (Ralph and Gademann, 2005) were recorded, representing a snapshot of the current light acclimation status of the monitored photosynthetic community in the ikaite matrix. In contrast, measuring pulses recorded with longer incubation periods (>5–10 min) at each irradiance yield steady state light curves (SLC), where the photosynthetic apparatus is allowed to acclimate to increasing irradiance. In this study, we focused on RLC measurements to document the in situ light acclimation of the phototrophs in the ikaite matrix. The light saturation coefficient, i. e., the irradiance at onset of light saturation of photosynthesis was calculated as E_k = rETR_max∕α, where rETR_max is the maximum activity and α is the initial slope of the rETR vs. irradiance curve; both parameters were obtained by curve fitting of rETR vs. irradiance curves with an exponential function (Webb et al., 1974) using a non-linear Levenberg-Marquardt fitting algorithm in OriginPro 2015 (OriginLab Corporation, Northampton, MA, USA).

Before in situ measurements of variable chlorophyll fluorescence, dense green patches inside the ikaite column apex were revealed by carefully scraping off the outermost ~2–6 mm of white ikaite with a sharp knife (Figures 3E,F). The fiber-optic cable from the Diving-PAM was then placed by the diver a few mm above the green patch (Supplementary Figure 1), and ϕ_PSII in the exposed pigmented ikaite matrix was measured as a function of nine consecutive increasing photon irradiances of actinic light, following a custom pre-programmed and calibrated (PAR) list of photon irradiance levels ranging from 0 to 258 μmol photons m⁻² s⁻¹.

Ikaite samples for molecular analysis of the microbial diversity were retrieved by collecting ikaite material from distinct colored zones in a larger cross section taken from the column apex. The microbial diversity in the samples was analyzed by pyrosequencing of 16S rRNA genes, and sequence analysis was carried out as previously described (Glaring et al., 2015). Briefly, DNA was extracted from ikaite material using a MO-BIO Powersoil DNA Extraction Kit (MO-BIO Laboratories, Carlsbad, CA, USA). A 466 bp fragment of the 16S rRNA gene from bacteria and archaea covering the V3 and V4 hypervariable regions was PCR amplified using the primers 341F (5′-CCTAYGGGRBGCASCAG-3′) and 806R (5′-GGACTACNNGGGTATCTAAT-3′). Adapters and tags for pyrosequencing were added to the gel-purified PCR products in a second PCR using the same primers carrying sequencing adapters and tags for multiplexing. The amplified fragments were sequenced on a Genome Sequencer FLX pyrosequencing system (454 Life Sciences, Roche, Branford, CT, USA).

Trimming and quality-filtering of 16S rRNA gene sequences was performed using the software suite Biopieces (www.biopieces.org). OTUs were clustered at 97% sequence identity using USEARCH (Edgar, 2010) and chimeric OTUs were removed using the UCHIME function by comparison to the chimera-free RDP Gold database (http://drive5.com/uchime/rdp_gold.fa). OTUs containing only one sequence (singletons) were discarded. Phylogenetic analysis was performed using the Quantitative Insights into Microbial Ecology (QIIME) pipeline version 1.9.1 (www.qiime.org) (Caporaso et al., 2010) by comparison to the Greengenes database (version 13_08; http://greengenes.secondgenome.com). In this study, we mainly focused on analyzing the microbial phototrophs (cyanobacteria and microalgal plastids) in ikaite. The complete dataset of 92,266 quality-filtered sequences in QIIME-compatible format is available from the MG-RAST server (http://metagenomics.anl.gov; ID 4653755.3).

Temperature measurements recorded at three different depths in 2013, showed a dynamic temperature variation at the tufa column surface that correlated with tidal changes (Figure 4A). Under rising water level at incoming tide, temperature decreased due to cold oceanic seawater entering and mixing with the warmer fjord water. This effect was reversed during outgoing tides, where relatively warmer surface waters reached the measuring location. During the 7 days of measurements, the recorded temperature mean at the column apex was 7.5°C with a maximum span of 4.4°C and maximum and minimum temperatures of 9.4 and 5.0°C, respectively. The average temperature variation during a tidal cycle was 2°C, with a maximum and minimum variation of 3.6 and 0.2°C, respectively. The tidal data did not take other factors such as changing wind and currents into account (Figure 4B).

All 32 pH profile measurements displayed similar characteristics despite being obtained from various locations on the column relative to sun angle and orientation (facing south, north, east, west and up) at various times of day, and thus shifting irradiance levels (data not shown). Thus, all pH measurements were pooled and plotted as the mean pH (Figure 5). The pH profiles revealed a strong increase from pH 8.25 in the seawater 2 mm above the ikaite surface to pH 9.1 at the porous ikaite surface, followed by a more gradual increase in pH vs. depth reaching pH 9.9 about 15 mm inside the ikaite matrix. It was very difficult to go deeper inside the ikaite crystal matrix without fracturing the delicate tip of the pH sensor, and the shown data point reaching pH 10.4 at a depth of 20 mm thus represents a single measurement.

The O₂ concentration profiles measured in the outermost 20–30 mm of the ikaite column apex displayed a great variability, not only from the column surface to deeper layers of the column, but also at different orientations on the column toward incident sunlight, as well as time of day (Figure 6). Although measurements were obtained at the same time of year, the ambient light conditions were rather different between 2012 and 2013. In 2012, dense fog often covered the fjord until midday, whereas foggy conditions were absent during the 2013 measuring campaign thus exposing the columns to more light prior to measurements. During daylight, a photon irradiance of ~100–300 μmol photons m⁻² s⁻¹ was measured at the column surface just before measurements in both years.

In 2012, O₂ levels increased with depth in the ikaite matrix due to photosynthesis driving O₂ concentration up to ~130% air saturation at the top and ~120% air saturation on the south side of the column apex at a depth of 20 mm inside the ikaite matrix (Figure 6A). In contrast, profiles of O₂ concentration on the northern side of the column apex in 2012 showed a relatively constant decrease from 100% at the column surface to 44% air saturation at a depth of 30 mm inside the ikaite matrix, with measurements at 20 mm showing an O₂ concentration close to 100% (Figure 6A). Measurements 1–3 h after sunset in 2012 (Figure 6B) showed a decrease in O₂ on both the north and south side of the column apex, going from 100% at surface level to 60% air saturation at a depth of 25 mm on the south side, with a dramatic drop between 20 and 25 mm inside the column.

The O₂ concentration profiles measured during the day in 2013 at the top of the ikaite column, facing directly toward the water surface, showed a slightly lower initial O₂ concentration of 133% air saturation at the ikaite surface and a relatively fast decrease in concentration, going from 115% at 5 mm depth to 27% at 10 mm, where after it decreased constantly toward 0.4% air saturation at 20 mm depth below the ikaite surface. South-oriented measurements displayed an almost linear increase in O₂ concentration from 175% at the ikaite surface to 330% air saturation at a depth of 15 mm below the surface before decreasing again to 202% at 20 mm depth. Measurements performed on the east-facing part of the ikaite column showed a similar trend albeit with slight lower O₂ levels, going from an average of 160% saturation at the surface to 202% at a depth of 10 mm, followed by a stronger decrease to 69% air saturation at a depth of 20 mm. From measurements on the northern side during the day in 2013, we observed an almost linear decrease in O₂ concentration with depth and the ikaite matrix exhibited anoxic conditions at 20 mm. On the west facing part of the column apex, we measured an even steeper decrease in O₂ concentration, from 175% air saturation on the column surface to 13% at a depth of 10 mm (Figure 6C). From about 10–20 mm inside the column surface, we measured low O₂ concentrations of < 27% air saturation. Night values in 2013 showed a relatively stable O₂ environment with depth inside the column around a level of 150–200% air saturation, whereas measurements performed on the top decreased to almost anoxic conditions 15–20 mm inside the column. O₂ levels of ~100% air saturation were measured in the surrounding seawater (data not shown).

In 2013, changes in O₂ concentration were recorded at a depth of 15 mm inside the ikaite matrix of the column apex with accompanying irradiance measurements on the ikaite surface during two distinct 24 h cycles; one with clear skies and one with 100% cloud cover (Figure 7). Under a clear sky and maximal sun exposure, high photosynthetic O₂ production was observed driving O₂ concentration in the ikaite to almost 300% air saturation and with photon irradiance levels (PAR, 400–700 nm) reaching 20–25 μmol photons m⁻² s⁻¹ at a depth of 15 mm inside the column during midday. With decreasing irradiance, we observed a steep drop in O₂ concentration rapidly reaching ~0% air saturation just after sunset. In the morning, O₂ levels in the ikaite matrix increased again with a short delay after sunrise (Figure 7A). The steep drop and rise in light levels seen around sunset and sunrise, corresponded to times where the sun disappeared behind the steep mountains in the west and appeared again in the east, respectively.

During an overcast day, photon irradiance in the ikaite matrix, did not exceed 3 μmol photons m⁻² s⁻¹ at a depth of 15 mm inside the column on the first day. However, we still observed a marked increase in O₂ concentration during daytime reaching levels close to 200% air saturation. During this 24 h measuring cycle, we did not see the development of anoxia in the ikaite matrix during the night. After sunrise on day 2 (Figure 7B), there was less cloud formation and we observed the beginning of a similar trend in higher irradiance and O₂ levels as observed in Figure 3A.

Scraping away the outermost ~15 mm ikaite matrix to expose the underlying pigmented zones, gas bubbles, presumably originating from intense photosynthesis, were sometime released from the ikaite matrix. In August 2013, in situ variable chlorophyll fluorescence measurements showed an active photosynthetic community in the ikaite matrix that responded to increasing irradiance with a saturation response typical for microalgae and cyanobacteria (Figure 8). The effective PSII quantum yield exhibited high initial values indicating an active community of microbial phototrophs within the ikaite matrix. We observed a relatively low light saturation coefficient, E_k, at a photon irradiance of ~24 μmol photons m⁻² s⁻¹, with rETR_max at 146 μmol photons m⁻² s⁻¹.

The diversity of phototrophs in 7 ikaite samples was analyzed by pyrosequencing of 16S rRNA genes and revealed operational taxonomic units (OTUs) that could be assigned to cyanobacteria and microalgal plastids. At the phylum level, cyanobacterial OTUs represented about 0.2–33%, and microalgal plastids about 0.1–27% of all sequences from the ikaite samples, with microalgal plastids dominating in the outer yellowish ikaite zones and cyanobacteria dominating in the green zones found deeper in the ikaite, with only little overlap between the two zones (Figure 9A). Microscopic observation of fresh ikaite samples as well as macroscopic observations of ikaite cross-sections showed that ikaite crystals and cells in the pigmented zones were commonly embedded in exopolymers (Supplementary Video 1).

A more detailed analysis of the distribution of different phototrophs across differently colored zones of ikaite revealed distinct community changes in composition of oxygenic phototrophs (Figure 9B, Table 1). The deeper zone, 1–2 cm inside the ikaite matrix was comprised of a community of unicellular cyanobacteria belonging to the orders Synechococcales and Chroococcales and the filamentous orders Nostocales, Oscillatoriales, and Pseudanabaenales. In the microalgal zone 0–1 cm inside the ikaite, we identified plastids of microalgae belonging to the phylum Chlorophyta, and orders Cryptophyta, Haptophyceae, and Stramenopiles. The green layers were dominated by cyanobacteria, whereas the white-yellowish, yellow, yellow-green and white-green layers were dominated by plastid-bearing organisms. Nostocales predominated in one of the two sampled green zones, and Chroococcales in the other. Cryptophyta dominated in the white-yellowish layer, and Stramenopiles dominated in the two sampled yellow and yellow-green layers.

The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.