Volume 35, Issue 2 e2019PA003722
Research Article
Open Access

Holocene Hydrographic Variations From the Baltic-North Sea Transitional Area (IODP Site M0059)

S. Ni

Corresponding Author

S. Ni

Department of Geology, Lund University, Lund, Sweden

Centre for Environmental and Climate Research, Lund University, Lund, Sweden

Correspondence to: S. Ni,

[email protected]

Search for more papers by this author
N. B. Quintana Krupinski

N. B. Quintana Krupinski

Department of Geology, Lund University, Lund, Sweden

Search for more papers by this author
J. Groeneveld

J. Groeneveld

MARUM, Bremen University, Bremen, Germany

Helmholtz Centre for Polar and Marine Research, Alfred Wegener Institute, Potsdam, Germany

Search for more papers by this author
A. S. Fanget

A. S. Fanget

Palaeoceanography and Palaeoclimate Group, Interdisciplinary Centre for Climate Change (iClimate), Department of Geoscience, Aarhus University, Aarhus, Denmark

Search for more papers by this author
M. E. Böttcher

M. E. Böttcher

Geochemistry and Isotope Biogeochemistry Group, Leibniz Institute for Baltic Sea Research (IOW), Warnemünde, Germany

Marine Geochemistry, University of Greifswald, Greifswald, Germany

Interdisciplinary Faculty, Department of Maritime Systems, University of Rostock, Rostock, Germany

Search for more papers by this author
B. Liu

B. Liu

Alfred-Wegener-Institute, Helmholtz Centre for Polar and Marine Research, Section of Marine Geochemistry, Bremerhaven, Germany

Search for more papers by this author
M. Lipka

M. Lipka

Geochemistry and Isotope Biogeochemistry Group, Leibniz Institute for Baltic Sea Research (IOW), Warnemünde, Germany

Search for more papers by this author
K. L. Knudsen

K. L. Knudsen

Palaeoceanography and Palaeoclimate Group, Interdisciplinary Centre for Climate Change (iClimate), Department of Geoscience, Aarhus University, Aarhus, Denmark

Search for more papers by this author
T. Naeraa

T. Naeraa

Department of Geology, Lund University, Lund, Sweden

Search for more papers by this author
M.-S. Seidenkrantz

M.-S. Seidenkrantz

Palaeoceanography and Palaeoclimate Group, Interdisciplinary Centre for Climate Change (iClimate), Department of Geoscience, Aarhus University, Aarhus, Denmark

Search for more papers by this author
H. L. Filipsson

H. L. Filipsson

Department of Geology, Lund University, Lund, Sweden

Search for more papers by this author
First published: 27 January 2020
Citations: 8


Deoxygenation affects many continental shelf seas across the world today and results in increasing areas of hypoxia (dissolved oxygen concentration ([O2]) <1.4 ml/L). The Baltic Sea is increasingly affected by deoxygenation. Deoxygenation correlates with other environmental variables such as changing water temperature and salinity and is directly linked to ongoing global climate change. To place the ongoing environmental changes into a larger context and to further understand the complex Baltic Sea history and its impact on North Atlantic climate, we investigated a high accumulation-rate brackish-marine sediment core from the Little Belt (Site M0059), Danish Straits, NW Europe, retrieved during the Integrated Ocean Drilling Program (IODP) Expedition 347. We combined benthic foraminiferal geochemistry, faunal assemblages, and pore water stable isotopes to reconstruct seawater conditions (e.g., oxygenation, temperature, and salinity) over the past 7.7 thousand years (ka). Bottom water salinity in the Little Belt reconstructed from modeled pore water oxygen isotope data increased between 7.7 and 7.5 ka BP as a consequence of the transition from freshwater to brackish-marine conditions. Salinity decreased gradually (from 30 to 24) from 4.1 to ~2.5 ka BP. By using the trace elemental composition (Mg/Ca, Mn/Ca, and Ba/Ca) and stable carbon and oxygen isotopes of foraminiferal species Elphidium selseyensis and E. clavatum, we identified that generally warming and hypoxia occurred between about 7.5 and 3.3 ka BP, approximately coinciding in time with the Holocene Thermal Maximum (HTM). These changes of bottom water conditions were coupled to the North Atlantic Oscillation (NAO) and relative sea level change.

Key Points

  • Foraminiferal geochemistry, faunal, and pore water δ18O data document changes in temperature, salinity, and oxygen concentration over the last 7.7 kyr
  • The bottom water in the Little Belt was warmer and hypoxic ~7.5-3.3 ka BP and became fresher, cooler and more ventilated since ~4.1 ka BP
  • Reconstructed bottom water conditions are linked to large scale climate variability related to the NAO and relative sea level change

Plain Language Summary

The Baltic Sea is an intracontinental sea connected to the North Atlantic Ocean with large economic and societal values. It is sensitive to salinity and temperature changes and low oxygen conditions in the bottom water. The Little Belt is a transitional area between the central Baltic Sea and the North Sea. It is an ideal area to study the changes of freshwater outflow and saline water inflow and how the dynamics vary in relation to larger-scale climate changes. We have reconstructed past seawater conditions (e.g., temperature, salinity and oxygenation) in the Baltic by geochemically analyzing microfossil (foraminifera) shells. We analyzed the trace elemental concentration and stable oxygen and carbon isotopes in calcite shells of low-oxygen tolerant foraminiferal species as well as the benthic foraminiferal assemblages from a sediment core originating from the Little Belt covering the past 7,700 years. The salinity increased dramatically ~7,700–7,500 years ago and decreased ~4,100–2,500 years ago. Variation in bottom water oxygen content and bottom water temperature coincide. Our study highlights the benefit of using modern marine monitoring data together with a multiproxy approach to establish the link between local hydrographic conditions and regional climate changes and to explain how the environment has developed over time.

1 Introduction

The Baltic Sea is one of the world's largest semi-enclosed brackish water basins with a catchment area of 2.13 × 106 km2, which is nearly 20% of the European continent (Nilsson, 2006). The catchment area has a total population of 85 million people across nine coastal and five upstream countries (International Council for the Exploration of the Sea (ICES), 2018). The sea experiences a considerable freshwater input from river runoff from the large surrounding catchment area (16,100 m3 s−1) but only a restricted inflow of saline water from the North Sea through the narrow Danish Straits (the Little Belt, the Great Belt, and the Öresund; Figure 1), making it a prime example of a restricted epi-continental sea. The restricted water exchange results in a permanent halocline that separates an upper layer of brackish water from more saline bottom waters. The permanent stratification in the deeper basins of the central Baltic plays an essential role in the bottom-water hypoxia and anoxia seen in many parts of the basin (e.g., Böttcher & Lepland, 2000). Here a strong halocline and a seasonal thermocline further weaken vertical circulation and ventilation of the bottom water (Lass & Matthäus, 2008). The salinity changes in the Baltic depend on the irregular inflow of North Sea water through the Danish Straits and the amount of freshwater runoff, which are in turn controlled by climatic factors in the Atlantic (Hänninen et al., 2000). The Baltic freshwater outflow into the Nordic Seas has a large potential influence on water mass transformation (Lambert et al., 2016, 2018; Winsor et al., 2001), by enhancing Polar Water outflow and suppressing Deep Water outflow (Lambert et al., 2018).

Details are in the caption following the image
Map of the southwestern Baltic Sea (basemap source: http://www.geomapapp.org). The position of IODP Site M0059 and long-term monthly monitoring hydrographic data (the Little Belt site, 55°00′N, 10°10′E) is shown as a red star. Water column samples taken during cruise MSM50 are shown as red open circles. Solid (surface water) and dashed (deep water) arrows indicate dominant current directions. The North Sea, Skagerrak, Kattegat, the three Danish Straits (Little Belt, Great Belt and Öresund), the Arkona Basin, the Drogden, and Darss Sills and the Baltic Sea are shown on the map.

Strong stratification due to warming and water mass salinity contrasts can lead to bottom-water hypoxia in the Baltic Sea. Like many other continental shelf seas, it is frequently affected by hypoxia (Diaz & Rosenberg, 2008), which is further amplified with increased temperatures (Murray & Riley, 1969). Hypoxia is defined as <1.4 ml/L dissolved oxygen concentration ([O2]) in the water. It has occurred intermittently, in particular in the deepest basins, in the Baltic Sea throughout the Holocene (Jilbert & Slomp, 2013; Kabel et al., 2012; van Wirdum et al., 2019), and the area of hypoxia has expanded sixfold from 1950 to 2000 CE (Carstensen et al., 2014). Baltic Sea hypoxic events have been steadily increasing with time and affect nutrient biogeochemical processes, ecosystem services, coastal habitats (Conley et al., 2009, 2011; Ning et al., 2018), and fisheries (Breitburg, 2002).

The Danish Straits hydrographically represent the transitional area between the Baltic Proper and the North Sea, and the salinity of these narrow straits is determined by the balance between the brackish water outflows and the intermittent saline water inflows from the North Sea through the Kattegat and Skagerrak (Figure 1; Lass & Matthäus, 2008). The straits play an important role in the global ocean as they connect the largest brackish water system in Northern Europe to the Nordic Seas and the Atlantic Ocean. The straits are thus ideal sites to study the water exchange between the Baltic Sea and the North Sea and the role of such narrow gateways for water exchange in general. The flushing ratio between the three straits (the Little Belt, the Great Belt, and the Öresund) is 1:7:3 (Jacobsen, 1980; Helsinki Commission (HELCOM), 1986). The Little Belt is the smallest of the three Danish Straits (cross-section area: ~16 × 103 km2; Hela, 1944), making it the most sensitive to changes in the hydrographical conditions in the Danish Straits. Therefore, the reconstruction of the hydrographic conditions in the Little Belt provides an indication of the conditions in the Danish Straits area as a whole.

Several studies have demonstrated that the North Atlantic Oscillation (NAO) (quantified by the NAO index) significantly influences the winter climate of the North Atlantic and northern European regions (e.g., Hurrell, 1995; Lu & Greatbatch, 2002; Trigo et al., 2002). Positive values of the NAO index are generally characterized by stronger westerly winds over northern Europe, usually causing the winter climate in northern Europe to be mild and wet. In contrast, a negative NAO index is associated with weaker westerly winds and more easterly/northeasterly winds, resulting in a cold and dry climate in northern Europe. In addition, variations in NAO drive sea-level changes (e.g., Marshall et al., 2001; Mehta et al., 2000; Wakelin et al., 2003), temperature (e.g., Hurrell, 1996; Hurrell et al., 2003), precipitation and freshwater fluxes (e.g., Hurrell, 1995; Hurrell et al., 2003). The wind field over Scandinavia controls the oscillations in water exchange in and out of the Baltic Sea and therefore also bottom water oxygenation. Westerly winds force more saline and oxygenated seawater from the North Sea-Skagerrak areas into the Baltic Sea, whereas easterly winds force brackish surface water out of the central Baltic Sea (Jakobsen & Ottavi, 1997; Lehmann et al., 2017). Major Baltic Inflows primarily occur between November and January and have shown very low frequencies during the summer months in the last century (Matthäus & Franck, 1992). During winter, strong easterly winds force the brackish surface water out of the Baltic Sea as a stronger Baltic Current and create a water level difference between the Baltic Sea and the North Sea. When this is followed by a period of stronger westerly winds (i.e., during a positive NAO phase), it results in saline inflows into the Baltic Sea through the Little Belt and increasing water depth (i.e., sea-level rise) and bottom water salinity. Furthermore, bottom-water cooling and increased oxygen content are attributed to these major inflow events, particularly if they occur between January and April (Matthäus et al., 2008). The importance of the outflows from the Baltic (their strength and quantity) has also recently been highlighted as a very important factor for the living quality of the benthic environment in the Öresund-Danish Strait-Kattegat region (Charrieau et al., 2019).

The Baltic Sea has undergone six major environmental stages (lacustrine and marine) since the last deglaciation (e.g., Andrén et al., 2011; Björck et al., 2008): the Baltic Ice Lake (~16–11.7 ka BP), the Yoldia Sea (11.7–10.7 ka BP), the Ancylus Lake (10.7–9.8 ka BP), the Initial Littorina Sea (9.8–8.5 ka BP), the Littorina Sea (8.5 ka BP to ~3 ka), and finally the present-day Baltic Sea. By ~10 ka BP the entire Baltic basin was deglaciated and sea-level rose above the straits' sills. The opening history of the three Danish Straits is complex, but it likely occurred between 9.3 and 8.1 ka BP (reviewed by Gyllencreutz et al., 2006). The marine ingression through the Great Belt occurred first ~9.0–8.7 ka BP and, through the Öresund slightly later (Gyllencreutz et al., 2006), with fully-brackish conditions reaching the southern Baltic at ~8.5 ka BP. The oldest dated marine shell from the Little Belt at IODP Site M0059 is from 7.7 ka BP, but brackish water conditions may already have been established between 8.6 and 8.3 ka BP (Bennike & Jensen, 2011). The long-term mean sea level in the Baltic Sea is mainly controlled by three factors: land uplift, eustatic sea level change, and water balance of the Baltic Sea (Johansson et al., 2001, 2003; Meier et al., 2004); this can further influence the water temperature and salinity in a relatively narrow and shallow water environment like the Little Belt.

The general history of the Baltic is thus well known; however, it is essential to investigate further and in detail how temperature, salinity, and oxygen conditions have varied in the past in order to improve our understanding of the causes and consequences of environmental changes in the Baltic Sea. In particular, knowledge of these parameters will allow us to evaluate the overall stability of water exchange with the open ocean and evaluate the role of NAO on water exchange and hypoxia. This will provide a better insight into the sensitivity of the Baltic Sea and other enclosed inland seas to external changes of the types expected to occur with current and future anthropogenic climate change.

To provide the strongest reconstruction of changes in temperature, salinity, and oxygenation in the Little Belt, we use a multiproxy approach based on benthic foraminifera, as summarized here. The occurrence and abundance of specific species of benthic foraminifera are indicators of bottom water conditions such as salinity and [O2] (e.g., Conradsen et al., 1994; Murray, 2006; Seidenkrantz, 1993). The oxygen isotopic composition of benthic foraminiferal calcite (δ18Oc) is a well established and the most widely used proxy for temperature and salinity of seawater (e.g., Elderfield et al., 2010; see Pearson, 2012 for review). In a semi-enclosed basin such as the Baltic Sea, it is possible to construct a relationship between δ18Oc and salinity that represents a mixing line among different water masses (e.g., Fröhlich et al., 1988; Harwood et al., 2008). The mixing line can reveal salinity changes resulting from evaporation and precipitation, freshwater run-off and sea ice processes (e.g., LeGrande & Schmidt, 2006; Rohling & Cooke, 1999). Mg/Ca in benthic foraminifera is used as a bottom water temperature proxy (e.g., Elderfield et al., 2006; Filipsson et al., 2010; Groeneveld & Filipsson, 2013; Katz et al., 2010; Kristjánsdóttir et al., 2007). In addition, the Ba/Ca in foraminiferal calcite has been proposed as a proxy for dissolved Ba/Ca ratios in seawater (e.g., Hönisch et al., 2011; Lea & Boyle, 1989, 1991), which in coastal areas generally reflects the amount of freshwater runoff from land (e.g., Groeneveld et al., 2018). Therefore, Ba/Ca of foraminiferal calcite provides information on salinity, nutrients, and alkalinity distributions in seawater. Manganese is a redox-sensitive element, and its concentration in bottom and pore waters increases under hypoxic and anoxic conditions, which result in incorporation of more divalent Mn into the calcite lattice (Böttcher & Dietzel, 2010; Boyle, 1983). Mn/Ca in benthic foraminiferal calcite is being developed as a proxy for low oxygen conditions (e.g., Glock et al., 2012; Groeneveld & Filipsson, 2013; Groeneveld et al., 2018; McKay et al., 2015; Koho et al., 2015, 2017). Stable carbon isotopes in foraminifera (δ13Cc) are controlled by productivity, carbon cycling, and water mass exchanges, (e.g., Filipsson & Nordberg, 2010; Schmiedl & Mackensen, 2006). Thus, both Mn/Ca and δ13Cc in benthic foraminifera can reflect past sea-water circulation and ventilation rates of the bottom water.

We apply this multiproxy approach to generate high-resolution reconstructions of Holocene bottom water conditions in the Little Belt by using the benthic foraminiferal geochemistry proxies and faunal assemblage analyses described above in a high-resolution sediment core from the Little Belt (Site M0059 from IODP Expedition 347) (Figure 1). A lower-resolution subset of some of these data (benthic foraminiferal faunal assemblages, Mg/Ca measured by solution-ICP-OES (Figure S1 in the supporting information), and foraminiferal δ18O) was published by Kotthoff et al. (2017). In the present study, we substantially increase the temporal resolution of the data sets (benthic foraminiferal faunal assemblages and carbon and oxygen isotopes) and we present new proxy variables (trace element/Ca measured by Laser Ablation Inductively Coupled Mass Spectrometry (LA-ICP-MS) and pore water oxygen isotopes (δ18Opw)). We have selected samples spanning the time since our site became brackish-marine, that is, the last 7.7 kyr, which include major salinity changes and warming periods and, therefore, may provide a valuable analogue for future changes. The oldest part of the study period (~7.7 ka BP) includes the final portion of the transition from fresh-water conditions into the mid-Holocene brackish-marine Littorina Sea stage in the study area. We investigate the linkage between modern environmental conditions (i.e., temperature, salinity, and [O2]) in the Little Belt and the NAO index and evaluate whether a similar linkage existed in the past. Our study's aims are to investigate the drivers and consequences of local environmental changes and to evaluate local environmental conditions (the Little Belt) as an indicator of regional climate changes (the Baltic Sea and the North Atlantic) over the past 7.7 kyr.

2 Modern Hydrographical Setting

Relatively low-salinity surface water exits the Baltic Sea through the Little Belt, while higher salinity inflows pass the main sills (Drogden and Darss Sills, Fischer & Matthäus, 1996) and become subsurface water when entering the Baltic Proper. Freshwater input in the Little Belt area is highest in May and June and lowest in January and February (Jakobsen, 1995). Long-term monthly monitoring hydrographic data covering the period CE 1975–2018 are available for the Little Belt (55°00′N, 10°10′E) from the Baltic Nest Institute, Stockholm University Baltic Sea Centre (Figure 2).

Details are in the caption following the image
Little Belt monthly hydrographic monitoring data from Baltic Nest Institute, Stockholm University Baltic Sea Centre, 1975–2018, of (a) surface water (0–7 m) temperature, (b) bottom water (26–37 m) temperature, (c) surface water salinity and (d) bottom water salinity, (e) buoyancy frequency, and (f) bottom water dissolved [O2], the dashed line indicates hypoxic level ([O2] <1.4 ml/L). The dots indicate outliers with 1.5 Interquartile Rule.

2.1 Seasonal Variability in the Little Belt

The surface (0–7 m) and bottom waters (26–37 m) in the Little Belt area show strong seasonal variability (Figure 2). The average surface water temperature is ~17 °C in summer and 4 °C in winter. The monthly average bottom water temperature varied between 3 and 12 °C, and maximum and minimum temperatures occur in October and March, respectively. The temperature difference between surface and bottom water is largest between May and September (>5 °C). From November to February, the monthly average temperature in the bottom water is 1–2 °C higher than that in the surface water, whereas during the other months the bottom water is cooler than the surface water. The monthly average salinity is 14–19 in the surface water and 21–25 in the bottom water. The salinity difference between surface and bottom water is the largest (~9) from May to August. As an estimation of water column stability/stratification, we calculate the Brunt-Väisälä (buoyancy) frequency (N) following Godhe and McQuoid (2003),
where g is gravitational acceleration (m s−2), z is water depth (m), and ρ is density (kg m-3). Depths used in the calculation were 5 and 30 m. Density is calculated from temperature, salinity, and pressure according to UNESCO (1981) and Massel (2015).

The average buoyancy frequency during the summer is 0.057 s-1, whereas during the winter the buoyancy frequency is much lower (~0.032 s-1). The large difference in salinity and thereby a strong halocline during summer time is the primary contributor to the high summer buoyancy frequency in the Little Belt. It indicates a stronger stratification in the water column, which may result in low dissolved [O2] in the bottom water during summer. A strong pycnocline develops from May to September in the Little Belt, which also limits vertical mixing and promotes hypoxic bottom water. Seasonal hypoxia occurs in the bottom water and dissolved [O2] is typically lowest, <1.4 ml/l, in September, while the bottom water is generally fully oxygenated during the winter and spring (Figure 2).

3 Materials and Methods

3.1 Study Site, Lithology, Age model

Samples were collected during IODP Expedition 347 from Site M0059 (55°0.29′N, 10°6.49′E; 37.1 m water depth), located in the southern section of the Little Belt (Figure 1). IODP Expedition 347 obtained high-resolution sediment cores from the Little Belt covering the Holocene. The sedimentary sequence of Site M0059 has been divided into seven lithostratigraphic units (I–VII) (Andrén et al., 2015). Our study focuses on Subunit Ia (0 to 49.37 m composite depth (mcd)), which is the interval since the start of the Early to mid-Holocene brackish-marine Littorina Sea in the region. Subunit 1a is composed of black to greenish black, organic-rich clay with millimeter-scale lamination and minor bioturbation (Andrén et al., 2015).

We use the age-depth model of van Helmond et al. (2017) for the depth interval 0–48.6 mcd at Site M0059, which gives a calibrated age of −63 to 7437 year BP (where present is CE 1950). The age model is based on 16 radiocarbon (14C) datings of intact bivalve specimens and bivalve fragments, and the sedimentation rate is generally steady with no abrupt changes in Subunit Ia (0.66 cm/year). In the bottom meter of Subunit Ia (48.6–49.37 mcd), ages were extrapolated linearly with the constant sedimentation rate of the interval above since no datable material was found.

We analyzed 127 depth intervals for foraminiferal δ18Oc and δ13Cc, 37 depth intervals for trace elemental geochemistry (a total of 215 foraminiferal specimens) and 264 depth intervals for foraminiferal faunal assemblages at Site M0059. This resulted in an average age resolution of 56 years for δ18O and δ13C samples, 182 years for trace element samples, and 32 years for faunal assemblage samples. We analyzed 42 pore water samples for δ18O and subsequently modeled the results to account for sediment compaction and pore water transport.

3.2 Benthic Foraminiferal Faunal Assemblages and Grain Size

We prepared foraminiferal samples following standard processing methods (Murray, 2006) via wet-sieving of 20 cm3 sediment samples at mesh diameters 63, 100, and 1,000 μm. We used the 100 to 1,000 μm fraction for foraminiferal faunal assemblage analyses. For the sediment samples with a high concentration of mineral grains, we performed foraminiferal separation by using tetrachlorethylene (C2Cl4) with a specific density of 1.6 g cm−3 (for detailed description see Kotthoff et al., 2017). In each sample, we counted at least 300 foraminiferal specimens, identifying them to species level based on Feyling-Hanssen et al. (1971) and Feyling-Hanssen (1972), when possible. We assigned Elphidium clavatum Cushman and Elphidium selseyensis (Heron-Allen and Earland) to an E. clavatum-selseyensis complex due to the difficulties of reliably distinguishing the gradational morphologies (Groeneveld et al., 2018); however, our foraminiferal assemblage analyses show that E. selseyensis is the dominant of the two species. We calculated the relative abundance of each species, and if the relative abundance was >5% in at least one sample, we considered it a major species. We calculated the benthic foraminiferal concentrations as the number of individuals per cm3 sediment and the Shannon index H by using PAST3 software to describe the faunal diversity. We obtained grain size data from the mass of the sand-sized fraction (>63 μm) from wet sieving during sediment sample preparation relative to the dried bulk sediment mass. Foraminiferal photographs (Figure S6) were prepared using the Leica Application Suite v. 4.12 and the Helicon Focus 7.0 stacking software.

3.3 Trace Element Ratios

For trace elemental analyses, we picked foraminifera from the size fraction 125–355 μm. We selected between three and seven specimens with complete final chambers from 37 depth intervals. We measured trace element ratios in the species E. clavatum-selseyensis complex on single foraminiferal chambers using LA-ICP-MS. Before trace element analysis, we briefly sonicated the foraminifera in MilliQ water and then methanol following Vetter et al. (2013).

A potential question in the study of any chemical proxy in marine sediments is the extent to which the primary “fossil” chemical signature was altered by postdepositional processes. This is particularly important to consider in hypoxic environments that are experiencing redox-related chemical reactions. We chose to use LA-ICP-MS analysis because it allows us to distinguish between trace metals associated with diagenetic coatings on shells (tests) and the primary calcium carbonate of the test. We conducted LA-ICP-MS analyses at Lund University using a Bruker Aurora Elite (quadrupole) ICP-MS and a 193 nm Cetac Analyte G2 excimer laser installed with a two volume HelEx2 sample cell. We used Helium as carrier gas (approximately 0.8 l/min) and combined this with Ar (approximately 0.95 l/min) downstream of the sample chamber. We tuned the instrument, using NIST612 to obtain high and stable signal counts on relevant elements, low oxide production (below 0.5% monitoring 238U/238U16O and 232Th/232Th16O), and 232Th/238U ratios close to 1.

We ablated all foraminiferal tests from the outside toward the inside. Whenever possible, we selected the final (f), penultimate (f-1), and prepenultimate (f-2) chambers for ablation. We quantified element/Ca ratios using counts of the isotopes 24Mg, 43Ca, 55Mn, 88Sr, 27Al, and 137Ba. We calculated the element ratios based on averaging measured concentrations during each ablation after selecting the noncontaminated part of the ablation profile (removing the outer and inner layers) in order to avoid elevated element concentrations on the surface of the test (Figure S2). We used 43Ca as the internal standard and NIST SRM 610 glass standard as the external calibration material (using established values from Jochum et al., 2011) because the NIST SRM 610 glass standard has been demonstrated to be more accurate and stable for elements with 193 nm laser than existing carbonate standards (Hathorne et al., 2008). We conducted laser ablation in manual mode and analyzed primary standards to correct for instrumental drift at the beginning of each sequence and throughout the run after roughly each 10 spot analyses. We measured baseline elemental levels before each analysis for minimum 30 s. We used NIST SRM 612 glass (Jochum et al., 2011) and calcium carbonate pellets of MACS-3 (Jochum et al., 2012), JCp-1, and JCt-1 (AIST Japan) as quality control material. We measured NIST SRM 610 and 612 glass standards at higher energy density (3 J/cm2), whereas for all carbonate material we used 1 J/cm2. We set the laser energy to 4 Hz and used a 30 × 30 μm spot size for both samples and standards. The changing of energy density (fluence) does not influence relevant element concentrations (Dueñas-Bohórquez et al., 2011). We converted raw counts to element concentrations (ppm) using the software package Igor Pro 6.37, Iolite v3.5.

In total, we used 402 ablation profiles on foraminiferal chambers for the trace element ratio calculations. All the ablation profiles we used fulfil the following criteria: (1) >12 sweeps (i.e., 3 s) of laser ablation, (2) averaged 27Al concentrations of selected signal portions were below the limit of detection (the average of the background plus three times the standard error) in order to eliminate the spots with clay contamination. We observed that the trace element/calcium (TE/Ca) ratios of exterior and interior diagenetic coatings can be up to 40 times higher than the ratios of the original calcite (Figure S2). By omitting these coatings from the sample integration peak in data processing, Mg/Ca ratios of foraminiferal tests decrease by 30%, 72% for Mn/Ca and 29% for Ba/Ca. We first calculated mean values of trace element ratios of f, f-1, and f-2 chambers for each foraminiferal specimen and then calculated the median values of foraminiferal specimens at each depth interval with standard error.

We employed the Mg/Ca-temperature calibration established for Melonis barleeanus (Williamson) for semiquantitative temperature reconstruction (Kristjánsdóttir et al., 2007):
following Kotthoff et al. (2017) and Groeneveld et al. (2018). We did not apply the E. clavatum calibration of Barrientos et al. (2018) as it only covers the temperature range −2–0.5 °C and therefore is well outside of the range of temperatures expected at our study site. It should be noted that using a temperature calculation created for a different species than the one on which analyses are performed introduces some uncertainty.

3.4 Stable Oxygen and Carbon Isotopes in Foraminifera

We measured δ18Oc and δ13Cc on 124 samples (10–20 pooled specimens from the 100–1000 μm fraction) of the E. clavatum-selseyensis complex using a Thermo Finnigan MAT 251 gas isotope ratio mass spectrometer at MARUM, Bremen University, Germany. We express stable isotope values in per mil (‰) relative to the VSMOW and V-PDB standards in δ notation. We monitored accuracy and reproducibility by replicate analysis of the in-house standard (Solnhofen limestone) calibrated to NBS19, and one standard deviation of this in-house standard is ±0.03‰ for δ13Cc and ±0.06‰ for δ18Oc.

3.5 Oxygen Isotopes in Sea Water and Pore Water (δ18Opw)

Forty-seven water column samples were collected during cruise MSM-50 of the RV Maria S. Merian in 2016 on a transect between the Skagerrak and the southern Baltic Sea (Bathmann et al., 2017) to establish the salinity-oxygen isotope covariation in modern waters within the North Sea-Baltic Sea transition. Water samples were taken at each station at three water depths (surface, middle and bottom) covering water depths between 1.3 and 379.1 m.

Water sampling took place using a Seabird SBE911plus CTD on an SBE-32 multiwater sampler equipped with 24 (10 l) Niskin bottles. Temperature and conductivity were measured with SBE-3plus and SBE-4C sensors, respectively (von Bröckel et al., 2018). Interstitial and pelagic water samples were stored in glass vials after on-board sampling until further processing in the IOW laboratory.

Pore water was extracted from the sediment immediately after recovery, using either rhizon samplers or squeezers, and subsampled under inert gas for stable water isotopes and further hydrochemical analyses according to standard IODP procedures (Andrén et al., 2015; Egger et al., 2017).

All stable oxygen isotope measurements (δ18O) were conducted by means of cavity ring-down spectroscopy (Picarro L2140-i) at the Leibniz Institute for Baltic Sea Research (IOW), with a reproducibility of better than ±0.05‰ (1 standard deviation; Böttcher et al., 2014). Results are presented in the δ notation versus the international V-SMOW standard, using the IAEA standards VSMOW, SLAP, GISP, and USGS48, besides in-house standards, to scale the isotope measurements.

3.6 Pore Water Oxygen Isotope Modeling (δ18Ompw)

We modeled advection and diffusion of pore water constituents since time of burial, as shown by H218O, using a one-dimensional diffusion/advection equation under steady compaction (Berner, 1980; Boudreau, 1997):
where z is the depth, t is the time, ϕ is the porosity, τ2 is the tortuosity, ω is the sedimentation velocity at depth, and D0 is the molecular diffusion coefficient of H218O. We calculated porosity as a function of depth (z),
where ϕ0 is the porosity at the water-sediment interface and ϕ is the depth below which porosity does not change. Υ is the porosity e-folding distance. By fitting the measured porosity data in Site M0059, we get ϕ = 0.63, ϕ0 = 0.90, and Υ = 50 m. We used the mean sedimentation rate ω of 6.6 mm year−1 from the M0059 age model of van Helmond et al. (2017). We used the function of tortuosity from Boudreau (1996),
We calculated the molecular diffusion coefficient of H218O as a function of temperature by using the equation given in Boudreau (1997),
where T is the temperature. The mean temperature was around 8 °C based on the multiproxy-based reconstruction of paleotemperature, paleosalinity, and paleoecosystem changes from the Little Belt (Site M0059) over the past ∼8,000 years (Kotthoff et al., 2017). With a1 = 4.2355605, a2 = 0.1642818, a3 = -0.3525408, and averaged temperature 8 °C, we get D0 = 0.0413 m2/year. The parameters of porosity, temperature, and sedimentation rate used in the model are slightly different from those used by Egger et al. (2017).

We solved the differential equation (equation 3) by finite difference scheme. We set the sediment column height to 50 m and the duration of the diffusion/advection simulation to 7.2 kyr. We assumed the initial vertical geochemical profile to be homogeneous δ18Ofw = −12 at the freshwater lake period all along the length domain. We forced the model top boundary condition with the time evolution of the water-sediment interface concentration using the tracer of δ18O(t). At each time step, the model accounted for the new boundary condition at the top and vertically advected and diffused the tracer (δ18O), according to equation 3. δ18O(t) increased linearly from δ18Ofw = −12 to maximum δ18Omax = −1 within 400 years (7.7 to ∼7.3 ka BP) and remained at δ18Omax for another 3,300 years (until 4.0 ka BP). Then δ18O(t) gradually decreased to the current δ18Ocurr = −2.3 within 2.9 kyr (4.0 to 1.7 ka BP) and kept at δ18Ocurr up to 0.5 ka BP. We did not apply a diffusive boundary condition at the model bottom boundary.

We applied modeled pore water δ18O data to the paleotemperature equation (Shackleton, 1974):
in order to solve for the influence of temperature on oxygen isotope fractionation in foraminiferal calcite.

3.7 Statistical Analyses

Because the sampling depth intervals in this study differ for each proxy, we interpolated between data points in higher-resolution datasets in order to evaluate correlation with lower-resolution datasets. All correlation analyses of two parameters use measured data and interpolated data (linear interpolation) with same age intervals. We applied the Pearson's correlation test and t test and considered P < 0.05 as significant.

4 Results

We have subdivided the Holocene record of the Little Belt (LB) into three local zones (LB1-3; Figure 3) based on our proxy records: LB1 (7.7–6.8 ka BP) is characterized by high and variable Mg/Ca and Mn/Ca, decreasing Ba/Ca values, highly variable occurrences of Elphidium incertum (Williamson), and increasing δ18Oc and δ18Opw; LB2 (6.8–3.3 ka BP) is distinguished by higher trace element/Ca, and more negative δ18Oc, and a continued increase in δ18Opw; LB3 (3.3 ka BP to present) is characterized by lower and more stable trace element/Ca, δ18Oc, and δ18Opw.

Details are in the caption following the image
Trace element/Ca: (a) Mg/Ca, (b) Mn/Ca, (c) Ba/Ca, and stable isotopes: (d) δ18O and (e) δ13C measured on foraminifera of E. clavatum-selseyensis complex, (f) measured pore water δ18O over the past 7.5 kyr, (g) local zones (LB1-3) based on foraminiferal geochemistry. Each trace element (TE) data point represents the median value of several foraminiferal individuals at each depth. The error bars of TE/Ca ratios represent one standard error (SE). The stars indicate modern typical values of TE in E. clavatum-selseyensis complex from the southern Baltic Sea (Groeneveld et al., 2018).

4.1 Benthic Foraminiferal Trace Element/Ca

The Mg/Ca ranged from 0.8 ± 0.7 to 6.0 ± 2.0 mmol/mol through the record (7.5–0.8 ka BP) (Figure 3a). The values were relatively higher and more variable during LB1 and LB2, with an average value of 2.9 ± 1.3 mmol/mol in this period, and the highest values around 5.3 ka BP (6.0 ± 2.0 mmol/mol). There were two periods with relatively high Mg/Ca values, that is, 7.3–6.8 and 5.7–3.3 ka BP. During LB3, the Mg/Ca was lower and varied less, in the range of 0.7 ± 0.4 to 1.7 ± 0.3 mmol/mol.

During LB1 and LB2, the Mn/Ca values were more variable and much higher than values established in modern calibrations (up to 15 times higher than values in, e.g., Groeneveld et al., 2018; Figure 3b) and only occasionally lower than 5 mmol/mol. Two periods show increasing Mn/Ca values: 6.7–5.7 and 4.7–3.3 ka BP, when Mn/Ca values increased from 1.9 ± 0.7 to 14.3 ± 6.0 mmol/mol and 4.0 ± 1.4 to 28.6 ± 2.4 mmol/mol, respectively. Mn/Ca values were relatively stable and low during the last ~2 kyr, varying between 0.2 ± 0.1-3.4 ± 2.5 mmol/mol.

Ba/Ca data show three clear phases: an initial rapid decrease during LB1, a moderate phase during LB2, and then the lowest values during LB3. Ba/Ca values vary from 3.9 ± 0.0 to 39.5 ± 11.8 μmol/mol from 7 ka BP to present (Figure 3c).

4.2 Benthic Foraminiferal Stable Oxygen and Carbon Isotopes

Foraminiferal δ18O and δ13C values in the Little Belt became more positive through time, showing a range from −2.2‰ to 1.0‰ for δ18Oc and −3.4‰ to 1.1‰ for δ13Cc (Figures 3d and 3e). The δ18Oc values in LB3 are less variable and more positive than the values during LB1 and LB2. The most rapid permanent change in δ18Oc occurred during LB1, when δ18Oc increased from −2.2‰ to 0.4‰. The average value was ~−0.2‰ in LB2, varying from −1.4‰ to 0.6‰. There was a steady increase in δ18Oc from 3.4 ka to 2.5 ka BP. After 2.5 ka BP, δ18Oc became more stable, ranging from −0.8‰ to 0.8‰ with an average value of 0.4‰. In general, δ13Cc increased throughout the record. During the first 3 kyr, the δ13Cc values show a clear increasing trend from −3.4‰ to −1.1‰. From around 4 ka BP to present, the increase of δ13Cc slowed down, and the average value is −1.3‰.

4.3 Water Column and Pore Water Oxygen Isotopes

In the modern water column of the North Sea-Baltic Sea transition (Figure 1), we determined a linear covariation between salinity and the oxygen isotope composition of sea water (δ18Owc) (Figure 4):
Details are in the caption following the image
Covariation of the oxygen isotope composition of the modern water column (δ18Owc) as a function of salinity on a transect between the Skagerrak and the southern Baltic Sea (cruise MSM 50). Sampling sites are shown in Figure 1.

The salinity of the water column varied between 35.2 and 20.6 from the Skagerrak to the Kattegat (sampling station numbers 014–024 in Figure 1), and δ18Owc shows a range of −3.1‰–0.4‰. In the Baltic Sea (station numbers 025–030) the salinity was 21.4–8.6, and the δ18Owc varied in a range between −6.1‰ and −3.0‰.

Pore water oxygen isotopes (δ18Opw; Figure 3f) show a trend of increasing δ18Opw values with time from −5‰ to −2‰ VSMOW, with the most rapid increase between 7.5 and 5.0 ka BP, a slower increase in LB2, and relatively stable values in LB3. The modeled pore water oxygen isotopes (δ18Ompw) present a very rapid increase between 7.7 and 7.5 ka BP from −2.5‰ to −1.0‰ and reached a relatively high, stable value of −1.0‰ in the following 3.4 kyr. Between 4.1 and 3.0 ka BP, δ18Ompw rapidly decreased from −1.0‰ to −2.1‰ and then remained fairly stable around −2.2‰ ± 0.1‰ in the most recent ~2.5 ka.

We calculated the δ18O-based salinity by using the linear correlation between salinity and δ18Owc we established in equation 8.

4.4 Benthic Foraminiferal Faunal Assemblages and Grain Size

In general, the foraminiferal fauna is dominated by the E. clavatum-selseyensis complex and the faunal assemblage diversity of the Little Belt is relatively low throughout the time interval (Figure 5). Between 7.5 ka and 6.0 ka BP, the fauna consisted of a more diverse assemblage (Shannon H index = ~0.8) consistently dominated by Elphidium incertum (Williamson) and E. clavatum-selseyensis complex (>90%). During this period, Ammonia batava (Hofker) (previously assigned to Ammonia beccarii by Kotthoff et al. (2017) but here reassigned to A. batava following Hayward et al. (2019)) occasionally became a dominant species, and Elphidium magellanicum (Heron-Allen & Earland) and Miliolinella subrotunda (Montagu) had relatively higher concentrations (>5%), but overall the concentration of specimens was low (~37 individuals per cm3). After approximately 6 ka BP, the faunal assemblage changed, and the diversity decreased around 5.5 ka BP (Shannon H index = ~0.3), indicating a gradual environmental change. Total abundance of benthic foraminifera generally increased over time (except for the last ~500 years), and they were most abundant around 5.5 ka BP and from 4.0 ka to 0.5 ka BP, potentially caused by a more preferable environment or better preservation. The percentage of sand grains of sediment generally increased through time, and in particular, the sediment became sandier after 1.3 ka BP.

Details are in the caption following the image
Faunal assemblage data showing relative abundances (%) of the foraminiferal major species (>5% in at least five samples), benthic foraminiferal diversity (Shannon H index) and concentration, as well as sediment grain size (sand weight %). Note the different scales on the x axes (Foraminiferal major species with >5% relative abundances in at least one sample are shown in Figure S5, light microscope images of five major species are shown in Figure S6).

5 Discussion

5.1 Salinity and Temperature Changes in the Little Belt

Foraminiferal calcite δ18Oc is generally influenced by two parameters, that is, temperature and seawater oxygen isotope composition, which is correlated with salinity (Figure 5). Salinity changes are caused by the combined effects of evaporation, precipitation and runoff, and mixing processes are coupled with changes in seawater δ18Ow (Gat, 1996; Harwood et al., 2008). The mixing line determined in our study between salinity and modern water column δ18Owc is similar to that of Fröhlich et al. (1988): Salinity = (δ18Owc, SMOW + 8.91)/0.272, but with the benefit of a wider salinity range in our study (8–35). Water masses in the Little Belt, which originate in the Skagerrak, the North Atlantic and the central Baltic Sea, are mixed and renewed constantly. Due to the complex hydrographic setting of the Danish Straits, it is complicated to distinguish the respective influence of temperature and salinity on δ18Oc values, and therefore, estimating paleosalinity from δ18Oc records is not straightforward. However, with δ18Owc and δ18Opw data, it becomes possible to calculate bottom water salinity according to a mixing line among different water masses and to calculate bottom water temperature, considering oxygen isotope fractionation in equilibrium between δ18Oc and seawater.

We use two types of geochemical salinity proxies from Site M0059. One is the modeling of advection and diffusion of the pore water δ18O (δ18Opw), which shows that salinity increased from ~24 to 30 during the time interval 7.7–7.5 ka BP (Figure 6e). From 7.5 to 4.1 ka BP, the δ18Opw indicates bottom-water salinity environments of ~30 in the Little Belt. The bottom-water salinity decreased from ~30 to ~24 in the time interval from ~4.1 to ~2.5 ka BP (Figure 6e). The other paleosalinity proxy type is the foraminiferal salinity proxies, Ba/Ca and δ18Oc, which show an initial salinity increase during LB1, subsequently followed by a higher bottom-water salinity stage during LB2. The salinity continued to increase during LB3. Of these two groups of paleosalinity proxies, the δ18Ompw reconstructed salinity is similar to the salinity trends of the past 8 kyr estimated by Gustafsson and Westman (2002), who found that the salinity of the Baltic Sea reached Holocene maximum values between 6 and 4 ka BP and decreased after ~4 ka BP reaching modern day salinity after ~2 ka BP.

Details are in the caption following the image
Temperature, salinity, and oxygen content proxies and calculations from pore water and foraminiferal results. Calculated bottom water temperature (BWT) based on foraminiferal calcite: (a) Mg/Ca and (b) δ18O (using δ18Oc and δ18Ompw); salinity proxies from foraminiferal calcite: (c) foraminiferal Ba/Ca (μmol/mol); reconstructed bottom water salinity: (d) absolute salinity calculated from TMg/Ca and δ18Oc, (e) modeled bottom water changes before modification via advection and diffusion processes in the sediment column (δ18Ompw) and the calculated salinity based on pore water δ18Ompw (section 2.6); (f) Inferred NAO circulation patterns NAOPCA3 (modified from Olsen et al., 2012); oxygenation proxies from foraminiferal calcite: (g) foraminiferal calcite δ18Cc (‰-V-PDB) and Mn/Ca (mmol/mol); (h) local zones (LB 1-3) based on foraminiferal trace element and stable isotope data and reconstructed temperature and salinity. The blue shadow indicates negative phases of NAO. The red and blue box charts on top of (a), (b), (d), and (e) show the modern temperature and salinity variation range, respectively. Note that x axes of Ba/Ca and Mn/Ca are reversed.

We calculated bottom water temperature using two independent methods: (1) by combining δ18Oc and δ18Opw data following Shackleton (1974) and (2) by using Mg/Ca and the calibration of Kristjánsdóttir et al. (2007). Both Mg/Ca-based and δ18O-based bottom water temperature reconstructions (Figures 6a and 6b) provide similar trends in the Little Belt over the past 7.5 kyr, though absolute values are generally somewhat higher for the δ18O-based results (average ~1 °C higher) than for the Mg/Ca-based results. In general, δ18O-based bottom-water temperature reconstruction (Tδ18O) is less variable than Mg/Ca-based bottom water temperature reconstruction (TMg/Ca). Both temperature proxies give a large range of bottom water temperature values, varying between ~21 and 3 °C for Tδ18O and between ~16 and 0 °C for TMg/Ca over the past 7.5 kyr, although this range is only slightly larger than present day bottom water variations throughout the year. The warming period from 7.5 to 4.1 ka BP coincides with the timing of the HTM as traditionally defined for the Baltic (between 8 and 4 ka BP, Sohlenius et al., 2001; Zillén et al., 2008), which is slightly longer than the typical timing of 7 to 6 ka BP in most of Europe (Renssen et al., 2012). The mean bottom water Tδ18O and TMg/Ca during the time interval 6.5–4.1 ka BP were stable and warm at about 12.5 ± 1.7 °C and 10.3 ± 3.2 °C, respectively. This warm period was also characterized by more negative δ13C values (Figure 6g). The negative δ13C values of benthic foraminifera have also been found during both Medieval and recent warm periods and may suggest high primary productivity (Filipsson & Nordberg, 2010; Polovodova Asteman et al., 2018).

The reconstructed temperature decrease after ~4.1 ka BP (Figures 6a and 6b) is associated with a salinity decrease, shown by our δ18Ompw and foraminiferal geochemistry salinity reconstructions (Figures 6d and 6e). Salinity and temperature remained at ~24 and ~6 °C (Tδ18O) during the last ~2.5 kyr. These values are in the range of modern annual mean bottom water salinity and temperature (Figures 2b and 2d). The salinity decrease starting ~4.1 ka BP is not clearly observed in Ba/Ca. Instead, the Ba/Ca shows that the salinity increased slightly after ~4.1 ka BP, perhaps due to other factors controlling foraminiferal Ba/Ca such as nutrient supply, local freshwater source Ba/Ca variations and organic matter decomposition at the sea floor. The ostracod faunal assemblage also reflects a shallow brackish environment with reduced salinity after ~4 ka BP in the Little Belt (Stepanova et al., 2019). The salinity decrease between ~4.1 and 3 ka BP agrees with existing salinity reconstructions in the central parts of the Baltic Sea (e.g., Emeis et al., 2003; Ning et al., 2017).

Although the absolute temperature values of Mg/Ca-based and δ18O-based reconstructed bottom water temperature differ, both show a steady temperature decrease of ~7 °C from 4.1 to 2.5 ka BP. This coincides with the period of salinity decrease, indicating substantial environmental change. Tδ18O increased slightly around 2.5 ka BP, following the previous decrease, and became relatively less variable, with a mean Tδ18O value of 5.2 ± 1.2 °C. This is in the same range as modern bottom water temperature during winter or spring, which is known as the period when a majority of typical shallow-water foraminiferal calcification occurs (e.g., Gustafsson & Nordberg, 2001). Mean temperature from both proxies was lowest between 2.5 and 0.5 ka BP (3.0 ± 2.1 °C).

Based on equations 2, 7, and 8, we calculated the salinity from TMg/Ca and δ18Oc (Figure 6d), and the results broadly agree with δ18Ompw-based salinity. During LB1, there was a rapid salinity increase followed by a relatively saline period (LB2). After a subsequent decrease in LB3, salinity reached the modern values at ~1.5 ka BP.

The true temperature variations may not be as large as indicated by the proxy reconstructions (~7 °C) (Figure 6). This could be attributed to at least one of the following reasons: (1) Decreased water depths in the Little Belt may have reduced warm Atlantic water inflows, such that the bottom water was replaced by cooler and fresher water from the Baltic Sea; (2) there were changes in seasonality of foraminiferal calcification, and the bottom water seasonal temperature difference can be up to ~7 °C as shown in modern oceanographic records (Figure 2b); (3) as species-specific temperature calibration equations for E. clavatum-selseyensis complex are lacking, the use of a calibration equation of another species may introduce larger variability; (4) the Tδ18O reconstruction also contains uncertainties from δ18Opw and δ18Ompw. The modeled pore water profiles are not sensitive to the time when the salinity started to decrease (Figure S3d) but show some variation with different decay rates (Figure S3a). However, the time and decay rate affect the isotopic temperature change (Figure S3b). In this study, we set δ18Omax = −1.0 at 4.1 ka BP (Figure S3c). The error bar of δ18Omax is ±0.5, which leads to a temperature uncertainty of 2 °C around 4.1 ka BP according to equation 7. Therefore, the temperature difference between 4.1 and 2.5 ka BP is around 7 ± 2 °C; and (5) a slight influence of diagenesis cannot be excluded. The laser ablation allows to avoid surface coatings in data processing but does not exclude diagenetic imprints present within the tests, for example, in pores (Figure S4).

The pattern of foraminiferal faunal assemblages also highlights the salinity changes at Site M0059. The species diversity decreased (Figure 3) when temperature and salinity changed dramatically in the first ~1 to 2 kyr of our record. Ammonia batava had relatively high frequencies from 7.5 to 6.3 ka BP, varying from 20% to 90%, indicating a relatively low salinity environments during this first millennium after the marine inundation of the southern Little Belt, as this species is typical of low salinity environments (Murray, 2006; Seidenkrantz, 1993). There is a significant inverse correlation between the E. clavatum-selseyensis complex and E. incertum (r2 = 0.94, p << 0.01) throughout the record. From 7.5 to 5.5 ka BP, A. batava, E. clavatum-selseyensis complex and E. incertum alternatingly dominated the benthic foraminiferal fauna, presumably due to salinity variations during this period. From 5.5 ka BP to present (which includes the end of the HTM around ~4 ka BP), the diversity and the relative abundance of E. incertum decreased and the E. clavatum-selseyensis complex dominated the assemblage, presumably due to its high tolerance to relatively large variations in salinity (Charrieau et al., 2018; Murray, 2006). From 7.4 to 4.1 ka BP, freshwater indicators (i.e., diatoms, green algae, and ostracods) also rapidly decreased or disappeared (Andrén et al., 2015; Stepanova et al., 2019). Increases in the abundance of E. incertum and in diversity likely indicates periods of saline influx.

In the Baltic Sea, the first large salinity increase in the present study interval was at the very end of the transitional stage (the Initial Littorina phase) between the fresh-water of the Ancylus Lake and the following brackish-marine Littorina Sea stage (9-7 ka BP, Zillén et al., 2008; Kotthoff et al., 2017; Stepanova et al., 2019), subsequently followed by a temperature rise. During the beginning of Littorina Sea stage (~7.5–4.5 ka BP), the southern Baltic experienced multiple marine transgressions when the ongoing sea level rise outpaced glacioisostatic rebound (Björck et al., 2008). The mean sea level reached its highest level in the Baltic during the Littorina Sea stage (~7.5–~5 ka BP in Berglund et al., 2005). The water depth in the Little Belt increased considerably (~5–8 m) during these transgressions (Berglund et al., 2005), which led to a significant increase in North Sea water inflows due to deeper sills and straits. The Baltic Sea salinity increase during this stage may therefore be a result of the sea level rise and the increased cross-sectional area in the Danish Straits, permitting increased inflow of marine water from the Kattegat (Westman et al., 1999), which would likely apply to the Little Belt as well. The high salinity in the Little Belt may also be related to reduced brackish water outflow from the Baltic Sea between 6.2 and 4.7 ka BP (Gyllencreutz et al., 2006). The transgressions continued but were less extensive in the southernmost Baltic after 6 ka BP, while in the central and northern Baltic, the uplift was still ongoing, resulting in a renewed regression in the central Baltic (Björck et al., 2008). During the late Littorina Sea stage, mean sea level of southern Baltic coastal area steadily fell until reaching present day sea level (Berglund et al., 2005).

We compare our data with the inferred NAO circulation pattern over the past 5.2 kyr in order to test a link between water exchange through the Little Belt and the NAO on longer time scales. Reconstructions of the NAO index are based on tree rings and speleothem data (Trouet et al., 2009), as well as high resolution lake sediment geochemistry (Olsen et al., 2012). Principal component analysis 3 (PCA3) was used in Olsen et al. (2012) to integrate the redox variability (palaeo-redox proxy: Mn/Fe ratio), which can be associated with NAO-like atmospheric circulation patterns (Olsen et al., 2012). PCA3-inferred NAO patterns (Figure 6f) show that the NAO changed to variable and intermittently negative conditions during the time interval 4.3–~2 ka BP, coinciding with the temperature and salinity decreases documented in the Little Belt. The salinity decrease during this period may likely be ascribed to both a sea level drop in the Baltic Sea and often negative NAO conditions, following intermittent positive NAO phases. A similar positive correlation between NAO and sea level in the Skagerrak and Kattegat was also observed for the time period 1955–2000 by Wakelin et al. (2003).

During the generally negative NAO phase (4.3–~2 ka BP), the water depth decreased due to a sea level drop in the Little Belt, which led to a salinity decrease (Figures 6d–6f) due to fewer inflows from the North Sea and more brackish water in the region. Meanwhile, oxygen conditions improved when salinity decreased in the Baltic Sea during this period as the halocline was relatively deeper and stratification was weaker in the water column (Gerlach, 1994). However, the sea level variability was less likely to have made substantial contributions to the temperature variations in the Little Belt as the shallower water coincided with bottom water temperature decrease during 4.3–2 ka BP.

From ~2–0.5 ka BP, a continuous positive NAO phase led to a milder and less variable Baltic Sea climate compared to the previous period with intermittently negative and positive NAO conditions. The water depth and the bottom water conditions in the Little Belt became more stable. It should be noted that the various NAO reconstructions show somewhat different results (Nesje et al., 2000; Olsen et al., 2012), and consequently, we only focus on the general patterns where the various NAO reconstructions are largely similar. Both NAO reconstructions indicate that the NAO index was variable and intermittently negative in the midlate Holocene, which coincides with the period of variable water conditions (Figure S7). The relative abundance of E. incertum and the Shannon diversity index are positively correlated with NAO (both p < 0.01) reconstructed by Nesje et al. (2000) (Figure S7).

5.2 Oxygen Reconstruction

Generally, the increasing trend of foraminiferal δ13Cc and the decreasing trend of foraminiferal Mn/Ca (Figure 5f) suggest that the bottom water in the Little Belt became less stratified and more ventilated and therefore more oxygenated over time. Bottom water exchange and renewal can result in rapid increases of the δ13CDIC (e.g., Filipsson et al., 2017) and significantly alter the δ13C signal in foraminiferal carbonate. The δ18Oc and δ13Cc data show a weak but significant positive correlation (r2 = 0.14, p < 0.01); this correlation presumably indicates changes in water mass composition, as a new water mass would usually be associated with changes in δ18Owater and δ13CDIC (Filipsson et al., 2017). Moreover, more negative δ13Cc values during LB1 and LB2 (Figure 6g) indicate higher primary productivity and more organic matter deposition, which may cause oxygen depletion and therefore hypoxia. The relatively positive δ13Cc in LB3 most likely indicates lower productivity and more oxygenated bottom water conditions.

High foraminiferal Mn/Ca in zone LB1 and LB2 (7.5–3.3 ka BP) suggests hypoxic conditions during these intervals, which coincide with the HTM in the Baltic region (Zillén et al., 2008). The hypoxic periods in the Little Belt overlap in time with hypoxic events in the Baltic Proper, as indicated by Corg/Ptot and sediment lamination and bioturbation intervals (Jilbert & Slomp, 2013; Zillén et al., 2008). In the most recent stage LB3 (~3–0.8 ka BP), Mg/Ca and Mn/Ca in foraminiferal calcite are similar to modern calibration values of benthic foraminifera (Groeneveld et al., 2018) (see stars in Figures 4a–4c). Foraminiferal Mg/Ca and Mn/Ca show a relatively warm period with low bottom water oxygen content from 7.5 to 6.8 ka and from 5.3 to 3.3 ka BP. Significant positive correlation between foraminiferal Mg/Ca and Mn/Ca ratios (r2 = 0.73, p < 0.01) suggests a strong interaction between temperature and dissolved [O2] in the bottom water, which is supported by modern hydrological correlation between bottom water temperature and dissolved oxygen concentration. However, there are also significant correlations between Ba/Ca and Mg/Ca and Ba/Ca and Mn/Ca. Solution-based TE/Ca measurements (Kotthoff et al., 2017) show similar TE/Ca results compared to the ones measured by LA-ICP-MS (Figure S1). We cannot completely exclude some potential contaminations of authigenic carbonate minerals inside the pores or that authigenic carbonates have replaced some primary calcite in some samples that cannot be removed by using LA profiles or standard cleaning procedures. To assess this, we compare our oxygen proxy results to the study of Site M0059 by van Helmond et al. (2017), which used bulk sediment geochemistry to evaluate past hypoxia in the Little Belt. They concluded that the presence of hypoxia is indicated by elevated Fe-carbonate concentrations (Fecarb) in the bulk sediments in the time interval ~7–2.5 ka BP and after ~1 ka BP. In addition, the bulk sediment molybdenum (Mo) concentrations at Site M0059 suggest that (seasonal) hypoxia occurred continuously after the Ancylus/Littorina transition. In general, their bulk sediment hypoxia indicators agree with our oxygen proxies, providing support for our results.

5.3 Modern Hydrographical Setting versus NAO

The observed salinity variations of surface waters in the modern hydrographical data from the Little Belt (Figure 2c) can be explained by the seasonality of freshwater input, as surface water salinity is directly affected by river inflow and precipitation. However, the bottom water masses are more complex in such a shallow transitional area. In addition to river inflow and precipitation, the bottom water is also strongly affected by inflow of saline and oxygenated water from the North Sea, with the water inflow in the Little Belt generally being strong and highly turbulent, causing distinct vertical mixing (Jakobsen & Ottavi, 1997). Therefore, the stratification is often not very strong in the Little Belt compared to the Baltic Proper, but the pycnocline is still comparatively stronger during summer due to the increased freshwater influx and summer heating of surface waters, resulting in an increased temperature and salinity difference between surface and bottom waters. Both annual and monthly mean bottom water temperature and oxygen concentration over a 20-year period (1989–2018) show significant inverse correlation (r2 = 0.29, p = 0.002; r2 = 0.59, p << 0.01, respectively) (Table S1). This reveals that the dissolved bottom water [O2] decreases as water temperature increases (also seen in Figure 2). We applied these correlations to our past reconstructed bottom water salinity, temperature and dissolved [O2] with the reconstructed NAO index (Olsen et al., 2012), with the limitations in mind, aiming to investigate the mechanisms behind the environmental condition changes in the past. This inverse correlation is similar to that we have observed in reconstructed temperature and oxygen conditions in the bottom water in the Little Belt over the past 7.5 kyr.

The correlation between the winter-NAO index (Jones et al., 1997) and winter bottom water salinity from instrumental data (Baltic Nest, http://nest.su.se) is weak but significant (r2 = 0.12, p = 0.036) (Table S2), presumably due to the profound effect of NAO on winter climate in the region. However, no significant correlation is seen between monthly mean bottom water salinity and the NAO index, and there is no significant linear correlation between the winter NAO and temperature or oxygen concentration of deep water in the past four decades. The positive correlation is also observed in reconstructed bottom water salinity and reconstructed NAO index (Olsen et al., 2012).

The winter NAO has a significantly positive influence on the inflow of saline water and therefore on the bottom water salinity, that is, higher salinities occur in the straits during winters with less wind from the north and northeast. However, the transport of water (and thereby indirectly salinity) in the Danish Straits is also to a large degree controlled by the short-term variability in relative sea level (Jakobsen et al., 2010), a variability that we do not register with the above analyses. Several studies suggest a NAO impact on Baltic sea level variations and demonstrate a high correlation between winter NAO index and the Baltic Sea and Kattegat winter mean sea level (e.g.,Andersson, 2002 ; Kuijpers et al., 2012 ; Różyński, 2015 ; Wakelin et al., 2003). Water depth variations attributed to North Atlantic atmospheric circulation further influence the bottom water salinity and bottom water circulation.

6 Conclusion

By combining the foraminiferal trace elemental, isotopic, and faunal assemblage data with pore water δ18O data, we have quantitatively reconstructed bottom water temperature, salinity and oxygenation at the entrance to the Baltic Sea since 7.7 ka, making it possible to evaluate the link between overall climatic conditions, water mass exchange and bottom-water hypoxia.

At the start of our record (the beginning of the brackish-marine Littorina Sea stage) between 7.7 and 7.5 ka BP, salinity increased markedly from ~24 to ~30. The bottom water temperature, the dissolved [O2] and the primary production varied between 7.5 and ~6.8 ka BP, indicating intensive environmental changes in this entrance area to the Baltic Sea. Relatively warmer and more saline bottom water conditions developed between 6 and 3 ka BP, that is, the main Littorina Sea stage. Meanwhile, primary productivity became high, and combined with increased stratification, this caused hypoxia in the bottom water in the Little Belt. The period with high bottom water temperature and salinity and low dissolved [O2] (7.5–3.3 ka BP) coincides in time with the HTM and with transgression and high sea level, and greater water depth in the Little Belt. After the HTM, salinity and temperature decreased in the time interval ~4.1–2.5 ka BP, coinciding with variable and intermittently negative NAO index. This was followed since 2.5 ka BP by cooler and fresher bottom water conditions until the present. The bottom water temperature decreased ~7 °C, while the salinity decreased six units between 4.1 and 2.5 ka BP. The salinity decrease after 4.1 ka BP can be recognized by modeled pore water δ18O results, but cannot be seen independently in foraminiferal δ18Oc and Ba/Ca data. However, when δ18Oc and Mg/Ca data are combined to calculate salinity, a similar decrease is observed. The different temperature proxies used in this study (foraminiferal Mg/Ca and δ18O) show similar trends over the past 7.5 kyr with an absolute temperature difference of about ~1 °C.

Modern bottom water salinity in the Little Belt and winter NAO index show a significant positive correlation. A similar, but somewhat weaker correlation is seen in our paleorecord, where temperature and salinity of the bottom waters decreased during intermittently negative NAO phase (4.3–~2 ka BP). In addition, the salinity decrease at ~4 ka BP may also be linked to a well-known regional glacio-isostatic sea-level fall during the late Littorina Sea stage. Broadly, more oxygenated bottom water conditions in the past 4 kyr are likely caused by a combination of shallower water depth, and lower temperature and salinity. Our multiproxy record demonstrates that the changes in bottom water conditions at the boundary between the brackish Baltic Proper and marine Kattegat/North Sea are noticeably influenced by larger scale climate and topographical change (i.e., NAO and eustatic sea level), and they are also linked to the changes in the other parts of the Baltic Sea.


We wish to thank IODP and ESSAC for funding and organizing the expedition and sampling, Git Ahlberg (Lund University) for assistance with sediment sample processing, Iris Schmiedinger (IOW) for expert mass spectrometry support, Jacob Carstensen (Aarhus University) for hydrographic data advice, Henning Kuhnert for stable isotope analyses, and Christof Pearce (Aarhus University) for aid in foraminiferal photography. We thank the Crafoord Foundation, Kungliga Fysiografiska Sällskapet i Lund, Linnaeus centre of excellence “Lund University Centre for studies of Carbon Cycle and Climate Interactions (LUCCI),” and the Centre for Environmental and Climate Research (CEC), Lund University for funding. Work during cruise MSM-50 was supported to MEB and ML by BMBF during Küno Projects SECOS I and II (03F0666 and 03F0738) and Leibniz Society. Foraminiferal studies were funded by the Carlsberg Foundation (IVAR-347 project), Geocenter Denmark (DAN-IODP-SEIS project), and the Independent Research Fund Denmark/Natural Science (7014-00113B, G-Ice). We also thank the associate editor and two anonymous reviewers for their suggestions that helped improve the manuscript. All data are available online at the Data Publisher for Earth and Environmental Science, PANGAEA (https://doi.pangaea.de/10.1594/PANGAEA.911232).