Microfaunal Recording of Recent Environmental Changes in the Herschel Basin, Western Arctic Ocean

The 21^st century marks the lowest level of sea-ice extent in the Arctic Ocean of the last 1400 years (Kinnard et al., 2011; IPCC, 2019). This decrease has led to increased net primary production, essentially due to both a longer open water season and an increased area of open water (Arrigo et al., 2008; Barber et al., 2012; Arrigo & van Dijken, 2015), at least until ca. 2008 CE. However, since then, new nutrients imported to the Arctic Ocean from the Pacific and the Atlantic oceans have become a major contributor to the increase in net primary production (Lewis et al., 2020). Partial ice cover also fosters upwelling of nutrient-rich waters onto the Arctic shelves (Barber et al., 2012; Schulze & Pickart, 2012) that act as a hotspot for phytoplankton blooms (Tremblay et al., 2011, 2012; Walkusz et al., 2012). However, a transition to a seasonally ice-free Arctic Ocean raises questions about the long-term effects of changes in the plankton community (Li et al., 2009; Comeau et al., 2011; Dolan et al., 2014; Lefebvre et al., 2016) and the impact of the stratification and vertical mixing of the water masses (McLaughlin & Carmack, 2010; Blais et al., 2017) on the Arctic marine food web. Open waters also contribute to the exchange of heat with the atmosphere and to a generally more unstable climate due to increased wind speed and storm events (Simmonds & Keay, 2009; Mioduszewski et al., 2018). Along Arctic coastlines, higher air temperatures and reduced sea-ice cover enhance coastal erosion, which increases the flux of carbon and sediment to the continental shelves, also affecting the food web (Fritz et al., 2017). The role of terrigenous nutrients linked to coastal erosion is rarely considered (Ardyna & Arrigo, 2020; Lewis et al., 2020), but could contribute up to 21% of the primary production in the Arctic Ocean (Terhaar et al., 2021). However, coastal erosion also leads to increased turbidity in the nearshore areas (Klein et al., 2019, Jong et al., 2020), with a documented negative impact on primary production (Anthony et al., 2004; Retamal et al., 2008; Bonsell & Dunton, 2018; Lewis et al., 2020). It may also contribute to ocean acidification through the degradation of terrestrial organic matter (Semiletov et al., 2016; Fritz et al., 2017).

In coastal Arctic environments, it is often difficult to disentangle the impacts of the recent climate change from interannual and local variations because direct measurements over several decades are rare, especially when it comes to the biota. Micropaleontological tracers can be used to develop such biological time series and to reconstruct longer-term changes in environmental conditions. Among microfossils, benthic foraminiferal assemblages are largely controlled by water depth, salinity, temperature (e.g., Polyak et al., 2002, 2003), and food availability (carbon fluxes; e.g., Wollenburg & Kuhnt, 2000). Moreover, some species thrive in sea-ice marginal conditions (Scott et al., 2008a; Seidenkrantz, 2013; Seidenstein et al., 2018), while others may tolerate reduced dissolved oxygen availability (Seidenkrantz, 2013; Moffitt et al., 2014). In shallow benthic environments, ostracods are also abundant and represented by numerous species (Gemery et al., 2017). Their assemblages relate to salinity and temperature, sea level, and river inputs (McDougall et al., 1986; Reimnitz et al., 1993; Stepanova et al., 2003, 2007; Gemery et al., 2013, 2017, 2021; Cronin et al., 2017), in addition to sea-ice cover (Cronin et al., 2010). Tintinnid microfossils may also occur in marine sediment of the Arctic. They belong to ciliates, which are important grazers in the western Arctic Ocean microzooplankton community (Sherr et al., 2009). Previous studies of tintinnids have highlighted their potential use as tracers of salinity and sea level (Echols & Fowler, 1973; Burkovsky, 1976; Rogers et al., 1981; Pierce & Turner, 1993).

Several studies have documented the spatial distribution and ecological affinities of benthic foraminifera on Arctic continental shelves (e.g., Hald & Steinsund, 1996; Wollenburg & Mackensen, 1998; Wollenburg & Kuhnt, 2000; Polyak et al., 2002; Lloyd, 2006; Scott et al., 2008a, 2008b; Husum et al., 2015). However, high-resolution temporal reconstructions of past nearshore environments of the Arctic using benthic foraminifera are rare due to the lack of continuous sedimentological records. In this study, the Herschel Basin provides us with the opportunity to establish multiannual time series (Grotheer et al., 2020) of microfaunal changes in a nearshore Arctic area that is presently influenced by sea-ice loss (Frey et al., 2015) and increased terrestrial supply from coastal erosion and rivers (Doxaran et al., 2015; Ehn et al., 2019; Klein et al., 2019; Jong et al., 2020). The Herschel Basin is located off northern Yukon, Canada (Fig. 1), where benthic foraminiferal and ostracod assemblages have not hitherto been documented. Thus, our first objective is to describe the microfaunal assemblages that characterize the last decades at two different depths in the basin, which may be used as a baseline for future paleo studies on the Beaufort Sea continental shelf. The second objective is to document interannual variations of the microfaunal records and eventually, to relate these to recent environmental changes and develop their use as tracers of biogenic productivity and diversity in nearshore Arctic areas.

The Herschel Basin is located on the continental shelf of the Beaufort Sea offshore the northern Yukon coastal plain, around 10 km east of Herschel Island – Qikiqtaruk and 135 km west of Shallow Bay (Fig. 1B, D). This 40-km long, 15-km wide, and up to 70-m deep basin (EBA Engineering Consultants Ltd., 1992; Fig. 1D) allows continuous sediment accumulation (Grotheer et al., 2020). A shallow sill with a water depth of 13 m forms the northern limit of the basin (EBA Engineering Consultants Ltd., 1992). The sill prevents large stamukhi, or sea-ice rubble, and other floating ice from reaching the basin. The land-fast ice is generally 2 m thick on the shelf (Carmack & MacDonald, 2002), which also limits ice scouring at depth in the basin.

The regional hydrography of the Herschel Basin area is complex and unstable as it is controlled by winds that vary in direction and strength (see below; Kulikov et al., 1998; Lin et al., 2020). In summer, the predominantly eastward coastal flow (Fig. 1B, purple) carries relatively warm and low salinity waters alongshore (>0°C and <31 psu, respectively) that derive from the Bering summer waters (Lin et al., 2020). The shelf current (Fig. 1B, green arrow) flows predominantly westward and carries cold (<–1°C) and relatively fresh (<31 psu) surface waters (<30 m) mainly formed from sea-ice melt and meteoric waters (Kulikov et al., 1998; Weingartner et al., 2017; Lin et al., 2020). In contrast, at the shelf break, the waters flow eastward as the “Beaufort shelfbreak jet” (Fig. 1A, B, blue arrow) transporting predominantly the relatively saline (>31.5 psu) and nitrate-rich Pacific winter water mass known as the remnant winter waters (hereafter referred to as Pacific waters; Pickart, 2004; Lin et al., 2016, 2020). These waters sit on top of dense and saline subsurface Atlantic waters (>33.5 psu; Lin et al., 2016, 2020; Ardyna et al., 2017; Fig. 1C). Further to the north, the Beaufort Gyre circulates clockwise (Fig. 1A, B, yellow arrow) under the influence of the Beaufort Sea High (Serreze & Barrett, 2011). Part of the southern gyre recirculates into the Mackenzie Trough (Lin et al., 2020; Fig. 1B, thin yellow arrows). A strong anticyclonic regime generates dominant easterly winds in the marginal seas that induce sea-ice drift to the central Arctic Ocean usually resulting in reduced sea-ice cover (Ogi & Wallace, 2007; Ogi et al., 2008). Increased easterly winds in the Beaufort Sea continental slope area under a strong Beaufort Sea High would also reduce the entry of Pacific waters through the Beaufort shelfbreak jet (Brugler et al., 2014).

The Mackenzie River is the most important freshwater source in the area (Carmack et al., 2006), and it has the fourth largest discharge of freshwater into the Arctic Ocean (Holmes et al., 2011). Its main freshwater flux usually occurs at the beginning of June during the spring freshet (Carmack & MacDonald, 2002). The Mackenzie River plume is very rich in dissolved organic carbon (Holmes et al., 2011; Juhls, 2021) and terrestrial suspended particles (Doxaran et al., 2015), both of which are dispersed up to hundreds of kilometers westward during the freshet and depending on the winds during summer (Wood et al., 2015; Juhls, 2021). The Firth and Babbage estuaries are located approximately 20 km and 40 km, respectively, from our study site (Fig. 1B). However, their impact on the hydrography of the study area is low as their mean summer flow is significantly less than that of the Mackenzie River (Environment and Climate Change Canada Historical Hydrometric Database, 2020).

The water masses in summer are strongly stratified on the Beaufort Sea shelf due to the low salinity of the surface waters (Figs. 1C, 2A, light gray zone). The halocline depth ranges between 5 and 10 m depth in summer (Carmack & MacDonald, 2002), and could reach deeper levels according to the summer 2016 Conductivity Temperature Depth (CTD) measurements (Fig. 2A). The temperature and salinity of the surface water layer in summer are 8–9°C and 21–22 psu, respectively, while they are close to the freezing point and 32–33 psu below the halocline (Fig. 2A).

The cold and saline bottom waters of the Herschel Basin originate from upwelling of waters at the shelf break (Fig. 1C) via the Mackenzie Trough (Fig. 1B) and brine rejections from sea-ice formation (Williams & Carmack, 2012; Fig. 2, dark gray zones). As the land-fast ice forms in October up to the freshet season in June, the stratification breaks down as the water column homogenizes (Fig. 2B).

Local winds are an important parameter in nearshore ecosystems, as they act as the main drivers of water and sea-ice transport (Kulikov et al., 1998; Frey et al., 2015; Lin et al., 2020), thus controlling sediment and nutrient transport. The important intra- and inter-annual environmental variability in our study area reflects this close relationship to local winds. Around Herschel Island – Qikiqtaruk, strong winds originate either from the northwest or from the east-southeast (Radosavljevic et al., 2016; Fig. 1B). During periods of dominant westerly winds, the Mackenzie River plume and the sea ice remain nearshore. In contrast, dominant easterly winds lead to offshore spreading of the Mackenzie River plume and foster sea-ice spreading and upwelling (Carmack & MacDonald, 2002; Pickart et al., 2011; Wood et al., 2015), thus enhancing regional primary productivity (Tremblay et al., 2011, 2012; Pickart et al., 2013). The sea ice that interacts with marine currents, upwelling, and vertical mixing (Barber et al., 2012; Schulze & Pickart, 2012; Williams & Carmack, 2015) also plays a protective role along the coastline (Günther et al., 2015).

We calculated the mean summer sea-ice concentration in the study area as the average daily sea-ice concentrations in June through September of each year. A breakpoint analysis on the 1979–2017 data highlights a shift in the mean in 2006 CE. Between 1979 and 2006 CE, summer sea-ice concentrations varied from 55% down to 5% with a mean of 31%. After 2006 CE, it stabilized around 7–16% with a mean of 13% (Fig. 3).

This study is based on the analysis of two sediment cores collected in the Herschel Basin: core YC18-HB-GC01 (hereafter HBGC01) and core PG2303-1. The HBGC01 core is from the western part of the Herschel Basin (69.544°N, 138.970°W; 18 m water depth), while core PG2303-1 was retrieved from the deeper part of the basin (69.513°N, 138.895°W; 32 m water depth; Fig. 1D).

Hydro-acoustic data were acquired using a Knudsen sub-bottom profiler (12 kHz; Figs. 1D, A.1, A.2). The acoustic image crossing the shelf (NW–SE transect; Fig. 1D) displays faintly laminated to chaotic reflections above 16 m water depth that evolve downslope into finely laminated reflections below the slope break. This lateral variation of acoustic facies suggests a transition from littoral coarse sediments to fine sediments (Fig. A.1). Thus, the HBGC01 core was collected at the slope below the littoral platform. For the PG2303-1 core site (E–W transect; Fig. 1D), acoustic data show high-amplitude and parallel reflections interpreted as clastic mud (Fig. A.2).

The 40-cm long HBGC01 core was retrieved using a UWITEC gravity corer deployed from a zodiac during the 2018 Nunataryuk Yukon Coast expedition on August 9^th. The core was stored at 3°C and cut in halves lengthwise at Université Laval, Canada, where the lithological analyses were performed. One half-core was subsampled at 1-cm intervals for micropaleontology and dating (this study). The 20-cm long PG2303-1 core was retrieved on April 20^th, 2016, with a UWITEC gravity corer from a tripod on the sea ice during a winter expedition led by the Alfred Wegener Institute. The core was stored cool in a refrigerator and subsampled at 1-cm intervals with a vertical extrusion system. Thus, the short core PG2303-1 was not cut lengthwise, and a descriptive lithology is not available.

The HBGC01 core was photographed and scanned for 3D imagery and density by X-rays Computed-Tomography Scanner (CT-scan) at INRS-ETE facilities using a SIEMENS Somatom Definition AS128 (e.g., Fortin et al., 2013, Figs. 4A, A.3) with the settings adjusted to 165–175 kV and 250–300 mA s^–1. We used the ImageJ software (Rasband, 2018) to export and process the raw DICOM format images.

The samples of core HBGC01 were analyzed for grain size at the Laboratoire de géomorphologie et sédimentologie of the Université Laval using a Spectrex Particle counter paired with a Horiba laser diffraction particle size analyzer allowing for grain-size measurements between 0.01 and 3000 µm. The samples of core PG2303-1 were processed at the Université du Québec à Montréal (UQAM) using a Microtrac MRB Bluewave laser diffraction instrument, which can detect grain sizes from 0.01 to 2000 µm. Prior to the measurement, the samples were treated with hydrogen peroxide (30%) to dissolve organic matter. Data were processed in GRADISTAT (Blott, 2010), which defines sand as particles between 2 mm and 63 µm, silt between 63 µm and 2 µm, and clays <2 µm.

The age models are primarily based on ²¹⁰Pb analyses (raw data in Carnero-Bravo et al., 2021). The measurements procedure of ²¹⁰Pb followed Baskaran & Naidu (1995), assuming its secular equilibrium with ²¹⁰Po and using a spike of ²⁰⁹Po. Sediment samples were digested with three acid solutions, successively (HNO₃; HCl, HF, and HNO₃; H₂O₂ and HNO₃). Residues were dissolved in a solution of HCl 0.5 N with ascorbic acid as a reducing agent for Fe. Both ²¹⁰Po and ²⁰⁹Po were finally deposited (plating) on a silver disk. The measurement was performed by alpha spectrometry using ORTEC^TM detectors and the Maestro^TM data acquisition software.

We obtained ²²⁶Ra and ¹³⁷Cs activity (γ-emitters) by high-resolution gamma spectrometry using a germanium detector (ORTEC). The efficiency calibration was calculated based on the Columbia River Basalt reference material (BCR-2; USGS). ²²⁶Ra was used to estimate the supported ²¹⁰Pb fraction (²¹⁰Pb_sup), in which ²¹⁰Pb_sup is the average of all measurements of ²²⁶Ra. The unsupported ²¹⁰Pb (²¹⁰Pb_ex) is the result of the total ²¹⁰Pb (²¹⁰Pb_tot) minus the ²¹⁰Pb_sup.

The age models were made using the Plum software (Aquino-López et al., 2018) under R (R Core Team, 2021) and compared with the Constant Flux Constant Sedimentation (CFCS) and the Constant Rate of Supply (CRS) model using the serac package (Bruel & Sabatier, 2020). A first order estimate of the accumulation rate (in a cm^–1 in Plum) of each core was specified prior to running the software. Plum takes into consideration the ²¹⁰Pb_tot activity and density (g cm^–3) of each sample, the ²¹⁰Pb_sup value, and the date of coring. Ultimately, the mass accumulation rate (g cm^–2a^–1) of each sample was calculated by multiplying the sedimentation rates (cm a^–1) derived from the Plum runs and the densities (g cm^–3) of the sample section (Table A.1). Sedimentation rate uncertainties were obtained by calculating the sedimentation rates of the highest and lowest ages given by the model (Table A.1). Then, the uncertainties of mass accumulation rate were obtained by following error propagation principles.

Sixty samples, 40 from core HBGC01 and 20 from core PG2303-1, were analyzed for their microfossil content, with special attention paid to benthic foraminifera, ostracods, and tintinnids. For each sample, 6–7 g of dry sediment were wet-sieved at 63 µm at Geotop-UQAM and dried again. Then, the microfossils were observed and picked with a fine brush under a stereo microscope at 40× to 115× magnification. Each sample was scanned in its entirety, but to ease the work process, the 63–106 µm and >106 µm fractions were counted separately. In samples containing high concentrations of fine sand, we used a solution of tetrachlorethylene (C₂Cl₄) with a specific gravity of 1.6 g cm^–3 to separate mineral particles from biogenic remains (Table A.2).

Benthic foraminifera, including calcareous and agglutinated taxa, were identified based on Feyling-Hanssen et al. (1971), Polyak et al. (2002), and Scott et al. (2008b). Identification was made at species level (Table 1 ; Figs. 5 –7), except for the Polymorphinoidea superfamily (hereafter polymorphinids), which has a high diversity of species with poorly known ecology.

Ostracods were identified to genus or species level (Table 2 ; Figs. 8 , 9) based on Gemery et al. (2017) and Stepanova et al. (2003). Specimens identified only to genus level were mainly early instars (molts on juvenile specimens). The ostracod taxa that can tolerate low salinity (<30 psu) and the taxa that prefer salinity >30 psu (euhaline) were distinguished based on Olausson (1982), McDougall et al. (1986), Stepanova et al. (2003, 2019), Tian et al. (2020), and Gemery et al. (2017, 2021; see Table 2).

The agglutinated loricas of tintinnid were all identified as belonging to Tintinnopsis fimbriata (Meunier, 1919), which was re-described by Agatha (2008; Table 3). It is important to mention that the lorica of T. fimbriata, usually 48–74 µm in length (Agatha, 2008), can be smaller than the sieving mesh size used here (63 µm). Hence, our T. fimbriata counts probably provide an underestimation of the population.

Other microfossils recovered in lower numbers in the >63 µm fraction include bisaccate pollen grains (Picea spp.), diatoms (Coscinodiscus spp.), radiolarians, and planktic foraminifera (Table 3; Fig. 10). The radiolarians were identified as Spongotrochus glacialis (Itaki et al., 2003). Being part of the zooplankton dwelling in the open ocean, the radiolarians might be reworked here. Finally, we encountered coccoliths in radiolarian alveoli in one sample (35–34 cm, core HBGC01). The detailed counts of all microfossils are reported in the appendix (Table A.3).

We calculated the concentrations (number of microfossils per gram of sediment; # g^–1) for the calcareous and agglutinated benthic foraminifera, ostracods, and tintinnids. Since the size of the observed pollen grains was close to the minimum mesh size (63 µm), if not smaller, we do not report their abundance. The other identified microfossils (diatoms, radiolarians, and planktic foraminifera) were too sparse to calculate concentrations. Fluxes (# cm^–2a^–1) were obtained by multiplying the mass accumulation rates (g cm^–2a^–1) with the microfossil concentrations (# g^–1). Minimum and maximum fluxes were calculated, taking into consideration the accumulation rate uncertainties.

The assemblages of benthic foraminifera and ostracods are presented in terms of relative abundance (%). Fatela & Taborda (2002) concluded that a minimum of 100 foraminiferal shells are satisfactory for assemblage assessments of dominant taxa (>5%), but that minimum counts should be 300 specimens for the non-dominant taxa. In all samples, the total counts of foraminifera including calcareous and agglutinated is >100 specimens, except for two samples of core HBGC01 (62 and 94 counted specimens in samples 5–4 cm and 4–3 cm, respectively). The required minimum number for ostracod valves was set at 10.

We calculated the Shannon entropy on the benthic foraminifera and ostracods Hellinger-transformed raw counts in the vegan package (Oksanen et al., 2013) under R. The exponential of the Shannon entropy is then used to obtain the Shannon diversity. This diversity considers the abundance and evenness of the species in one sample and is thus less sensitive to rare species, which seemed appropriate for an Arctic Ocean shelf, where transported allochthonous microfossils from nearshore currents or ice rafting may occur (Wollenburg & Kuhnt, 2000). Among the foraminifera, the polymorphinids remained grouped together for the analysis. Benthic foraminifera not identified to species level were removed from the diversity index calculation but included in the percentage and flux calculations. For ostracods, the Shannon diversity was calculated using identification at the genus level.

Redundancy Analyses (RDAs) were performed using the vegan package under R with the Hellinger-transformed raw counts of foraminifera as the response matrix and standardized concentration of sand (%) and microfossil fluxes (for core HBGC01) or microfossil concentrations (for core PG2303-1) as the explanatory variables. For both cores, the RDAs were used to help discriminate ecozones and major shifts in the assemblages. RDAs were not performed on the ostracods due to their low counts especially in core HBGC01 (see section on the Microfossil concentrations and fluxes of core HBGC01).

The sediments in core HBGC01 are brownish gray with sparse presence of millimetric dark laminations. The CT-scan reveals homogenous sediment from the bottom of the core to 10 cm with signs of bioturbation marked by pale mottles and holes (Figs. 4A, A.3). The upper 10 cm show faintly laminated sediments that are discontinuous and of unequal thickness. The content of sand, silt, and clay is almost uniform throughout the record with a mean of 10±5%, 81±4% and 9±4%, respectively (Fig. 11). Sand content increases to a mean of 17±7% in the upper 7 cm with a peak of 31% in the surface sample. The samples mean sediment density is 1.06±0.20 g cm^–3 (Table A.1).

The ²¹⁰Pb_sup (²²⁶Ra average) in core HBGC01 is 1.60±0.05 dpm g^–1 (disintegrations per minute per gram, Fig. 4B). A log-decrease of ²¹⁰Pb_ex can be evidenced between 40 and 10 cm depth (Fig. 4C). Such a behavior supports continuous accumulation of sediments. The ²¹⁰Pb_tot in the upper 10 cm shows oscillations of large amplitude (Fig. 4B). Laminations in this layer (Figs. 4A, A.3) suggest pulses of reworked sediments with variable ²¹⁰Pb_ex through coastal erosion and nearshore processes (tides, waves, storms, currents, etc.) of shallow-water environments (<20 m; Hill & Nadeau, 1989; Corbett et al., 2006; Hanna et al., 2014). The ¹³⁷Cs data record a minimum of 0.05±0.02 dpm g^–1 near the surface and an increasing trend downcore to a maximum of 0.23±0.03 dpm g^–1 at 38–37 cm. For the Plum model, we considered a mean accumulation (acc. mean) of 1.23 a cm^–1 by dividing the ²¹⁰Pb disintegration constant with the slope of the linear regression of log ²¹⁰Pb_ex observed between 40 and 10 cm depth (Sanchez-Cabeza & Ruiz-Fernández, 2012; Fig. 4C).

Based on these data, the Plum model provides an age of ∼1969 CE at the bottom of the core up to the year of coring (2018 CE) at the surface (Fig. 4D). The mean sediment accumulation rates average 0.8±0.1 g cm^–2a^–1 below 10 cm and are higher but variable above (Fig. 11). The mean estimated age at 38–37 cm is 1973±9 CE (minimum is 1963 CE; Fig. 4D), which coincides with increased ¹³⁷Cs activity (Fig. 4B) assigned to the maximum nuclear fallout (Aoyama et al., 2006). Although this ¹³⁷Cs peak is weakly recorded in core HBGC01, it is clearly recorded in a well-dated core collected in the nearby Ptarmigan Bay (Carnero-Bravo et al., 2021; YC18-PB-SC01; Fig. 1D). The Plum output is represented in Figure 4D with the CFCS and the CRS models for which a mixed layer of 10 cm was included. The CRS model was also forced to fit with the 1963 CE ¹³⁷Cs peak (CRS-piecewise). All three models yield comparable results. Hence, the age model presented here is as robust as possible given all the available radiogenic and lithological data.

Two ecozones are defined in core HBGC01 based on the redundancy analyses of the benthic foraminiferal assemblage, specifically the scores of the first RDA axis (Figs. 12 , 13). They are named Ecozone HBGC01-B (1969–2000 CE) and Ecozone HBGC01-A (2000–2018 CE). Details on the two ecozones can be found below.

Counts of calcareous benthic foraminifera were >100 in most samples of core HBGC01 (288±134; Table A.3). This corresponds to concentrations of calcareous benthic foraminifera ranging between 8 and 87 foraminifera g^–1 throughout the record. Agglutinated foraminifera were found in lower numbers (17±14; Table A.3) with concentrations between 0 and 10 g^–1. The quasi-continuous presence of agglutinated and calcareous foraminifera suggests generally good preservation.

From 1969 to 2000 CE (Ecozone HBGC01-B), the calcareous foraminiferal concentration averaged to 55±17 g^–1. After 2000 CE (Ecozone HBGC01-A), this average decreased to 30±13 g^–1. Estimated foraminiferal fluxes are about 37±19 cm^–2a^–1 throughout the record (Fig. 11). Agglutinated foraminifera record a mean concentration of 3±2 g^–1, which corresponds to fluxes between 0 and 7 cm^–2a^–1 (Fig. 11). Ostracod concentrations range between 0 and 5 g^–1 in Ecozone HBGC01-B. Ostracod fluxes are low throughout the record (<3 cm^–2a^–1), particularly in the upper section (2000–2018 CE), where they number <1 g^–1 of sediment. The mean concentration of Tintinnopsis fimbriata is 3±4 g^–1. From the bottom of the core to 1996 CE, T. fimbriata is five times less abundant (1±1 g^–1) than in the 1996 CE to present interval (5±4 g^–1). This corresponds to low fluxes until ∼1991 CE (<3 cm^–2a^–1) and a mean of 6±6 cm^–2a^–1 afterwards (Fig. 11), with maxima between 2008 and 2014 (11±6 cm^–2a^–1).

The most important feature in the microfaunal record of core HBGC01 is the opposite shifts in T. fimbriata and ostracod concentrations at ca. 2000 CE (at the boundary between ecozones HBGC01-B and HBGC01-A; see the last section of the Discussion). We also note that the 2013–2015 CE sediment layer is characterized by maximum calcareous benthic foraminiferal, ostracod, and T. fimbriata occurrences, with fluxes reaching up to 119±23 cm^–2a^–1, 2 cm^–2a^–1, and 24±5 cm^–2a^–1, respectively (Fig. 11).

Twenty-six (26) species of calcareous benthic foraminifera were identified in core HBGC01 (Table 1; Figs. 5, 6). In addition, some specimens were identified to the genus level only (i.e., Valvulineria spp.) or grouped as polymorphinids. The dominant species are Elphidium clavatum and Cassidulina reniforme, with Eoeponidella pulchella and Haynesina nivea as occasional accompanying species (Fig. 13). A total of seven agglutinated foraminiferal taxa were identified: Spiroplectammina biformis and Textularia torquata were dominant, while Textularia earlandi and Recurvoides turbinatus were found occasionally. Rare occurrences of Lagenammina difflugiformis, Portatrochammina karica (also called P. bipolaris), and Sigmoilopsis schlumbergeri were noted (Tables 1, A.3; Fig. 7).

The first RDA axis explains 46.5% of the variance. It shows an opposite direction of Eoeponidella pulchella, Haynesina nivea, and Quinqueloculina stalkeri that record negative scores and Islandiella norcrossi, Textularia torquata, Stainforthia feylingi, and Pyrgo williamsoni that record positive scores (Fig. 12; see Table 4 for abbreviation corresponding names). Axis 1 also shows an opposite direction of Tintinnopsis fimbriata fluxes and percentage of sand on the negative side and ostracod abundance on the positive side. The second RDA axis explains 33.2% of the variance and shows an opposition between most of agglutinated benthic foraminiferal taxa, which have positive scores, and Elphidium clavatum, Islandiella norcrossi, and Cassidulina reniforme, which have negative scores. Axis 2 is possibly related to microfossil abundances since all microfossil fluxes record positive scores, unlike the sand content.

The species diversity of benthic foraminiferal assemblages is relatively high (Shannon index of 5±1, on average) in core HBGC01 (Fig. 13). Elphidium clavatum remains dominant throughout the record (40–65%; Fig. 13). However, the relative abundance of accompanying species records a significant change around 2000 CE when the RDA axis 1 shifts from positive to negative values (Figs. 12, 13) at the ecozone HBGC01-B/A boundary. Haynesina nivea and Eoeponidella pulchella percentages peak synchronously at ∼2013 CE, when they together reached 28% of the assemblage (Fig. 13).

A total of 11 taxa of ostracods were identified in core HBGC01 (Table 2; Figs. 8, 9). Among those, Paracyprideis pseudopunctillata is a recurrent low salinity tolerant (<30 psu) taxon, while Cluthia cluthae and Cytheropteron elaeni are the dominant euhaline species. In Ecozone HBGC01-B, from the bottom of the core to ∼ 2000 CE, Sarsicytheridea bradii, Paracyprideis pseudopunctillata, and Cluthia cluthae are frequent (Fig. 14) and Rabilimis mirabilis, Eucythere spp., and Semicytherura complanata occur occasionally. This interval is characterized by relatively high diversity with a Shannon index of 4±1. After 2000 CE (Ecozone HBGC01-A), the occurrence of Eucytherura delineata increased (Fig. 14), but the general low counts of ostracod valves do not permit statistical assessments.

Although the sediment of core PG2303-1 was not described, the visual description of a long piston core collected at the same location (core PG2303; Fig. A.2) revealed homogenous gray-brownish sediment (Pfalz, 2017). Sediment density measured on core PG2303-1 revealed almost uniform values with a mean of 1.1±0.2 g cm^–3 (Table A.1), suggesting even water content and low sediment compaction. The core is composed of coarser sediments from 20 to 12 cm, corresponding to ∼1998–2005 CE (Fig. 15B), where the sand content reaches a mean of 15±9%, but the average grain size distribution is 9±8% of sand, 72±3% of silt, and 18±5% of clay (Fig. 16).

In core PG2303-1, the ²¹⁰Pb_tot activity ranges between 1.8 to 3.0 dpm g^–1 with an average of 2.6±0.3 dpm g^–1 (Fig. 15A). ²¹⁰Pb_tot decreases down to 1.8±0.1 dpm g^–1 at 14–13 cm, coinciding with a maximum of sand (31%; Fig. 16). The average ²²⁶Ra (²¹⁰Pb_sup) is 1.77±0.03 dpm g^–1. The ¹³⁷Cs activity is <0.14 dpm g^–1 and all measurements overlap taking in consideration uncertainties.

For the calculation of a Plum age model in core PG2303-1, we used the average age at which the incomplete inventory of ²¹⁰Pb_ex at 20 cm in core PG2303-1 (17.5 dpm cm^–2) was reached in cores HBGC01 and YC18-PB-SC01 (Fig. 1D; Sanchez-Cabeza & Ruiz-Fernández, 2012; Carnero-Bravo et al., 2021) and estimated a mean accumulation rate of 0.83 a cm^–1. The Plum model suggests that core PG2303-1 is younger than ∼1998 CE (Fig. 15B), indicating that the entire core corresponds in age only to Ecozone HBGC01-A. However, due to the high apparent sediment mixing the uncertainty in mean accumulation rate, and the short length of the record, we did not interpret the record in terms of age and thus report all results against depth. For the same reason, the mass accumulation rate and microfossil fluxes were not calculated.

Despite the short time span covered by the core, the high sample resolution permitted the definition of three ecozones based on the RDA (Figs. 17 , 18). Ecozone PG2303-1-C corresponds to the 15–20 cm core depth interval, Ecozone PG2303-1-B covers from 12 to 15 cm, and Ecozone PG2303-1-A encompasses the upper 12 cm.

Counts of calcareous foraminifera range between 208 and 646 specimens per sample (∼6–7 g of dry sediment; mean of 330±149 specimens; Table A.3). More than 50 agglutinated foraminifera were counted in each sample (mean of 80±28; Table A.3). A mean of 11±7 ostracod valves was identified per sample with a maximum of 34. Tintinnopsis fimbriata was abundant with >49 specimens counted in each sample. These counts produce mean concentrations of 62±26 calcareous foraminifera g^–1 and 16±8 agglutinated foraminifera g^–1, which suggest good microfossil preservation with depth in the Herschel Basin. Mean ostracod and Tintinnopsis fimbriata concentrations are 2±2 g^–1 and 27±13 g^–1, respectively.

There are three abundance peaks of calcareous foraminifera (>80 g^–1): at the bottom of the core, at 15–13 cm (approximately corresponding to Ecozone PG2303-1-B), and at 3–1 cm. The agglutinated foraminiferal concentrations do not vary, but the ostracod abundance seems to diminish towards the top of the core, while Tintinnopsis fimbriata records maximum concentrations in the upper part of the core (Fig. 16).

All calcareous benthic foraminiferal taxa observed in core PG2303-1 also occur in core HBGC01 (Table 1; Figs. 5, 6), with 24 taxa in total. As in core HBGC01, Elphidium clavatum and Cassidulina reniforme are the dominant species throughout the record, comprising 20–35% and 10–20% of the assemblages, respectively (Fig. 18). Accompanying species are Haynesina nivea (5–20%) and Eoeponidella pulchella (5–20%) and to a lesser extent Stainforthia feylingi (1–10%) and Buccella frigida (0–5%). There are nine species of agglutinated foraminifera, including Spiroplectammina biformis that dominate, together with Textularia earlandi, Textularia torquata and Recurvoides turbinatus, which are the main accompanying species. The occurrence of other species including Cribrostomoides crassimargo and Eggerelloides advenus is occasional (Tables 1, A.3; Fig. 7). The Shannon diversity index is higher than in core HBGC01 with a mean of 9±2 (Fig. 18).

The RDA reveals an inverse relationship of the abundance of the agglutinated taxa Spiroplectammina biformis and Textularia earlandi and the calcareous taxa Elphidium clavatum, Haynesina nivea, Eoeponidella pulchella, and Buccella frigida on the first axis (40.5% of the variance; Fig. 17; see Table 4 for the abbreviation corresponding names). Accordingly, the concentrations of agglutinated foraminifera have a negative score, while calcareous foraminiferal concentrations score positively. The second axis, which explains 11.5% of the variance, shows an opposite direction of ostracod concentrations, sand (%) and the abundance of numerous foraminiferal taxa on the negative side with Tintinnopsis fimbriata concentrations, Cassidulina reniforme and Elphidium clavatum on the positive side (Fig. 17).

In Ecozone PG2303-1-C, from 20 to 15 cm, the diversity is the highest of the core (Shannon diversity index >10; Fig. 18). Thus, the RDA Axis 2 scores negatively as several taxa reach maximum percentages, including Islandiella norcrossi, Islandiella helenae, polymorphinids, and Recurvoides turbinatus (Figs. 17, 18). In Ecozone PG2303-1-B, from 15 to 12 cm, the RDA axis 1 is positive due to abundant Haynesina nivea, Eoeponidella pulchella, and Buccella frigida (Figs. 17, 18). Finally, in Ecozone PG2303-1-A, from 12 cm to the top of the core, the RDA axis 1 is close to nil, while there is a slight increase of Cassidulina reniforme and Spiroplectammina biformis, Quinqueloculina stalkeri, Stainforthia feylingi, and Textularia earlandi are also present. In the 8 to 1 cm section, the continuous occurrence of Stetsonia horvathi is notable (mean of 3±1%; Fig. 18). The uppermost sample differs from all others with negative RDA axis 1 score, abundant agglutinated taxa and low percentages of Elphidium clavatum, together with the disappearance of Buccella frigida and Eoeponidella pulchella. This atypical assemblage recorded in the upper cm of the core might relate to the winter season rather than mean annual conditions, taking into consideration the very high sedimentation rates and the date of coring (April 2016).

The ostracod assemblages in core PG2303-1 contain 11 taxa (Table 2; Figs. 8, 9). Among those, Cytheropteron brastadensis (discoveria) and Cytheropteron sulense are restricted to this core, while Cluthia cluthae and Sarsicytheridea bradii are exclusive to the HBGC01 core. The most common species in core PG2303-1 are Cytheropteron spp. and Cytheropteron suzdalskyi. The mean Shannon diversity index is 2±1. The three ecozones as defined by the RDA of the foraminiferal assemblages (Figs. 17, 18) also reflect changes in the ostracod assemblages. The 20–15 cm interval, Ecozone PG2303-1-C, is dominated by euhaline taxa (Fig. 19), notably Semicytherura complanata, Cytheropteron suzdalskyi, and Cytheropteron spp., and ostracod concentrations reach maximum values (Figs. 16, 17). From 15 to 12 cm, in Ecozone PG2303-1-B, low-salinity tolerant taxa, including Heterocyprideis sorbyana and Loxoconcha venepidermoidea, constitute most of the assemblages (Fig. 19). The species diversity is maximum in this interval, as the samples contain both low-salinity tolerant and euhaline taxa. In the first part of Ecozone PG2303-1-A, from 12 to 6 cm, the assemblages are dominated by euhaline taxa, but with low occurrence of Semicytherura complanata. Finally, in the upper seven centimeters of Ecozone PG2303-1-A, corresponding to the level where Stetsonia horvathi increases in the foraminiferal assemblage (Fig. 18), the ratio between low-salinity tolerant and euhaline ostracods is approximately even.

With a water depth of 18 m, site HBGC01 is located close to the depth of the summer halocline (Fig. 2A). Vertical mixing in response to storm events and tidal dynamics may cause the upper layer (∼22–24 psu and >0°C) to mix with the subsurface waters (30 psu and ∼0°C; Hill & Nadeau, 1989; Rainville & Woodgate, 2009; Fig. 2A). Hence, the bottom waters at this site may be influenced by the low salinity surface waters that also carry a high concentration of suspended particles either from the Mackenzie River (Doxaran et al., 2015; Ehn et al., 2019; Juhls, 2021) or from terrestrial sources in relation with coastal erosion and resuspension (Klein et al., 2019; Jong et al., 2020). In contrast, core PG2303-1 was retrieved from 32 m water depth, which is significantly deeper than the summer halocline. Consequently, this deep and more offshore site is less likely to be affected by turbidity than the nearshore shallower core HBGC01 (Fig. 1D; see the last section of the Discussion). Turbidity in surface waters could have negative impacts on primary production (Anthony et al., 2004; Retamal et al., 2008; Bonsell & Dunton, 2018; Lewis et al., 2020). In winter, however, below the two-meter-thick land-fast ice, subsurface waters return to saline (32–33 psu) and frigid (<0°C) conditions and sites HBGC01 and PG2303-1 are both bathed by the same water mass, recording values close to the annual mean of the 32 m deep site PG2303-1 (Figs. 2A, B).

In addition to bottom water salinity and turbidity in summer, the two study sites are subject to different bottom current speeds and dynamics. The higher clay content in core PG2303-1 compared to core HBGC01 suggests a higher proportion of suspended sediments deposited by settling in the deeper part of the Herschel Basin, and thus a quieter depositional environment than at the basin margin (Figs. 11, 16). Moreover, the thin laminations in the top 10 cm of the shallow core HBGC01 (Figs. 4A, A.3) could suggest a more dynamic depositional environment, possibly linked to events of resuspension and reworking (Hill & Nadeau, 1984; Corbett et al., 2006; Hanna et al., 2014) during the last decade. Laminations in sediment cores may also refer to oxygen depletion, but supplementary information would be needed to address this assumption.

In general, the benthic foraminiferal and ostracod assemblages of the two studied cores compare well with assemblages described from other Arctic shelf settings (McDougall et al., 1986; Reimnitz et al., 1993; Wollenburg & Mackensen, 1998; Wollenburg & Kuhnt, 2000; Polyak et al., 2002; Scott et al., 2008b, Husum et al., 2015; Gemery et al., 2017). The dominance of Elphidium clavatum, the abundance of agglutinated taxa such as Spiroplectammina biformis and Textularia spp., and the presence of taxa tolerant to low salinity such as Haynesina orbicularis, Buccella frigida, and polymorphinids, are typical of Arctic shelf environments (Polyak et al., 2002; Scott et al., 2008a; Figs. 13, 18). However, in our study, we report for the first time the occurrence of the accompanying taxon Haynesina nivea on the Beaufort Sea shelf. This species has been associated with shallow and low salinity environments (Madsen & Knudsen, 1994; Luoto et al., 2011; Voltski et al., 2015; see section below on sea ice and salinity).

The two sites also show some important differences in the microfossil assemblages. The shallow core HBGC01 holds high relative abundance of taxa tolerant to river proximal and low salinity environment such as Elphidium clavatum (cf. Hald & Steinsund, 1996; Hald & Korsun, 1997; Polyak et al., 2002; Knudsen et al., 2008), Islandiella helenae (cf. Cage et al., 2021), and Elphidium bartletti (cf. Polyak et al., 2002). The occurrence of Elphidium asklundi in core HBGC01 probably also suggests a tolerance to low salinities. In contrast, Cassidulina reniforme, related to cold and saline conditions (Hald & Korsun, 1997; Polyak et al., 2002), is more abundant at the deeper site of core PG2303-1 (Fig. 18), where euhaline ostracods also dominate (Fig. 19). Stainforthia feylingi is also more abundant in core PG2303-1 than in core HBGC01, which might be due to the higher salinities and closer proximity to productive sea-ice margins (cf. Seidenkrantz, 2013) evident for the central Herschel Basin compared to the near-coastal area. The differences in microfaunal assemblages at the two study sites probably reflect the water depth, with increased seasonal salinity gradient and bottom water instability at the shallower site HBGC01, proximal to the summer halocline and wave action. Salinity variations at site HBGC01 may partly control the changes in foraminiferal assemblages, which is illustrated by the low-salinity tolerant taxa Haynesina nivea and Elphidium bartletti that vary in opposite to Islandiella norcrossi and Cassidulina reniforme (Fig. 12). However, changes in foraminiferal assemblages in the Herschel Basin would primarily be linked with food supply as shown by the occurrence of Eoeponidella pulchella, a foraminiferal taxon indicative of high primary productivity (Knudsen et al., 2008; Wollenburg & Kuhnt, 2000), highly weighted on the first axis of both cores (Figs. 12, 17). This relationship is especially clear in core PG2303-1, as E. pulchella is inversely related to Textularia earlandi, a low trophic requirement taxon (Jennings et al., 2020; Fig. 17).

The sediment substrate and bottom currents may also affect the assemblages. The coarser-grained sediments of the shallow core HBGC01 go together with abundant Elphidium bartletti (Polyak et al., 2002). The ostracod Sarsicytheridea bradii, which is only observed in core HBGC01 (Fig. 14), may also prefer coarser substrates (Gemery et al., 2021). Moreover, Sarsicytheridea bradii can tolerate a wide range of salinity, temperature, and oxygen levels (Stepanova et al., 2019; Gemery et al., 2021), which better corresponds to the more variable conditions that likely characterize the shallow HBGC01 core site. Early instars of ostracods are less abundant in core HBGC01 than in core PG2303-1 (mainly Cytheropteron spp.; Figs. 14, 19), which also suggest a higher-energy environment in the basin margin (Whatley, 1983).

High sand content could potentially point to preservation issues and/or unfavorable enhanced bottom currents at the shallow site HBGC01, as it is opposed to microfossil abundances, especially agglutinated foraminifera (Fig. 12). This relationship is expressed in a few samples at the core top and in the 1969–1975 CE interval (Figs. 11–13). Similarly, particularly abundant sand in Ecozone PG2303-1-B (>∼10%; Fig. 16) may have negatively affected the preservation condition or productivity of the agglutinated foraminifera (Figs. 16, 18). However, the sand content at the deeper site PG2303-1 seems to be associated to a certain extent with increased primary productivity, as it reaches maximum values in Ecozone PG2303-1-B concurrently with high calcareous foraminiferal concentrations and E. pulchella abundances (Figs. 16–18). The sand content is also relatively high in Ecozone PG2303-1-C, accompanied by high ostracod concentrations (Fig. 16) and abundant Semicytherura complanata (Fig. 19). Because Semicytherura complanata has been associated with polynyas (Stepanova et al., 2003), we hypothesize the presence of a polynya-type environment in Ecozone PG2303-1-C with enhanced brine rejections that lead to vertical mixing (Smith & Morison, 1998) and oxygenated bottom waters, in addition to high salinity. Accordingly, Islandiella norcrossi, which prefers saline stable bottom waters (Cage et al., 2021), reaches its maximum occurrence (Fig. 18). Such conditions might have contributed to maximum foraminiferal species diversity at the bottom of the Herschel Basin (Fig. 18).

An important shift is recorded in the shallow core HBGC01 at the transition between ecozones HBGC01-B and HBGC01-A (∼2000 CE). It is marked by a transition from abundant Islandiella norcrossi and Cassidulina reniforme that have preference for salinity >30 psu (Polyak et al., 2002; Cage et al., 2021) to assemblages characterized by Elphidium bartletti and Haynesina nivea that are both tolerant to low salinity (Fig. 13; Madsen & Knudsen, 1994; Polyak et al., 2002; Luoto et al., 2011; Voltski et al., 2015). Hence, the shift at ∼2000 CE seems to correspond to a transition from a relatively stable and saline environment to an environment marked by lower salinity and unstable conditions. The occurrence of H. nivea seems particularly significant to explain this transition. Hence, in modern assemblages of the Scoresby Sound and the White Sea, H. nivea was retrieved in very shallow environments (<10 m), where high seasonal fluctuations occur with warm and low saline summer surface waters and currents fostering high sand content (Madsen & Knudsen, 1994; Voltski et al., 2015). In paleoceanographic studies, this species was often related to shallow marine environments (Feyling-Hanssen & Ulleberg, 1984; Hansen & Knudsen, 1995; Luoto et al. 2011). In Ecozone PG2303-1-B, H. nivea reaches maximum values concomitantly with low salinity tolerant ostracod taxa such as Loxoconcha venepidermoidea and Heterocyprideis sorbyana (Figs. 18, 19).

At the shallow site HBGC01, sea ice may foster stable conditions in the bottom water, as it is a natural barrier for waves and winds (Overeem et al., 2011; Schulze & Pickart, 2012). The transition towards lower mean summer sea-ice concentrations after 2006 CE (Fig. 3) follows slightly the shift in microfaunal assemblages dated at 2000±4 CE and appears closely synchronous with the near-disappearance of Islandiella norcrossi (Fig. 13). The benthic foraminifera I. norcrossi in the lower part of core HBGC01 could thus be an indirect indicator of sea ice-induced stratification and increased bottom-water stability at this site prior to 2006 CE. In contrast, the high sand fraction may indicate increased bottom water instability from wind-induced nearshore currents and wave action in longer ice-free seasons towards the present (Fig. 11), which could have generated reworking and resuspension in the surface sediment (10–0 cm; Figs. 4A, A.3; see the Radiogenic isotope and age model section of core HBGC01).

A significant microfaunal change in Ecozone HBGC01-A is indicated by the high percentages of Eoeponidella pulchella in this zone (Fig. 13). In modern assemblages, E. pulchella is common on Arctic shelves and shelf breaks (Wollenburg & Mackensen, 1998; Wollenburg & Kuhnt, 2000; Scott et al., 2008b; Griffiths, 2010), but was also found in the Gulf of Mexico (Poag & Tresslar, 1981), and is often linked to high primary productivity. In surface sediments of the Canadian Arctic Ocean, E. pulchella constitutes ∼9% of the benthic foraminiferal assemblage off Cape Bathurst (59 m; Scott et al., 2008b), known for the advection of saline nutrient-rich upwelled waters (Tremblay et al., 2011; Walkusz et al., 2012). Eoeponidella pulchella was also noted as an important species (20% of the living fauna) in the surface sample of core PS2480 in Vilkitsky Strait, Laptev Sea (51 m; Fütterer, 1994; Wollenburg & Kuhnt, 2000), which is characterized by high biogenic silica content (Nürnberg, 1996), chlorophyll-a concentrations (Boetius et al., 1996) and marine organic carbon (Boucsein & Stein, 2000). Furthermore, in 1993 CE, the year that core PS2480 was collected, the wind conditions favored upwelling in the Laptev Sea (Janout et al., 2015; Osadchiev et al., 2020). The distribution of E. pulchella in surface sediment samples of the eastern Arctic Ocean documented by Wollenburg & Kuhnt (2000) also supports a strong link to high primary productivity.

In our study, the maximum foraminiferal concentrations occurred concurrently with peaks in E. pulchella at ∼2013 CE in core HBGC01 (Figs. 11–13) and in Ecozone PG2303-1-B (Figs. 16–18), which suggests high benthic foraminiferal fluxes linked to high primary productivity. Wollenburg & Kuhnt (2000) argued that assemblages of benthic foraminifera provide a better indication of trophic level than foraminiferal concentrations due to preservation issues. Still, in our study, the good preservation of foraminiferal shells throughout the sequences allows us to consider the total concentrations and fluxes of benthic foraminifera as a valuable productivity indicator in the Herschel Basin. We also conclude that E. pulchella is an excellent indicator of productivity in the Herschel Basin. However, at present it is not clear if the high productivity is due to longer sea ice-free seasons (Fig. 3), the replenishment of nutrient-rich Pacific waters, increased organic carbon from rivers and coasts, or a mix of all the above (more specifications in the last section of the Discussion).

A special feature in the micropaleontological records of the shallow core HBGC01 is the high Tintinnopsis fimbriata fluxes, particularly pronounced after 2008 CE. This coincides with enhanced T. fimbriata concentrations at the end of Ecozone PG2303-1-A (<9 cm; Figs. 11, 16). Concentrations of T. fimbriata can be underestimated due to their small size. However, specimens tend to be bigger when they are abundant (Rogers et al., 1981), as observed for several tintinnid species (Dolan et al., 2014). The lower abundance of T. fimbriata before 1996 CE is likely not due to poor preservation as it occurs downcore together with abundant agglutinated foraminifera (Fig. 11). Hence, the increase in T. fimbriata fluxes after 1996 CE at the basin margin presumably reflects a real increase in tintinnid abundance.

According to previous studies, T. fimbriata occurs in nearshore areas near river mouths (Echols & Fowler, 1973; Burkovsky, 1976; Rogers et al., 1981). It was probably observed in sediments from the Beaufort Sea shelf by Scott al. (2008a, b), who reported it as Tintinnopsis rioplatensis (Souto, 1973). Tintinnid abundances in the water column have shown positive correlations with chlorophyll-a concentrations (Dolan et al., 2007). Additionally, maximum abundance of T. fimbriata occurred in relation to maximum chlorophyll-a concentration in a Hudson Bay inlet (Rogers et al., 1981) and early spring bloom in the Bedford Basin (Paranjape, 1987; Li, 2014). However, T. fimbriata also reaches peak occurrence during winter in the White Sea (Burkovsky, 1976), when primary production is limited (Chernov et al., 2018), which suggests it could be a low-temperature tolerant opportunistic species. It has also been proposed that the abundance of tintinnids might depend more upon the nature of the prey than the quantity of the food (Dolan et al., 2014).

Preferential occurrence of T. fimbriata in river proximal areas could be related to the availability of small mineral particles to build agglutinated lorica (Pierce & Turner, 1993). Scott et al. (2008b) associated the agglutinated tintinnids with the Mackenzie River water discharges, thus to freshwaters rich in suspended particulate matter. Its association with suspended particles is consistent with the concomitant increase of Quinqueloculina stalkeri, which is known to tolerate turbidity and to be a bacterial-feeder rather than an algal-feeder (Guilbault et al., 2003; Figs. 11, 13). Hence, T. fimbriata might be an indicator of turbid freshwater discharge, but a relationship to primary productivity remains equivocal. The significant increase of T. fimbriata after 2008 CE may reflect a western spreading of the Mackenzie River plume to the study site, which could also be related to reduced sea-ice concentration (Fig. 3). The detrital particles used to build their agglutinated lorica might also originate from a more local source related to coastal erosion that has increased since ∼2000 CE (Jones et al., 2009; Obu et al., 2016; Radosavljevic et al., 2016).

The occurrences of E. pulchella and H. nivea are closely correlated (r = 0.68 and r = 0.78 in cores HBGC01 and PG2303-1, respectively). Both species show increased occurrences in ecozones HBGC01-A and PG2303-1-B, which leads us to suggest a change towards increased productivity concurrently with low salinity excursions. Kutos et al. (2021) already proposed a relationship between benthic productivity and freshwater discharge in the Mackenzie Trough. In our study, the influence of low salinity waters from river discharge is further supported by the abundance of Tintinnopsis fimbriata fluxes over the last two decades in the shallow HBGC01 core (Fig. 11).

The main source of freshwater is the Mackenzie River. However, the Mackenzie River is not particularly rich in nutrients compared to other Arctic rivers (Holmes et al., 2011), especially at distal locations from the plume (Emmerton et al., 2008; Tremblay et al., 2014). Nevertheless, the terrigenous organic carbon and/or the microbial activity therein, either from the more distal Mackenzie River or the proximal permafrost thaw, may provide large food sources (Bell et al., 2016; Terhaar et al., 2021). Another source of nutrients originates from the upwelling of nutrient-rich Pacific waters onto the shelf during easterly winds (Tremblay et al., 2011, 2012; Pickart et al., 2013). The Mackenzie River plume reaching the study site and upwelling of deeper waters may both occur during periods of strong easterly winds, which could explain the duality between freshwater and productivity and the co-occurrence of H. nivea and T. fimbriata with E. pulchella.

The rapidly changing environment of the 21^st century may also have negative impacts on the microfauna, as suggested by the decrease in ostracod fluxes at the shallow site of core HBGC01 (Fig. 11). From the inverse relationship between ostracod and T. fimbriata concentrations (Figs. 11, 16), we suggest that an increased supply of low salinity waters rich in suspended particulate matter at the HBGC01 site is unfavorable for ostracods. Because low salinity is not a limitation for ostracods, as several taxa can tolerate extreme salinity variations (for ex., from 6.5 to >30 psu for Sarsicytheridea bradii; Stepanova et al., 2019), the turbidity is a more likely limiting factor. Vertical mixing of the surface waters rich in suspended particulate matter may lead to turbid bottom waters, especially at the shallow site of core HBGC01 (Hill & Nadeau, 1989; Rainville & Woodgate, 2009; see the bottom water masses section). However, turbidity might not be the direct cause. Various physico-chemical effects derived from increased suspended particulate matter supply, like acidification (Semiletov et al., 2016), could also affect ostracod abundance.

This study presents micropaleontological data of benthic foraminifera, ostracods, and tintinnids that provide time series with annual to multiannual time resolution in an Arctic nearshore environment. The three groups of microfossils provide complementary information on pelagic and bottom water conditions at two sites in the Herschel Basin, one near the summer halocline (HBGC01, 18 m depth) and the other below the halocline (PG2303-1, 32 m depth). The microfaunal communities of the two cores illustrate the sensitivity of micropaleontological tracers to water conditions at local scales, especially with regards to salinity and productivity. The 40-cm-long core HBGC01 covers the period ca. 1969–2018 CE, while a satisfactory age model could not be established for PG2303-1 (20 cm long). A redundancy analysis performed on the foraminiferal assemblages in both cores permitted us to identify ecozones that were also consistent with the ostracod assemblage data and Tintinnopsis fimbriata abundance. Thus, the three studied microfossil groups responded simultaneously to decadal environmental changes and can be used as complementary paleo-tracers to distinguish the impacts of recent climate changes over long-scale environmental variations in Arctic nearshore areas.

Among the most important microfaunal changes is an increase in the benthic foraminifera Eoeponidella pulchella, which indicates enhanced primary productivity since ∼2000 CE. The last two decades are also marked by the high occurrence of the shallow, low-salinity species Haynesina nivea, while Islandiella norcrossi, which requires more stable saline conditions, decreased in abundance. The microfaunal record of the shallow core HBGC01 thus suggests a transition from stable saline bottom waters to more unstable bottom water salinity. The decrease in summer sea-ice concentration that occurred after 2003 CE and more clearly after 2006 CE might best explain this transition at the basin margin, as the loss in summer sea-ice cover may have led to increased vertical mixing and thus to lower and varying bottom-water salinity. Moreover, the high fluxes of the tintinnid T. fimbriata combined with increased sand content, the abundance of Quinqueloculina stalkeri, and a reduction in ostracod concentrations during the most recent decade points to enhanced turbidity of the water. Turbid waters could derive from increased inputs of particulate matter-rich freshwaters related to Mackenzie River discharge, local rivers and/or coastal erosion, which could be facilitated in the context of reduced sea-ice cover. We suggest that increased productivity might be caused by enhanced terrigenous organic fluxes and/or upwelling during easterly wind events, which would be amplified under reduced summer sea-ice concentration.

We wish to thank the Fonds de recherche du Québec Nature et technologies (FRQNT) for their financial support to JF through scholarships. Additional thanks are due for technical support provided by the Geotop Research Center. This study has been supported by the Natural Sciences and Engineering Research Council (NSERC) of Canada through a Discovery grant to AdV, a CREATE grant to ArcTrain [grant no. 432295] and a National Council of Science and Technology (CONACYT) postdoc grants from Mexico [CVU no. 174856] to VCB. Fieldwork was also supported by Polar Knowledge Canada through the Northern Scientific Training Program. M.-S. Seidenkrantz is funded by the Danish Council for Independent Research [grant no. 7014-00113B G-Ice project, and 0135-00165B GreenShelf] and through the European Union's Horizon 2020 research and innovation program under Grant Agreement [grant no. 869383; ECOTIP]. We want to thank the EU Horizon 2020 project Nunataryuk [BG-2017-1] and the Aurora Research Institute for making the fieldwork possible and the Qikiqtaruk Territorial Park, chief ranger Richard Gordon and the other park rangers, who have welcomed our team on Herschel Island – Qikiqtaruk. We want to thank all the colleagues who helped in the coring and mapping activities, mostly George Tanski and Konstantin Klein. We acknowledge many other colleagues and collaborators who participated in this study: Charles Brunette for the sea-ice data, Diogo Barnetche for the rental of equipment, Anna To for lab work, Tiffany Audet for managing the laboratory, Pierre Francus and Patrick Lajeunesse for providing access to CT-scan and laser granulometer facilities, Nicole Sanderson for her support with the Plum software and Pierre Poitevin for the photographs with the Keyence. TMC and LG were funded by the U.S. Geological Survey Climate Change Research Program. Any use of trade firm, or product names is for descriptive purposes only and does not imply endorsement by the U.S. Government. Finally, we are grateful to the anonymous reviewers of the Journal for their thorough examination of the manuscript and constructive comments. Appendices can be found linked to the online version of this article.