Re‐investigation of the Bispingen palaeolake sediment succession (northern Germany) reveals that the Last Interglacial (Eemian) in northern‐central Europe lasted at least ~15 000 years

Investigating past interglacial climatic and environmental changes can enhance our understanding of the natural rates and ranges of climate variability under interglacial boundary conditions. However, comparing past interglacial palaeoclimate records from different regions and archives is often complicated by differing and uncertain chronologies. For instance, the duration of the Last Interglacial in Europe is still controversial as southern European palaeoclimate records suggest a duration of ~16 500–18 000 years, while a length of only ~11 000 years in northern‐central Europe was previously inferred from the analysis of partly annually laminated (varved) palaeolake sediments recovered at Bispingen, northern Germany. To resolve this discrepancy, we here present sediment microfacies, geochemistry and pollen data from a new sediment core from the Bispingen palaeolake sediment succession, covering the entire Last Interglacial (Eemian) and the earliest part of the Last Glacial (Weichselian). In particular, we provide evidence that the duration of the Last Interglacial at Bispingen must have been hitherto underestimated due to the investigation of an incomplete sediment core. Using microscopic varve counting and sedimentation rate estimates for non‐varved sections on the new sediment core, we show that the Eemian in northern‐central Europe probably lasted at least ~15 000 years, about 4000 years longer than previously thought. This new duration estimate is in much better agreement with results from southern European palaeoclimate records, clarifying the enigma of a steep trans‐European vegetation gradient for several millennia at the end of the Last Interglacial.

The Last Interglacial, ~125 000 years ago, was characterized by a smaller global ice volume as well as higher air temperatures and sea levels than today (Bintanja et al. 2005;Dutton & Lambeck 2012;Fischer et al. 2018;Fox-Kemper et al. 2021), a function of strong eccentricityprecession forcing despite lower atmospheric CO 2 levels (Schneider et al. 2013).As one of the warmest interglacials during the last ~800 000 years (Past Interglacials Working Group of PAGES 2016) and the most recent periodwith climate conditions similar to those expected in the near future under rising atmospheric greenhouse gas concentrations and continuing global warming, it is therefore suitable to investigate environmental responses to warmer-than-present climate conditions as well as natural rates and ranges of interglacial climate variability.Consequently, numerous palaeoclimate proxy records from this time interval have been obtained from different types of sedimentary archives in Europe and the adjacent North Atlantic.However, the establishment of robust internal chronologies for records from this region as well as their absolute dating and their alignment still pose a major challenge.To date, only very few complete and accurately dated proxy records of Last Interglacial climate variability in Europe are available from speleothems (e.g.Drysdale et al. 2005Drysdale et al. , 2007;;Tzedakis et al. 2018;Wilcox et al. 2020;Luetscher et al. 2021) and annually laminated (varved) (palaeo)lake sediments (e.g.Sirocko et al. 2005;Brauer et al. 2007;Allen & Huntley 2009).In contrast, the vast majority of regional terrestrial palaeoclimate proxy records, mostly originating from non-varved (palaeo)lake sediments, suffers from substantial chronological uncertainties.This includes records that have (i) no age control at all (e.g.Mamakowa 1989;Litt 1994;Kołaczek et al. 2012;Bober et al. 2018;Malkiewicz 2018a, b;Kupryjanowicz et al. 2021;Pidek et al. 2021), (ii) only weakly constrained age models (e.g.Hahne et al. 1994;Tzedakis et al. 2003;Sinopoli et al. 2018), or (iii) chronologies that are only based on tentative correlations to marine sediment records (e.g.Kukla et al. 1997Kukla et al. , 2002a;;Milner et al. 2013;Salonen et al. 2018;Kern et al. 2022).In contrast, the chronologies of marine palaeoclimate records from the eastern DOI 10.1111/bor.12649 Ó 2024 The Authors.Boreas published by John Wiley & Sons Ltd on behalf of The Boreas Collegium.
This is an open access article under the terms of the Creative Commons Attribution License, which permits use, distribution and reproduction in any medium, provided the original work is properly cited.
North Atlantic rely either on orbital tuning (e.g. S anchez Goñi et al. 1999), the calibration of benthic foraminifera stable oxygen isotope (d 18 O) data against radiometrically dated coral and speleothem records of past sea level changes (e.g.Shackleton et al. 2002Shackleton et al. , 2003;;S anchez Goñi et al. 2008), or correlation to absolutely dated terrestrial palaeoclimate records (Tzedakis et al. 2018).The equivocal reliability, precision and accuracy of these various chronological approaches seriously hamper the reliable alignment of individual proxy records and the robust assessment of the duration of the Last Interglacial at different sites and consequently (i) the reliable quantification of the rates of Last Interglacial climatic and environmental changes and (ii) the identification of regional leads and lags in climate development.
The first estimate for the duration of the Last Interglacial (i.e. the Eemian in northern-central European pollen stratigraphies) in the circum-Atlantic region was based on a tentative correlation between a phase of low foraminifera d 18 O values in North Atlantic sediment cores and the occurrence of interglacial vegetation in Europe.This led to the conclusion that the Last Interglacial comprised only the oldest part of Marine Isotope Stage (MIS) 5, i.e. substage MIS 5e, and had a duration of ~11 000 years (Shackleton 1969).The exact duration and timing of the Last Interglacial in continental Europe remained, however, elusive because the d 18 O analyses on the deep-sea sediments were not accompanied by pollen data.Nevertheless, the idea of a Last Interglacial duration of ~11 000 years was supported shortly thereafter by evidence from the partly varved Eemian palaeolake sediment succession of Bispingen, northern Germany (M€ uller 1974).In contrast, later studies on pollen records from eastern and southerncentral France suggested a substantially longer duration of the Last Interglacial (~19 000-23 000 years) and its extension into MIS 5d (Kukla et al. 1997(Kukla et al. , 2002a, b) , b) as well as a millennia-long vegetation gradient between southern and northern-central Europe at its demise (Kukla et al. 2002b;Tzedakis 2003).However, these duration estimates were controversially discussed as they were only based on tentative correlations to the marine d 18 O stratigraphy and not on independent dating (Turner 2002).Since then, considerable progress has been made in refining the chronology of the Last Interglacial in Europe.For instance, parallel pollen and d 18 O analyses on marine sediments from the Iberian margin yielded a length of ~16 400 years for the pollen-defined Last Interglacial and revealed a considerable asynchrony between global ice volume-driven d 18 O changes in the marine realm and terrestrial vegetation development (S anchez Goñi et al. 1999;Shackleton et al. 2002Shackleton et al. , 2003)), while pollen data from the varved lake sediment record of Lago Grande di Monticchio, southern Italy, provided evidence for a Last Interglacial duration of 17 700AE200 years (Brauer et al. 2007;Allen & Huntley 2009).Furthermore, a precisely dated speleothem d 18 O record from Corchia Cave, northern Italy, indicated that the Last Interglacial lasted ~16 600 years (Drysdale et al. 2005(Drysdale et al. , 2007)), which has recently been revised to ~18 000 years, providing also a new chronological framework for pollen records from the Iberian margin (Tzedakis et al. 2018).These latest findings leave the ~50year-old results from the partly varved Bispingen palaeolake sediment succession, which determine the duration of the Last Interglacial in northern-central Europe to be ~5500-7000 years shorter than further south, as a notable exception.However, whether this reflects chronological deficiencies of the individual records, difficulties in correlating regionally different vegetation successions and/or proxies (e.g.pollen vs. d 18 O) or a prolonged vegetation gradient between southern and northern-central Europe at the end of the Last Interglacial, remains unresolved.
To re-evaluate the biostratigraphy and chronology of the Eemian in northern-central Europe we analysed a new sediment core from the Bispingen palaeolake sediment succession.Here we present the results of detailed sediment microfacies analysis, including microscopic varve counting, as well as geochemical and palynological analyses on this sediment core.By comparing these data with results from the classic sediment core KS 186/70 (M€ uller 1974), we revise the duration estimates of individual pollen zones, establish a new floating chronology for the Eemian in northern Germany and discuss this in relation to other Last Interglacial palaeoclimate records from Europe.

Study area
The Bispingen palaeolake sediment succession is located in northern Germany, ~1.3 km southwest of the village of Bispingen (approx.latitude 53°05 0 N, longitude 10°00 0 E) and ~50 km south of Hamburg (Fig. 1).It was discovered in the early 1970s during a large-scale prospection for diatomite deposits and belongs to a series of spatially confined palaeolake sediment successions in the upper valley of the Luhe River, some of which had been exploited for diatomite since the 19th century (Benda & Brandes 1974).The diatomite deposit that is associated with the Bispingen palaeolake sediment succession covers an area of ~0.1 km 2 (estimated volume ~150 000 m 3 ) and is characterized by a wedge-shaped geometry, reaching maximum thickness in the southeast and rapidly thinning towards the west and north/northeast, as deduced from a suite of spatially distributed sediment cores obtained during the early 1970s prospection (Benda & Brandes 1974).Detailed pollen analyses on one of these sediment cores (KS 186/70) from the southeastern margin of the palaeolake sediment succession (Fig. 1) revealed that the former lake existed throughout the entire Last Interglacial (Eemian) as well as during the early Last Glacial (Weichselian) (M€ uller 1974).The similarity of the pollen record to other sites in northern-central Europe (e.g.Menke & Tynni 1984;Mamakowa 1989;Hahne et al. 1994;Litt 1994;Kołaczek et al. 2012;Malkiewicz 2018a, b;Kupryjanowicz et al. 2021;Pidek et al. 2021;Suchora et al. 2022) with respect to characteristic vegetation elements and their succession thereby unequivocally proves the Last Interglacial to early Last Glacial age.The Bispingen palaeolake developedlike all former lakes in the upper valley of the Luhe Riverin an ENE-WSWoriented subglacial meltwater channel that had been formed during the Drenthe Stage ice advance, the main phase of the Penultimate Glacial (Saalian), as indicated by petrographic analysis of the sediments that constitute the hills surrounding the Luhe River valley (Benda & Brandes 1974).As the following Warthe Stage ice advance, the last major glacier advance of the Saalian, did not reach south of Bispingen (Ehlers et al. 2004), the then-exposed high-relief subglacial channel was partly filled with reworked glacifluvial sands and gravels during the terminal phase of the Saalian.This initial phase of channel filling was followed by the formation of several small lakes in the remaining basins and the deposition of calcareous muds, gyttjas and diatomites during the Eemian (Benda & Brandes 1974).As the glacier advances of the following Weichselian did not reach the Saalian ice margins (Ehlers et al. 2004), all former lake basins in the upper valley of the Luhe River were covered during the Weichselian by several metres of glacifluvial sands, which are mainly composed of reworked Drenthe Stage material (Benda & Brandes 1974).
core segments from these two sites was impossible, consecutive core segments from site BIS-3 were used for BIS-2000, considering gaps of 2 cm between the consecutive core segments to account for possible sediment loss during core retrieval.BIS-2000 has a total length of 3123 cm and the palaeolake sediments between the basal Saalian and overlying Weichselian sands cover the section between 3068 and 1555 cm (Fig. 2).

Bulk sediment geochemical analyses
Bulk sediment geochemical analyses were carried out at the GFZ on freeze-dried and homogenized sediment samples that were taken from BIS-2000 between 3072 and 1555 cm (n = 200; average sampling resolution 10 cm, locally increased to 2-4 cm).The total inorganic carbon (TIC) content was measured on ~50-mg aliquots by coulometric titration using a STR € OHLEIN Coulomat 702 (CO 2 released by treatment with hot H 2 PO 4 ).As microscopic inspection revealed that calcite is the dominant carbonate mineral in the sediments (see below), the CaCO 3 content was calculated stoichiometrically from the TIC content (CaCO 3 = TIC 9 8.333).Total carbon (TC) and total nitrogen (TN) contents were measured by burning ~100-200 mg aliquots in a LECO CNS-2000 elemental analyser in an oxygen gas flow at 1350 °C and measuring TC and TN contents by infrared and heat conductivity detection, respectively.Total organic carbon (TOC) contents of each sample were finally calculated as the difference between TC and TIC contents.TOC and CaCO 3 contents are expressed as per cent of sediment dry weight and the C/N ratio was calculated as the TOC/TN mass ratio.

X-ray fluorescence core scanning
To characterize the element composition of the BIS-2000 sediments, non-destructive X-ray fluorescence (XRF) core scanning was carried out at the GFZ.First, a fresh core surface was prepared for all split core segments of the composite profile, which was covered with foil before XRF measurements were conducted with an ITRAX XRF core scanner.Measurements were performed every 2 mm by irradiating the split core surface with a Rh Xray source (30 kV, 55 mA) for 15 s.As element intensity records obtained by XRF core scanning provide information about relative changes in element composition that are affected by down-core variations of the physical properties of the sediments, core surface geometry and matrix effects (Tjallingii et al. 2007;Weltje & Tjallingii 2008), the geochemical characterization is best presented by log-ratios of the element intensities (Weltje & Tjallingii 2008).Therefore, logratios of the silicon, calcium and titanium intensities, i.e. log(Si/Ti) and log(Ca/Ti), are presented here to visualize relative variations in the content of biogenic silica (i.e.diatoms) and calcite with respect to detrital components throughout the sediment record.

Palynological analyses
Pollen analyses on ~1-cm 3 sediment samples from BIS-2000 (n = 138; average sampling resolution 10 cm, locally increased to 2-4 cm), covering the section between 3068 and 1819 cm, were carried out at the University of Bonn, Germany, and subsequently at the University of the Witwatersrand, South Africa.Sample preparation followed the standard method described by Berglund & Ralska-Jasiewiczowa (1986), including treatment with cool HF and HCl, 3 min acetolysis, staining with safranine, and mounting in glycerol.At the University of the Witwatersrand, semi-permanent slides were prepared using glycerine jelly.In total, 442-1016 (median: 517) terrestrial pollen grains per sample were counted using a light microscope at 40009 magnification (1009 oil immersion objective).Identification of individual pollen taxa was aided by modern pollen reference collections of central European plants and descriptions of the European pollen flora (Punt 1976;Punt & Clarke 1980, 1981, 1984;Punt et al. 1988Punt et al. , 1995Punt et al. , 2003;;Punt & Blackmore 1991;Beug 2004).Pollen percentages were calculated following the standard method based on the sum of trees/shrubs (arboreal pollen -AP) and herbs (non-arboreal pollen -NAP; except aquatic and wetland plants) and the pollen diagram was plotted using TILIA (Grimm 1992).Additionally, we applied the percentage calculation originally used for KS 186/70 (Corylus and non-tree pollen percentages based on tree pollen values only; M€ uller 1974).Based on this calculation, the BIS-2000 pollen record was subdivided into individual pollen assemblage and abundance zones (PAAZ; Table 1) following the classic Eemian zonation scheme for northern Germany (Selle 1962).Being exactly the same zonation as used for KS 186/70 (M€ uller 1974), this enables a direct comparison between the two sediment cores with respect to the thickness/duration of individual PAAZ.However, if not stated otherwise, all pollen percentages presented here are based on the standard method.

Sediment microfacies analysis, varve counting and sedimentation rate estimates
Large-scale petrographic thin sections for sediment microfacies analysis were prepared at the GFZ according to the method described by Brauer et al. (1999) from sediment slabs (100 9 20 9 10 mm) that were taken continuously from the core segments included in BIS-2000.Microscopic analyses were carried out at different magnifications (25-4009) and optical conditions (planeand cross-polarized light, dark field illumination) using a NIKON SMZ-U stereoscopic microscope and a ZEISS Axiophot polarization microscope.
To establish a floating chronology for BIS-2000, microscopic varve counting was carried out between 3047 and 2922 cm.Within this section, varves were counted three times and classified into three categories according to their preservation state and the variance between the individual counts: (i) sections with wellpreserved and clearly distinguishable varves (no variance between individual counts and therefore negligible counting uncertainty), (ii) varved sections where the discrimination of the boundaries of individual varves was partly difficult (5% counting uncertainty), and (iii) faintly varved sections, where the discrimination of individual varve boundaries was hardly possible and the number of varves was interpolated using the average varve thickness/sedimentation rate of adjacent well-varved sections (estimated uncertainty 20%).The results of microscopic varve counting (including local varve thickness-based interpolation) were used to determine the duration of individual PAAZ in the section 3057-2891 cm (see sections 'Sediment microfacies, varve counting and sedimentation rate estimates' and 'Pollen zonation and duration of individual PAAZ').In contrast, to determine the duration of individual PAAZ below 3057 cm and between 2891 and 2046 cm, i.e. in sections of BIS-2000 where reliable varve counting was not possible, we used for convenience exactly the same average sedimentation rates as deduced for these PAAZ from their thickness and estimated duration in KS 186/70 (M€ uller 1974) (see Table 2 and sections

Sediment composition and geochemistry
Based on sediment composition, geochemistry and microfacies, seven sedimentological units (1 to 7 from bottom to top) can be distinguished in BIS-2000 between 3068 and 1555 cm (Fig. 2).
Unit 1 (3068-3047 cm) is composed of dark brownish grey calcite mud.It represents the initial stage of lacustrine sedimentation and directly overlies yellowish grey, fine to medium sands with occasionally intercalated coarse sand and gravel layers that were deposited during the Saalian (see Benda & Brandes 1974).The main characteristic of unit 1 is the distinct increase in CaCO 3 from ~2% at the base to ~80-90% at the top, also reflected by rapidly increasing log(Ca/Ti) values (Fig. 2).However, the sediments still contain abundant reworked fine sand-to clay-sized siliciclastics (mainly quartz and feldspar).Further components are diatoms and amorphous organic matter (TOC ~1-4%; Fig. 2) as well as lm-scale pyrite framboids, which occur finely dispersed within the matrix above 3055 cm.
Unit 2 (3047-2922 cm) consists of brownish grey calcite mud that is, in contrast to unit 1, characterized by a distinct and mostly well-preserved sub-mm-scale lightdark lamination (Fig. 3A, B).CaCO 3 gradually decreases from >80 to ~60%, also reflected by a decrease in log(Ca/Ti), while the TOC content ranges between ~3 and ~5% (Fig. 2).C/N values of ~15 (Fig. 2) indicate that the organic matter consists of a mixture of algae, aquatic macrophytes and higher terrestrial plants (see Meyers & Lallier-Verg es 1999).Diatoms become increasingly abundant while the amount of silt-to clay-sized siliciclastics distinctly decreases.Nest-like aggregates of lm-scale pyrite framboids, partly forming discrete layers of ~50-100 lm thickness, are frequent in the lower part of unit 2. The boundary to unit 3 (2922-2129 cm), which is composed of calcitic diatomite, is defined by the disappearance of the continuous sub-mm-scale lightdark lamination; instead, occasional diatom layers and calcite lenses form a faint lamination (Fig. 3C).CaCO 3 remains at ~50-60% until ~2850 cm, followed by a distinct decrease to ~30% between ~2850 and ~2770 cm and a subsequent gradual decrease to ~3% at the top of unit 3, also reflected by gradually decreasing log(Ca/Ti) values (Fig. 2).The decreasing calcite content is accompanied by a gradual change in sediment colour from brownish grey to dark greyish brown.In contrast, the diatom content remains relatively constant throughout unit 3, reflected by stable log(Si/Ti) values (Fig. 2), and also the content of amorphous organic material and plant debris (TOC ~5-11%) as well as the C/N values (~12-15) remain largely unchanged (Fig. 2).Only very few silt-to clay-sized siliciclastics as well as pyrite are diffusely dispersed within the sediment and above 2879 cm vivianite occurs.An intercalated sand layer at 2239-2221 cm with lumps of regular lacustrine sediment most likely reflects instantaneous input of catchment material that apparently dislocated lacustrine sediments from the basin slope.Relatively high log(Si/Ti) values across the sand layer (Fig. 2) are related to its high quartz content, obscuring the reflection of the otherwise largely constant diatom content by log(Si/Ti).
Unit 4 (2129-1886 cm) is composed of nonlaminated, dark brown diatomite that is still relatively organic-rich (TOC ~5-9%) but, in contrast to unit 3, virtually calcite-free (CaCO 3 ~1-3%), which is also reflected by lower log(Ca/Ti) values (Fig. 2).Diatoms are still abundant as indicated by largely unchanged log(Si/Ti) values (Fig. 2) and horizontally bedded organic macro-remains as well as very few finely dispersed siliciclastics, pyrite and vivianite occur, the latter occasionally forming discrete layers of up to 1 mm thickness.
Subunit 5a (1886-1849 cm) is characterized by a faint mm-scale alternation of clayey-siliciclastic and organicdetrital layers, indicating a significantly changed depositional environment.The organic matter content is lower (TOC <6%) than in unit 4 (Fig. 2), vivianite is absent, but diatoms are still abundant.The disappearance of diatoms in overlying subunit 5b (1849-1697 cm) apparently marks the end of lacustrine deposition.Sediments in this subunit are characterized by a cmscale alternation of layers that contain silt-sized siliciclastics mixed with fine-grained organic material and layers that consist of sand mixedwith coarse plant debris.The increasing contribution of terrestrial plant material is reflected by a gradual C/N increase from ~15 to >30 (Fig. 2; see Meyers & Lallier-Verg es 1999), while slightly higher log(Si/Ti) values reflect the increased amount of siliciclastics in absence of diatoms.Strong variations in TOC (~1-6%) above 1771 cm reflect the random sampling of either more clastic or more organic layers.
Unit 6 (1697-1606 cm) is characterized by another sharp shift in sediment composition.Following the deposition of mixed organic-clastic sediments in subunit 5b, the recurrence of diatoms and vivianite in unit 6 most likely reflects the recommencement of lacustrine sedimentation.The amount of fine-grained siliciclastics and coarse-grained organic debris, both scattered within the dark brown to greyish brown diatomite, is lower than in subunits 5a and 5b.TOC values of ~6-9% (Fig. 2) mainly reflect amorphous organic matter.
The boundary to unit 7 (1606-1555 cm) is characterized by an abrupt re-increase in the amount of siliciclastics and organic macro-remains, deposited as alternating mm-scale layers of either silt-sized siliciclastics or organic debris.Diatoms occur only sporadically, partly as discrete mm-scale layers, up to 1569 cm.The subsequent disappearance of diatoms together with a drop in TOC from >9 to <4% (Fig. 2) and an increase in siliciclastics most likely reflects the terminal silting-up of the lake.This is corroborated by relatively high C/N values (>40) in the upper part of unit 7 (Fig. 2), indicating a strong contribution of higher terrestrial plants (see Meyers & Lallier-Verg es 1999).
Above 1555 cm, the lacustrine sediments are covered by yellowish to light grey, quartz-rich fine to medium sands (reflected by high log(Si/Ti) values) with some intercalated organic-rich layers, which were deposited during the Weichselian (see Benda & Brandes 1974).

Sediment microfacies, varve counting and sedimentation rate estimates
The sediments of unit 2 in the basal part of BIS-2000 are characterized by a distinct sub-mm-scale light-dark lamination.The first well-defined light-dark laminae couplets occur between 3047 and 3024 cm and each couplet in this section is composed of two sublayers (Fig. 3A).The basal light sublayer contains, besides some diatoms, mostly micritic calcite.As biogeochemical calcite precipitation is mainly driven by (i) a shift in the CaCO 3 /H 2 CO 3 balance due to photosynthetic removal of CO 2 from the water by algal blooms in early spring and (ii) rising water temperature, affecting the solubility of CO 2 in the water and hence its CaCO 3 saturation (Brunskill 1969;Kelts & Hs€ u 1978;Zolitschka et al. 2015), the calcite sublayer is interpreted to represent late spring and summer deposition.The distinct upward grading of the calcite crystals (e.g.Geyh et al. 1971;Kelts & Hs€ u 1978;Lotter 1989;Brauer et al. 2008) most likely reflects gradually changing saturation levels during the period of calcite precipitation (Brauer 2004), the successive deposition of coarse-to fine-grained calcite crystals due to size-dependent sinking velocity, and/or the partial dissolution of the fine-grained crystals during sinking (Kelts & Hs€ u 1978).The overlying dark sublayer contains in contrast to the light sublayer predominantly organic material and some siliciclastics as well as abundant pyrite framboids, which are likely early diagenetic and partly form discrete layers (Fig. 3A).This sublayer is interpreted to represent deposition during autumn and early spring.In general, the light sublayers are almost twice as thick as the dark sublayers and the thickness of the laminae couplets ranges between ~0.4 and >1.0 mm with thinner couplets at the base of this section.
Between 3024 and 3002 cm, the abundance of diatoms strongly increases and discrete diatom sublayers are occasionally observed below the calcite sublayers.However, the laminae couplets in this section are relatively poorly preserved and the boundaries between individual sublayers are mostly faint.
Above this poorly laminated section, a different laminae microfacies with three instead of two sublayers is observed between 3002 and 2922 cm (Figs 3B, 4) (Tables 1, 2).The sublayer succession thereby resembles that of laminated lake sediments from northern Germany and Poland, where seasonally different deposition was proven by sediment trap studies (see Roeser et al. 2021).The base of each sublayer succession in this section is formed by a light diatom sublayer with only very little biogeochemically precipitated calcite (Figs 3B, 4).It is consequently considered to reflect early spring diatom growth in response to increased nutrient availability following spring circulation (Bluszcz et al. 2008;Maier et al. 2018) and rising air and water temperatures (Zolitschka et al. 2015).The light diatom sublayer is overlain by another light sublayer that is composed of biochemically precipitated micritic calcite.It is similar to the calcite sublayers observed between 3047 and 3024 cm (Figs 3B, 4) and consequently interpreted to also reflect deposition during late spring and summer.The calcite sublayer is again overlain by a dark sublayer that mainly contains amorphous organic matter but only little calcite and few diatoms   4).This sublayer is interpreted to reflect resuspension processes during the autumn and winter stagnation period (see Roeser et al. 2021).The average thickness of the laminae couplets in this section is ~0.8 mm and couplets become slightly thicker (~1.0 mm) towards the top, being mainly related to a thickening of the dark sublayers.
Based on the composition and succession of the individual sublayers, which are similar to those of proven annually laminated sediments from other (palaeo)lakes in carbonate-rich catchments (e.g.Geyh et al. 1971;Kelts & Hs€ u 1978;Lotter 1989;Brauer et al. 2008;Lauterbach et al. 2011Lauterbach et al. , 2019;;Tylmann et al. 2013;Roeser et al. 2021), the sub-mm-scale light-dark laminae couplets in the basal part of BIS-2000 can be classified as carbonaceous-biogenic varves (see Zolitschka et al. 2015).Microscopic counting across the entire section between 3047 and 2922 cm revealed 1572AE160 varves.This includes 308AE45 varves between 3047 and 3024 cm, 275AE46 varves in the poorly laminated section between 3024 and 3002 cm and 989AE69 varves between 3002 and 2922 cm.These results were in the following applied to the pollen zonation of BIS-2000 to infer the duration of individual PAAZ in the distinctly varved lower part of the sediment succession (see section 'Pollen zonation and duration of individual PAAZ').
In contrast, the duration of individual PAAZ in the faintly/non-varved parts of BIS-2000 (below 3047 cm and between 2891 and 2046 cm) was determined by adopting the average sedimentation rate estimates for these PAAZ from KS 186/70 (M€ uller 1974; Table 1).
Based on local varve counts in PAAZ IIIc and the observation that the ratio between diatoms and organic matter in the faintly/non-varved middle and upper part of KS 186/70 (PAAZ IV-VIb) remained fairly constant while the calcite content concomitantly decreased, M€ uller (1974) assumed lower sedimentation rates for the PAAZ IV-VIb section than for the varved part of KS 186/70 (PAAZ IIa-IIIc).Geochemical analyses on the BIS-2000 sediments indicate that, in agreement with the observation of M€ uller (1974), the proportions of organic matter, diatoms and siliciclastics remain also more or less stable in the middle and upper part of the sediment succession, i.e. between ~2850 and 2046 cm (Fig. 2).In contrast, the calcite content continuously decreases from ~55-65% at ~3000-2850 cm (including the uppermost part of the distinctly varved section) to ~30% at ~2770-2700 cm and further to ~10-30% at ~2700-2300 cm, ~5-10% at ~2300-2130 cm and <3% above ~2130 cm (Fig. 2).Due to the consistent sedimentological/geochemical evidence from BIS-2000 and KS 186/70, we therefore also assume lower sedimentation rates for the faintly/non-varved section of BIS-2000 above ~2700 cm and adopted for the PAAZ in this interval for convenience the same average sedimentation rates as those calculated from the thickness and estimated duration of the respective PAAZ in KS 186/70 (Table 2; M€ uller 1974).

Pollen zonation and duration of individual PAAZ
Using the same zonation criteria as for KS 186/70 (M€ uller 1974; Table 1), the pollen diagram for the ).As a detailed interpretation of the palynological results in terms of climatic/environmental changes is beyond the scope of the present study, the pollen data are used here only for the biostratigraphical characterization of the Eemian deposits and the establishment of a chronology.PAAZ I (Betula; 3068-3057 cm) is characterized by high percentages of Betula (~45-58%) and Pinus (~20-49%) as well as decreasing Juniperus and Salix percentages (Fig. 5), reflecting the spread of pioneer forests at the onset of the Eemian.Open-land indicators, most prominently Poaceae, fluctuate but generally decrease from ~13 to ~7% (Fig. 5).Grassy communities with Artemisia dominate in drier habitats, whereas wetter habitats are characterized by Cyperaceae and ferns.The boundary to PAAZ IIa is defined by the intersection of the Betula and Pinus pollen curves and the increase of Quercus and Ulmus to >1% (M€ uller 1974; Table 1).Using the same average sedimentation rate as in KS 186/70 (~0.70 mm a À1 ; see section 'Sediment microfacies, varve counting and sedimentation rate estimates'), PAAZ I lasted ~160 years (Table 2).
PAAZ IIa (Pinus-Betula; 3057-3035 cm) is characterized by high Pinus percentages (~70%) and a parallel decrease of Betula from ~39 to ~21%, reflecting the replacement of pioneer woodlands by boreal forests (Fig. 5).Gradual soil stabilization due to the spread of boreal forests could explain the decreasing content of allochthonous siliciclastics at the transition between sedimentological units 1 and 2 (3047 cm).Quercus spreads and reaches ~3%, whereas Salix and Juniperus are only weakly present (Fig. 5).As already observed by M€ uller (1974), Ulmus increases before Quercus and reaches ~2-9%.Fraxinus generally remains <1% and open-land indicators, including grasses, reach a maximum of ~3%.The boundary to PAAZ IIb is defined by the intersection of the Betula and mixed oak forest pollen curves (M€ uller 1974; Table 1).Microscopic varve counting, including varve thickness-based interpolation for the lowermost 10 cm, yielded a duration of 365 years for PAAZ IIa (Table 2).
PAAZ IIIa (Quercetum mixtum-Corylus; 3005-2975 cm) is characterized by a steep increase of Corylus from ~15 to ~42%, while Pinus drops to <30% and Ulmus and Quercus moderately decrease (Fig. 5).Alnus culminates at the end of PAAZ IIIa and Tilia and Taxus are recorded for the first time.Following the percentage calculation applied to KS 186/70 (M€ uller 1974; Table 1), the boundary to PAAZ IIIb is defined by the increase of Corylus to >100% (Figs S1, S2).Applying this calculation method to BIS-2000, similar Corylus percentages are reached at the top of PAAZ IIIa (Fig. S1).As inferred from microscopic varve counting, PAAZ IIIa lasted 406 years (Table 2).
PAAZ IV (Carpinus; 2731-2303 cm) can be subdivided into a lower (PAAZ IVa; 2731-2571 cm) and an  1).The pollen count data are available at the European Pollen Database (https://epdweblog.org)via the Neotoma Paleoecology Database (http://www.neotomadb.org;dataset ID 40169).Lithological symbols as in Fig. 2. upper part (PAAZ IVb; 2571-2303 cm), separated by a weakly defined Carpinus minimum and a parallel Picea increase (M€ uller 1974).PAAZ IVa represents the latetemperate phase of the Eemian and is characterized by the expansion of Carpinus (from ~18 to 35-40%) and Picea (from ~6 to ~17%).Parallel, a decline of Quercus, Ulmus, Corylus, Taxus and Tilia is observed (Fig. 5).Ilex and Hedera are well distributed, suggesting high summer temperatures and oceanic climate conditions (e.g.Malkiewicz 2018a, b).Low percentages of dwarf shrubs, herbs and wetland elements indicate a closed forest canopy.The overlying PAAZ IVb is characterized by the strongest presence of Carpinus (maximum ~48%) and Picea (maximum ~24%) and a steady increase of the Betula and Quercus percentages.Corylus, Ulmus and Tilia percentages decline and oceanic components (Ilex, Hedera, Buxus) occur only sporadically.Calluna, regularly present with up to 4%, reflects the presence of a reed belt.The boundary to PAAZ Va is marked by the intersection of the Carpinus pollen curve with the Pinus and Picea pollen curves (M€ uller 1974; Table 1).Applying the same average sedimentation rate as in KS 186/70 (~0.60 mm a À1 ; see section 'Sediment microfacies, varve counting and sedimentation rate estimates'), PAAZ IV lasted ~7130 years (Table 2).
PAAZ VIa (Picea-Pinus; 2175-2134 cm) reflects the terminal phase of the Eemian and is characterized by a major decline of thermophilous elements.Pinus increases to ~76%, while Picea decreases to ~9%, Carpinus and Quercus drop to <5% and Alnus decreases to <10% (Fig. 5).Dwarf shrubs and herbs fluctuate around 5% while Calluna increases to ~4%.The boundary to PAAZ VIb is defined by the intersection of the Betula and Picea pollen curves and the increase of dwarf shrubs and herbs (NAP) to >5% (M€ uller 1974; Table 1).Although NAP values already varied between ~3 and ~6% in PAAZ Vb when applying the percentage calculation of M€ uller (1974) (Fig. S1), we placed the upper boundary of PAAZ VIa only at the significant NAP increase between 2145 and 2124 cm.Using the same average sedimentation rate as in KS 186/70 (~0.74 mm a À1 ; see section 'Sediment microfacies, varve counting and sedimentation rate estimates'), PAAZ VIa lasted ~550 years (Table 2).
PAAZ VIc 1 +c 2 (high NAP-Betula-Pinus; 2046-1819 cm), which biostratigraphically already belongs to the Weichselian, is characterized by the continuing increase of NAP, reaching ~40% at the top of the investigated section (Fig. 5).Betula is strongly present (~9-34%) and Picea reaches ~1-5%.Cold-tolerant Salix and Juniperus as well as Calluna-dominated heathlands (maximum ~28%) spread while Pinus decreases and fluctuates between ~30 and ~50%.Poaceae reach ~5-10% and Artemisia and Amaranthaceae occur regularly.High NAP and Calluna percentages as well as a strong dominance of Calluna over Poaceae are characteristic of the early Weichselian Herning Stadial in northern Germany (Caspers & Freund 2001).Local moisture is indicated by the strong presence of Cyperaceae, fern spores (Fig. S1) and Sphagnum (maximum ~23%; the boundary between PAAZ VIc 1 and VIc 2 at 1886 cm is defined by the Sphagnum increase to >10%).Low percentages of thermophilous elements (Carpinus, Corylus, Alnus) and sporadic findings of Ilex and Hedera likely indicate partial reworking of older sediments, which is confirmed by the presence of dinoflagellate cysts and Neogene pollen like Liquidambar.

Discussion
Comparison of BIS-2000 and KS 186/70 and implications for the duration of the Eemian in northern Germany Detailed microfacies analysis revealed that the sub-mmscale light-dark laminae couplets between 3047 and 2922 cm in BIS-2000 represent carbonaceous-biogenic varves with individual sublayers reflecting seasonspecific deposition.Sediment microfacies information thereby confirms previous pollen analyses on individual sublayers in the KS 186/70 sediments.These revealed a dominance of pollen from taxa flourishing in early/middle spring (e.g.Corylus, Ulmus, Alnus, Betula, Fraxinus, Pinus, Quercus) in the dark sublayers, indicating formation during spring, but most probably including also deposition during the previous autumn and winter (M€ uller 1974).In contrast, pollen from taxa flourishing in late spring and summer (e.g.Picea, Tilia, Poaceae, Artemisia, Calluna) prevailed in the light sublayers, signifying them as predominantly formed during summer (M€ uller 1974).Given this combined sedimentological and palynological evidence for an annual origin of the sub-mm-scale light-dark laminae couplets, the number of varves/years in the laminated basal part of BIS-2000 and KS 186/70 should be similar.In fact, varve counting and varve counting-based sedimentation rate interpolation yielded a duration of 2098 years for the PAAZ IIa-IIIb section in BIS-2000 (Table 2).This confirms within an uncertainty of <5% the results from KS 186/70, where 2195 years were counted in the same section (Table 2; M€ uller 1974).Differences in the duration of individual PAAZ between the two sediment cores (Table 2) might be related to the lower resolution of the palynological analyses in BIS-2000 (n = 19) in this interval compared to KS 186/70 (n = 38), affecting the exact positioning of the PAAZ boundaries.
For the faintly/non-varved parts of BIS-2000, i.e.PAAZ I and PAAZ IIIc-VIb, the duration of individual PAAZ was determined by using exactly the same sedimentation rates as those in KS 186/70 (see section 'Sediment microfacies, varve counting and sedimentation rate estimates').By using this approach, we findin addition to the close similarity between the two sediment cores for the varve-counted PAAZ IIa-IIIb sectionalso a good agreement for the thickness/duration of PAAZ 1, which covers 100 years (7 cm) in KS 186/70 and 157 years (11 cm) in BIS-2000, and the PAAZ Va-VIb section, which covers 4000 years (255 cm) in KS 186/70 and ~3800 years (239 cm) in BIS-2000 (in total <4% difference for the duration of these PAAZ between the two sediment cores; Table 2).
In contrast to the good agreement between BIS-2000 and KS 186/70 regarding the thickness/duration of the PAAZ I-IIIb and PAAZ Va-VIb sections documented above (Fig. 7, Table 2), major differences between the two sediment cores are evident in the middle part of the Eemian sediment succession, i.e. in PAAZ IIIc-IVb.While this section covers 303 cm in KS 186/70 (M€ uller 1974), it includes 578 cm in BIS-2000 (Fig. 7, Table 2).Applying the average sedimentation rates deduced from KS 186/70 to BIS-2000, we find that PAAZ IIIc and IV respectively lasted ~1800 and ~7130 years in BIS-2000 compared to 700 and 4000 years in KS 186/70 (Table 2).As there is no reasonable explanation why the amount of sediment deposited (i.e. the sedimentation rate) at both sites should be fairly identical in the basal and upper part of the sediment succession but significantly different in the middle part, the clearly reduced thickness and consequently also the shorter duration of the PAAZ IIIc-IVb section in KS 186/70 is considered to be very likely related to one or more gaps in the KS 186/70 sediment record in this section.This interpretation is supported by distinct differences between the BIS-2000 and KS 186/70 pollen records.For example, the transition between PAAZ IIIc and IVa in BIS-2000 is characterized by a very gradual increase of Carpinus as well as steady declines of Pinus and Corylus, whereas these changes are very abrupt in KS 186/70.In particular, Carpinus increases from <5 to ~35% across ~135 cm in BIS-2000, whereas the same increase in KS 186/70 occurs rather abruptly within only 20 cm (Figs 5,6).Furthermore, the decrease of Tilia in BIS-2000 to <1% occurs rather gradually at the end of PAAZ IVa, whereas this appears to occur earlier in KS 186/70 (Figs 5,6).These notable differences in individual taxa occurrence between the two sediment cores are very likely indicative of sedimentologically unrecognized gaps in KS 186/70, similar to evidence from other sediment records (e.g.Leroy et al. 2000;Kupryjanowicz et al. 2023).The occurrence of larger gaps in KS 186/70, which would entail a previous underestimation of the true duration of the Eemian, might be explained by its slightly more marginal location in the former lake basin.The partial loss of larger sections of the KS 186/70 sediment succession could have been caused by local slumping at the steep slopes of the high-relief subglacial channel.This hypothesis is supported by the spatially highly variable thickness of the Bispingen palaeolake sediment succession (Benda & Brandes 1974).Another, and maybe even more reasonable, explanation for gaps in the KS 186/70 sediment succession could be unrecognized loss or non-recovery of sediment during the coring campaign in 1970 as only a single sediment core was obtained back then and not overlapping parallel sediment cores as used for the compilation of BIS-2000.
Taken together, the results of microscopic varve counting for PAAZ IIa-IIIb and the sedimentation ratebased duration estimates for PAAZ I and PAAZ IIIc-VIb determine the length of the entire pollen-defined Eemian in BIS-2000 as ~15 000 years (Table 2) compared to ~11 000 years initially reported for KS 186/70 (see M€ uller 1974).The new duration estimate of ~15 000 years derived from BIS-2000 must, however, be considered a minimum estimate that could still be biased by the uncertainty of the assumed sedimentation rates in the PAAZ IIIc-VIb section.For example, a reduction of the sedimentation rates in this interval by 15-20% compared to the original estimates derived from KS 186/70 would be sufficient to yield a duration of ~18 000 years for the pollen-defined Eemian at Bispingen, which would be very similar to the duration of the Last Interglacial in southern European records (e.g.Brauer et al. 2007;Allen & Huntley 2009;Tzedakis et al. 2018).

The revised duration of the Eemian in northern Germany in comparison with other European Last Interglacial proxy records
To date, the most robust estimate for the duration of the pollen-defined Last Interglacial in Europe originates from the varved lake sediment record of Lago Grande di Monticchio, southern Italy.Here, the Last Interglacial lasted 17 700AE200 years between 127.2AE1.6 and 109.5AE1.4ka BP (Brauer et al. 2007;Allen & Huntley 2009; Fig. 8) with a later chronological revision shifting the Last Interglacial by a few hundred years towards older ages, e.g. its end to 110.4AE0.7 ka BP (Martin-Puertas et al. 2014, 2019).Keeping in mind that vegetation changes at the onset of the Last Interglacial were not necessarily synchronous with climatic shifts observed in d 18 O records, not only the duration but also the absolute dating of the Last Interglacial inferred from the Lago Grande di Monticchio pollen record are confirmed within the dating uncertainty by the recently revised U/Th-dated speleothem d 18 O record from Corchia Cave, northern Italy, where full interglacial climate conditions lasted between 129.0AE0.5 and 111.0AE0.9ka BP, i.e. for ~18 000 years (Tzedakis et al. 2018;Fig. 8).This age/duration estimate was also transferred to marine sediment core MD01-2444 from the Iberian margin offshore Portugal (Tzedakis et al. 2018;Fig. 8), where the duration of the Last Interglacial was previously determined to be ~16 400 years from combined pollen and d 18 O analyses on marine sediment core MD95-2042, placing the start of reforestation at the onset of the Last Interglacial at 126.1 ka BP and the end of the interglacial vegetation succession at 109.7 ka BP (S anchez Goñi et al. 1999;Shackleton et al. 2002Shackleton et al. , 2003)).Given the revised chronology for the Iberian margin proxy records and the evidence from Lago Grande di Monticchio and Corchia Cave, Last Interglacial palaeoclimatic/-environmental changes in southern Europe occurred therefore apparently contemporaneously.
The spread of temperate vegetation in Europe at the onset of the Last Interglacial isdespite regional differences in vegetation compositioncommonly assumed to have been quasi-synchronous across a wide latitudinal/longitudinal range (Tzedakis 2003), similar to that at the onset of the Holocene (e.g.Brauer et al. 1999;Lohne et al. 2013;Engels et al. 2022).Consequently, the apparent ~2500-3000-year difference between our new minimum duration estimate of ~15 000 years for the Eemian at Bispingen and the latest duration estimates of ~18 000 years from southern Europe (Brauer et al. 2007;Allen & Huntley 2009;Tzedakis et al. 2018) could reflect a regionally different vegetation development at the demise of the Last Interglacial.The existence of such a vegetation gradient between southern and northern-central Europe at the end of the Last Interglacial could be related to the relative proximity of Bispingen to the successively growing Scandinavian ice sheet (see Ehlers & Gibbard 2004).This probably exerted a stronger impact on the vegetation in northern Germany during the final stage of the Last Interglacial, e.g. through stronger winter cooling and reduced precipitation (K€ uhl et al. 2007), terminating the phase of temperate vegetation earlier than at locations further south.However, based on our new floating duration estimate, this phase of contrasting climate conditions between southern and northerncentral Europe was significantly shorter than the multimillennia-long trans-European vegetation gradient that was previously suggested from the comparison between pollen records from France and the initial Last Interglacial duration estimate of ~11 000 years from Bispingen (Kukla et al. 2002b;Tzedakis 2003).Nevertheless, we actually cannot exclude a similar length of the Last Interglacial and therefore synchronous vegetation changes throughout Europe given the uncertainty of the sedimentation rate estimates in the PAAZ IIIc-VIb section of BIS-2000 (see above).Either way, the relatively long dominance of temperate deciduous tree vegetation, particularly Carpinus, until ~11 000 years after the onset of the Eemian (i.e. until the end of PAAZ IV) appears notable with respect to the continuous decline in summer insolation since the middle Eemian, which should have been accompanied by decreasing growing season warmth and thus an earlier disappearance of temperate vegetation in northern Europe compared to locations further south (Tzedakis 2003).However, the persistence of Carpinus into the terminal phase of the Eemian is also observed at other sites in northern-central Europe (e.g. Erd 1973;Hahne et al. 1994;Litt 1994;Pidek et al. 2021) as well as at high-altitude sites further south (de Beaulieu & Reille 1992), indicating that in addition to summer insolation/growing season temperature also other factors might have influenced the regional vegetation.
A robust absolute age determination for the Bispingen palaeolake sediment succession is, however, still lacking because absolute dating methods for (palaeo)lake sediments from the Last Interglacial still lack the necessary precision.Although a quasi-synchronous onset (within the dating uncertainty of the individual proxy records) of the pollen-defined Last Interglacial at ~127-129 ka BP is indicated by proxy records from southern Europe (e.g.Brauer et al. 2007;Allen & Huntley 2009;Tzedakis et al. 2018) (Fig. 8), caution should be exercised in directly transferring this date to the Bispingen pollen record as Last Interglacial vegetation successions from different latitudes and elevations are not necessarily directly comparable (see Sirocko et al. 2005).Furthermore, a simple alignment of the Bispingen pollen record with marine sediment and speleothem d 18 O records is also not recommendable as the onset of the Last Interglacial/MIS 5e in marine sediment and speleothem d 18 O records consistently predates the spread of temperate vegetation by up to several millennia (e.g.Shackleton et al. 2003;Luetscher et al. 2021;Honiat et al. 2022).Hence, a robust synchronization of the Bispingen pollen record with other Last Interglacial palaeoclimate archives from Europe remains subject to the identification of isochronous time markers like, for example, cryptotephras.

Conclusions
A continuous record of lacustrine deposition from the onset of the Last Interglacial (Eemian) into the early Last Glacial (Weichselian) is preserved in the partly varved palaeolake sediment succession of Bispingen, northern Germany, representing a unique archive of environmental changes in Europe during one of the warmest interglacials of the last ~800 000 years.Sedimentological and palynological features of the new composite sediment core BIS-2000 from this classic Last Interglacial site are largely consistent with data from the previously investigated sediment core KS 186/70 (M€ uller 1974).In particular, varve counting and sedimentation rate-based duration estimates reveal a very good agreement for the thickness and duration of individual pollen zones in the lower and upper parts of the two sediment cores.However, the distinctly longer middle part of the Eemian sediment succession (PAAZ IIIc-IVb) in BIS-2000 suggests previously unrecognized gaps of in total ~4000 years duration in KS 186/70.Consequently, the minimum duration of the pollendefined Last Interglacial in northern Germany is revised to ~15 000 years, which is still ~1500-3000 years shorter than in southern Europe, but substantially longer than previously reported.This goes a long way in resolving previous suggestions of a major asynchrony in vegetation changes across Europe at the demise of the Last Interglacial that were based on the previous estimate of an ~11 000-year-long Eemian at Bispingen.

Fig. 1 .
Fig. 1.Topographic and simplified geological map of the surroundings of the Bispingen palaeolake sediment succession in northern Germany (basic topographic map DGK5, Sheet 2925/12 -Bispingen (zone 3 cartesian Gauss-Kr€ uger coordinates); Lower Saxony State Office for Geographic Information and Land Development, Hanover, Germany).The insert map shows the locations of Bispingen and other Last Interglacial proxy records discussed in the text and displayed in Fig. 8. Positions of the sediment cores obtained in 2000 and the early 1970s (see Benda & Brandes 1974) are indicated by red and grey dots, respectively.The extent of the diatomite deposit as well as geological data and sediment core locations were obtained from the Lower Saxony Soil Information System (NIBIS) map server (Lower Saxony State Office for Mining, Energy and Geology, Hanover, Germany).

Fig. 2 .
Fig. 2. Sedimentology of the composite sediment core BIS-2000 and results of geochemical analyses.Numbers (3A-3C) next to the stratigraphical column indicate the positions of different varve types displayed in Fig. 3.

Fig. 7 .
Fig. 7. Comparison of the sedimentology and pollen zonation of sediment cores BIS-2000 and KS 186/70 (M€ uller 1974) from the Bispingen palaeolake sediment succession.Note that a direct sedimentological comparison between BIS-2000 and KS 186/70 is only possible to a limited extent due to the not very detailed sedimentological description of KS 186/70.

Fig. 8 .
Fig. 8.Comparison of the BIS-2000 pollen zonation (plotted on depth scale) with different European Last Interglacial palaeoclimate proxy records (plotted on their individual time scales; for the location of the individual records see Fig. 1). A. Planktonic foraminifera d 18 O and pollen data from marine sediment core MD01-2444, Iberian margin (Tzedakis et al. 2018).B. Pollen data from Lago Grande di Monticchio, southern Italy (Brauer et al. 2007; Allen & Huntley 2009).Note that the recent revision of the chronology, which shifted the Last Interglacial towards older ages, now placing its end at 110.4AE0.7 ka BP (Martin-Puertas et al. 2014, 2019), is not considered here.C. Speleothem d 18 O data from Corchia Cave, northern Italy (Tzedakis et al. 2018).D. Pollen data from F€ uramoos, southern Germany (Kern et al. 2022).Note that the chronology of this record is based on tuning to the age model of marine sediment core MD95-2042 from the Iberian margin (S anchez Goñi et al. 1999; Shackleton et al. 2002, 2003), while a new chronological framework for sediment records from the Iberian margin was recently proposed by Tzedakis et al. (2018).