Increased sea ice melt as a driver of enhanced Arctic phytoplankton blooming

Phytoplankton primary production in the Arctic Ocean has been increasing over the last two decades. In 2019, a record spring bloom occurred in Fram Strait, characterized by a peak in chlorophyll that was reached weeks earlier than in other years and was larger than any previously recorded May bloom. Here, we consider the conditions that led to this event and examine drivers of spring phytoplankton blooms in Fram Strait using in situ, remote sensing, and data assimilation methods. From samples collected during the May 2019 bloom, we observe a direct relationship between sea ice meltwater in the upper water column and chlorophyll a pigment concentrations. We place the 2019 spring dynamics in context of the past 20 years, a period marked by rapid change in climatic conditions. Our findings suggest that increased advection of sea ice into the region and warmer surface temperatures led to a rise in meltwater input and stronger near‐surface stratification. Over this time period, we identify large‐scale spatial correlations in Fram Strait between increased chlorophyll a concentrations and increased freshwater flux from sea ice melt.

the Arctic has been attributed to multiple factors including sea ice thinning, sea ice retreat timing, increasing wind-driven mixing, and increasing sea surface temperatures (SSTs; e.g., Mayot et al., 2020;Zhao et al., 2022). Sea ice retreat and thinning increases net solar irradiation and can thereby drive enhanced primary production.
Furthermore, sea ice melt leads to spring stratification of the upper water column. On the one hand, this can drive biological productivity by confining phytoplankton to the light-rich surface layers (Cherkasheva et al., 2014;Mayot et al., 2020). On the other hand, in nutrient-limited settings, stronger stratification may lead to more prevalent nutrient limitation as vertical fluxes of nutrients are reduced. The drivers of primary production have been shown to vary regionally across the Arctic Ocean (Arrigo et al., 2017;Frey et al., 2019;Krisch et al., 2020;Moreau et al., 2019;Randelhoff et al., 2019;.
The major export of Arctic water into the North Atlantic runs through the Fram Strait, largely in the form of solid sea ice (Haine et al., 2015). The role of sea ice and stratification on phytoplankton productivity here are likely distinct from other regions and may continue to shift with increasing temperatures and sea ice loss. The role of sea ice meltwater in determining bloom dynamics in Fram Strait and in the Arctic in general has come under interest in recent years (Lester et al., 2021;Lewis et al., 2020;Mayot et al., 2020;Moreau et al., 2019), as there appears to be a significant relationship between the distribution of sea ice meltwater and that of phytoplankton blooms. In this study, we explore this link in more detail. We note that the role glacial meltwater plays in driving bloom processes has also been discussed extensively (Arrigo et al., 2017;Bhatia et al., 2021;Gerringa et al., 2012;Krisch et al., 2020), with an emerging consensus that glacial waters can act as an important source of iron to often iron-limited polar marine ecosystems (Dierssen et al., 2002;Forsch et al., 2021). Furthermore, glacial discharge can drive upwelling of nutrient-rich bottom waters (Oliver et al., 2020(Oliver et al., , 2023. Alongside changes in the magnitude of phytoplankton blooms, satellite records of ocean color have also recorded shifts in the bloom timing across the Arctic (e.g., Harrison et al., 2013;Kahru et al., 2011;Manizza et al., 2023), with increased occurrence of fall blooms (e.g., Zhao et al., 2022) and signs of earlier spring blooms in the Fram Strait in particular. This can be seen in Figure 1, where the positive trend in chl-a has been notably steeper in May than in June in recent years.
In May 2019, an oceanographic research cruise conducted an interdisciplinary study of the Fram Strait, sampling both ice-edge and pelagic habitats. The timing of this cruise coincided with a record phytoplankton bloom event in the area. Here, we use the term "bloom" to simply describe a period of increasing chlorophyll a concentration. F I G U R E 1 (a) Trends of May-June mean surface chlorophyll a concentrations over the period 2000-2020 from the merged Ocean-Colour Climate Change Initiative (OC-CCI) product (Sathyendranath et al., 2019). Sea ice concentration (SIC) contour lines shown are averaged over the same time period and computed from the NASA Team processing algorithm of the SMMR-SSM/I satellite record (Cavalieri et al., 1996). The Fram Strait study area is enclosed in a black box. (b) Monthly OC-CCI chlorophyll a time series (green), averaged over the Fram Strait region (black box in panel a). Highlighted are the typical peak bloom periods in May (blue) and June (red). (c) May and June chlorophyll a, with yearly values as as in panel (b) (faint markers), and 3-year running mean overlaid (thick lines). The field work time period is indicated by the gray vertical line in panels (b, c). Note chl-a in panels (b and c) is shown on a logarithmic scale.
2 | DATA AND ME THODS

| In situ data
During 7-21 May 2019, we conducted an oceanographic research cruise on the MY Arctic Sunrise in the Fram Strait region between Svalbard and Greenland. The ship track consisted of a transect along the 79° N parallel followed by a series of sample collection stations along the sea ice edge. The ice edge was not compacted and featured locally shifting concentrations but on a larger scale remained roughly stationary over the study period (ship track and station locations are shown in Figure S2). The station locations ranged from within the marginal ice zone (sea ice concentration, SIC ≤ 50%) to roughly 100 km from the sea ice edge (SIC = 0). Measurements focused on physical environmental data and bottom-up processes.
No data were analyzed on phytoplankton growth or grazing and loss rates. Therefore, all chlorophyll data are approximations of phytoplankton biomass, not primary production.

| CTD measurements
At nine collection stations along the cruise track a rosette equipped with Niskin bottles and carrying a Sea-Bird Scientific SBE25 CTD measuring temperature, conductivity, and pressure was lowered by winch from the starboard side of the deck to maximum depth of ≈300 m. This primary CTD was also equipped with a SBE43 oxygen sensor, a Seabird FLNTU sensor measuring turbidity and in vivo chlorophyll a fluorescence (henceforth chl-a fluorescence), and a Biospherical/Licor PAR (photosynthetically available radiation) sensor. At 26 stations, a Sea-Bird SBE19plus SeaCAT CTD equipped with a SBE43 oxygen sensor, Seapoint Fluorometer measuring chl-a fluorescence, and a Seapoint Turbidity sensor was deployed to provide redundancy and additional coverage, with maximum depths ranging from 30 to 300 m. A handheld SonTek CastAway CTD was also deployed at most stations to provide further validation of profile results. Sensors were factory calibrated prior to the cruise. Chl-a fluorescence, converted to μg/L using factory calibration coefficients, was not corrected for non-photochemical quenching (NPQ; e.g., Xing et al., 2018). NPQ is a reduction in cellular fluorescent emission resulting from cell exposure to high light levels in near-surface waters (i.e., not associated with a decrease in phytoplankton biomass; Kiefer, 1973), and the near-surface (upper ≈15 m) chl-a fluorescence data discussed in the Supporting Information therefore represent lower bounds on these values.

| Pigment and nutrient measurements
Water was collected via Niskin bottles attached to the primary CTD rosette. Surface water samples were also collected using a bucket lowered from the deck of the ship. Phytoplankton from water samples collected for pigment analysis were immediately filtered onto 25 mm GF/F filters (nominal pore size 0.7 μm) and stored in cryovials, which were then frozen at −20°C while at sea and at −80°C on land for later analysis. Samples for 6 locations were analyzed via high-  (Hood et al., 2010;Hydes et al., 2010).
Other measurements were also collected, including sea ice cores, water samples for trace metal analysis, zooplankton casts and marine mammal observations, but are not reported on here.

| Remote sensing observations
To examine longer-term patterns, we focus primarily on the period from 2000 to 2020, for which there is high-resolution spatial and temporal data coverage over the study region. Chlorophyll a concentration data from the Ocean-Colour Climate Change Initiative (OC-CCI) merged weekly and monthly products (henceforth satellite chl-a) were analyzed for the period from 2000 to 2020 (Sathyendranath et al., 2019). Since the satellite chl-a data only contain information on surface concentrations, the associated discussion is limited to surface processes, and we are not able to shed light on the role of depth variations in this regard. SIC for the Arctic and Fram Strait spanning the years 1980-2020 were obtained from SMMR-SSM/I data using the NASA Team product (Cavalieri et al., 1996)

| State-estimate data
We investigated Fram Strait-wide trends of satellite chl-a and a number of climatic variables. Monthly net surface freshwater flux into the ocean from sea ice melt (FWF), sea surface salinity (SSS) and SST were taken from the "Estimating the Circulation and Climate of the Ocean Version 5" (ECCOv5) state estimate, which is computed at a nominal horizontal resolution of 1/3° (Zhang et al., 2018). We note that ECCOv5 spans the years 1992-2017, so the final 3 years of the study period (2018-2020) are not covered. We also examined the dependence of satellite chl-a, FWF, SSS, and insolation on the distance from the sea ice edge. To do so, we determined the distance to the sea ice edge (defined as the SIC = 0 contour) for each grid box on all monthly distribution maps. We then computed the values of satellite chl-a, FWF, SSS, and insolation at each location and binned the data as a function of distance from the ice edge (using 10 km bins). Finally, we averaged over the years 2000-2017 to get decadal-mean monthly curves for April, May, and June as functions of distance from the sea ice edge. We also considered SST from the observational NOAA 0.25-degree Daily Optimum Interpolation Sea Surface Temperature (OISST) product (Huang et al., 2020), to further validate our findings from the ECCO output. Finally, insolation was estimated as net surface shortwave radiation computed from the CERES EBAF Ed4.1 data set (Loeb et al., 2018).

| The May 2019 bloom event
The  meltwater and ambient ocean waters (Gade, 1979).
With regard to nutrient profiles, we find that as to be expected, (The sea ice contours did not vary greatly over the 2-week cruise period; a corresponding map for May 16 is shown in Figure S2). . Station locations correspond to those labeled in Figure 3.
T-S diagrams of before and after-storm conditions do not exhibit clearly distinct water mass regimes ( Figure S6).

| Decadal trends in sea ice, freshwater flux, and bloom characteristics
Sea ice has been in broad decline in the Arctic Ocean over the past four decades (Fetterer et al., 2017, see also   temperatures have increased in the region over the study period as discussed above ( Figure S7).
There has been no measurable trend in the absorbed sunlight in the Fram Strait near-ice environment between 2000 and 2020, as can be seen from CERES insolation timeseries shown in Figure S8. This is consistent with the steady SIC values in the region, as discussed above.
The regional spatial patterns of the increased FWF and reduced SSS (from ECCO, Figure 7a slightly larger (shown in Figure 1a).
In addition to regional and temporal patterns over the last two decades, we explored bloom characteristics in relation to distance from the ice edge using satellite chl-a and ECCO state-estimate data ( Figure 8). Figure 8a shows how satellite-derived chl-a concentrations Naturally, FWF is confined to the area that has at least partial sea ice cover, with peak FWF at roughly 50% sea ice concentration, which on average is located ≈80 km from the open ocean. This is accompanied by a freshening of the surface layer. Figure 8c shows SSS versus distance from the ice edge, illustrating that the impact of the FWF (injected within the ice-covered zone) is clearly observed in the form of depressed SSS to about 400 km from the ice edge. This freshening intensifies through the spring, with the maximum SSS perturbation occurring in June near the 50% sea ice contour. Finally, we calculate the absorbed solar radiation, shown in Figure 8d, from CERES data.
As expected, light intensity increases overall from April through June.
And due to the reflectivity of the sea ice, absorbed insolation is suppressed in the presence of ice. We observe an approximately linear reduction with increased SIC in the marginal ice zone (Figure 8d). For all months, the suppression of insolation at the 70% SIC location compared to the SIC = 0 line is approximately one third.

| D ISCUSS I ON AND CON CLUS I ON S
Multiple physical, chemical, and biological variables influence phytoplankton concentrations across the Arctic and the relationships between each of these drivers are complex and appear to be rapidly changing in this polar environment. Increasingly, the ecological dynamics at play in Fram Strait are emerging as distinct from other regions in the Arctic and specifically, are found to be strongly linked to the advection and melting of sea ice. Yet, changes in temperatures, wind, ice extent, ice export, freshwater transport, and nutrients make projections of phytoplankton production and phenology a moving target. While we cannot examine all of the variables that influence primary production here, below we discuss some of the environmental and ecological controls on production in the Fram Strait.
For much of the marginal Arctic Ocean, increasing phytoplankton concentration appears to coincide with a reduction in sea ice coverage (Arrigo et al., 2008;Arrigo & van Dijken, 2015;Lewis et al., 2020). In many regions, this reduction in ice extent is thought to provide increased habitat, light availability, and growing season for phytoplankton (Arrigo & van Dijken, 2015). In addition, this ice retreat is linked to increased mixing at the shelf break (Carmack & Chapman, 2003) that may deliver additional nutrients. In other studies within the Fram Strait (Mayot et al., 2020), and elsewhere in the Arctic (Arrigo & Van Dijken, 2003), it has been shown that high phytoplankton concentrations and primary productivity are often light limited (Schourup-Kristensen et al., 2021). Within the Fram Strait, little if any reduction in ice coverage has occurred ( Figure 6) and absorbed shortwave radiation has not increased ( Figure S8). Yet, we observe some of the most significant increases in satellite chl-a concentrations within the Arctic. We therefore suggest that here, enhanced light availability due to increased open water can be ruled out as a main driver of increased chlorophyll a.
Yet, light availability is also strongly influenced by near-surface stratification driven through temperature and salinity gradients. The importance of stratification in this region has been examined by previous studies. For example, Mayot et al. (2020) found that years of high sea ice flux through Fram Strait are correlated with stronger surface stratification, earlier blooming, and intensified primary productivity. The modeling study by Lester et al. (2021) considered an environment that is not nutrient limited and reduced bloom dynamics to just insolation, stratification, and advection. Simplifying bloom dynamics to these three variables, the authors found good agreement with observed relationships between chlorophyll concentration and distance from the ice edge. The importance of stratification in the Fram Strait is supported by the in situ and satellite-based results presented here, particularly before the storm event: In May of 2019, chl-a concentrations were correlated with salinity-driven stratification, declining with increasing salinity (Figures 3 and 4).
The in situ observations are consistent with decadal-scale trends in the region whereby increased sea ice import into and decreased export out of the Fram Strait, together with background warming ( Figure S7), has led to increased meltwater flux and a freshening of surface waters. This corresponds to a recently proposed regime shift toward thinner sea ice and faster ice advection into the region that occurred around 2007 (Sumata et al., 2023). We found that these physical changes have been accompanied by biological ones, with clear spatial correlations between freshwater flux and chlorophyll concentration trends. Thus, in the Fram Strait, freshwater flux, more than expansion of open water, appears to be a strong driver of increases in chlorophyll concentrations.
Other ways in which sea ice could drive blooming is as a source of nutrients such as iron or as a direct source of sympagic allochthonous phytoplankton species which seed the bloom. Krisch et al. (2020) showed that the eastern Fram Strait near Svalbard in particular can be iron limited. Findings by Arrigo et al. (2017) showed that the relationship between Greenland Sea blooming and glacial meltwater was one in which the meltwater provided an important nutrient source, and Moreau et al. (2019) found that blooming within several Southern Ocean polynyas was primarily driven by sea ice meltwater and circumpolar deep water concentrations, and not stratification. Lewis et al. (2020) also showed that primary production in the Arctic Ocean, while typically nitrogen limited, has more recently been driven by increases in phytoplankton biomass supported by new nutrient sources.  (Vernet et al., 2019), contributing more to biomass at a given location than biomass grown in situ. The more eastern (poststorm), sampling stations may in this way have been more highly influenced by northward flowing Atlantic waters, which may have resulted in different water column characteristics (including nutrient concentration, phytoplankton community composition, and productivity rates).
The controls on phytoplankton biomass in the Arctic are numerous and disentangling them to understand current mechanisms and predict future ones is challenging. The current study only focuses on some of these controls, namely bottom-up processes, and thus correlations and interpretations made here must be considered with these limitations in mind. Previous work has also documented how quickly the relative importance of ecological drivers on productivity can transition. For example, Lewis et al. (2020) illustrated a temporal shift in the relative importance of open water and nutrients in driving primary production over two decades. While further work is needed, this study provides support for a critical link between sea ice export and melt and chlorophyll concentration in the Fram Strait. Combining in situ, satellite, and state-estimate data, we find that sea ice drift into the Fram Strait has increased, sea ice drift out has decreased, warm temperatures have resulted in increased melting and freshwater flux, and this is spatially and temporally correlated with increases in chlorophyll concentrations. Further research will be needed to predict whether this trend will continue as sea ice is predicted to decline across the Arctic, including the Fram Strait. While increased freshwater flux appears, from the present data, to enhance chlorophyll concentrations, if freshening continues, the resulting stratification may also increase nutrient limitation. Thus, wind events could become increasingly important in eroding such structure and increasing sustained or secondary bloom events. Further work is needed, particularly focused on stratification and wind-induced mixing, nutrient sources, and nutrient utilization to clarify the mechanisms underlying the correlation between sea ice melt and phytoplankton concentrations in the Fram Strait.

CO N FLI C T O F I NTE R E S T S TATE M E NT
The authors declare no potential conflict of interest.

DATA AVA I L A B I L I T Y S TAT E M E N T
The data that support the findings of this study are openly available in GitHub at https://github.com/tillw agner and https://doi. org/10.5281/zenodo.7992493