A map of the study area with stations indicated by red dots is given in Fig. 1b. General information about the selected stations and samples used in this study is provided in Table S1 in the Supplement. Samples were taken along two transects to the south and three transects to the north of Davis Strait along the west Greenland shelf. Transect 1 was located in a shallow area (<120 m depth) north of Nuuk. Transects 2 and 3 started close to the mouth of Kangerlussuaq Fjord and Nassuttooq Fjord, respectively. Station 15, which is part of transect 3, had to be moved further south because of the presence of sea ice. Transect 4 started in Disko Bay approximately 70 km from the mouth of the Ilulissat Icefjord, where the Jakobshavn Isbræ terminates. Transects 2, 3, and 4 continued westward away from the coast towards the shelf edge. Transect 5 started close to Disko Island and continued towards the northwest.

Water column samples were taken at 25 stations as conductivity-temperature-depth (CTD) casts along five transects during ten consecutive sampling days between 18 and 28 July 2021 on board RV Dana (DANA 6/21, Fig. 1b, Table S1). A SBE 911 CTD+ unit (Sea-Bird Scientific; Bellevue, USA) was used to collect hydrographic profiles, including temperature, salinity, pressure, and dissolved oxygen. Two oxygen sensors were part of the CTD system and calibrated prior to the cruise. By cross-referencing the data of the two sensors, we were able to monitor their drift during the cruise and correct the data accordingly. Samples were collected in 10 L metal-free Niskin water samplers (Ocean Test Equipment; Fort Lauderdale, USA). Detailed information about each sample is given in Table S1, with sample IDs corresponding to the respective station and sampling depth.

The raw CTD data were processed and averaged into 0.5 dbar pressure bins. The processed CTD data were used to construct water depth profiles of potential temperature, salinity, oxygen (Fig. 2), and potential density (Fig. S1 in the Supplement). Calculations were performed using R (version 4.2.2; R Core Team, 2022) and RStudio (version 2023.06.0; Posit team, 2023). Bivariate, linear data interpolation in combination with a surface approximation using multilevel B-splines with the “interp” function (package akima; Akima and Gebhardt, 2022) and the “mba.surf” function (package MBA; Finley et al., 2022) was used before plotting.

For trace elements analysis, filtration duplicates were obtained and measured as triplicates (n=6). Outliers were identified using a Dean–Dixon outliers test and removed from further analysis. Mean concentrations (μ) and standard deviation (SD; σ) were estimated, considering twice the SD to represent μ±2σ. Limits of detection (LOD) and limits of quantification (LOQ) are given in Table S3 and were calculated according to DIN 32645:2008-11, based on measurements of filtrations blanks (n=8). The LOD is defined as three times the standard deviation (3 × SD) of the blank measurements and represents the lowest concentration that can be reliably distinguished from background noise, though not necessarily quantified with precision. The LOQ is defined as ten times the standard deviation (10 × SD) and represents the lowest concentration that can be quantitatively determined with acceptable accuracy and precision (DIN e.V., 2008).

The sum parameter NO_x for nitrate and nitrite (NO_x = nitrate + nitrite) was established and used throughout the discussion. The LOQs for the nutrient analysis are given in Table S3.

Parameters of the carbonate system (pH, pCO₂, Revelle factor, and Ω Aragonite) were calculated using the CO2Sys Macro (Pierrot et al., 2011) with salinity, temperature, AT and CT as input variables. As input parameters, we used the dissociation constant by Mehrbach et al. (1973), refit by Dickson and Millero (1987), the HSO4- dissociation constant by Dickson (1990) and the seawater pH-scale (Dickson, 1990; Dickson and Millero, 1987; Mehrbach et al., 1973). We used the CO2Sys Macro to calculate temperature-normalized CT (CT_temp) according to Wu et al. (2019). The calculation was based on the median potential temperature of WGSW (3.23 °C, n=465) from the CTD profiles, filtered for a depth range of 30 to 50 m. Further input parameters were observed salinity, AT, and the in situ pCO₂.

For statistical analysis, the data of the different parameters were merged according to station and sampling depth, obtaining one data matrix with 17 parameters and 41 observations. As a minor proportion of the total observations (<3 %) were below the LOD or LOQ, a single imputation method was applied (Helsel, 2005; Jain, 2016). Values below the LOD were replaced with a random value between zero and the LOD, while values below the LOQ were replaced with a random value between the LOD and LOQ. Each parameter was tested for normality (Shapiro-Wilk test, α=0.01) and transformed (log- or Box Cox-transformation) if the normality criteria was not met (Table S4). The data were standardized, and principal component analysis (PCA) with varimax rotation was performed using the “principal” function (package psych; Revelle, 2024). Three principal components (PCs) were selected based on the Guttman–Kaiser criterion and the scree plot, retaining only those with eigenvalues greater than 1 (Bro and Smilde, 2014). A broken stick analysis was performed to evaluate the significance of variable loadings and to identify which variables are associated with specific PCs (Peres-Neto et al., 2003).

The distribution of salinity, potential temperature, and oxygen are shown as water depth profiles for each transect in Fig. 2. Ranges of minimum and maximum values for salinity, potential temperature, oxygen and AOU are summarized in Table S5. Salinity was generally low in surface waters at stations close to the coast (stations 6, 11 and 17) and towards the west of each transect (stations 14, 15, 24, 29 and 30), with a minimum value of 29.70 recorded at the surface of station 14 (Fig. 2g). The highest salinity (34.89) was observed in deep waters at station 10, between 530 and 558 m depth (Fig. 2d). Potential temperature was highest in surface waters in the southeastern part of Davis Strait, particularly at transect 2 (stations 6, 7, 8, and 10), and at station 17 in Disko Bay. Lower potential temperatures were found further offshore in the northwestern part of Davis strait. The minimum potential temperature of -1.63 °C was recorded in intermediate waters (40 to 55 m) at station 30 (Fig. 2n), while the maximum value of 6.7 °C occurred in surface waters at station 8 (Fig. 2e). In general, higher oxygen concentrations and lower AOU values were found in the photic zone (1.5 to 40 m), either near the Greenlandic coast at the beginning of transects (stations 6 and 17) or at the western ends (stations 10, 14, 15, and 24). The lowest oxygen concentration (204 µmol L⁻¹) and highest AOU (147 µmol L⁻¹) were observed in deep waters at station 15 (645 to 673 m) (Fig. 2i). Conversely, the highest oxygen concentration (406 µmol L⁻¹) and lowest AOU (-72 µmol L⁻¹) were measured in intermediate waters (34 to 37 m) at station 10 (Fig. 2f).

The overall nutrient distribution is shown as vertical profiles in Fig. 3 and as sea surface concentration plots in Fig. S2. In surface waters, nutrient concentrations were uniformly low across the shelf. NO_x concentrations were particularly low (<LOQ=0.21 µmol L⁻¹) in surface waters and shallow shelf waters. Exceptions to this trend were observed at station 17 in Disko Bay and at stations 14, 15, 24 and 30 towards the west of each transect, where surface waters with higher nutrient concentrations were present (refer to Fig. S2). Below the biological productive zone, nutrient concentrations gradually increased with depth, reaching maximum concentrations at station 15 (667 m) (refer to Table 1). Sherwood et al. (2021) reported similarly high nutrient concentrations in these deep waters, which are further discussed in Sect. 4.1.2.

In comparison to data collected during 2015 (GEOTRACES Intermediate Data Product Group, 2023) and 2016 (Bruyant et al., 2022), maximum nutrient concentrations of this study are lower, which we attribute to regional differences in the study areas, particularly with respect to the maximum sampling depth. We measured maximum values at the shelf edge at 667 m, whereas the referenced studies reported maximum values at stations located in central Baffin Bay, where sampling extended to significantly greater depths.

The distributions of trace elements are shown as vertical profiles in Fig. 4 and surface water concentrations in Fig. S3. The vertical profiles of dV and dCu showed opposing trends (Fig. 4a and f). With depth, dV concentrations increased, while dCu concentrations decreased. The dV concentrations ranged from 1270±30 ng L⁻¹ (station 24; 3 m) to 1810±60 ng L⁻¹ (station 10; 551 m), while dCu concentrations ranged from 206±3 ng L⁻¹ (station 24; 3 m) to 85.4±2.5 ng L⁻¹ (station 10; 551 m), both matching minimum and maximum occurrences of salinity. Surface waters close to the coast of Greenland and at the western end of each transect showed lower dV concentrations (Fig. S3a) and higher dCu concentrations (Fig. S3f), respectively. The distributions of dFe, dNi and dCd showed low concentrations in surface and subsurface waters, which gradually increased with depth (Figs. 4c, e, and S3c, e, and g). Maximum concentrations were observed at depth, with dFe reaching 630±30 ng L⁻¹ (station 18; 256 m), dNi reaching 307±9 ng L⁻¹ (station 15; 667 m) and dCd reaching 52.2±1.2 ng L⁻¹ (station 15; 667 m). Minimum concentrations of dFe, dNi and dCd were present in surface waters (refer to Table 1), particularly at stations with some distance to the coast. At several stations, dFe concentrations in surface waters were below the LOQ (<50 ng L⁻¹). Surface concentrations of dFe and dNi (Fig. S3c and e) were elevated near the coast, decreased with increasing distance along each transect and increased again towards the west of each transect. In contrast, surface dCd concentrations were rather uniform across the shelf, with a slight increase toward the westernmost stations (Fig. S3g). The concentrations of dMn and dCo decreased from surface to subsurface waters and increased again with depth at stations located on the shelf (Figs. 4b, d and S3b, d). In contrast, at the shelf edge, dMn and dCo concentrations remained stable with depth, resulting in minimum values of 54.8±1.8 ng L⁻¹ for dMn (station 24; 317 m) and of 3.81±0.15 ng L⁻¹ for dCo (station 10; 551 m). Elevated surface concentrations of both dMn and dCo were observed at station 17 in Disko Bay and at stations 11 and 12, located near the mouth of the Nassuttooq Fjord (Figs. 4b and d; S3b and d). Surface concentrations of dMn and dCo generally decreased with increasing distance from the coast. The depth profile of dPb did not exhibit a consistent vertical pattern (Fig. 4h). South of Davis Strait, elevated dPb concentrations were observed across transect 2 (refer to Fig. S3h), including a maximum dPb concentration of 4.04±0.13 ng L⁻¹ in the deep waters of station 10. In contrast, north of Davis Strait, dPb concentrations were generally lower, falling below the LOQ (<0.98 ng L⁻¹) in the deep waters of station 15 and the intermediate to deep waters of Disko Bay (stations 17 and 18). The minimum dPb concentration of 0.99±0.08 ng L⁻¹ was observed in the deep waters of station 28.

Compared to literature values, our results for trace element concentrations are generally consistent with data collected during the 2015 Geotraces cruise GN02 (GEOTRACES Intermediate Data Product Group, 2023). Variations of minimum and maximum values are attributed to differences in sampling locations. While the present study took place on the shelf, capturing the influence of coastal runoff and sea ice meltwater, the Geotraces cruise focused on offshore stations located further north in central Baffin Bay.

The vertical profiles of AT and CT (Fig. 5a and b) show a gradual increase with depth. Minimum concentrations of AT and CT were present in the low-salinity surface waters at station 15 (AT: 2062 µmol kg⁻¹; CT: 1949 µmol kg⁻¹). While maximum CT concentrations of 2214 µmol kg⁻¹ were observed in the deep waters of station 15, AT concentrations followed the salinity gradient, reaching a maximum value of 2305 µmol kg⁻¹ in the deep waters of station 10. The vertical profiles of pH and pCO₂ (Fig. 5d and e) show opposing trends, with pH increasing and pCO₂ decreasing towards intermediate waters of the photic zone. Below the photic zone, pH gradually decreases and pCO₂ increases with depth. This trend results in a pH maximum (8.20) and a pCO₂ minimum (249 µatm) in the photic zone of station 15 at 32 m, while the deep waters of the same station show a pH minimum (7.86) and a pCO₂ maximum of (559 µatm). The vertical profiles of the CT : AT ratio, Revelle factor and Ω Aragonite (Fig. 5c, f and g) show that these parameters remain relatively stable within the upper 25 m of the water column. Below this depth, the CT : AT ratio and Revelle factor gradually increase, while Ω Aragonite decreases. The maximum CT : AT ratio (0.968) was observed in the deep waters of station 15, coinciding with a maximum Revelle factor of 17.80 and a minimum Ω Aragonite value of 0.90.

The surface water concentrations of the carbonate system parameters are given in Fig. S4. In general, stations closest to the Greenland coast exhibited lower AT and CT values, which increased with distance to the coast. Following this trend, surface water pH increased and pCO₂ values decreased along each transect with increasing distance from the coast. In coastal surface waters at stations 6 and 11, higher CT : AT ratios (0.920 and 0.925) coincided with higher Revelle factors (12.8 and 13.3) and lower Ω Aragonite values (1.93 and 1.82), indicating a reduced buffering capacity in coastal waters. In contrast, surface waters in Disko Bay (station 17; 3 m) exhibited a lower CT : AT ratio (0.907), driven by a low CT concentration of 1972 µmol kg⁻¹ and a corresponding low pCO₂ value of 255 µatm. This was accompanied by a high pH (8.19), a low Revelle factor (11.85), and a maximum Ω Aragonite value of 2.17, indicating the higher buffering capacity of this water. In the westernmost surface waters of each transect, AT and CT values were again lower, but CT : AT ratios were among the highest observed. As a result, pCO₂ values and Revelle factors were elevated at these stations, while pH and Ω Aragonite values were reduced.

Compared to AT and CT data collected in 2015 (Helmuth Thomas, personal communication, 28 June 2024) and 2016 (Miller et al., 2020), our results fall within a similar range. However, the minimum values observed in this study are slightly lower, likely due to the influence of sea ice meltwater, which was captured during our sampling.

The distribution of water masses along the west coast of Greenland has been the subject of considerable discussion over the past decades, leading to the development of multiple classification systems and a wide range of nomenclatures. In this study, we identify water masses based on characteristic salinity and potential temperature ranges, as illustrated in Fig. 6 (black boxes; based on Tang et al., 2004; Curry et al., 2011, 2014; Sherwood et al., 2021). This terminology is applied consistently throughout the discussion. For comparison, a more recent interpretation of the water mass structure along the west Greenland coast, as proposed by Rysgaard et al. (2020), is also included in Fig. 6 (orange labels). The West Greenland Shelf Water (WGSW; θ<7 °C; S<34.1) was the dominant water mass in the study area, occurring in each transect across the shelf above a water depth of 200 m. The WGSW is also referred to as Southwest Greenland Coastal Water (CW), originating from the East Greenland Current and modified by freshwater input from Greenlandic glacial runoff. The West Greenland Intermediate Water (WGIW; θ>2 °C; S>34.1) was present in deeper waters along the slope below 200 m. This water mass corresponds to the Subpolar Mode Water (SPMW), of which we identified a colder and less saline variant – the deep Subpolar Mode Water (dSPMW) – along the slope. Arctic Water (AW; θ≤2 °C; S≤33.7) was detected north of Davis Strait at transects 3, 4, and 5, forming an intermediate layer between WGSW and WGIW above 100 m on the western side of the transects. These cold, low-salinity AW pockets can be observed in Fig. 2h, k, and n. At transect 4, AW was found at station 17 approximately 70 km off the coast of Greenland, extending into Disko Bay. The updated classification refers to this water mass as Baffin Bay Polar Water (BBPW), of which a diluted form was found west of transects 4 and 5. The Transitional Water (TrW; θ≤2 °C; S>33.7) was primarily observed at transects 4 and 5 between 100 and 200 m, separating AW from WGIW. At station 15, the waters below 300 m were defined as Baffin Bay Water (BBW; θ≤2 °C , S>34.1), a shallower extension of the Baffin Bay Deep Water, presenting as a distinct tail on the θ–S diagram (Sherwood et al., 2021). Due to very low surface salinity values at stations 14, 15, and 24 (S<31.5), we introduced sea ice meltwater (SIM) as an additional source water type to distinguish areas affected by sea ice melt.

The extensive data set was processed using PCA to find physicochemical processes that alter the distribution of carbon, macronutrients, and trace elements. The results of the PCA analysis are summarized in Table S6, where significant PC loadings of each parameter are highlighted in bold. Overall, 82 % of the total variance in the normalized data set can be explained by three PCs. The majority of variance is explained by PC 1 with 52 %, followed by PC 2 with 18 % and PC 3 with 12 %. Salinity, depth, oxygen, AT, CT, dV, and dCu significantly load on PC 1; potential temperature, AOU, NO_x, silicate, phosphate, dNi, dCd, and dPb significantly load on PC 2; and dFe, dMn, dCo, dNi, and dCu significantly load on PC 3. The results of the PCA are provided as biplots in Fig. 7, with colors indicating the water mass affiliation of each sample.

The spatial distribution of parameters revealed significant differences in eastern and western shelf waters. To investigate these differences, we conducted a direct comparison of the three major water masses present along the west Greenland shelf: the northward-moving WGSW (1 to 220 m) and WGIW (130 to 560 m) with the southward-moving AW (2 to 165 m). For each parameter, the median and the mean ± SD values are summarized in Table 2. The median is used throughout the data discussion because of its robustness against outliers.

The median values of salinity and potential temperature reflect the origin of the water masses and the freshwater dynamics of the area. In the upper water column, AW is slightly fresher (33.48) and colder (0.38 °C) than WGSW (33.57; 2.68 °C). This east–west gradient in temperature and salinity has also been described by Tang et al. (2004) between Baffin Island and the Greenland coast. As the warmer WGSW flows cyclonically around Baffin Bay and mixes with adjacent waters from the CAA, it becomes progressively colder and fresher, eventually returning southward as AW (Tang et al., 2004). The WGIW is part of the deep water WGSC that transports warm (2.93 °C) and salty (34.45) waters from the North Atlantic (Curry et al., 2014).

The AOU concentrations are reflective of respiration and the biological activity of the water (Sarmiento and Gruber, 2006). As AOU is a depth-dependent variable, we examined a consistent depth range of 30 to 50 m to enable a more meaningful comparison between WGSW and AW. Within this range, the median AOU values are -6 µmol L⁻¹ for WGSW and 7 µmol L⁻¹ for AW. This suggests that warmer shelf waters (WGSW) are more biologically productive than colder AW, as indicated by greater oxygen production in the upper water column. Although primary production was not directly measured in this study, this interpretation is supported by Krawczyk et al. (2021), who reported the highest chlorophyll a concentrations along the southwest Greenland coast and the lowest values along Baffin Island. This pattern is consistent with overall lower nutrient concentrations observed in WGSW compared to AW (refer to Sect. 4.1.2 and Table 2), as nutrients are removed from the water column during PP. In Fig. 8a, NO_x concentrations are plotted against longitude within the 30 to 50 m depth range, including literature data from the Green Edge cruise in 2016 (Bruyant et al., 2022). A clear spatial gradient is evident, with lower NO_x concentrations in WGSW on the shelf and higher NO_x concentrations in AW further offshore, particularly in areas that were still ice-covered. This nutrient gradient reflects the east-to-west progression of sea ice retreat and has been linked to enhanced PP and species richness along the southern coast of west Greenland and in Disko Bay, compared to the more nutrient-rich but less productive regions near Baffin Island and northwest Greenland (Lafond et al., 2019; Krawczyk et al., 2021). We consider the timing of our sampling during the mid-summer season to be critical in capturing these pronounced differences in water mass composition. By this time, shelf waters had already become depleted in nutrients, whereas offshore regions were only beginning to experience increased biological processes following the retreat of sea ice (refer to Sect. 4.3). With increasing depth, respiration returns nutrients to the water column, as evidenced by high AOU values in WGIW (62 µmol L⁻¹) and overall higher nutrient concentrations at depth (refer to Table 2).

A comparison of the carbonate system parameters in WGSW and AW reveals that both AT and CT median concentrations were lower in WGSW. This observation aligns with a study by Burgers et al. (2024), who reported that the dominant source of freshwater for WGSW in eastern Baffin Bay is glacial meltwater, characterized by a significantly lower AT endmember (390±140 µmol kg⁻¹) compared to AW in western Baffin Bay, where the AT endmember is 1170±120 µmol kg⁻¹. The higher median CT : AT ratio in AW (0.941) compared to WGSW (0.922) is attributed to the CAA outflow, which carries Pacific-origin water characterized by elevated CT concentrations (Azetsu-Scott et al., 2010; Shadwick et al., 2011b). In contrast, Atlantic-derived waters transported to eastern Baffin Bay tend to have lower CT : AT ratios (Burgers et al., 2024). These slightly higher CT : AT ratios in AW are also reflected in a higher median pCO₂ concentration, lower median pH value and lower median Ω Aragonite concentration (refer to Table 2).

To minimize depth-related variability, we further examined CT values within a consistent depth range of 30 to 50 m. It is well established that colder sea surface temperatures at high latitudes contribute significantly to the latitudinal gradient in surface CT, due to the increased CO₂ solubility of colder waters (Li and Tsui, 1971; Wu et al., 2019). To assess the extent to which temperature differences between AW and WGSW influence CT concentrations, we calculated temperature-normalized CT (CT_temp) following the method described by Wu et al. (2019) (see Sect. 2.4). This normalization was based on the median potential temperature of WGSW (3.23 °C, n=465) within the 30 to 50 m depth range. The observed CT (CT_obs) and CT_temp values for AW and WGSW are shown in Fig. 8c and d. For comparison, we included literature data from the Green Edge cruise (Miller et al., 2020) and Geotraces GN02 in 2015 (BB1–BB3, CAA2) (Helmuth Thomas, personal communication, 28 June 2024), filtered to the same depth range of 30 to 50 m. Including literature data, the difference between the median CT values of AW and WGSW is ΔCTobsWGSW-AW=35 µmol kg⁻¹. After temperature normalization, the difference is significantly reduced to ΔCTtempWGSW-AW=12 µmol kg⁻¹. This indicates that the majority of the CT difference in AW and WGSW is driven by the temperature difference between the two water masses. The remaining CT difference may be attributed to higher biological productivity in WGSW, which reduces CT concentrations, and to the mixing of AW with the CAA outflow, which is known to transport additional CT to Baffin Bay (Azetsu-Scott et al., 2010; Shadwick et al., 2011b; Burgers et al., 2024).

In the deeper waters of WGIW, both AT and CT median concentrations increase, following the salinity gradient and reflecting the contribution of OM respiration to the release of CT to the water column. This is also evident in higher median pCO₂ concentrations and lower median Ω Aragonite concentrations in WGIW (refer to Table 2), consistent with observations by Burgers et al. (2024).

All median concentrations of trace elements exhibiting a nutrient-type profile (dFe, dNi, and dCd) were lower in WGSW compared to AW. This difference is attributed to biological uptake during PP, which we found to be higher in WGSW. The return of these elements to the water column through the remineralization of OM is evident in the elevated median concentrations observed in WGIW (refer to Table 2). In addition to biological cycling, variations in source water composition also contribute to differences in trace element concentrations. AW exhibited elevated concentrations for dFe, dMn, dCo, and dNi (refer to Table 2), which are likely linked to the CAA outflow into Baffin Bay. This outflow has been shown to have higher dFe and dMn concentrations due to sediment resuspension in the benthic boundary layer (Colombo et al., 2020, 2021), as well as higher dNi and dCu concentrations associated with Pacific-origin waters advected from the Canada Basin through the CAA (Jensen et al., 2022). In contrast to other trace elements, higher dPb median concentrations were observed in WGSW and WGIW compared to AW (refer to Table 2). As mentioned in Sect. 3.3, a latitudinal gradient in dPb concentrations exists across Davis Strait, with dPb concentrations decreasing along the west coast of Greenland. This trend is illustrated in Fig. 8b, where dPb concentrations are plotted against latitude within the 30 to 50 m depth range. The figure combines our data with literature values from the Geotraces GA01 cruise in 2014 (stations 64 to 71) and Geotraces GN02 in 2015 (BB1–BB3, CAA2) (both: GEOTRACES Intermediate Data Product Group, 2023). The elevated dPb concentrations in WGSW and WGIW are likely influenced by freshwater discharge from the GIS (Krisch et al., 2022b) and the advection of North Atlantic waters (Colombo et al., 2019), both enriched in dPb. In contrast, AW exhibited much lower dPb concentrations, which we attribute to the influence of the CAA outflow, as this region is relatively isolated from anthropogenic sources of dPb (Colombo et al., 2019). Our results demonstrate that the distribution of dPb is primarily controlled by the mixing of water masses with distinct dPb signatures. As such, dPb can serve as a useful tracer for illustrating the gradual mixing and transformation of water masses along the west Greenland shelf.

Baffin Bay and its surrounding areas are typically ice-free between July and October (Bi et al., 2019). Our statistical analysis highlighted the significant influence of SIM on the chemical composition of surface waters (refer to Sect. 4.1.1). To visualize the decline in sea ice concentration during the sampling period, we used the ASI sea ice concentration product provided by the University of Bremen (Spreen et al., 2008). In Fig. 9a, the change in sea ice concentration between 19 and 28 July 2021 is illustrated. At the beginning of our sampling period on 19 July 2021, two extensive areas with high sea ice concentration were present between 66–68° N and 70–74° N. By 28 July 2021, these areas had disappeared almost completely, indicating a rapid and complete sea ice retreat within just 10 d. This melt event significantly affected sea surface salinity and potential temperature, as shown in Fig. 9b and c. Both parameters decreased towards the west along transects 3, 4 and 5, with salinity reaching a minimum of 29.70 (station 14; 3 m) and potential temperature of -0.36 °C (station 24; 3 m). The category SIM (refer to Sect. 3.5) was identified in the upper 4 m of stations 14 and 24, and in the upper 8 m of station 15. Although not formally classified as SIM, a similar trend of reduced surface salinity and potential temperature values was observed at stations 10, 13, 23, 28, 29, and 30, adjacent to SIM stations (refer to Fig. 2). We suggest that this trend reflects the gradual advection of SIM from central Baffin Bay towards the southeast.

In Baffin Bay, Lafond et al. (2019) observed elevated surface nutrient concentrations at stations covered by sea ice, in contrast to open-water stations located east of the ice edge, where nutrients were nearly depleted. In this study, the distribution of surface nutrients follows a similar trend (refer to Sect. 3.2). Along each transect, nutrient concentrations were consistently higher towards the west, coinciding with the recent retreat of sea ice. These western areas also exhibited elevated oxygen concentrations and low AOU values (refer to Sect. 3.1), suggesting the onset of a phytoplankton bloom. The timing of the study enabled us to observe a transitional phase in the seasonal cycle: nutrients in the western shelf waters had not yet been fully consumed, while waters farther east had already become depleted due to earlier biological activity.

Our observation of the carbonate system indicates that surface waters of stations influenced by SIM exhibited lower AT and CT concentrations, along with higher CT : AT ratios. These conditions resulted in higher pCO₂ values and lower Ω Aragonite values (refer to Sect. 3.4). Similar results of sea ice meltwater increasing pCO₂ and decreasing CaCO₃ mineral saturation states have been reported along the east and west coast of Greenland (Henson et al., 2023), the Fram Strait (Tynan et al., 2016), and in other Arctic regions such as the Canada Basin (Bates et al., 2009). To investigate the effect of sea ice on the upper 50 m of the water column, which is the typical penetration depth of sea ice meltwater (Jones et al., 1983), we calculated salinity-normalized AT (ATsal=35 AT/S) and temperature-salinity-normalized CT (CTtempsal=35CTtemp/S), according to Wu et al. (2019) (refer to Sects. 2.4. and 4.2). These normalizations account for the effects of temperature and salinity on CO₂ solubility, while remaining sensitive to biological processes, air–sea gas exchange, and water mass mixing (Shadwick et al., 2011a). For AT_sal (Fig. 10a), most surface water concentrations ranged between 2330 and 2350 µmol kg⁻¹. We observed an AT addition of ≈25 to 55 µmol kg⁻¹ for SIM stations and of ≈5 to 45 µmol kg⁻¹ for SIM adjacent stations. These values are consistent with Jones et al. (1983), who reported an AT addition of ≈100  µmol kg⁻¹ further north in Baffin Bay, where the influence of sea ice meltwater is more pronounced. This AT enrichment is attributed to the release of alkalinity during the melting of sea ice that contains ikaite, as described by Rysgaard et al. (2011, 2012) and Jones et al. (1983). The expected increase in CT by ikaite dissolution should be half the corresponding change in AT (Jones et al., 1983), or approximately 15 to 30 µmol kg⁻¹. The CTtempsal values are shown in Fig. 10b, with most surface water concentrations ranging between 2140 and 2170 µmol kg⁻¹. At SIM stations, we observed an increase in CTtempsal of ≈60 to 75 µmol kg⁻¹, and at SIM adjacent stations of ≈10 to 60 µmol kg⁻¹. Hence, the observed CT addition exceeds the expected contribution from ikaite dissolution alone. Possible explanations for the observed surplus in CT include the release of CT from organic carbon remineralization under the ice (Bates et al., 2009; Tynan et al., 2016), or within the ice itself, as sea ice is known to contain higher concentrations in OM than the underlying seawater (Vancoppenolle et al., 2013). Additionally, the inflow of CT-enriched waters transported through the CAA into Baffin Bay may contribute to the elevated CT levels (Azetsu-Scott et al., 2010; Shadwick et al., 2011b; Henson et al., 2023). Our results indicate that the dissolution of ikaite in sea ice acted as a source of AT, providing a small geochemical buffer to surface waters (Jones et al., 2021). However, the inflow of Pacific-origin waters and remineralization under and within the ice contributed to elevated CT : AT ratios, resulting in higher pCO₂ and lower pH and Ω Aragonite values in surface waters. There is evidence that once the sea ice melt facilitated bloom fully develops and CT is removed from seawater by PP, higher pH and Ω Aragonite values can be expected along the sea ice edge (Tynan et al., 2016). This process is reflected in the photic zone of station 15 at 32 m, where an oxygen maximum and AOU minimum indicate high PP, resulting in a pH maximum (8.20) and pCO₂ minimum (249 µatm) (refer to Sect. 3.4).

We observed elevated surface water concentrations of dFe, dCo, dNi, dCu, and dCd in areas recently covered by sea ice, particularly towards the west of each transect (refer to Sect. 3.3). This observation aligns with a study by Tovar-Sánchez et al. (2010), who observed an enrichment of these trace elements in sea ice relative to surface waters, suggesting that sea ice meltwater is a significant source of these trace elements to surface waters in Baffin Bay. In particular, the additional input of dFe from melting sea ice may play a biologically important role in maintaining ice-edge blooms, as previously observed, e.g., in the Bering Sea (Aguilar-Islas et al., 2008). In addition to sea ice input, we suggest that, similar to macronutrients, trace elements had not yet been removed from surface waters by biological uptake at the time of sampling. We hypothesize that following the development of a bloom, concentrations of both macronutrients and trace elements would decline, as already observed in surface waters at shelf stations closer to the Greenlandic coast. Consistent with findings from other Arctic regions (Nakanowatari et al., 2007; Kanna et al., 2014), the progressive melting and retreat of sea ice in Baffin Bay and Davis Strait appears to significantly influence the biogeochemical cycling of carbon and trace elements, as well as the biological productivity of the upper water column.

The influence of coastal freshwater runoff on the chemical composition of surface waters was demonstrated by statistical analysis (refer to Sect. 4.1.3) and is evident in minimum salinity values observed at stations 6, 11, and 17 (refer to Sect. 3.1). Stations 6 and 11 are located near to the mouths of the Kangerlussuaq Fjord and the Nassuttooq Fjord, both of which receive glacial and riverine freshwater from the GIS (Monteban et al., 2020). Station 17 is located in Disko Bay, approximately 70 km from the mouth of the Ilulissat Icefjord/Jakobshavn Isbræ system, where glacially modified waters form a buoyant coastal current outside the fjord (Beaird et al., 2017). These low-salinity surface waters coincide with elevated macronutrient concentrations (refer to Sect. 3.2). A modeling study by Møller et al. (2023) demonstrated that a portion of the macronutrient input into Disko Bay is directly linked to freshwater runoff. Additional mechanisms that transport marine-sourced nutrients from deeper waters into the photic zone include vertical mixing fueled by wind and tide (Møller et al., 2023), as well as estuarine upwelling circulation forced by the glacier's freshwater input (Williams et al., 2021). Our results suggest that the continued replenishment of macronutrients into the photic zone of Disko Bay can sustain PP into the summer season. This is evidenced by high oxygen and low AOU concentrations in surface waters at station 17 (refer to Sect. 3.1). These conditions have significant implications for the carbonate system and the uptake of CO₂, as reflected by low Revelle factors and low pCO₂ values in surface waters of Disko Bay (refer to Sect. 3.4). A similar pattern was observed in the Godthåbsfjord system in southern Greenland, where biological processes were identified as the primary drivers of CO₂ uptake, resulting in low surface water pCO₂ (Meire et al., 2015). Based on these findings and supporting literature, we propose that nutrient cycling in Disko Bay is strongly influenced by GIS-derived freshwater input. This input stimulates PP and establishes Disko Bay as an important seasonal sink for atmospheric CO₂ well into the summer.

Apart from the Disko Bay area, surface waters at stations closest to the Greenland coast exhibited lower Ω Aragonite values with lower buffering capacities (refer to Sect. 3.4). Glacial freshwater discharge from the GIS is known to have a strong dilution effect on AT relative to CT, which reduces the buffering capacity of surface waters, resulting in lower Ω Aragonite saturation states and pH values (Henson et al., 2023; Burgers et al., 2024). However, surface water acidification can be mitigated by PP, which removes CT from the photic zone, consequently increasing both pH and Ω Aragonite (Chierici and Fransson, 2009), as observed in Disko Bay. At stations along the coast, surface waters exhibited negative AOU values, indicating CT drawdown via PP. Simultaneously, extremely low NO_x values (<LOQ) were detected, suggesting that PP was approaching nutrient limitation. This nutrient depletion likely constrained further CT uptake, thereby no longer compensating the AT dilution effect caused by the GIS freshwater input. As a result, surface waters exhibited lower pH values coupled with higher Revelle factors and pCO₂ values. Our findings suggest that as the summer season progresses and PP declines, surface waters along the west Greenland coast become increasingly vulnerable to acidification due to the input of poorly buffered glacial freshwater. These observations are consistent with Henson et al. (2023), who observed that AT dilution from glacial meltwater drives corrosive conditions in Greenlandic coastal surface waters, leading to a general decline in both Ω Aragonite and pH (refer to Sect. 3.4).

In addition to influencing the cycling of macronutrients and carbon, coastal freshwater input significantly affects the distribution of trace elements. Surface water concentrations of dMn, dFe, dCo, dNi, and dCu were generally highest close to the coast and decreased with increasing distance along each transect (refer to Sect. 3.3). This trend was especially pronounced at stations 11 and 12 (mouth of Nassuttooq Fjord), and station 17 (Disko Bay, Ilulissat Icefjord). While surface water concentrations of dNi in Disko Bay remained relatively constant, elevated concentrations were observed at the mouth of Nassuttooq Fjord. Similar spatial variability in dNi has been reported by Krause et al. (2021), who related these trends to the presence of outcrops of Ni-rich minerals along the Greenland coast. Overall, our work is consistent with previous studies from various regions around Greenland, which identified GIS-derived freshwater as a source of trace elements including dMn, dFe, dCo, dNi, and dCu (Colombo et al., 2020; Hawkings et al., 2020; Krause et al., 2021; Chen et al., 2022; Krisch et al., 2022a;). The input of bioactive trace elements, e.g., dMn, dFe, and dCo, via GIS freshwater runoff has been suggested to enhance primary productivity in adjacent shelf areas (Bhatia et al., 2013; Hawkings et al., 2020). Our study provides further evidence that freshwater from the GIS may support biological processes by supplying bioactive trace elements to surface waters.