Regional setting

Currents in the Pacific Arctic region are dominated by a northward advection of water crossing the Bering shelf, converging in the Bering Strait and moving into the Chukchi Sea (Fig 2A). Different water mass components influence the transfer of associated heat content, organic matter and nutrients to the ecosystem [38–40]. There are three primary current pathways during the open water season: the nutrient-rich Anadyr Current to the west, Bering Sea water with summer and winter variants, and the warmer, nutrient-poor and seasonal Alaska Coastal Current to the east [38, 41, 42]. The northward flowing hydrography brings nutrient rich Pacific waters into the euphotic zone and supports persistent localized in situ production and advection and deposition of organic carbon to the benthos, and this productivity plays a role in the maintenance of benthic biological “hot spots” in the Bering Strait region [15].

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 2. Study site in the Pacific Arctic region.

A) The surface sediment sampling locations in the northern Bering and Chukchi Seas occurred within the framework of the Distributed Biological Observatory (DBO) regions (black boxes). The DBO regions in this study from south to north include: The St. Lawrence Island polynya (SLIP), Chirikov Basin (CHIR), southeast Chukchi Sea (SECS), northeast Chukchi Sea (NECS) and Barrow Canyon (BARC). B) The northeast Chukchi Sea region with the locations of the Chukchi Ecosystem Observatory (CEO) moored sediment trap and Haps core locations. Reprinted from Ocean Data View under a CC BY license, with permission from R. Schlitzer, original copyright 2020.

https://doi.org/10.1371/journal.pone.0231178.g002

The shallow shelf that spans from the northern Bering Sea to the northeast Chukchi Sea averages 40 meters in depth and has in recent years been seasonally ice covered for 0–3 months in the Bering Sea and 6–9 months in the Chukchi Sea [32]. The maximum median sea ice extent (1981–2010) has historically occurred in March in the northern Bering Sea and the minimum ice extent in September in the Chukchi Sea near the shelf break (Fig 2A). More recently, the minimum extent has shifted northwards away from the shelf break into the basin. The delayed freeze up in the Chukchi Sea ultimately impacts the winter sea ice extent and shifts the sea ice coverage in this entire region [36]. Throughout the sea ice cycle, primary production typically initiates with the ice algae bloom prior to sea ice melt, followed by or possibly partially seeding a pelagic phytoplankton bloom [8, 14, 43].

Permitting

No national or international permitting was required as part of the sample collection efforts. Concerns regarding sampling in waters near Indigenous subsistence hunting areas was addressed by provision of cruise plans to the Arctic Waterways Safety Committee and some samples were imported into the United States from Canada using a US Fish and Wildlife Service Declaration for Importation or Exportation of Fish or Wildlife (USFWS Form 3–177).

Sediment trap deployment

A sequential sediment trap (Hydro-Bios, Germany; 24 cups) was moored at 37 m depth, 8 m above the seafloor, as part of the Chukchi Ecosystem Observatory (CEO) located on the southeastern flank of Hanna Shoal (71.6°N 161.5°W, Fig 2B). The sediment trap was deployed in August 2015 and recovered in August 2016. Collection cups rotated at pre-programmed intervals ranging from one week during spring and summer to one month during winter. The last sample was excluded from the study as the sediment trap was recovered before the completion of the last rotation when the cup was still open. Before deployment, collection cups were filled with filtered seawater, adjusted to a salinity of 38 with NaCl to create a solution denser than ambient seawater to ensure material remained in the cup while open, and poisoned with formalin (4% final solution) to preserve samples during deployment and after recovery. In preservation tests of marine samples, formalin did not affect HBI proportions or indices relative to wet/dry freezing [44]. Trap samples were stored in the dark at room temperature until analysis, but we note that the effects of storage temperature on HBI degradation in formalin preserved samples have not been investigated [45].

Diatom identification and quantification

Subsamples (0.1–3 mL) from the sediment trap bottles were adjusted to a volume of 3 mL with filtered seawater for the enumeration and identification of algal cells in an Utermöhl chamber [46]. A minimum of 300 phytoplankton cells were counted and identified to the lowest taxonomic level possible by inverted light microscopy at 100X, 200X or 400X depending on cell size using the Utermöhl method [46]. Empty algal cells (without chloroplasts) were distinguished from intact cells (with chloroplasts) assumed to be alive at the time of collection and resting spores [11]. Algal measurements were converted to daily fluxes depending on the subsampled volume and open cup duration of each sample.

Two sea ice exclusive diatom species, Nitzschia frigida (Grunow in Cleve and Grunow) and Melosira arctica (Dickie), were selected as indicators of the ice algae bloom. The Gyrosigma/Pleurosigma/Haslea group were selected to be the source of sympagic HBIs based on the currently known species that produce these lipids, which include Pleurosigma stuxbergii var. rhomboides (Cleve in Cleve and Grunow) Cleve, Haslea kjellmani (Cleve) Simonsen, H. crucigeroides (Hustedt) Simonsen, and H. spicula (Hickie) Lange-Bertalot [19, 47]. This broader group is not exclusively associated with sea ice. Another caveat is that Pleurosigma spp. includes species that produce the pelagic HBI III, including P. intermedium [48]. The diatom genera Rhizosolenia was selected as an indicator of the potential sources of HBI III. Species known to produce HBI III include R. hebetata, R. polydactyla f. polydactyla and R. setigera [20]. A more detailed analysis of the major diatom taxa and fluxes is discussed in Lalande et al. [49]

Subsamples for chlorophyll a (chl a) measurements were filtered onto GF/F filters (0.7 μm), extracted in 90% acetone for 24 h at -20°C and measured on a Turner Design Model 10-AU fluorometer following the methods outlined in Welschmeyer [50]. Samples were kept cool and in the dark prior to chl a measurements.

Surface sediment collection

Surface sediment sampling was conducted on six annual expeditions from 2012 to 2017 on board the USCGC Healy (HLY; 2012, 2013, 2017) and the CCGS Sir Wilfrid Laurier (SWL; 2014, 2015, 2016). Sample collections from 2014 to 2017 were made at Distributed Biological Observatory (DBO) program sites (https://www.pmel.noaa.gov/dbo/), where long-term monitoring has been established in the Bering, Chukchi, and Beaufort seas [16, 51]. These sites are in the vicinity of five DBO long-term sampling station grids that were selected on the basis of having high productivity and/or biodiversity specifically in the north Bering Sea, the St. Lawrence Island polynya (SLIP), and the Chirikov Basin (CHIR), and north of Bering Strait, the southeast Chukchi Sea (SECS), the northeast Chukchi Sea (NECS) and Barrow Canyon (BARC) (Fig 2A). Sample collection in 2012 and 2013 focused primarily on the NECS region near Hanna Shoal (Fig 2B), but extended to all of the long-term Bering and Chukchi DBO benthic sampling sites in other sampling years. Surface sediments were collected by a van Veen grab (0.1 m²), with a trap door on the top that was opened prior to opening the grab in order to obtain relatively undisturbed sediments that were assayed for total organic carbon (TOC) and HBIs in the surface sediments. Samples were stored frozen (-20°C) until analysis.

Sediment core collection

Sediment cores were collected using a multi-HAPS corer (area = 133 cm²) with stainless steel barrels and acrylic inserts deployed from the USCGC Healy in 2017 at station DBO 4.6 (71.62°N 163.77°W) and station NNE-14 (73.29°N 160.04°W, Fig 2B). A single core was collected at station DBO 4.6 on the shelf from a bottom depth of 43 m (Table 1). This core was sectioned shipboard for the first two centimeters at 1-cm intervals and the remaining length of the core at 2-cm intervals. A pair of cores were collected at station NNE-14 (1200 m depth; Table 1). Both cores were sectioned in 1 cm intervals at sea. Sections from core sections were immediately frozen and stored at -20°C until analysis.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 1. Sediment coring locations and parameters.

https://doi.org/10.1371/journal.pone.0231178.t001

Sediment core radiocesium measurements

The sectioned core from NNE-14 was analyzed for radiocesium (¹³⁷Cs) by gamma spectroscopy using a Canberra GR4020/S reverse electrode closed-end coaxial detector at the Chesapeake Biological Laboratory following established protocols [52]. Sedimentation data from another core collected in 2009 at station UTX13-23 (71.39°N 166.28°W), approximately 50 nautical miles from DBO4.6, was used in lieu of gamma analysis of the single core from DBO4.6 [52], which was instead used for analysis of IP₂₅ and other biomarkers. The ¹³⁷Cs profile from core UTX13-23, which has been presented elsewhere [52] was used as a sedimentation proxy for DBO4.6, based on similarities in deposition [52]. DBO4.6 and UTX13-23 have similar grain sizes (50–75% ≥ 5 phi) and TOC (0.5–1%) [52], which have been found to be significantly correlated with radiocesium activity in surface sediments [53]. Additionally, we expected DBO4.6 to be highly influenced by bioturbation, as are most cores collected from this area of the Chukchi shelf [52]. This substitution was expected to be reasonable for the purpose of comparing cores collected in the biologically productive NECS region on the shelf relative to a core collected on the less productive continental slope (NNE-14).

Biomarker extraction

HBIs were extracted from surface sediment samples (n = 184; S1 Table), sediment trap sample cups (n = 23), and two sectioned sediment cores. Surface sediment and core samples were freeze dried for 48 hours, homogenized by mortar and pestle, followed by subsampling of approximately 1 g dried sediment. Sample cups from the sediment trap were gently mixed before subsamples were extracted with a modified pipette to enable the collection of larger particles for the measurement of HBIs. Sample volumes varied from 10 to 30 mL to accommodate the fluctuating particle flux through the year. These aliquots were filtered on Whatman GF/F filters (0.7 μm) and rinsed with deionized water. The filters were frozen overnight in petri dishes and placed into 8 mL vials for biomarker extraction.

HBIs were extracted following the methods of Belt et al. [54] and Brown et al. [29]. An internal standard (10 μL) of 9-octylheptadec-8-ene (9-OHD, 1 μg mL^-1) was added to the sample before extraction to facilitate yield quantification. Samples were first saponified in a methanolic KOH solution and heated at 70°C for one hour. Hexane (4 mL) was added to the saponified solution, vortexed, and centrifuged for 3 minutes at 2500 RPM for three iterations. The supernatant with the non-saponifiable lipids (NSLs) was transferred to clean glass vials and dried under a gentle N₂ stream to remove traces of residual methanolic KOH.

Elemental sulfur was removed from the sediment samples due to analytical interference with HBI III (m/z 346.3). This was accomplished by re-suspending the NSLs in 2 mL hexane with the addition of 1 mL of a tetrabutylammonium (TBA) sulfite reagent and 2 mL of 2-propanol. The solution was shaken for one minute and repeated, if necessary, until a precipitate formed. MilliQ water (3 mL) was added and the mixture centrifuged for 2 minutes at 2500 RPM. The hexane layer was removed into a clean vial with the hexane extraction and centrifugation repeated three times. The extract was dried under a gentle N₂ stream at 25°C and removed immediately once the solvent had evaporated.

Following sulfur removal, the extracts were re-suspended in hexane and fractionated using open column silica gel chromatography. The non-polar lipids containing the HBIs were eluted while the polar compounds were retained on the column. The eluted compounds were dried under N₂. 50 μL of hexane was added twice to the dried extract and transferred to amber chromatography vials.

Biomarker analysis

The extracts were analyzed using an Agilent 7890A gas chromatograph (GC) coupled with a 5975 series mass selective detector (MSD) following methods outlined by Belt et al. [54]. Samples were analyzed on an Agilent HP-5ms column (30 m x 0.25 mm x 0.25 μm). The oven temperature was programmed to ramp up from 40°C to 300°C at 10°C/minute with a 10-minute isothermal period at 300°C. HBIs were identified using both total ion current (TIC) and selective ion monitoring (SIM) techniques. TIC chromatograms and mass spectral output data were used to identify individual HBIs while SIM chromatograms were used to quantify the abundances by peak integration with ChemStation software. A purified standard of known IP₂₅ concentration was used to confirm the mass spectra, retention time and retention index (RI). Authentic HBI standards were also measured alongside the internal standard 9-OHD to determine the instrument response factor (RF, Table 2). For experimental purposes, samples were reanalyzed on an Agilent DB-5ms column (30 m x 0.25 mm x 0.25 μm) to determine the column-specific retention indices of these compounds. The HBIs were identified by their mass ions and RI including IP₂₅ (m/z 350.3), HBI II (m/z 348.3) and HBI III (m/z 346.3). To the best of our knowledge, the RIs for these HBIs have not been previously reported in the literature on a DB-5ms column (Table 2). A procedural blank was run every 9^th sample.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 2. HBI parameters for gas chromatograph-mass spectrometry.

https://doi.org/10.1371/journal.pone.0231178.t002

Individual HBI concentrations in the surface sediment samples were normalized by TOC on an organic gram weight basis (S1 Table). TOC data from HLY12 (2012), HLY1702 (2017) and SWL 14–16 (2014–2016) cruises were accessed through the National Science Foundation’s Arctic Data Center [55–59]. TOC data from the HLY13 (2013) cruise are available through another data archive, the National Oceanic and Atmospheric Administration National Centers for Environmental Information [60]. HBI concentrations from sediment trap samples were converted to daily fluxes depending on the subsampled volume and open cup duration of each sample and integrated over a 365-day period to annual fluxes.

The relative abundances of the sympagic HBIs (IP₂₅ and HBI II) to the pelagic HBI (HBI III), were quantified in order to determine the proportions attributable to different organic carbon sources. An HBI fingerprinting index, termed “H-Print”, was used to estimate the relative organic carbon contributions of sea ice algae versus phytoplankton sources [29]. The H-Print method (Eq 1), is calculated using the relative abundances of IP₂₅, HBI II and HBI III, as determined by GC-MSD methods: (1) The estimated organic carbon contribution resulting from the H-Print analysis varies from 0% to 100%, with lower values indicative of proportionally greater sympagic inputs and higher values indicative of proportionally lower sympagic inputs (i.e. substantial pelagic diatom sources). Analytical error from replicate control tests was determined to be less than 14% (relative standard deviation, RSD) for HBI quantification and less than 12% (RSD) for H-Print values.

Sea ice concentration and snow cover

At the sediment trap location, daily averaged sea ice concentrations were retrieved at a 12.5-km resolution from the National Snow and Ice Data Center (NSIDC, https://nsidc.org) using the Defense Meteorological Satellite Program (DMSP) Special Sensor Microwave Imager/Sounder (SSMIS) passive microwave data. Snow depth on top of sea ice was retrieved at a 25-km resolution from the Northern Hemisphere snow depth files derived from the SSMIS data. Daily sea ice concentration and snow depth were averaged for a delimited region above the mooring (44 x 44 km; 71.4–71.8°N; 161.4–161.9°W).

The spring sea ice concentration (SpSIC) for each year of the study was averaged from monthly (April-June) sea ice concentration using DMSP SSMIS data [28]. The mean sea ice concentration at each of the sediment sample locations was extracted from the pixel containing the station location. The sea ice break-up dates were determined at each of the surface sediment sample locations. The sea ice break-up date was defined as the date when the pixel containing the station registered two consecutive days of sea ice concentration ≤15%, a common threshold for open water conditions in sea ice studies [32]. The sea ice break-up date was then subtracted from the sample collection date to determine the ice-free period prior to sampling at each specific location of interest.

Statistical analysis

Spatial analysis of the biomarker concentrations and H-Print values were conducted with ODV using DIVA (Data-Interpolating Variational Analysis) gridding methods [61]. All other statistical analyses were performed in R v. 3.6.1 [62] and plots were produced using the package ggplot2 [63]. Multiple linear regressions were used to investigate correlations between sea ice data and H-Print. One-way ANOVA testing and Tukey Honest Significant Difference (HSD) multiple pairwise comparisons were used to analyze the differences in relative HBI concentrations by DBO region. Principal components analysis (PCA) was used to analyze the impact of individual relative biomarker abundances at each location. Pearson product moment correlations were used to test for relationships among biomarker, diatom and chl a fluxes.

Annual cycle of sea ice concentration, biomarker and diatom fluxes

At the CEO mooring site, open water conditions persisted from the initial deployment in mid-August through mid-November 2015 (Fig 3A). The increase in sea ice concentration in late November 2015 indicated a rapid sea ice freeze-up and the site remained ice-covered through mid-July 2016 (Fig 3A). Snowmelt first occurred in May 2016 and sea ice melt initiated in June 2016 (Fig 3A). Some sea ice (>15%) however remained present above the sediment trap until the end of deployment.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 3. Sea ice concentration, snow depth, and annual fluxes of diatoms and biomarkers at the Chukchi Ecosystem Observatory 2015–2016.

The parameters measured from the CEO sediment trap from August 2015 –August 2016 included: A) sea ice concentration (%) and snow depth (cm). The blue-dashed line indicates the 15% sea ice concentration threshold defining open water, B) chlorophyll a fluxes (mg m^-2d^-1) and POC fluxes (g C m^-2 d^-1). POC and chl a data from Lalande et al. 2020 [49] C) Nitzschia frigida and Melosira arctica fluxes (sea ice exclusive diatoms), D) Gyrosigma/Haslea/Pleurosigma fluxes (group containing HBI-producing species), E) Rhizosolenia spp. fluxes (group containing HBI III-producing species), F) IP₂₅ fluxes (ng m^-2d^-1), and G) HBI III fluxes (ng m^-2d^-1). All panels indicate the ice-covered period within the blue shaded boxes and the onset of snow melt is depicted by the red-dashed line.

https://doi.org/10.1371/journal.pone.0231178.g003

Chl a fluxes ranged from 1.5 to 1.9 mg m^-2 d^-1 from August through September 2015. Chl a levels remained relatively low (below 0.2 mg m^-2 d^-1) from December 2015 through April 2016. Chl a rapidly increased in late June 2016 and the maximum flux occurred in late July 2016 at 4.9 mg m^-2 d^-1 (Fig 3B). Similarly, POC fluxes were highest from August through September 2015 (1.09 to 1.18 g C m^-2 d^-1), a decline through the winter months and steady increase beginning in April 2016. The POC flux reached 1.04 g C m^-2 d^-1 in late July 2016 before the trap was recovered (Fig 3B).

The sympagic diatom fluxes are indicated by N. frigida and M. arctica (Fig 3C). N. frigida was first detected in the sediment trap in early April 2016, increased through late May 2016 and was no longer detected in early June 2016. N. frigida reappeared in mid-June and the maximum flux occurred in late June 2016. M. arctica was detected in the trap in early September 2015 and did not reappear until the maximum flux occurred in June 2016, corresponding to the peak flux for the exclusively sympagic species. M. arctica resting spores were present in August and September 2015, reappeared in May and remained consistently present until the end of the deployment. The Gyrosigma/Pleurosigma/Haslea group was detected in the trap throughout most of the year with the exception of early September through early November 2015 (Fig 3D). This group steadily increased starting in April 2016 and reaches a maximum in early July 2016. Rhizosolenia fluxes were only detected as intact cells from September through November 2015 (Fig 3E) although there were substantial fluxes of fragments year round (data not shown). The peak flux occurred in mid-November 2015.

IP₂₅ was detected throughout the entire sampling period (Fig 3F). IP₂₅ fluxes in the initial winter months (December 2015 through February 2016) occurred without the corresponding diatom groups recorded in the traps (Fig 3D and 3F). IP₂₅ fluxes began to increase in mid-May and reached a maximum in early July 2016 at 1331 ng m^-2 d^-1 (Fig 3F and Table 3). IP₂₅ sharply declined to 119 ng m^-2 d^-1 in late July (Table 3). This precipitous decline coincided with the peak chl a flux (Fig 3B). Overall, IP₂₅ fluxes mirrored the export of the Gyrosigma/Pleurosigma/Haslea taxonomic group. HBI III was also detected throughout the year (Fig 3G). The HBI III peak flux corresponded to the maximum Rhizosolenia spp. flux. HBI III fluxes reached a maximum flux of 799 ng m^-2 d^-1 in September 2015 (Fig 3G and Table 3). As indicated by the H-Print index, the sympagic diatom signal was present but low from September 2015 to late November 2015 with H-Print values ranging from 48–70% (Table 3), representing a mixed to pelagic diatom composition. H-Print values indicated a strong sympagic diatom signal in late March through late July 2016, with the strongest sympagic indicators during mid-May, late June and early July. The annual flux of IP₂₅ was 60 μg m^-2 yr^-1, HBI II fluxes were 278 μg m^-2 yr^-1, and HBI III fluxes reached 87 μg m^-2 yr^-1 (Table 3).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 3. Sediment trap summary data.

https://doi.org/10.1371/journal.pone.0231178.t003

A Pearson correlation test was conducted on the assigned diatom groupings, chl a fluxes, and HBI fluxes (Table 4). The group containing sympagic-HBI producing species (Gyrosigma/Pleurosigma/Haslea) was strongly correlated with IP₂₅ fluxes (r = 0.80, p<0.001). The group containing pelagic-HBI producing species (Rhizosolenia spp.) was not significantly correlated with HBI III fluxes. Chl a was positively correlated with IP₂₅ fluxes (r = 0.60, p<0.01) and Gyrosigma/Pleurosigma/Haslea spp. (r = 0.56, p< 0.01). IP₂₅ and HBI III were also positively correlated (r = 0.61, p<0.01). IP₂₅ was positively correlated with the sea ice diatom flux (N. frigida and M. arctica, r = 0.58, p<0.05).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 4. Pearson product-moment correlation matrix for flux data.

https://doi.org/10.1371/journal.pone.0231178.t004

Distribution and variation of biomarker deposition

IP₂₅ was detected in all of the surface sediment samples (Fig 4). Localized high concentrations occurred in the NECS and BARC regions in 2013 and 2017 and in the Chirikov Basin in 2016. IP₂₅ concentrations were generally higher (>3 μg g^-1 TOC) overall in the NECS and BARC regions relative to the lower latitude DBO regions. The SLIP region in 2015 was an exception with IP₂₅ concentrations reaching 12 μg g^-1 TOC at the SLIP3 station (S1 Table), which was the highest concentration observed of all years and stations. IP₂₅ data were only available for the SLIP region from 2015 through 2017, however, the concentration decreased over this time. Values exceeded 6 μg g^-1 TOC in four samples total (8–12 μg g^-1 TOC), which were determined statistically to be outliers by the IQR (Interquartile Range) method, and were incorporated as the maximum value (6 μg g^-1 TOC) rather than omitted for DIVA gridding. HBI III values were relatively consistent from year to year, with the highest concentrations found in the southeast Chukchi Sea (SECS) and northern Bering Sea (SLIP and CHIR) and minimal concentrations in the NECS (Fig 4).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 4. IP₂₅ and HBI III biomarker distributions.

Spatial distribution of the relative abundances of IP₂₅ and HBI III concentrations (μg g^-1 TOC) in surface sediments from 2012–2017. The white and grey bounding boxes indicate the DBO regions from south to north (SLIP, CHIR, SECS, NECS and BARC). Not all sampling stations and DBO regions were able to be occupied every year due to sea ice or weather, indicated by grey boxes (no data collected). IP₂₅ and HBI III values were used as sympagic and pelagic diatom proxies, respectively, for the H-Print analysis. Reprinted from Ocean Data View under a CC BY license, with permission from R. Schlitzer, original copyright 2020.

https://doi.org/10.1371/journal.pone.0231178.g004

The spatial distribution of H-Print index followed the general pattern of spring sea ice retreat each season, with weaker sympagic signatures (H-Print > 60%) in the northern Bering Sea, particularly south of St. Lawrence Island and in the Chirikov Basin (Fig 5). The NECS and BARC regions displayed an elevated to moderate sea ice signal each year, with mean H-Print values ranging from ~21–59% for NECS and ~38–49% for BARC (Table 5). Spring sea ice concentrations derived from satellite data indicated sea ice persistence through July in the northeast Chukchi Sea for 2012 to 2017 and an increasing open water period from the Chirikov Basin north to the southeast Chukchi sea from 2014 to 2017. By the spring months (April-June), the lower latitude stations were consistently ice-free. Four stations in 2017 had duplicate surface sediment samples from Haps core tops and Van Veen grabs. The maximum difference in H-print was 6%, within the margin of error (12%).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 5. H-Print index and satellite-derived sea ice concentration.

The spatial distribution of H-Print (%) in surface sediments from 2012–2017 and the spring sea ice concentration (SpSIC%) derived from April–June mean sea ice concentrations collected from SSMIS passive microwave data (NSIDC). The white and grey bounding boxes indicate the DBO regions from south to north (SLIP, CHIR, SECS, NECS and BARC). Not all sampling stations and DBO regions were able to be occupied every year due to sea ice or weather, indicated by grey boxes (no data collected). H-print ranges from 0–100%, where low values indicate elevated sea ice algae contributions while high values indicate higher contributions of pelagic diatoms. Reprinted from Ocean Data View under a CC BY license, with permission from R. Schlitzer, original copyright 2020.

https://doi.org/10.1371/journal.pone.0231178.g005

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 5. Regional summary of H-Print sea ice index spatial distributions.

https://doi.org/10.1371/journal.pone.0231178.t005

To assess the relationship between the H-Print index and sea ice, linear regressions of two sea ice metrics were examined, including the SpSIC and sea ice break-up date relative to sample collection (Fig 5). Both relationships were significant at the 99% confidence level but the SpSIC relationship was a better fit (R² = 0.46 versus R² = 0.34, n = 184). The SpSIC and H-Print regression shows that the locations with ice-coverage through spring had more substantial sympagic HBI contributions (Fig 6A). The number of ice-free days before sampling shows longer relative periods of open water were associated with reduced sympagic and elevated pelagic organic matter inputs (Fig 6B). There was a linear gradient and association between higher H-Prints and extended open water periods (lower spring sea ice concentration) at lower latitudes and lower H-Prints with higher spring sea ice and shorter ice-free periods at higher latitudes. There was a large degree of variability in both relationships.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 6. Latitudinal variation and correlation of the H-Print index with sea ice.

The 2012–2017 H-Print values were compared with two different metrics for sea ice to determine the influence on the biomarkers. A) Linear regression of H-Print and the mean Spring Sea Ice Concentration (SpSIC) derived from April-June monthly sea ice concentration values. B) Linear regression of H-Print and the ice free period determined by the sea ice break-up date relative to sample collection date. Both relationships are shown with respect to latitude.

https://doi.org/10.1371/journal.pone.0231178.g006

To further explore the relationship between latitude and H-Print, the H-Print values were grouped by DBO region and plotted by latitude (Fig 7A). The box-and-whisker plots show the transition of increasing sea ice algal signature from south to north. There is also a greater degree of variability in the Chukchi Sea regions (SECS, NECS and BARC). The principal components analysis with individual HBIs (IP₂₅, HBI II, and HBI III) and grouped by DBO region also depict a divergence between the SLIP-CHIR-SECS and the NECS-BARC regions.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 7. H-Print index by DBO region.

Statistical analysis of the H-Print values from surface sediments in relation to the location A) boxplot of H-Print variability by DBO region and latitude B) Multivariate separation of surface sediments visualized by principal components analysis (PCA) of individual HBIs (IP₂₅, HBI II isomers, and HBI III) grouped by DBO region.

https://doi.org/10.1371/journal.pone.0231178.g007

A one-way ANOVA test for the H-Print values grouped by DBO region suggests that the mean values were statistically different (p<0.001, F-value = 55.97). A Tukey multiple-pairwise comparison indicates that the differences between NECS-BARC, SECS-CHIR, CHIR-SLIP, and SLIP-SECS were not significant. In other words, the northern regions (NECS, BARC) are similar to each other and the southern regions (SLIP, CHIR, SECS) are similar to each other, but both of the northern stations differ from each of the southern stations (p<0.001). The H-Print index varied by latitude, with the greatest amount of variability among NECS and BARC locations in addition to a stronger sea ice carbon signature at the higher latitudes (71–73°N) and stronger pelagic influence at the lower latitudes (62–68°N; Fig 7A). The PCA of the relative abundances of individual HBIs grouped by DBO region also supports this divergence in H-Print between the northern Bering and northeast Chukchi Seas (Fig 7B). The first principal component (PC1) accounted for 83.3% of the variation, with primary contributions from HBI III, and the second principal component (PC2) accounting for 16.6% of the variation, with HBI II and HBI III as the primary contributors (Fig 7B).

The annual mean H-Print and SpSIC values for the two distinct regions were grouped to assess temporal trends over the study period (Fig 8). Based on regression analyses, the only significant trend identified was for the SpSIC in the northeast Chukchi Sea, with a decline of 5.8% per year (p<0.001). However, the patterns are consistent for both regions, where the SpSIC is declining and the H-Print is increasing from 2012–2017 in the northeast Chukchi (Fig 8A) and from 2014–2017 in the northern Bering and southeast Chukchi Seas (Fig 8B).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 8.

Annual Trends in H-Print and Spring Sea Ice Concentration (A) the northern Chukchi DBO regions (NECS and BARC) for 2012–2017 and (B) the Bering-southeast Chukchi DBO regions (SLIP, CHIR and SECS) for 2014–2017. The bold dashed line shows the only significant trend (p<0.01).

https://doi.org/10.1371/journal.pone.0231178.g008

HBI profiles in sediment cores

The core collected on the Chukchi shelf break, NNE14, showed minimal signs of bioturbation, based visually on three distinct layers of sediment and validated by ¹³⁷Cs measurements indicating a single subsurface peak in the upper 5 cm (Fig 9) that can be interpreted as corresponding to the bomb fallout peak in 1963 [52]. The top 3 cm of the core consisted of oxidized red-brown sediment, the next 5 cm consisted of brown sediments with similar consistency as the shelf sediments, and the remaining length of the core was composed of grey, fine-grained sediments. The H-Print values for this core were generally homogenous and less than 30%, indicating a high and consistent degree of sympagic organic carbon contributions (Fig 9). The core collected on the Chukchi shelf, DBO 4.6, on the other hand, was subject to significant bioturbation, including by polychaete worms present in the core when it was sectioned (sometimes spanning multiple core intervals). The H-Print values from this core were higher than the slope core, with values ranging from 30–55%, representing a greater pelagic contribution compared to the slope core but still having substantial sympagic inputs.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 9. H-Print and radiocesium profiles in sediment cores.

H-Print profiles for a core collected on the Chukchi slope, NNE-14 (blue), and a bioturbated core collected on the shallower Chukchi shelf at DBO 4.6 (red). ¹³⁷Cs profiles for NNE-14 and UTX13-23, a core collected in close proximity to DBO 4.6, depict the consistent sedimentation or bioturbation of deposited material.

https://doi.org/10.1371/journal.pone.0231178.g009

Seasonal variations of HBI and diatom export in the northeast Chukchi Sea

The IP₂₅ and sea ice diatom fluxes observed at the CEO indicated an early summer sea ice algal bloom on the Chukchi shelf. Peak IP₂₅ fluxes during early July 2016 (1331 ng m^-2 d^-1), coincided with the largest flux of the diatom group Gyrosigma/Pleurosigma/Haslea (Fig 2D), which account for ~1% of the relative abundance of major diatom taxa groups [49]. This value is higher than maximum values observed in August 2008 and August 2009 in the Chukchi Borderland (46 ng m^-2 d^-1 and 33 ng m^-2 d^-1, respectively) [37]. This is not surprising given that generally shallow Arctic shelves are more productive than the basin and slope regions [11, 14]. The presence of sea ice associated species, such as N. frigida and M. arctica, provided additional indicators of ice algae release. Export of N. frigida was first detected at the CEO sediment trap in early April 2016 (Fig 3C), which was followed by the highest detected relative abundance of Gyrosigma/Pleurosigma/Haslea (~5%) in late April [49]. The slight increase of IP₂₅ fluxes from January to February 2016 (47 to 223 ng m^-2 d^-1, Table 3) corresponds to the reappearance of the Gyrosigma/Pleurosigma/Haslea taxonomic group [49]. This timing would correspond to the first seasonal deposition by known IP₂₅ producers contributing to the algal flux from the CEO site. This period also corresponds to an increase in HBI III (220 to 1143 ng m^-2 d^-1). The sharp decline in IP₂₅ in late July, along with an increase in pelagic diatom species (Chaetoceros and Thalassiosira spp.) [49], likely signified the end of ice associated diatom export as a result of the bottom few centimeters of sea ice melting that contain the most organic material and nearly all sea ice algae [7].

There were limitations in this study for comparing the selected diatom fluxes with the HBI fluxes, as taxonomy to the desired level was not feasible. The flux of the Gyrosigma/Pleurosigma/Haslea group into the sediment trap was used to compare the IP₂₅ fluxes, given the potential inclusion of the three or four known species that produce IP₂₅ from the genera Haslea and Pleurosigma [19]. However, this genus cluster also includes species that are not HBI producers and diatoms that are not considered exclusively sympagic. HBI-producing species are a minor taxa (ca. 1–5%) and only represent a small fraction of the abundances observed in this group [19]. Therefore, these results are presented with caution in regards to being the direct source for IP₂₅. However, the onset of increasing levels of IP₂₅ strongly corresponded to the increasing levels of the sympagic diatoms and the Gyrosigma/Pleurosigma/Haslea group as indicated by the Pearson product-moment correlation (r = 0.73 and 0.80, p<0.001; Table 4). The strong correlation of IP₂₅ with the sea ice diatom group and the Gyrosigma/Pleurosigma/Haslea genus group strengthens our interpretation that IP₂₅ is an appropriate sea ice proxy on the Chukchi shelf. In an example that echoes the complexities we observed in HBI source attribution, in a study conducted in an ice-covered fjord in Greenland, all known HBI-producing species were detected in ice cores and algal fluxes at 37 m but IP₂₅ production could only be attributed to H. spicula [47]. Limoges et al. [47] also found that H. crucigeroides and H. vitrea were producing both the diene (HBI II) and triene forms (HBI III), meaning that it is unclear what promotes synthesis of IP₂₅, including the sea ice conditions and other parameters that may promote or depress the synthesis of this compound. A recent study found that sea ice diatoms increase HBI concentration up to ten-fold when nutrients are a limiting factor [65]. We had to address similar uncertainties in determining the HBI III producing species in this study. The decision to investigate the correlations between Rhizosolenia fluxes with HBI III was made as this genus contains several HBI III producing species, but again would not encompass all potential sources. However, we found no correlation between the Rhizosolenia spp. and HBI III flux (Table 4).It is also noteworthy that the HBI III flux decreased in late July as chl a reach a maximum flux during a pelagic under-ice phytoplankton bloom (Fig 3B and 3G). There was also a positive correlation between IP₂₅ and HBI III fluxes that we cannot unambiguously interpret. This correlation could be due to the potential overlap of the taxa and broad assignments of possible biomarker producers (i.e. Pleurosigma). The weaker correlations with HBI III overall may suggest that HBI III perhaps is not reliable as a pelagic productivity indicator at this location and that Rhizosolenia did not adequately capture the source of the HBI III. This finding was also discussed in a recent study in the sea ice, raising similar complexities in assigning this to a pelagic source or the potential of regional implications [66]. An association of HBI II and HBI III was also observed in the Antarctic where it was suggested that HBI III is an indicator of the intense phytoplankton blooms that emerge in the marginal ice zone (MIZ) rather than open water, meaning that HBI III was more suitable as a proxy for sea ice seasonality or MIZ duration [67].

The HBI fluxes obtained at the CEO suggest that a combination of processes including production, resuspension and advection led to the persistent IP₂₅ signature in the algal flux recorded on the northeast Chukchi Sea shelf. The high export event at the end of June and beginning of July in 2016 (Fig 3B–3G) corresponded to the declining sea ice concentration rather than the early snowmelt in May (Fig 3A). This could also mean that IP₂₅ producing diatoms were present below the ice, rather than within the ice matrix. While pelagic diatoms were largely responsible for the chl a signal in the NECS region from August to October 2015, based upon taxonomic analysis, IP₂₅ and HBI II fluxes observed under-ice during April and May 2016 reflected a large proportion of sea ice algae in diatom export. Photosynthetically active radiation (PAR) measured at 33 m depth at the CEO began to increase (>1 uE cm^-2 s^-1) in March 2016, reaching upwards of 15 uE cm^-2 s^-1 in May 2016 [68]. The onset of increasing PAR, along with snow melt (Fig 3A), triggered the initiation of the sea ice algae export, as reflected in the HBI fluxes and the first maxima of the sympagic diatom (N. frigida) in the trap material in May 2016. Export occurring prior to melt events was possibly due to the detachment of ice algae by currents and/or grazing processes. Diatoms that may have been incorporated into the ice matrix seeding a phytoplankton bloom (e.g. Fragilariopsis spp., Pseudonitzschia/Nitzschia spp.) dominated algal fluxes in July before complete sea ice retreat, along with the exclusively pelagic diatom Chaetoceros spp. [49]. A Bering Sea study previously found a fluid reciprocity between sympagic and pelagic diatoms through the melt season, with both groups incorporated into the sea ice matrix and a gradual transition of assemblages throughout the season [8].

The most notable finding from the sediment trap analysis was the detection of IP₂₅ fluxes year round, likely the result of both new production and resuspension events. Our observations are consistent with continuous fluxes of organic matter, which were recorded under land-fast ice from winter through late spring on the Mackenzie Shelf of the Beaufort Sea, although particulate organic carbon fluxes in winter were not consistent with diatom export [7]. In our samples, there was a lack of chloroplast-containing Haslea spp. during most of the winter months (October through February; Fig 3D). The winter sympagic HBI signal may be the result of resuspension, as supported by the low export of diatoms with chloroplasts recorded during winter [49].

Pelagic HBI III fluxes increased from 495 to 799 ng m^-2 d^-1 from August to September 2015, reflecting the export of an autumn phytoplankton bloom and/or resuspension due to storm activity (Table 2). The large flux of HBI II (2654 ng m^-2 d^-1; Table 3) at this sampling interval suggested resuspension as more likely than in situ production given the absence of sea ice. During this period, there was also a large flux of chloroplast-containing Cylindrotheca closterium, a rapid growing diatom when resuspended in the euphotic zone and common on shallow shelves, among other diatoms that suggested a resuspension-driven autumn bloom as sea ice was absent and sunlight sufficient for growth [49]. In addition, water temperatures, salinity and nutrient data collected at the CEO as well as the meteorological record from the US National Weather Service station in Utqiaġvik indicated an increase in storm frequency and intensity during this period [68]. These fall storms generally lead to a mixing of the water column, bringing remineralized nutrients to the surface, and allowing for the possibility of an autumn bloom [69].

Latitudinal gradients of sympagic HBIs and declining sea ice

While core tops can provide more reliable collection of undisturbed surface sediments, comparisons of surface sediments collected by Van Veen grabs and Haps core tops in this biologically productive region were found to have no significant difference in radiocesium activity, suggesting similar recent deposition [53]. Therefore, we are confident that the results from the surface sediment analysis present recent deposition with some degree of interannual variability, but unlikely to represent a single year due to the mixing on the shelf.

IP₂₅ and HBI II were detected throughout our study sites in the Bering and Chukchi Seas. The range of IP₂₅ concentrations in the surface sediments (0–12 μg g^-1 TOC), are comparable with the range of previously reported pan-Arctic observations (0–10 μg g^-1 TOC) [70]. The largest concentration observed (12 μg g^-1 TOC), in addition to samples with values exceeding 10 μg g^-1 TOC (n = 4), suggest there were localized areas of elevated ice algal export in the Pacific Arctic. One prior study of IP₂₅ in the Pacific Arctic indicated comparable concentrations (0–5 μg g^-1 TOC [37]). However, direct comparisons with our data may be equivocal because of the less productive location further offshore near the Chukchi Borderlands. In addition, there were methodological differences in the prior study because instrument response factors were not taken into account in their IP₂₅ estimates.

When the H-Print index was compared with two satellite-derived sea ice metrics (mean spring sea ice concentrations and ice-free period before sample collection), there was general agreement regarding the periods of open water and sea ice cover for each season (Figs 5 and 6). The H-Print was a slightly better predictor of the mean spring sea ice concentration rather than break-up date, likely due to the scale of this measurement and the resolution of the satellite data. As was the case with the sediment trap analysis, the snowmelt period prior to break-up was the event that initiated the biomarker flux consistent with an ice algae bloom. This parameter likely signified melt pond formation and melting of the bottom few centimeters of sea ice. This represents an advantage over satellite-based observations that do not indicate whether there was significant production occurring beneath the ice.

H-Print indices from 2014–2017 show significant sea ice algal deposition, and increasing proportions of sympagic inputs on a latitudinal gradient (Figs 3 and 6A). Pelagic influences were significantly greater in the northern Bering Sea and southeast Chukchi Sea than in the northeast Chukchi Sea. However, individual biomarkers provide a more nuanced perspective of localized areas of elevated ice algae markers. Sea ice algal material deposition was increasingly significant throughout the northeast Chukchi shelf, southeast of Hanna Shoal and in upper Barrow Canyon (Fig 3).

Sympagic HBI burial through bioturbation and sedimentation

The H-Print levels from the sediment core collected on the slope (NNE-14) were dominated by sea ice carbon biomarkers throughout the entire 20 cm core depth (Fig 9). The location of this core, near the median minimum limit of summer sea ice extent (1981–2010, Fig 2A), means it is likely representative of late season export and a shorter duration of open water relative to the shelf. Sedimentation rates for this core based on estimates from peak ¹³⁷Cs activity (0.04 cm yr^-1) were similar to the estimate from ²¹⁰Pb (0.02 cm yr^-1, data not shown), suggesting a core spanning centuries of deposition. Based on radiocesium measurements throughout the shelf region, maximum ¹³⁷Cs activity occurs between 6–10 cm depth, suggesting the surface sediments represent years and not decades or centuries of deposition [52]. While core tops can provide more reliable collection of undisturbed surface sediments, comparisons of surface sediments collected by Van Veen grabs and Haps core tops in this biologically productive region were found to have no significant difference in radiocesium activity, suggesting similar recent deposition [53]. The H-Print values were slightly higher in the top 6 cm (>20%, Fig 9), where the sediment characteristics were similar to the shelf, although still predominantly sympagic, suggesting a possible recent increase in pelagic phytoplankton deposition. In the bottom 8–20 cm of the core, where the composition consisted of grey, fine-grained sediments, the H-Print is relatively homogenous and strongly sympagic (8–20%, Fig 9). By comparison to the slope, unambiguous sedimentation rates cannot typically be estimated from cores collected on the Chukchi shelf due to the high degree of bioturbation [13, 52, 78]. The ¹³⁷Cs profile from the station UTX 13–23 on the shelf (Fig 9) suggests a well-mixed profile and a somewhat mixed composition of HBIs at nearby DBO4.6 (H-Prints between 40 and 60%). However, there is an increasingly sympagic signature (~30%) at the bottom of the core, suggesting a persistence of the sympagic sourced organic matter at depth or possibly a reduction of sympagic production associated with sea ice declines. Since the shelf has higher nutrient loads, levels of productivity [14], and an earlier retreat of sea ice, it is not surprising the core collected at DBO4.6 indicates a greater influence of pelagic production than NNE14. The H-print data from NNE14 also reflects the limits of phytoplankton deposition relative to sea ice algal deposition on the slope, since this core was collected from a slope area that was historically close to the minimum extent of the ice edge or is ice-covered for most of the year.

The sediment core H-Print data collected near DBO4.6 supports the assumption that there is rapid burial of sea ice algae relative to phytoplankton. The propensity of ice algae to form aggregates, facilitated by microbial exopolymeric substances and the rapid sinking of the pennate diatom N. frigida, may indicate greater relative pulses of ice algae to the seafloor despite a larger relative proportion of pelagic productivity [79, 80]. These processes have also been suggested to support the greater burial potential of sympagic lipid biomarkers [66, 81]. The H-Print values also suggest there is a greater source of ice algae lipids available to the benthic infaunal communities that occupy these sediment horizons. HBI burial data are not available for cores spanning the entire shelf, but it can be expected from the surface sediment data presented in this study that it is likely that ice algal lipids are stored in sediments throughout the Bering and Chukchi shelf. The persistence and potential availability of labile ice algal lipids mixed to depth in the sediments is an important consideration for assessing the ecosystem response to the loss of seasonal sea ice. It is important to note that despite the high degree of bioturbation, the preservation of these biomarkers is still robust. IP₂₅ in particular has proven to be controlled more by climatic conditions rather than degradation processes [82]. According to Rontani et al. [83], autoxidation of lipids in the oxic layers of sediments can be particularly important in regions of low accumulation rates, where near-surface sediments can represent decades to centuries of deposition. There is relatively high deposition based on ¹³⁷Cs sediment profiles throughout the Chukchi Shelf, where the ¹³⁷Cs maxima associated with peak bomb fallout deposition (1963) averaged 7–8 cm in depth. Radiocesium based sedimentation estimates determined from these previous studies on the shelf ranged from 0.1 up to 0.3 cm yr^-1 [52], suggesting deposition on the scale of years in near-surface sediments.

Mechanisms for HBI distribution throughout the Pacific Arctic

The gradient of HBIs throughout the northern Bering and Chukchi Sea sampling locations and the seasonal succession of sympagic to pelagic diatoms as determined through export fluxes at the CEO [49], suggests a general regionally-specific HBI production mechanism (Fig 10). In similarity to the use of HBIs in the Antarctic MIZ [67], the HBI distribution in the Pacific Arctic may be a proxy for relative sea ice persistence rather than proportions of production of sea ice algae and phytoplankton organic matter. In the more southerly latitudes of the northern Bering Sea (62–65°N), sea ice persistence typically occurs 0–3 months of the year. Sea ice retreat historically initiated early in the year (March-April), allowing for a spring sea ice algae bloom. The ice algae bloom is thought to seed a phytoplankton bloom as the ice retreats, with a gradual transition of sympagic to pelagic assemblages [8]. The more recent extended open water period in the northern Bering Sea region and a deepening of the mixed layer allows for a second phytoplankton bloom in the fall before sea ice freeze up, which may be particularly relevant during warmer years [84]. Therefore, sympagic HBI (IP₂₅ and HBI II) production likely occurs during the brief period in early spring with two possible pulses of HBI III production throughout the late spring and fall. This results in a greater relative proportion of the apparent pelagic HBIs relative to the sympagic-origin HBIs. There are also likely to be years with no new IP₂₅ or HBI II production due to the timing of sea ice retreat or lack of formation. The current flow over the Bering shelf, through Bering Strait and into the Chukchi shelf promotes the advection of HBIs northward, potentially elevating the HBI III proportionally in the southeast Chukchi Sea as currents slow north of the Strait. HBI flux data in the northern Bering Sea do not yet exist but could help to refine some of these assumptions.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 10. Conceptual diagram for the production, flux and fate of HBIs in the Pacific Arctic.

Sea ice persistence increases from the northern Bering Sea to the northeast Chukchi Sea. There is a brief opportunity for sympagic production (yellow shading) in the Bering Sea due to the timing of sea ice retreat and return of sunlight, followed by extensive ice-edge and open water phytoplankton blooms (green shading) in the spring and fall. Sympagic production can occur over a longer period in the Chukchi Sea. Sympagic IP₂₅ production (yellow circles) occurs in much lower proportions to pelagic HBI III (green circles) owing to the extensive open water period in the northern Bering Sea. In the Chukchi Sea, there is a greater proportion of IP₂₅ to HBI III. This relative proportionality is observed in the surface sediments when sampled in the summer (pie chart). There is rapid burial of the sympagic HBIs (yellow spiral) owing to aggregation and rapid sedimentation in both regions, with a greater proportion available on the Chukchi shelf. Resuspension (upward arrows) plays a larger role in the Chukchi Sea, sustaining the suspension of IP₂₅ and in the water column. Advection (horizontal arrows) is also likely to be a more prominent contribution to the HBI signal in the Chukchi than the northern Bering Sea. Symbols courtesy of the Integration and Application Network, University of Maryland Center for Environmental Science (ian.umces.edu/symbols/) and reprinted under a CC BY license, permission from B. Walsh, original copyright 2020.

https://doi.org/10.1371/journal.pone.0231178.g010

In the northeast Chukchi Sea, sea ice coverage extends into the summer months (July-August), with some regions of localized persistence throughout the summer, particularly near Hanna Shoal [41]. Sea ice persistence at these higher latitudes typically occurs for 6–9 month intervals. Advection of HBIs from more southerly locations is likely but ultimately may be a minimal source deposited to the northern shelf sediments, due to the aggregation and rapid sinking of diatoms closer to the point of production [9]. The sympagic production initiates with increasing PAR followed by the release of ice algae in April-May, and an under-ice bloom composed of sympagic and pelagic diatoms from June to August as open water is initiated (Fig 10). The presence of exclusively pelagic diatoms reflected the development of an under-ice bloom, as observed in June and July 2016 [49], with HBI III export that coincides with pelagic-sourced production. However, the peak export of HBI III should occur after ice break up during the open water period. In this study, IP₂₅ export occurs year-round through both new production and resuspension. The appearance of M. arctica resting spores following the ice algae bloom and through the fall months supports the prevalence of sympagic diatom persistence in a sediment “seed bank”, which can be resuspended in the fall [85]. The presence of IP₂₅ throughout the year may suggest that Haslea and Pleurosigma resting cells persist until the return of sea ice on the Chukchi shelf. Supporting evidence of this was observed in laboratory cultures of H. crucigeroides and H. vitrea maintained in complete darkness for over six months, where the cells remained viable and with their HBI content the same as when grown in light (unpublished data). Owing to the shallower conditions on the Chukchi shelf (40–50 m), it seems clear that resuspension during the open-water period plays an important role in the persistent IP₂₅ signal.