Introduction

The Arctic Ocean is undergoing rapid change. Summer areal sea-ice extent has been steadily decreasing, reaching a low of 3.41 million km2 in 2012, equivalent to 48.5% less than the mean for the period 1979–20001. It is predicted that the Arctic will have ice-free Septembers before 2050, and consequently it is expected that the receding and thinning permanent sea-ice cap will be replaced by seasonal annual ice2,3. Several biogeochemical processes are coupled to these shifts, including increased surface water productivity because of less light limitation4; increased CO2 drawdown from enhanced productivity in some regions5 and increased air-sea gas exchange, leading to increased acidification6; and changes in hydrodynamics because of the removal of a permanent ice cap7. As the Arctic shifts to a system dominated by annual ice, the complete seasonal melt will lead to intensified seasonal freshening of Arctic surface waters in ice melt zones. The impact on Arctic Ocean biogeochemistry remains unclear.

Although the Arctic represents only 3% (14.06 million km2) of the global sea surface area, it accounts for 10% of the global ocean carbon uptake8,9. A major concern for future predictions is the impact these changes in sea ice extent will have on the Arctic as a sink for rising atmospheric CO2 levels. The influence of intensifying annual ice melt processes on carbonate system parameters is not well understood.

Atmospheric CO2 uptake is impacted by processes altering the differences in partial pressure of CO2 (pCO2) across the air-sea boundary. The pCO2 in surface waters is dependent on a cascade of carbonate system reactions and the resulting buffering capacity of the waters. Carbonate system variables including pH, dissolved inorganic carbon (DIC), and total alkalinity (AT) are used in combination to define the CO2 chemistry of a water parcel. To isolate bicarbonate and carbonate contributions, most carbonate system studies use well-established ratios and parameterizations to correct for other constituents that may add to AT10. Boron in the form of borate (orthoborate) ions (B(OH)4-) is the third greatest contributor to AT in marine systems after bicarbonate and carbonate.

Boron (B) is ubiquitous in the environment, occurring naturally in more than 80 minerals. It is a major element in seawater at concentrations in the mg kg−1 range. Because of the low dissociation constant of B in seawater, the majority of marine B is found in the form of boric acid, B(OH)3, although approximately 20% of this is ionized to borate. The exchange between boric acid and borate is described by its equilibrium reaction (1), which is used to determine the contribution of borate to seawater alkalinity (the capacity of water to resist a change in pH or acidification):

$${{{{{{\rm{B}}}}}}({{{{{\rm{OH}}}}}})}_{3}+{{{{{{\rm{H}}}}}}}_{2}{{{{{\rm{O}}}}}}\rightleftharpoons {{{{{{\rm{H}}}}}}}^{+}+{{{{{{\rm{B}}}}}}({{{{{\rm{OH}}}}}})}_{4}^{-}$$
(1)

In seawater B is a conservative element, and the boron to salinity ratio (B/S) has been shown to be uniformly consistent across ocean waters (North Atlantic and North Pacific) at salinities of 33–36 g kg−1 (0.1336 ± 0.0005 mg kg−1 ‰−1)11. The B/S ratio is known to shift in some coastal regions including the Baltic Sea12, and in low salinity waters where variations in freshwater endmember B concentrations may become important13. As the Arctic receives 11% of total riverine inputs to the global ocean but represents only 1.4% of the ocean volume14, it too undergoes considerable freshening. Olafsson et al.15 reported a B/S ratio of 0.1324 ± 0.0008 mg kg−1 ‰−1 in the eastern Arctic in water having salinities as low as 30 g kg−1, and found that the B/S ratio, though slightly lower, was reasonably consistent with that of open ocean values. However, there are additional sources of freshwater from sea ice melt that contribute to Arctic freshening, and the B/S ratio in these sources is highly uncertain. There is little known about the B/S ratio in low salinity (<30 g kg−1) marginal ice zones (MIZs), though this is critical to our ability to quantify AT when predicting future CO2 uptake in Arctic waters. Here we present results of analysis of B/S ratios in 100 samples collected between 20 May and 12 June 2021 along the MIZ of the Bering and Chukchi Seas, in a study to evaluate the B/S ratio in these dynamic, low salinity regions.

Results and discussion

Boron in open ocean water: Samples were collected in a northerly direction along a transect across the Bering Sea (stations 1–4) into the Chukchi Sea MIZ (stations 5–9), and then in a southerly direction along the same transect (stations 10–16) (see Supplementary Table 1). Total B concentrations were measured in open ocean, ice core, and brine (the interstitial water of the sea ice matrix) samples. Figure 1a, b shows the total B concentration and B/S for surface water samples (1–2 m) across the transects. Total B ranged from 3.7 to 4.4 mg kg−1 and the B/S ratios ranged from 0.1280 to 0.1330. Notably, a region of relatively low B concentration and B/S ratio occurred in ice-free waters having substantial freshening associated with the Yukon–Kuskokwim River plume. This was associated with characteristically higher temperatures (4oC) and lower salinities (28 g kg−1) at this station. The low values of this region were also apparent in the depth profiles of total B and borate (Fig. 2a, b).

Fig. 1: Station location on the RV Sikuliaq between May 20th to June 14th, 2021.
figure 1

Ice covered stations are depicted with square markers and open water stations are represented in round markers. a Boron concentrations in surface water samples across transect in mg kg−1 and b boron to salinity ratios for all surface water stations.

Full size image
Fig. 2: Depth profiles.
figure 2

a Total boron concentrations in mg kg−1 and b alkalinity due to borate (B-Alk) in µmol kg−1.

Full size image

Lower water temperatures and lower salinities tend to increase the pKa in Eq. (1)16, thus decreasing the proportion of borate ions. The resulting acidity constants for the boric acid/borate acid-base pair ranged from 8.844 to 8.951, and borate ranged from 3–14 % of total boron. The total borate contribution to AT ranged from 11 µmol kg−1 in the lower pH waters (pH = 7.300, AT = 2297 µmol kg−1) and up to 57 µmol kg−1 in the relatively alkaline waters (pH = 8.096, AT = 2085 µmol kg−1).

Boron in Snow: Supplementary Fig. 1 identifies locations of snow and ice stations (5,6,8,9 and 12). Snow samples across all ice stations (n = 4) were consistently low in B, with concentrations that were either below the detection limit or <0.151 mg kg−1, and all had zero salinity values. These B values are within the range reported for a various snow sample types (0–0.32 mg kg−1, n = 79)17 in regions of North America that are less remote than the Arctic and may be a useful reference endmember in Arctic atmospheric wet deposition.

Boron in Ice Cores and Brine: Fig. 3 shows the total B in ice cores and brine for the five ice stations in this study. All stations had a general trend of increasing total B with depth. The B/S ratios at stations 5 and 6 remained consistent across depth and between ice and brine, but were slightly lower than those reported for open ocean samples11,16. Samples from these stations were obtained during the northward sampling along the transect. Samples from stations 8 and 9 were obtained during the southward sampling along the transect and had B/S ratios that were much less consistent. Fig. S1 shows the location of the ice sheet before and during sampling at each station. On 2 May the ice sheet was still attached to the Alaskan coast, but by 9 May it had detached and moved westward. The ice sheet underwent a flow reversal, and therefore the location of the stations in our study were subject to a short period of no ice cover before the sheet returned. Station 5 was the earliest ice station, furthest into the ice sheet relative to the ice edge and was likely the least disturbed station in terms of interactions with open ocean water and advancement of melt. The subsequent stations were closer to the ice edge and the results obtained for these may reflect the mechanical impacts of the MIZ movements.

Fig. 3: Vertical profiles of total boron and boron to salinity ratios (B/S) across all five ice stations.
figure 3

Wide bars represent total boron in ice (dark blue) and brine (light blue) and thin, solid bars (black) represent the B/S ratios. The horizontal dashed line represents the open ocean B/S ratio from Lee et al.11. Error bars represent uncertainty in replicate measurements (n = 3).

Full size image

It is not clear why the B/S ratios at stations 8 and 9 were higher (50%) or lower (25%) than the established open ocean ratio of 0.1336, although likely mechanisms are suggested from other studies. As seawater freezes, many of the salts are rejected. However, within the ice crystals that form, selected cations (i.e., Ca2+) tend to concentrate in the liquid regions (brine channels) between the crystals18,19. As these ions accumulate to levels at which solubility limits are reached, calcium carbonate minerals begin to precipitate inorganically20. In these unique microenvironments, B may concurrently precipitate with CaCO3 in the brine channels21. The ice conditions resulting in the inorganic precipitation of CaCO3 in conjunction with B are unknown, but we hypothesize that such inorganic processes (precipitation, re-dissolution) may have occurred at stations 8 and 9. In addition, there are several shifts in brine channel chemistry associated with upward and downward flushing of the brine network, changes associated with the porosity of the brine and coalescence of brine networks during melt22,23.

The B/S ratios in the sampled ice cores and brine (n = 19) averaged 0.1333 ± 0.0231. Surprisingly, this is similar to the open ocean B/S ratio, although with a much larger range. The average and range suggest a decoupling of B from the salinity matrix from which it is derived during the evolution of the brine network. This has important implications for the determination of the B contribution to alkalinity in melt regions, and implies that for studies where highly resolved alkalinity contributions are required (±2 µM), either direct B measurements should be made or greater uncertainty should be incorporated into estimates of borate contributions to AT. This becomes important for melt waters where the salinity is <30 g kg−1 and very important for ice and brine studies.

Boron to Salinity Ratios: Fig. 4a, b shows the B/S ratios across all sample types. There were discernable differences between the open ocean B/S ratio of 0.133611,16 and our mean open ocean value of 0.1305 ± 0.0008 (Fig. 4c). The difference can lead to as much as about 9 μmol kg−1 of total boron concentration. This translates to a difference of only 2 μmol kg−1 in total alkalinity. Nevertheless, it is important that investigators are aware of these uncertainties in future carbonate system studies. Comparison of B measurements in this study with those derived using the open ocean B/S ratios resulted in total B uncertainties of 0–21 µM B for ice cores and 0.4–87 µM B for brine samples. The resulting uncertainty in borate alkalinity (assuming borate is <12% of total boron) is 0−10 µM. Such variations in carbonate system calculations conducted in MIZ regions could be substantial, and this warrants further investigation. In this study, the overall B/S ratio in the MIZ across this large range of salinities (see Fig. 4a, b) was 0.1312 ± 0.0008 mg (kg ‰)−1 and in ice and brine alone was 0.1333 ± 0.02313 mg (kg ‰)−1.

Fig. 4: Variation in the boron to salinity ratios (B/S) across all sample types.
figure 4

Values are shown relative to the established values of Lee et al.11 for 5 open ocean regions and Olafsson et al.15 in the North Atlantic and polar regions: (a) Total B (all samples from this study) versus salinity, b B/S ratio (all samples from this study) versus salinity, and (c) B/S ratio in open seawater samples from this study versus salinity where the gray shaded area represents 1 standard deviation.

Full size image

This study showed important shifts in B/S ratios in the Arctic MIZ, which is critical information for future AT interpretations and ultimately CO2 uptake studies in ice melt environments. The shifts may lead to inaccuracies in carbonate system analyses of between 0 and 2 μmol kg−1 in open melt waters (salinities 28–34 g kg−1) or between 0 and 10 μmol kg−1 in ice and brine. All the ice and brine samples in this study were derived from annual ice of a single ice sheet. Consequently, it is important that this work be expanded to include: (i) annual ice across Arctic regions; (ii) annual ice during summer as melt progresses; and (iii) multi-year ice profiles, to ascertain the impact they may have as they progressively thin and add to Arctic water chemistry. This evidence for the non-conservative behavior of B in ice and brine has not previously been documented in the microenvironments within the sea ice. The observed non-conservative nature of B that we have observed in some ice samples and the impacts of our observations on air-sea CO2 flux and carbonate calculations in open waters with some meltwater is likely small. However, as these systems are currently not well understood and melt intensification is inevitable, we strongly believe that such non-conservative patterns during ice formation and melting are highly noteworthy.

Methods

Sample collection

All samples for B were collected in 100 mL polycarbonate bottles. Open water samples were collected using rosette casts and Teflon-lined Niskin bottles (15 stations). The remaining samples were collected at five ice stations (See details in Supplementary Table 2). Snow samples were collected during on-ice sampling from clearly loose surface snow that lay above the ice sheet. Brine samples were collected by drilling with an Eskimo 10 inch HC40 ice auger to the desired depth within the ice and then allowing brines from the surrounding ice to drain into the hole (<5 min).Brine was collected using a submersible sump pump through tubing placed in the sack hole. Ice cores were collected using a Kovacs Mark II coring system, which retrieved a 9-cm diameter ice core that was sectioned into 10-cm subsections while on the ice. Under ice water was collected once the core was removed and pumped into 5 L compressible LDPE Exetainers.

Total Boron

Boron samples were shipped to Kitack Lee at Pohang University of Science and Technology, Korea. Samples were analyzed using the four-step curcumin method by Upstrom24 as refined by Liu and Lee25. The results indicated considerably better precision and accuracy for B concentration measurements than both the conventional curcumin method and other analytical techniques (Supplementary Fig. 2 and Supplementary Table 3). Briefly, water samples were collected in three 100 mL polyethylene bottles and placed in a water bath at 25 °C. A 500 μL aliquot of sample was pipetted into a reaction bottle along with 1.0 mL of glacial acetic acid and 3.0 mL of propionic acid anhydride, followed by the dropwise addition of 250 μL of oxalyl chloride. After the mixture had equilibrated at room temperature (20 min), 3.0 mL of a sulfuric–acetic acid mixture and 3.0 mL of curcumin reagent were added to the reaction bottle, and the solution was mixed, and reacted to completion (70 min). During the reaction time an orange-colored boron–curcumin complex (rosocyanine) formed. For analysis of low S samples, the amounts of sample and reagents were increased; 1000 μL for sample, 2.0 mL for glacial acetic acid, 5.0 mL for propionic anhydride, 500 μL for oxalyl chloride, and 4.0 mL for a sulfuric–acetic acid and curcumin reagent26. The absorbance of the final solution was measured at 543 nm using a 1 cm quartz cell and corrected for the absorbance of reagent blanks (%RSD range 0.3–0.7%).

Salinity

Salinity (Practical Salinity Scale) was measured onboard in accordance with IOC, SCOR and IAPSO27, using a Portasal Salinometer 8410A. The instrument was calibrated at 25 °C against IAPSO standard seawater. The precision was at least ±0.002 g kg−1.

pH

The concentration of hydrogen ion (pH) was determined spectrophotometrically using an Agilent Cary 60 UV-Vis systems spectrophotometer and m-cresol purple indicator dye following Carter et al.28. 40 ml samples were temperature equilibrated in the instrument’s thermostatted cell holder at 25 °C and precision and accuracy monitored throughout the cruise using triplicate samples from the same Niskin bottle and TRIS buffers. The precision of the pH measurements is estimated to be approximately ± 0.001429. Further details on the methods and quality control are described in Millero et al.30 and Woosley et al.29. All open water samples and under ice water were analyzed for pH. Brine and ice samples were not included in the analysis due to limited sample availability.

Borate

Acidity constants for boric acid/borate equilibria were calculated after Dickson16 using the temperature and salinity of each discreet sample (Supplementary Table 1). Note borate alkalinity for ice and brine could only be estimated due to the lack of accurate pH profile data and is therefore not used in this analysis.

Reporting summary

Further information on research design is available in the Nature Research Reporting Summary linked to this article.