Benthic foraminifera have been studied in sediment cores PS72/340-5 and PS72/396-5 from the Amerasian Basin west of the southern Mendeleev Ridge and in core PS2185-6 from the central Lomonosov Ridge (Fig. 1). Primarily, the uppermost core sections that are characterized by normal magnetic polarity (Brunhes Chron, Frederichs, 1995; Bazhenova, 2012; Elkina et al., 2023) were studied because this interval comprises the foraminiferal events used previously for stratigraphic correlation (e.g., Backman et al., 2004; Polyak et al., 2004). The sediment cores were sampled at variable depth intervals in 1 cm-thick slices at the Polarstern Core Repository in Bremerhaven (Germany) depending on the expected foraminifera content. The mid-depth of the sample is used in this study. The brown layers were sampled at a higher resolution because these layers are usually more productive for foraminifera (e.g., Polayk et al., 2004; Adler et al., 2009). All sediment cores referred to in the text are shown in Fig. 1 and the meta data are listed in Table 1.

Arctic Ocean sediments are characterized by a pronounced lithological variability and previous studies revealed that the stratigraphic occurrence of calcareous foraminifera is related to specific lithological layers (e.g. Polyak et al., 2004; Adler et al., 2009). Therefore, the lithostratigraphic schemes used for stratigraphic correlation in the Arctic Ocean are shortly reviewed. For the western Arctic Ocean, Clark et al. (1980) proposed thirteen lithostratigraphic units (standard arctic lithostratigraphic units A to M) based on grain size composition and presence of detrital carbonate-rich pink-white layers. This scheme has been routinely applied in subsequent studies and slightly modified for cores from Alpha Ridge (Minicucci and Clark, 1983; Mudie, 1985; Darby et al., 1989; Clark et al., 1990; Poore et al., 1993, 1994; Ishman et al., 1996; Phillips and Grantz, 1997; Stein et al., 2010a, b). In the past years, a simplified lithostratigraphy has been preferred for visual core description counting downcore the alternation of brown, calcareous foraminifer- and manganese-rich and coarser-grained grey to olive, foraminifer- and manganese-poor layers (Polyak et al., 2004). These layers were termed brown beds B1 to B7 and grey beds G1 to G6, and are interpreted to reflect interglacials/interstadials and glacials/stadials, respectively. In the western Arctic Ocean, lithological Unit M comprises brown beds B1 to B6, whereas the top of Unit L is marked by brown bed B7 (Figs. 3, 4). This scheme was extended downcore into older sediments but the number of brown layers in specific stratigraphic intervals varies between cores (Stein et al., 2010b; Wang et al., 2018; Dong et al., 2020; Park et al., 2020). März et al. (2011) note that diagenetic processes may lead to post-depositional dissolution and/or formation of Mn-rich brown layers. Therefore, they emphasize that these layers should not be used for stratigraphic correlation without independent age control.

The combination of the standard lithostratigraphic units and colour stratigraphy has been applied on cores from southern Mendeleev Ridge including PS72/340-5 and PS72/396-5 (Figs. 3, 4) (Stein et al., 2010a, b). However, the link between brown layers and calcareous foraminifera abundances is not straightforward because when sedimentation is extremely slow (mm kyrs⁻¹) as at site PS72/396, brown beds are difficult to distinguish because of extensive bioturbation blurring the record (Fig. 3) (Stein et al., 2010b). Moreover, brown bed B 5 is barren of calcareous foraminifera and contains only agglutinated foraminifera in core PS72/340-5 (Fig. 4). This is also a common feature in Northwind Ridge cores (Poore et al., 1993,1994; Ishman et al., 1996; Phillips and Grantz, 1997; Yurco et al., 2010).

Thin layers enriched in pink-white and whitish lenses or clasts of detrital dolomite (white and pink-white layers) may be useful for stratigraphic correlation in the western Arctic Ocean but the relative high number of these layers in sediment cores from the Northwind and Mendeleev ridges make an unequivocal correlation difficult (Clark et al., 1980; Mudie and Blasco, 1985; Minicucci and Clark, 1983; Poore et al., 1993; Poore et al., 1994; Phillips and Grantz, 1997; Polyak et al., 2004; Adler et al., 2009; Stein et al., 2010a; Stein et al., 2010b; Cronin et al., 2014; Bazhenova et al., 2017). These diamictons reflect synchronous rapid sedimentation events from melting icebergs. These icebergs had calved at the grounding line of the Laurentide ice sheet, at the Arctic margin, then drifted with prevailing currents across the Arctic basins and released predominantely calcareous debris while melting (Darby et al., 2002; Polyak et al., 2004; Matthiessen et al., 2010; Stein et al., 2010a, b; Bazhenova et al., 2017).

The lithostratigraphic scheme of Clark et al. (1980) cannot be applied on Lomonosov Ridge sediments (Sellén et al., 2008) and a separate lithostratigraphic scheme has not been developed yet for Eurasian Basin cores. Non-destructive physical property records, such as magnetic susceptibility (ms) and wet bulk density (wbd) enable correlation within similar depositional environments (Sellén et al., 2010; O'Regan et al., 2019; Vermassen et al., 2021; Razmjooei et al., 2023). Thus, Razmjooei et al. (2023) used wbd maxima of two diamictons in the upper part of sediment cores from the western Lomonosov Ridge for stratigraphic correlation, including core PS2185-6 (Fig. 5).

Since foraminifer-rich brown layers are also widespread in the eastern Arctic Ocean (e.g., Jakobsson et al., 2001; Backman et al., 2004; Polyak et al., 2004; Löwemark et al., 2014), these may be applicable for a basin-wide lithostratigraphic correlation if they were coeval across the Arctic Ocean. Visual inspection of sediment core images revealed that some intervals in core PS2185-6 with high planktic foraminifera concentrations (Spielhagen et al., 1997a, 2004) and increased manganese contents (Schoster, 2005) have a brown colour (Grobe and Fütterer, 2003), comparable to those in adjacent core 96/12-1pc (Jakobsson et al., 2001; Backman et al., 2004).

There are possibly some lithological marker sequences that may be used for stratigraphic correlation across the Arctic Ocean. In the Mendeleev Ridge area, the base of lithological Units M and J are marked by the dolomite-rich pink-white diamictons PW 2, and PW 1, respectively (Figs. 3, 4; Clark et al., 1980; Poore et al., 1993; Polyak et al., 2004; Adler et al., 2009; Stein et al., 2010a, b; März et al., 2011). Pink-white layer PW 2 is interbedded in two brown layers which were correlated to foraminifer-rich brown layers in core 96/12-1pc on Lomonosov Ridge (Backman et al., 2004; Polyak et al., 2004). A similar sequence may be observed in core PS2185-6 where detrital carbonates in the 125–500 µm size fraction (Spielhagen et al., 1997b) may represent fine-grained detrital dolomites between two foraminifer-rich brown layers in the lower part of the uppermost interval with normal polarity (Fig. 5). Pink-white clasts in a diamicton in adjacent core PS70/358-4 (Stärz, 2008) were used by Stein et al. (2010a, Fig. 9) to tentatively infer the presence of diamicton PW 2. Based on this correlation, Stein et al. (2010b) suggest that brown bed B7 must be present below this layer in cores PS70/358-4 and PS2185-6. At about the same stratigraphic level, Kaparulina et al. (2016) observed a maximum of dolomite in the heavy mineral fraction >63 µm in core 96/12-1pc, interbedded into two brown layers. Previously, Morris et al. (1985, p. 904) used the presence of whitish blotches at the base of the Theta member of their Makarov Basin Formation in LOREX core B8 to infer the presence of the base of Unit M (see also Sellén et al., 2008). These observations suggest that sediments equivalent to the base of Unit M and the top of Unit L may also be recorded in the central Lomonosov Ridge area.

However, sedimentation on the shallow Lomonosov Ridge may be disrupted by glacial erosion and/or slow sedimentation during times of extensive glacial ice cover (e.g., Jakobsson et al., 2001, 2010; Frank et al., 2008) complicating the construction of age models. Thus, an erosional unconformity is observed below a pink layer in core 96/09-1pc that is correlated by wbd and ms to core PS2185-6 (Jakobsson et al., 2001; Razmjooei et al., 2023). This correlation suggests that this pink layer is synchronous with the inferred pink-white layer PW 2 above brown bed B 7 in PS2185-6 (Fig. 5).

The discontinuous occurrence of foraminifera in the studied cores is related to the lithology (Figs. 3–5). Planktic and calcareous benthic foraminifera show absolute abundance maxima rather in brown layers than in the grey to olive layers, as it has been previously observed (e.g., Polyak et al., 2004; Backman et al., 2004; Adler et al., 2009). Core PS72/396-5 has lower sedimentation rates and thus comprises older sediments in the lower core section than core PS72/340-5. Below brown bed B 7 absolute abundances of calcareous foraminifera are much lower in brown layers than in younger sediments. On Lomonosov Ridge, only some brown layers in the uppermost interval with normal magnetic polarity are rich in calcareous foraminifera. The few calcareous shells in the grey to olive layers of the three cores may be bioturbated from the brown layers, as indicated by extensive brown mottling (see digital images and x-radiographs; Fütterer and Grobe, 2003; Matthiessen and Stein, 2008a, 2008b; Matthiessen, 2013a, b), or may be of allochthonous origin (Geibert et al., 2021). Coarse-grained intervals, such as the diamictons in core PS2185-6 are barren (Fig. 5).

Planktic foraminifera assemblages are almost exclusively composed of Neogloboquadrina pachyderma (not shown) in cores PS72/340-5 and PS72/396-5. In most benthic foraminifera assemblages in the intervals with calcareous taxa Stetsonia horvathi contributes more than 50 % to the assemblages (Figs. 3-5), whereas stratigraphically important taxa such as Bolivina arctica exceed only in a few samples more than 50 %. Calcareous foraminifera bioevents occur in Unit M and in the upper part of Unit L in brown bed B 7.

Agglutinated benthic foraminifera do not show a consistent distribution in the three cores. Where calcareous and agglutinated foraminifera are preserved, absolute abundances of agglutinated foraminifera are generally two orders of magnitude lower than those of the calcareous taxa. Brown layers in the lower part of core PS2185-6, and brown bed B5 in core PS72/340-5 comprise exclusively agglutinated foraminifera. In core PS72/340-5 the thick interbedded grey to olive layers comprise few to no calcareous foraminifera but sometimes low abundances of agglutinated foraminifera. Core PS72/396-5 is almost devoid of agglutinated taxa.

Species of Pullenia are present at different stratigraphic levels in the studied sediment cores. Pullenia bulloides only occurs in the B. aculeata acme of core PS2185-6 (Fig. 5), and in core PS72/340-5 in brown bed B 3 (Fig. 4). Pullenia quinqueloba and P. osloensis are sporadically present in cores PS2185-6 and PS72/340-5, respectively. Pullenia bulloides shows a variable distribution in sediment cores from the Arctic Ocean (Fig. 7).

Benthic taxa which are potentially important for biostratigraphy and ecology are shown in Figs. 3, 4, and 5. Additionally, the stratigraphic occurrence of these taxa in cores from Alpha Ridge, Northwind Ridge, Mendeleev Ridge, and Lomonosov Ridge (Fig. 1; this study; Scott et al., 1989; Lazar and Polyak, 2016) is compiled in Figs. 6 to 11 to evaluate their distribution across the Arctic Ocean.

The agglutinated benthic foraminifera Siphotextularia rolshauseni, a stratigraphic marker for MIS 2 in Norwegian-Greenland Sea sediments, has not been observed in the sediment cores from the CAO but it occurs sporadically in the Fram Strait area and at the adjacent Barents Sea continental slope (Wollenburg et al., 2001b). This species has not been observed in any other study on agglutinated foraminifera from the Arctic Ocean (e.g., O'Neill, 1981; Scott et al., 1989; Evans and Kaminski, 1998).

Bulimina aculeata is abundant only in core PS2185-6 where an absolute abundance maximum is restricted to a short interval in a brown layer (Fig. 5). It is present in very low numbers at the deep-water sites, in a single sample in brown bed B 4 in core

PS72/340-5, and in two samples in brown bed B 3 in core PS72/396-5 (Figs. 3, 4). This species occurs at different stratigraphic intervals in sediment cores and a distinct maximum of absolute abundances has been only observed in cores from relatively shallow water depths (<∼2000 m), occurring usually above pink-white layer PW 2 in lithological Unit M in the western Arctic Ocean (Fig. 6).

Species of Pullenia are present at different stratigraphic levels in the studied sediment cores. Pullenia bulloides only occurs in the B. aculeata acme of core PS2185-6 (Fig. 5), and in core PS72/340-5 in brown bed B 3 (Fig. 4). Pullenia quinqueloba and P. osloensis are sporadically present in cores PS2185-6 and PS72/340-5, respectively. Pullenia bulloides shows a variable distribution in sediment cores from the Arctic Ocean (Fig. 7).

Oridorsalis umbonatus (often referred to O. tener, see Appendix A) has a distinct lowest maximum of absolute abundances above PW 2 in a brown layer across the Arctic Ocean (Figs. 3, 4, 5, 8). A distinct relation to a specific brown layer is not observed in the western Arctic Ocean where the oldest maximum occurs in brown beds B 3 and B 6 in cores PS72/340-5 and PS72/396-5, respectively. However, brown beds B 4, B 5 and B 6 are difficult to distinguish in core PS72/396-5 because of intensive bioturbation, and at site PS72/340 brown bed B 5 is barren in calcareous foraminifera. The oldest abundance maximum coincides with that of B. aculeata in cores from shallow water depths (Figs. 5, 8) as it has also been observed in the size fraction >150 µm in cores 96/12-1pc and the composite record of cores NP26-5 and NP26-32 (as O. tener; Jakobsson et al., 2001; Polyak et al., 2004). Below the oldest maximum, O. umbonatus can be present but never reaches high absolute abundances.

Cassidulina neoteretis occurs in certain brown layers in core PS2185-6 and is particularly abundant in the B. aculeata acme above the lowest maximum of absolute abundances of O. umbonatus (Fig. 5). A comparable assemblage has been observed in the composite record of cores NP26-5 and NP26-32 and core 96/12-1pc (as C. teretis; Jakobsson et al., 2001; Polyak et al., 2004). In cores PS72/340-5 and PS72/396-5 from deeper waters, C. neoteretis is only rare in some intervals (Figs. 3, 4). This species might be a good additional marker for the B. aculeata acme in shallow water cores, but C. neoteretis has not been consistently distinguished from C. teretis making it currently difficult to assess its stratigraphic potential (Appendix A).

Epistominella arctica and E. exigua do not show a consistent stratigraphic occurrence in the studied sediment cores (Figs. 3–5). Epistominella arctica is more abundant than E. exigua and occurs with variable absolute and relative abundances in certain brown layers. Hereby, E. arctica is more frequent above PW 1, and E. exigua below PW 2 (Figs. 9, 10).

Bolivina arctica occurs in variable absolute and relative abundances through the cores (Figs. 3–5). Absolute abundance maxima are generally restricted to the lower part of the interval with normal magnetic polarity. The youngest pronounced absolute abundance maximum is terminated by a pronounced decline to low abundances. This is located at the top of Unit L in brown bed B7 just below pink-white layer PW 2 in the western Arctic Ocean, and at the same stratigraphic level on Lomonosov Ridge based on the proposed lithostratigraphic correlation (Fig. 11, Stein et al., 2010b).

The extinct agglutinated foraminifera Haplophragmoides obscurus dominates the benthic assemblages in the lower part of core PS2185-6 (Fig. 5) and disappears above the top of the youngest absolute abundance maximum of B. arctica. This species is only observed in traces in PS72/340-5 (Fig. 4). In core PS72/396-5 H. obscurus is absent but likely the stratigraphic occurrence of the extant Pyrgo rotalaria is comparable with that of H. obscurus (Fig. 3).

The changeover in predominance from agglutinated to calcareous benthic foraminifera has been observed in core PS2185-6 slightly above the change to normal magnetic polarity (Fig. 5). This substantial decline of agglutinated foraminifera has been previously described for this core (Evans et al., 1995; Evans and Kaminski, 1998). Below the changeover, assemblages only comprise foraminifera firmly agglutinated with high iron content in the cement (Schröder, 1988; Hedley, 1963; Bender, 1989). These include taxa accessory in recent assemblages (Cribrostomoides subglobosus, Glomospira spp., Rhabdammina spp., Jacullela spp., Hyperammina spp., Saccammina socialis, S. sphaera, Psammosphaera fusca, Reophax spp.), and H. obscurus and Cyclammina trullissata that are unknown from the modern Arctic benthic foraminifera fauna. Cyclammina trullissata disappeared in core PS2185-6 in a brown layer above the proposed pink-white layer PW 2, and in core PS72/340-5 in brown bed B 5. In younger sediments above the HOs of H. obscurus and C. trullisata the agglutinated fauna is dominated by modern taxa, and predominantly firmly agglutinated Cribrostomoides subglobosus occurs sporadically in certain intervals. Agglutinated foraminifera are almost absent in the studied interval of core PS72/396-5 whereas in core PS72/340 brown layers and some laminated interbedded sequences comprise predominantly agglutinated tubular taxa.

A robust chronostratigraphic framework for Pleistocene sediments in the CAO has not been developed yet, and beyond the range of radiocarbon dating there are few chronological tie-points because of discontinuous microfossil and sparse radionuclide records. Moreover, variable accumulation of sediments and stratigraphic breaks at the Arctic Ocean sea-floor hamper linear interpolation between tie-points and thus calculation of numerical ages for bioevents (Hillaire-Marcel et al., 2017; Matthiessen et al., 2018).

Since the turn of the century the Pleistocene chronostratigraphy in the western Arctic Ocean primarily relied on stratigraphic correlation to the central Lomonosov Ridge (e.g., Backman et al., 2004; Polyak et al., 2004; Spielhagen et al., 2004; Adler et al., 2009; Park et al., 2024). Three brown layers that comprise absolute abundance maxima of planktic foraminifera interbedded in two diamicts (diamict 1 and 2 sensu Razmjooei et al., 2023) in cores PS2185-6 (Fig. 5) and 96/12-1pc have been dated to MIS 5 by coccolith biostratigraphy and optical stimulated luminescence (OSL) ages (Jakobsson et al., 2000, 2003; Spielhagen et al., 2004). These ages served with AMS ¹⁴C dates as backbone for the Arctic Ocean MIS 1 to 6 chronostratigraphy (e.g., Jakobsson et al., 2001; Jakobsson et al., 2003; Backman et al., 2004; Spielhagen et al., 2004). New radiocarbon, coccolith, amino acid razemisation and radiometric data from various Lomonosov Ridge cores now require a substantial revision of these age assignments (Hillaire-Marcel et al., 2017; Geibert et al., 2021; Razmjooei et al., 2023; Song et al., 2023; West et al., 2023; Wollenburg et al., 2023a).

Chronological tie points for core PS2185-6 are now radiocarbon ages for the Holocene to late glacial interval (MIS 1–3) and a ²³⁰Th_xs extinction age for the Middle Pleistocene (Wollenburg et al., 2023a; Song et al., 2023). These ages confirm that the uppermost interval with normal magnetic polarity in core PS2185-6 corresponds to the Brunhes Chron (Fig. 5).

Although the extinction age has large uncertainties with respect to the stratigraphic interval and age (ca. 226±54 ka (MIS 7) at 174±76 cm, Song et al., 2023), the core section with the three foraminifera maxima between 160 and 240 cm is older than MIS 6 even if the true extinction age is close to the youngest possible ²³⁰Th_xs age (Fig. 5). The previous coccolith biostratigraphy of Spielhagen et al. (2004) for core PS2185-6 based on the presence of Emiliania huxleyi (assigned to Geophyrocapsa huxleyi by Bendif et al., 2023) in these foraminifera maxima could not be confirmed by a detailed taxonomic restudy (Razmjooei et al., 2023).

The new age constraints complicate the identification of MIS 5 in Lomonosov Ridge sediments. The radiocarbon and ²³⁰Th_xs extinction ages are supplemented by correlating coccolith bioevents from sediment cores on the western Lomonosov Ridge to core PS2185-6 by means of wbd (Razmjooei et al., 2023). This correlation suggests that E. huxleyi must have appeared below diamict 1 in core PS2185-6 indicating an age younger than the MIS 8/9 transition based on its global first appearance (290 ka, Backman et al., 2012; Anthonissen and Ogg, 2012), which agrees with the extinction age below this possible LO (Fig. 5). Amino acid razemization data on N. pachyderma and C. wuellerstorfi indicate a MIS 6–9 age for the LO of E. huxleyi in core LOMROG12-PC03 (West et al., 2023). In contrast, the extinction age of ²³¹Pa_xs of 140 ka below the correlated LO in core PS87/030-1 indicates an age younger than MIS 6 (Hillaire-Marcel et al., 2017; Razmjooei et al., 2023). Thus, E. huxleyi might have appeared in the Arctic Ocean not before MIS 5 (Razmjooei et al., 2023) but this datum should be calibrated in a core with a detailed independent chronostratigraphy.

Unfortunately, the interval between the base of diamict 1 and the foraminifera maximum at 170 cm is barren of calcareous microfossils in core PS2185-6. In contrast, a foraminifera-bearing layer has been recognized in adjacent core 91/12-1pc at the stratigraphic level of the proposed LO of E. huxleyi (Jakobsson et al., 2001; Backman et al., 2004; Razmjooei et al., 2023) that might correspond to MIS 5.

The top of the lower foraminifera-rich layer at ca. 220 cm, tentatively assigned to brown bed B 7, may be of late MIS 9 age if the detrital carbonate maximum in core PS2185-6 is equivalent to a pink-white layer in core PS115/2-2-2 dated to 282 ka based on ²³⁰Th_xs data (Fig. 5; Stein et al., 2010a, 2025).

Based on the wbd correlation the coccolith Pseudoemiliania lacunosa must have disappeared slightly below diamict 2 in core PS2185-6 (Fig. 5; Razmjooei et al., 2023), the former MIS 6 (Spielhagen et al., 2004). This stratigraphic level might be placed at the absolute abundance maximum of foraminifera at ca. 310 cm that has been previously assigned to MIS 7 (Spielhagen et al., 2004). However, the disappearance of P. lacunosa indicates an age older than uppermost MIS 12 (ca. 430–440 ka) for this stratigraphic level (Backman et al., 2012; Anthonissen and Ogg, 2012). Razmjooei et al. (2023) suggest that P. lacunosa disappeared in MIS 13 (>478 ka) because they assume that this coccolithophore was not present in the Arctic Ocean in a glacial period such as MIS 12. However, a pronounced temperature decline in a glacial stage might have been a more likely trigger for extinction than a relatively warm interglacial. The correlated HO of P. lacunosa in core PS2185-6 indicates that the first downcore change in magnetic polarity occurred in the early Middle Pleistocene, and is older than MIS 11 but probably younger than the Brunhes/Matuyama boundary (Fig. 5).

The new stratigraphic data appear to be rather consistent for the central Lomonosov Ridge. In particular, the revised coccolith biostratigraphy now aligns much better with the ²³⁰Th_xs data. However, this calls into question the interpretation of the OSL ages from core 96/24-1sel located close to core 96/12-1pc (Fig. 1; Jakobsson et al., 2003) that apparently disagrees with the new chronology.

The chronostratigraphy of the western Arctic Ocean cores is more debatable. Magnetostratigraphy provides the basic age model for cores PS72/340-5 and PS72/396-5 (Bazhenova, 2012; Elkina et al., 2023). The entire studied interval in core PS72/340-5 is assigned to the Brunhes Chron supported by radiocarbon ages (Fig. 4) (Bazhenova, 2012; Elkina et al., 2023) while ²³⁰Th_xs data (Geibert et al., 2021) suggest that the change to reverse magnetic polarity in core PS72/396-5 corresponds to the Brunhes/Matuyama boundary, located in the middle part of Unit K, while the Jaramillo Subchron is placed at the transition of Unit K to J (Fig. 3, Elkina et al., 2023). This agrees with the correlation of lithological units to the magnetic polarity pattern based on the re-analysis of paleomagnetic data obtained on sediment cores recovered from ice island T-3 in the Mendeleev Ridge area (Jones, 1987). Song et al. (2023) used the ²³⁰Th_xs data to tentatively propose a MIS 5e/late Termination II age at ca. 28 cm, and an age older than MIS 7 and younger than MIS 9 at ca. 65 cm in core PS72/396-5. At the latter stratigraphic level the diamicton PW 2 has been dated to 282 ka (top MIS 9) based on the correlation to a pink-white layer in core PS115/2-2-2 (Stein et al., 2025). Both age estimates are quite close indicating a MIS 9 age for the biogenic carbonate-bearing brown bed B7. In contrast, the interval which may have a MIS 5e/late Termination II age (Song et al., 2023) is barren of foraminifera in core PS72/396-5 (Fig. 3).

Here, we critically discuss the benthic foraminifera bioevents to select those that are useful for stratigraphic correlation in the CAO. These bioevents are tentatively calibrated to independent lithological and chronological data. We refrain from assigning numerical ages or marine isotope substages to bioevents because of a lack of a robust independent chronostratigraphy. We rather stick to marine isotope stages that imply a certain duration and not a single definite age. This leaves some degree of freedom for chronostratigraphic interpretations based on other proxies.

Since evolutionary turnover does not play a role in the stratigraphic occurrence of most benthic foraminifera species in the Pleistocene of the Arctic Ocean, the composition of assemblages is determined by the complex interaction of ecologic requirements and taphonomic processes (Appendix Table B). Thus, the formation of bioevents is controlled by a set of factors rather than a single environmental variable (Martin, 2003; Loubere et al., 1993; Loubere and Rayray, 2016).

The spatial distribution of living benthic foraminifera and their preference for specific bathyal water depths is essentially controlled by their food and oxygen requirements (Jorissen, 2003; Jorissen et al., 1995). To a lesser extent competition, grain size of sediments, current activity, bottom water pH, and hydrostatic pressure determine the bathyal faunal composition (Gooday and Jorissen, 2012; Wollenburg et al., 2015).

However, foraminifera do not exclusively live at the sediment surface but can also survive at significant sediment depths if labile organic matter and oxygen is still available (Jorissen, 2003). Mean modern carbon export in the permanently ice-covered CAO is considered amongst the lowest in the world's oceans (Honjo et al., 2008; Nowicki et al., 2022) resulting in particulate organic carbon (POC) fluxes of 0.17–1 g C m⁻² yr⁻¹ at depths >1000 m (Harada, 2016; Roca-Martí et al., 2016). The low amount of POC reaching the seafloor in the CAO is usually immediately consumed at the sediment surface and not buried to sustain living foraminifera below the surface centimeter under a permanent ice cover (Wollenburg and Mackensen, 1998b). Moderate to deep-infaunal living taxa like Melonis zaandami and Nonionellina labradorica are only sustained where food flux is seasonally high in the seasonally ice-free areas. Species with an even higher food-demand like Bulimina aculeata (Jorissen et al., 1995) are absent from the modern Arctic Ocean (Wollenburg and Mackensen, 1998a, b; Wollenburg and Kuhnt, 2000; Husum et al., 2015).

Benthic foraminifera have been used to reconstruct past sea ice conditions (Cronin et al., 2008; Polyak et al., 2013; Seidenkrantz, 2013). However, benthic foraminifera are only indirectly linked to sea-ice conditions because primary production and sedimentation of organic matter is related to light-penetration through sea ice, upwelling processes at the ice margin and release of ballast material from melting sea ice (Anderson et al., 2003; Mar, 2014; Swoboda et al., 2024). The ice-covered bathyal Arctic Ocean is characterized by opportunistic shallow-infaunal and epilithic/-phytic foraminiferal taxa adapted to low to very moderate carbon flux (Wollenburg and Kuhn, 2000). In seasonally ice-free areas the surplus of labile organic matter provided by export from algae blooms (Swoboda et al., 2024) at the ice edge cause an increase in the number of species that dwell on the sea floor in the accumulated phytodetritus (Faiezieva et al., 2026). Such species rapidly reproduce after algae export events and are termed phytodetritus species (Gooday, 1988; Gooday and Lambshead, 1989; Thomas et al., 1995; Wollenburg and Kuhnt, 2000; Wollenburg et al., 2001b, 2004; Moodley et al., 2002; Polyak et al., 2013).

Among the taphonomic processes, dissolution strongly affects benthic foraminifera assemblages in the Arctic Ocean and its marginal seas (Hunkins et al., 1971; Herman et al., 1989; Steinsund and Hald, 1994; Wollenburg and Kuhnt, 2000; Wollenburg et al., 2001b, 2004; Loubere and Rayray, 2016). Thus, Hunkins et al. (1971, p. 234) assume that foraminifer-poor intervals in CAO sediments indicate conditions more favourable for complete dissolution of calcareous material. In our cores the abundance of thin-shelled specimens and the preservation status of more robust shells indicate when dissolution had a significant impact on the assemblage composition in a specific sample. Many epifaunal taxa, in particular phytodetritus species, which form a major component of living benthic foraminifera in the Arctic Ocean (Wollenburg and Mackensen, 1998a), are thin-shelled and are therefore susceptible to dissolution. These taxa are often lost in the bioturbated uppermost 5 to 10 cm sediments (Loubere and Rayray, 2016). Consequently, the Arctic fossil record usually consists of infaunal and occasional thick-shelled epifaunal taxa (e.g. Cibicides-types). In contrast, high percentages of epifaunal taxa like Epistominella exigua are unusual and cannot be explained with modern geochemical processes in near-surface sediments (Loubere and Rayray, 2016). The extensive loss of thin-shelled epifauna species is often caused by massive sedimentation of labile organic matter after algal blooms (Wollenburg and Kuhnt, 2000). The oxidization of this labile organic matter often leads to lowered pH and carbonate aggressive conditions in near-surface sediments and pore water resulting in a dominance of agglutinated taxa even in the living fauna (Scott and Vilks, 1991; Wollenburg and Mackensen, 1998a; Wollenburg and Kuhnt, 2000; Seidenkrantz, 2013).

If the bottom water pH drops seasonally or periodically to values ≤7.8, thin-shelled epifaunal foraminifera shells dissolve first. Therefore, assemblages enriched in robust infaunal species such as Bolivina arctica, Oridorsalis umbonatus, and especially Bulimina aculeata reflect a significant taphonomic loss in associated thin-shelled epi- and shallow-infaunal species (Figs. 12–13). Such dissolution-affected assemblages are common in the brown layers especially during the transition from glacial to interglacial conditions. Here the whitish and edged shells of thick-shelled calcareous infaunal taxa are accompanied only by shell fragments of a diminishing number of thin-shelled Stetsonia horvathi and Epistominella arctica in the small size fraction (Fig. 12).

The identification of MIS 5, especially the last interglacial MIS 5e, is a controversial issue, and the respective foraminifera fauna might have been lost by diagenetic processes in some cores from Lomonosov Ridge. The last interglacial was significant warmer than today, and on the northern Barents Sea continental slope this resulted in primary and export production exceeding todays values (Matthiessen and Knies, 2001; Wollenburg et al., 2001). The degradation of labile organic matter at the sea floor causes a drop in pH at the sediment-water interface (Steinsund and Hald, 1994; Wollenburg and Kuhnt, 2000) and is the main reason for the partial or complete dissolution of calcareous foraminifera at sites of high primary and export production in the modern and marginal Arctic Ocean (Steinsund and Hald, 1994; Wollenburg and Kuhnt, 2000; Wollenburg et al., 2001, 2007). We may thus presume also at site PS2185-6 a sea-ice cover being only seasonal and/or thinner and enabling a higher MIS 5e primary and export production than today. In such a scenario all calcareous foraminiferal shells were likely dissolved and not just a significant proportion as on the northern Barents Sea continental slope (Wollenburg et al., 2001). In core PS2185-6 we observe a few agglutinated C. subglobosus at ∼125 cm that could indicate interglacial or interstadial conditions, but this remains speculative until supported by additional data.

In the CAO calcareous foraminifera are usually restricted to Quaternary sediments, and were rarely reported from the Pliocene. Often sediments older than the Brunhes Chron exclusively comprise agglutinated foraminifera (O'Neill, 1981; Mullen and McNeil, 1995; Cronin et al., 2008b; Kaminski et al., 2009; Kender and Kaminski, 2013). Modern assemblages composed exclusively of agglutinated foraminifera occur only where carbonate dissolution is prevalent, in the deep-sea below the CTD, mud-flats or deep shelves (Murray and Alve, 2011; Murray, 2014). Nonetheless, these are residual assemblages resulting from post-depositional dissolution of calcareous shells in an originally mixed assemblage due to carbonate-aggressive past bottom- and/or pore water pH values (Murray and Alve, 2011). The transition from agglutinated assemblages or wide-spread barren core sections to predominantly calcareous faunas is characterized by the absence of thin-shelled and corrosion traces on more robust taxa (Figs. 12, 13).

The modern CAO benthic foraminifera fauna is rich in loosely agglutinated taxa like Crithionina spp., Rhizammina spp., and Aschemonella spp. that disintegrate soon after death, and robust agglutinated taxa (Wollenburg and Mackensen, 1998). Robust species, firmly agglutinated with ferruginous cement and often thicker shells like Rhabdammina spp., Saccammina spp., Psammosphaera fusca, Adercotryma glomerata, Glomospira spp., Reophax spp., Cribrostomoides subglobosus, Cyclammina spp. (Bender, 1989; Schröder, 1988), and the extinct Haplophragmoides obscurus, are often the only foraminifera preserved in residual sediment core assemblages. It is assumed that the preservation of agglutinated foraminifera is favoured by a rapid burial and/or exposure of sediments to reducing pore waters. Under oxidizing conditions, the iron in the cement is mobilized and shells disintegrate (Schröder, 1988). High organic-bound iron content (Gooday et al., 2008a) is further suggested to be responsible for preservation of organic-walled Placopsilinella aurantiaca down to >400 cm sediment depth in core PS2185-6 and in Alpha Ridge core CESAR 83-014 (Scott et al., 1989). As H. obscurus possess firmly agglutinated thick-shelled tests, we presume that it is the least vulnerable taxon susceptible to disintegration (e.g. this study, Scott et al., 1989) explaining their predominance in the intervals with exclusively agglutinated foraminifera. The fact that the original agglutinated residual assemblage with abundant H. obscurus had a higher diversity is obvious in samples where associated agglutinated species are observed. Accordingly, the absence of agglutinated foraminifera in core PS72/396-5 can likely be explained by sedimentation rates being too low to prevent iron mobilization from agglutinated shells. The occurrence of Saccorhiza ramosa, e.g. in laminated sediments of PS72/340-5, can be explained by its preference for moderate turbiditic water (Schröder, 1988). It is a dominant species in troughs of the Kara Sea where it is positively related to organic- and natural radionuclide-rich sediments (Domanov et al., 2017), indicating a preferred preservation at sites/times with increased sedimentation rate. Secondary enrichment by hydrodynamic sorting as it has been observed in distal turbidites on abyssal plains of the Northwest Atlantic, may have happened.

Environmental conditions changing with time may also have caused the formation of benthic foraminifera-rich intervals. The observed abundance maxima often occurred after intervals that are barren or contain only a few individuals. Even if one assumes that episodically most foraminifera were lost due to carbonate dissolution or disintegration of agglutinated foraminifera, recolonization must also have occurred after times of adverse living conditions that have caused the partial or total death of faunas. Dispersal of foraminifera following e.g. anoxia or extinction is usually rapid on geological time scales (Buzas and Culver, 1991; Alve, 1999; Murray, 2006), however, to restore the full faunal biodiversity after disturbance may take several millennia indicating different species-dependent capabilities of dispersal (Schmiedl et al., 2003). Based on studies on shallow-water foraminifera and occasional net catches, it is assumed that the dispersal and recolonization of foraminifera in the deep realm occurs via propagules, juvenile individuals smaller than 32 µm (Alve and Goldstein, 2010; Alve and Goldstein, 2003; Murray, 2006; Gooday and Jorissen, 2012). The possible time span of a viable transport of propagules is different for different species and up to two years (Alve and Goldstein, 2014; Alve and Goldstein, 2010). As Fram Strait (sill depth ∼2500 m) is the only deep-water connection of the Arctic to the world's ocean, any recolonization has and had to occur through Fram Strait. Propagules are advected by inflow of waters from subpolar latitudes, then circulating anticlockwise as Atlantic Water (∼200–600 to 850 m) and Upper Polar Deep Water (Rudels and Carmack, 2022; Timmermans and Marshall, 2020). The fact that Atlantic Water advection is a main factor controlling the distribution of certain foraminiferal species in the Arctic Ocean has been previously suggested based on the restricted distribution of certain taxa in areas close to the Atlantic water inflow (Wollenburg and Kuhnt, 2000). Combined sediment and water DNA analyzes around western and northern Svalbard now support these observations and indicate effective foraminiferal propagule dispersal by Atlantic Water (Nguyen, 2022; Nguyen et al., 2025). From entrance to exit the mean transit time of modern Atlantic Water circulation in the Arctic Ocean is 15–55 years for a full circle (Wefing et al., 2021). In the past the maximum reachable location for a settlement of Atlantic-derived foraminifera species within the Arctic Ocean depended on the species, the local environmental conditions, and the strength of Atlantic water advection during that time. In particular the availability of food played then a major role for the successful colonization at a particular site, not only for the invading species but also the species endemic to the CAO (H. obscurus, B. arctica).

The modern Arctic Ocean shares the majority of foraminiferal species with the Norwegian-Greenland Sea (Lagoe, 1977; Mackensen et al., 1985; Scott and Vilks, 1991; Bergsten, 1994; Struck, 1995; Wollenburg and Mackensen, 1998a; Wollenburg and Kuhnt, 2000; Rasmussen et al., 2003b; Rasmussen and Thomsen, 2008; Husum et al., 2015). In pre-Brunhes sediments this was different, e.g. Oridorsalis umbonatus occurred in Late Miocene sediments on the Vøring Plateau (Osterman and Qvale, 1989) but likely much later in the Arctic Ocean. As other calcareous taxa like E. exigua, that possess a thinner shell than O. umbonatus or C. teretis/neoteretis with an approximately similar shell-thickness, are preserved in much older sediments, this indicates that living propagules of O. umbonatus had not been successfully advected to the investigated sites for considerable time before the event.