Omnipresent authigenic calcite distorts Arctic radiocarbon chronology

Abstract

As calcareous foraminifera precipitate their shells from the surrounding water mass, they are the basis of most marine radiocarbon chronologies and paleo-proxies. Nevertheless, post-mortem alteration of shells, especially addition of authigenic calcite, impact proxy records. In the Arctic Ocean, authigenic calcite overgrowth on foraminifera has been attributed to hydrocarbon release, with a single report on ¹³C-enriched authigenic calcite, indicating a different carbon source. Here, we use comparative radiocarbon, carbon and oxygen isotope measurements to show that this ¹³C-enriched authigenic calcite impacts a large proportion of Holocene and the majority of last glacial planktonic foraminifera in the Arctic Basin. This authigenic precipitated calcite is ¹⁴C-depleted, so overgrowth results in invariably older ¹⁴C-ages. We show that, in comparison with published data, the true chronology of Arctic basin sediments can deviate by more than 10,000 years in critical parts of the last deglaciation and that stable oxygen and carbon isotopes, as likely all calcite-based proxy-records are affected with potential implications for paleoclimate models.

Introduction

As the Arctic Ocean warms twice as fast as the lower latitude oceans¹, the timing and extent of past climate changes recorded in Arctic sediments is of utmost importance to ground-truth climate model projections^2,3 of future Arctic and world-ocean climate scenarios. Radiocarbon (¹⁴C) dating on calcareous foraminifera^4,5,6,7,8,9 provides the most trustworthy and stringent chronology for such ground-truthing in the last 45 ka.

Radiocarbon forms in the stratosphere, enters the ocean in oxidized form and finally becomes incorporated in shells, where radioactive decay results in decreasing ¹⁴C levels over time, ideally recording the date of death of the animal¹⁰. However, calcareous shells are prone to postmortem diagenetic modifications^{10,11,12,13,14}. Shells first and foremost encompass dissolution at the sediment-water interface, and within the uppermost cm of the sediments, these modifications are largely driven by metabolic CO₂ released from the mineralization of organic matter¹⁵.

Diagenetic alteration of foraminifera shells by authigenic calcite overgrowth (hereafter overgrowth) has often been linked to situations where CO₂ is provided by either the microbial oxidation of methane or the microbial anaerobic degradation of labile organic matter^11,16. Precipitation then involves a subsequent increase in pH and carbonate saturation state to pH >8.0 and Ω >5 by the anaerobic oxidation of methane or sulfate or iron reduction^17,18. Therefore, observations of authigenic carbonate precipitation in the world’s oceans are essentially limited to regions with methane emanation or high labile organic carbon accumulation^18,19,20,21. Notably, carbonate precipitates resulting from both the oxidation of methane and organic matter are expected to feature depleted δ¹³C and ¹⁴C signatures, the latter resulting in an age bias to older ages^11,22,23.

Overgrowth also alters oxygen, boron isotope^19,24,25,26 and trace metal/Ca^16,25 ratios by incorporation of respective exogenous elements. This is extremely critical for the Arctic Ocean because here, >45 ka-chronology, and most assumptions about the extent and direction of paleoenvironmental change essentially rely on these paleoceanographic proxies^8,27,28.

Here, we provide the first comprehensive documentation on the late Pleistocene to Holocene preservation state of the dominant (>90%) calcareous planktonic foraminifera Neogloboquadrina pachyderma, used both for establishing a chronology and palaeoceanographic change in the Arctic Ocean^5,27,28.

We use paired stacked stereo (hereafter stereo) and scanning electron microscopy (SEM) images to illustrate progressive overgrowth with depth. Using a mini carbon dating system (MICADAS)^29,30 it is possible to obtain ¹⁴C ages on samples consisting of only 20–60 N. pachyderma specimens in contrast to approximately 1 mg C equivalent to a minimum of 1000 specimens required for standard AMS analyses. Such small sample sizes allow to carefully inspect and document all shells prior to ¹⁴C measurement, and to carry out multiple comparative measurements on shells with different degrees of overgrowth. We also include comparative studies on stable carbon and oxygen isotopes in order to exemplify the impairment of paleoceanographic proxies by overgrowth.

This study focuses on the consistently foraminiferal-rich core PS72/413-3 supplemented by analyses on PS2185-6 to investigate the impact of overgrowth on the chronology of a core previously used for Arctic-wide correlations⁴ (Fig. 1 and Supplementary Tables 1 and 2). Qualitative inspections on surface sediments document the widespread phenomenon.

Fig. 1: Map with the investigated core sites (see Supplementary Table 1 for the core locations).

Studied sediment cores (white), sediment records whose chronologies are probably affected by overgrowth (red). Surface sediments with overgrowth on foraminifer specimens (black). Bathymetric map created by Ocean Data View version 5.6.2⁹⁸.

Results

Morphological description of overgrowth and authigenic precipitates

Neogloboquadrina pachyderma precipitates a porous, translucent, multichambered, Mg-poor calcite shell³¹. The primary shell is multilamellar, and with each new chamber added, an additional layer is deposited on the older chambers^32,33. On the outside conical protuberances protrude above the smooth porous microcrystalline shell, the inner surface of the wall is smooth (Fig. 2a–c)^34,35. Primary shells are prone to dissolution and therefore are rare in Arctic Ocean sediments^36,37,38,39. Here, most specimens have precipitated an outer secondary calcite crust on the primary shell (Fig. 2d–g)^32,35,40,41. Two principal types of secondary calcite crust are observed on the shell surface. The reticulate type is characterized by polygonal ridges that surround a variable large smooth area around the pore pits⁴² (Fig. 2e, g). In the crystalline type, large euhedral crystals with growth facets on the crystal faces form around the pores⁴² (Fig. 2e, f). The reticulate and crystalline form, and intergrades between these two, are often found on the same shell.

Fig. 2: Axio Zoom and SEM images of overgrowth.

a–c Translucent N. pachyderma with primary shell and little overgrowth in the penultimate chamber and sediment-infilling in the last chamber. d–g Translucent N. pachyderma specimen with secondary calcite crust and minor ≤1–2-µm-sized authigenic crystallites in depressions on the crystalline (f) and reticulate crust (g). The overgrowth often clogs the pores and contrasts with the smooth reticulate surface texture and the euhedral biologically precipitated crystals. h Cassidulina neoteretis (benthic foraminifera) patches of ≤1–2-µm-sized authigenic crystallites on the otherwise smooth shell. i–k N. pachyderma encrusted by porous overgrowth. k At magnifications >4500 the porous structure of the overgrowth becomes apparent. l Cross-section through the last chamber of a different specimen shows tetrahedral overgrowth on the outer and inner shell wall (see the deviation from the usually smooth inner wall), but also authigenic calcite penetrating into the originally porous shell wall⁹⁹. m Closed tetrahedral overgrowth. a, d, i stacked Axio Zoom, b-d, g–h, j–m SEM images. a–k, m PS413-3, l (PS2185-6). Core depth (cm) a–c 3.5, d–g 0.5, h 3.5, i–k 22.5, l, m 10.5. Arrows: green = last chamber, blue = penultimate chamber, red = authigenic crystals, purple = pore. Yellow line highlights the rough shell interior.

Whether primary or secondary shell type, any unaltered biogenically precipitated N. pachyderma shell is translucent, and AMS, isotope and trace metal data should only be obtained from such shells^12,13,25. However, such ideal shells are rare in Arctic sediment cores where samples are usually dominated by white¹² or even brownish discolored shells (see https://www.pangaea.de/?q=doi:10.1594/PANGAEA.938246, plate 2 in Eynaud et al.³⁸). Hereby, not only in shells with overgrowth but also in those affected by dissolution the radiocarbon age, and the carbon and oxygen isotope value should differ^{10,12,13,14,25,43,44,45}.

Consequently, in this paper, only a translucent N. pachyderma shell with surface textures as described is regarded as unaltered by diagenetic processes, as pristine specimen. In contrast, an increasingly whitish discoloration of the shell reflects increasing diagenetic change (dissolution or initial overgrowth) (Fig. 2i), and a color alteration to the light yellow-brown spectrum indicates major overgrowth (Fig. 3a, b)⁴⁶.

Fig. 3: Surface texture of heavily encrusted N. pachyderma tests and sedimentary authigenic crystals.

a, b Brownish discolored overgrowth of tetrahedral calcite crystals. c–g Authigenic envelopes: c–e Microcrystalline envelope, d close-up of (c), e N. pachyderma with porous overgrowth in envelope. f, g Tetrahedral envelope around juvenile planktonic foraminifera (f), g close-up of (f). h–o Selection of authigenic crystals. h Authigenic spheres of needle-like calcite crystals, likely pseudomorphoses after high Mg-calcite. i Wheat-sheaf-like bundles of needle-like crystals, likely pseudomorphoses after high Mg-calcite. j Crystal crust. k, l Aggregate of small crystals. l Close-up image of (k). m Authigenic crystals resembling glendonite. n, o Cluster of crystals. Needle bundle is likely a pseudomorphosis after aragonite, the dumbbell-like clusters are pseudomorphoses after high Mg-calcite. b, j Stacked Axio Zoom, a, c–h, k–o SEM images. d PS2185-6, a–c, e–o PS72/413-3. Core depth (cm) a, b 22.5, c–g 17.5, h 20.5, i–o 22.5.

Initial overgrowth is characterized by minute discrete authigenic crystallites (<0.5–1 µm) precipitated at various places on the shell, best seen on the otherwise perfectly smooth shell of the benthic foraminifera Cassidulina neoteretis (Fig. 2c, f–h). Thereafter, the number of crystallites increases first laterally to form closed sheets around the shells which do not alter the original shape of the foraminifera, then in thickness (Fig. 2h–m). Overgrowth usually shows a porous appearance, less commonly a sheet of tetrahedral calcite crystals (Fig. 2k–m) or larger (>10 µm) crystals sticking out from the shell.

Aksu and Vilks³⁷ reported authigenic Mg-calcites on foraminiferal shells from surface sediments of the CESAR, LOREX and FRAM sites. In this study, shells from randomly selected surface sediments of the central Arctic Ocean reveal overgrowth and crystals at all sites (Fig. 1).

In contrast to overgrowth, co-occurring authigenic calcite crystals (hereafter crystals) display a large morphological variability from xenomorphic or microcrystalline envelopes surrounding clusters of planktonic foraminiferal shells to microcrystalline spheres, dumbbell-like forms, wheat-sheaf-like bundles, rhombohedra, needle-like forms or even crusts (Fig. 3c–o).

Optical and geochemical characterization of overgrowth

Energy-dispersive X-ray spectroscopy (EDS) analyses identify the biogenic shell, overgrowth and sedimentary crystals as low Mg-calcite (see https://www.pangaea.de/?q=doi:10.1594/PANGAEA.938246). This is in contrast to the findings by Aksu and Vilks³⁷, who reported authigenic high Mg-calcites on large foraminiferal shells from central Arctic surface sediments. But individual measurements were not reported and their maximum value of 10 mol% Mg-calcite (corresponding to about 2.5 weight % Mg) corresponds to the minimum value for this mineral. Occasional Mg concentrations >2.5 weight % observed in some of our specimens with porous overgrowth are likely attributed to clay impurities, not to Mg-calcite, due to simultaneously elevated Al and Si contents, and we assume that such a contamination also affected some of the reported specimens of Aksu and Vilks³⁷.

In the Arctic Basin, sediments are derived either from sea-ice, icebergs or collapsing ice-sheets⁴⁷. Whereas sea-ice mainly incorporates fine-grained shallow-water sediments, barren or low in planktonic foraminiferal shells⁴⁸, icebergs and ice shelves can erode shelf and upper slope sediments and release incorporated fossil shells during their passage⁴⁹. Consistent with the important role of Arctic shelves on iron supply⁵⁰, an increasing brownish discoloration is observed in, e.g., Pliocene benthic foraminifera fossilized there and transported by glacial iceberg/ice shelves to be released in the Arctic Basin (e.g., Figure 2 of Geibert et al.⁴⁹).

High abundances of such brownish discolored Pliocene benthic species indicate that also the planktonic assemblage and proxies connected to them are unreliable. Consequently, brownish discolored planktonic foraminifera were excluded, as they could either be allochthonous or encrusted by obvious overgrowth²². However, in no analyzed sample, the benthic foraminifera community indicated a strong allochthonous component.

Sediment or authigenic coatings of various kinds may also be deposited on the shell’s interior (termed “infillings” here), and were excluded to the best of our knowledge.

In this study, we did comparative measurements on the best-preserved translucent to white and similar-colored foraminifera with and without overgrowth in each sample.

Preservation statuses of measured shells (Fig. 4) are categorized as follows: Pristine, translucent specimen visually free of overgrowth = Preservation Index (PI)_1 category; specimens visually free of overgrowth (PI_2); and those with overgrowth (PI_3). PI_1 specimens are generally extremely rare and at depths >5 cm contrast strongly with the predominant foraminifera shell assemblage in both cores. Due to distinctively lower abundances of planktonic foraminifera, PI_1 radiocarbon age could not be studied in core PS2185-6. Our high-resolution measurements, therefore, focused on the predominant translucent-white to white shells classified as PI_2 and PI_3 regarded as autochthonous shell assemblage. If possible, only translucent-white shells were used for measurements; however, in both cores below 10 cm sediment depth, progressively more white shells had to be picked to meet the weight requirements for the measurements. In all samples, the PI_2 and PI_3 specimens measured matched in terms of shell coloration and we consider them to be deposited at the same time.

Fig. 4: Images on the classification of the preservation status of Neogloboquadrina pachyderma.

a Pristine translucent PI_1 specimen. b Close-up view of primary shell showing smooth wall with low protuberances. c, d Specimens with no visible overgrowth PI_2. c Translucent-white specimen, d white specimen. e, f Specimen with visible overgrowth PI_3, both with distinct crystals, on translucent-white (e), and white shell (f). Specimen from core PS72/413-3. Core depth: a, b 6.5 cm; c, d 4.5 cm; e, f 11.5 cm. Stacked Axio Zoom images. cr= crystal.

The impact of overgrowth on isotope (Fig. 5 and Supplementary Tables 3 and 4) and radiocarbon measurements (Fig. 6 and Supplementary Tables 5 and 6)

Carbon isotope analyses in both cores reveal that PI_3 specimens are enriched in ¹³C compared to PI_2 and the majority of crystals have δ¹³C values of >2.5–5.6‰ (Fig. 5a, b, e, f and Supplementary Tables 3 and 4). Thus, the maximum and mean offset of PI_3 vs PI_2 specimens are 0.75 and 0.31‰ and 0.88 and 0.44‰ for cores PS72/413-3 and PS2185-6, respectively. However, as the mean offset of crystals vs PI_2 species is 3.89 and 3.99‰ for cores PS72/413-3 and PS2185-6, respectively, the δ¹³C value of each PI_3 specimen essentially depends on the extent of overgrowth.

Fig. 5: Carbon and oxygen isotope values in the investigated cores.

a–d PS72/413-3. a, b Carbon isotope values of N. pachyderma (a), and incl. crystals, envelopes and detrital carbonate (b). c, d Oxygen isotope values of N. pachyderma (c), and incl. crystals, envelope and detrital carbonate (d). e–h PS2185-6/3. e, f Carbon isotope values of N. pachyderma this study and Nørgaard-Pedersen⁴ (e), and incl. crystals (f). g, h Oxygen isotope values of N. pachyderma this study and Nørgaard-Pedersen⁴ (g), and incl. crystals (h).

Fig. 6: Extent of overgrowth and radiocarbon ages of the respective preservation index categories.

a–c PS72/413-3. a Mean relative estimates of overgrowth (no, minor, moderate, major). b, c ¹⁴C age vs depth of PI_1-3 shells (b), and incl. crystals (c). d–f PS2185-6/3. d Mean relative estimates of overgrowth (no, minor, moderate, major). e, f ¹⁴C age vs depth of PI_1-3 shells (e), and incl. crystals (f). Please note that PI-measurements (b, c and e, f) are based on stereomicroscope investigations, mean relative estimates (a and d) additionally on SEM analysis. However, major overgrowth was usually noticed on discolored brownish specimens, that were not measured, thus are not included in PI_3 results. Error bars represent ±2σ analytical uncertainty.

Different to carbon, in core PS72/413-3 the direction of oxygen isotope PI_3 vs PI_2-offset is not constant but maximum values of −1.63 and +1.03, respectively, are observed. In core PS2185-6 the PI_3 vs PI_2 -offset is usually positive with maximum and mean values of 1.83 and 0.66‰, respectively.

The extreme δ¹⁸O differences in small vs large crystals within one sample in core PS72/413-3 with mean offsets of 8.05‰ and extreme low values of up to −13.5‰ for large crystals are striking. In core PS2185-6, only large crystals were measured with the lowest values of −0.45‰ at 20.5 cm sediment depth. In PS2185-6, only the carbon and oxygen isotopes of PI_3 specimens resemble published values⁴ whereas those of PI_2 specimens deviate. Carbon and oxygen analyses on different crystals in core PS72/413-3 reveal large scatter even within a single sample; thus, the N. pachyderma offset due to overgrowth cannot be predicted.

In core PS72/413-3, a few single and replicate radiocarbon measurements on PI_1 specimens from Holocene sediments (Fig. 5b and Supplementary Table 5) reveal radiocarbon ages (hereafter ages) comparable to those of PI_2 specimens. PI_1-plateaus at 15.5–19.5 and 20.5–23.5 cm are attributed to downcore displacement by bioturbation (for discussion see methods).

The ages of PI_2 and PI_3 specimens increase with core depth, hereby, PI_2 shells are invariably younger than PI_3 shells (Fig. 5a, d). The age offset ranges from a few hundred to thousands of years.

Due to a slight brownish discoloration²² different from the other PI_2 shells, the ages between 18.5 and 23.5 cm are considered unreliable (see https://www.pangaea.de/?q=doi:10.1594/PANGAEA.938246).

At site PS2185, the ages of PI_3 shells resemble published conventional AMS ages on N. pachyderma from the box corer⁴, whereas PI_2 specimens below 10 cm, less affected by overgrowth, show distinctively younger ages (Fig. 5d and Supplementary Table 6). Below 20 cm, only PI_3 shells are found, and between 30 and 118 cm, a barren interval prevents any measurements. The structure of the radiocarbon chronology based on PI_3 specimens resemble those in a number of cores in the Arctic Ocean (Figs. 1 and 7).

Fig. 7: Comparative radiocarbon ages and their implication for the assumption of missing late glacial sediments.

a PI_2 and PI_3 specimens in the analyzed cores vs published data. b Exemplified radiocarbon records from the Mendeleev, Lomonosov Ridge and Morris Yesup Rise (see also Supplementary Fig. 1). See the almost linear age development downcore of PI_2 specimens and the distinct age jump of PI_3 specimen and published records in glacial sediments.

The radiocarbon measurements of crystals of various morphologies reveal ages between >20 and 50 ¹⁴C ka, with the oldest ages obtained from large, the youngest from aggregates of small crystals (Fig. 5c, f). The ages of the small crystal aggregates are probably distorted by adhering sediments on the rough surface.

While the age of the idiomorphic crystals increases with depth, that of envelopes (without foraminifera) shows no relation to depth. Duplicate measurements on a broken yet single idiomorphic large crystal from the sediment surface at site PS72/404 (Fig. 1) reveal ages of 30.32 and 12.19 ka. This indicates that authigenic calcite forms discontinuously by repeated rather than a single precipitation event.

The occurrence of PI_1 specimen in grayish sediments (15–25 cm)

In core PS72/413-3, the N. pachyderma shells between 15 and 25 cm core depth are almost exclusively whitish to brownish with major overgrowth (Fig. 2b, sample documentation in https://www.pangaea.de/?q=doi:10.1594/PANGAEA.938246). Selecting PI_1 specimens from this section required to screen more than a hundred thousand shells of each sample. We assume that they originate from overlying brownish sediments, usually assigned to interstadial to interglacial sediments⁵¹, and were downcore displaced by burrowing organisms into the glacial grayish sediments. This interpretation is supported by: (1) the extreme rarity of this preservation state, (2) trace fossils in the form of brownish mottles in the grayish sediments, and (3) PI_1-radiocarbon age plateaus in these sediments.

Discussion

The process of authigenic carbonate precipitation in Arctic sediments

Previous studies have reported authigenic overgrowth on foraminifera shells primarily at seeps where methane is subsequently oxidized under anaerobic condition¹⁶. Methane as a classic source of old ¹⁴C in porewater and authigenic calcite is a well-known feature in Arctic shelves, and often linked to thawing permafrost^52,53. For the Arctic basin, hydrothermal seeps are numerous along the Gakkel Ridge⁵⁴, and can be suspected due to the many plutonic intrusions in the crustal area of the Mendeleev Ridge⁵⁵. However, while the ¹⁴C-depleted nature of PI_3 shells in principle supports methane as a carbon source, the very high δ¹³C values recorded from our authigenic crystals are in sharp contrast to the extremely depleted δ¹³C values recorded from DIC and from authigenic calcite precipitates in sediments close to the sea floor at methane emanation sites^{17,20,21,22,52}. Furthermore, sites PS2185 and PS72/413 are at a distance from the Gakkel Ridge and show no faunal evidence of hydrothermal activity at sites. Moreover, basin-wide no δ¹³C or δ¹⁸O anomalies indicative of hydrocarbon sources can be detected at the bottom water/sediment interface⁵⁶ and sediment cores examined reveal oxygen-rich pore waters in the uppermost 400 cm⁵⁷.

Degradation of labile organic matter would likewise result in depleted δ¹³C signatures, while radiocarbon age offset would likely be less extreme than what we observe in our records. The central Arctic Ocean is an oligotrophic area with an extremely low mean labile organic carbon export production^58,59,60. Labile organic components of particulate organic carbon are only present in surface and near-surface sediments where they are immediately digested by benthos, whereas the majority is inert⁶¹. In porewater, dissolved carbon decreases with increasing sediment depth and labile components can only be detected in the uppermost 5 cm⁶⁰. In addition, the low central Arctic sedimentation rates do allow for extended oxygen exposure of near-surface sediment and bottom water where most labile organic carbon is immediately respired⁵⁹. This is reflected in low sedimentary organic carbon contents⁶² as in core PS72/413-3 (Fig. 8 and Supplementary Table 2). Porewater was not analyzed in the cores of this study but in core PS72/410-1 (Fig. 1) close to PS72/413-3, and PS72/248-6 on the Lomonosov Ridge porewater analyses reveal well-oxygenated pore waters with low TOC in the upper 400 cm ^{(PS72/248-6)57}.

Fig. 8: Chronostratigraphy and sediment characteristics cores PS72/413-3 and PS2185-3⁶⁶.

a Published stratigraphy of core PS2185-3. b Chronostratigraphic interpretation based on ¹⁴C ages of PI_2, non-calibrated and without reservoir correction (see Fig. 6b, e). c, d Relative abundance of carbonates quantified from X-ray diffraction analyses of bulk sediments¹⁰⁰, and planktonic foraminifera numbers for cores c 2185-3⁶⁶ and d PS72/413-3⁹⁴. e Total organic carbon PS72/413-3.

As Arctic Ocean deep-water has a modern to glacial δ¹³C_DIC of ~1–1.5‰ Vienna Pee Dee Belemnite (VPDB)^63,64, lower than the observed δ¹³C signature of PI_3 specimens, and any inorganic calcite precipitated from this water mass should reflect this in a 1:1 ratio, we rule out direct precipitation from bottom water. Calcite must therefore precipitate from porewater, possibly even close to the sediment surface, as indicated by filigree star-shaped crystals. Moreover, the huge variability in δ¹⁸O values of especially large authigenic crystals, with lowest values equal to δ¹⁸O values of Arctic glaciers or atmospheric precipitation⁶⁵, suggest a periodically low saline crystallization fluid for the large, thus slow-growing crystals. Carbon and oxygen analyses of epifaunal (Cibicidoides wuellerstorfi) and shallow-infaunal (Cassidulina neoteretis) dwelling in bottom (C. wuellerstorfi) respectively porewater (C. neoteretis), show especially for PI_3 C. neoteretis a strong positive δ¹³C but negligible δ¹⁸O-offset (Supplementary Table 3). As these foraminifers have precipitated their shells in these habitats, a continuous freshwater influence at the sediment-water interface or in the respective porewater can be ruled out.

In core PS72/413-3, the occurrence of thick overgrowth (major overgrowth Fig. 6a) and sedimentary authigenic calcite crusts peak in the pink-white ice-rafted sediments below 27 cm (Fig. 9). Similarly, colored ice-rafted calcite clasts distributed throughout both cores fall into the carbon and oxygen isotope range of crystals from the same sediment depth (Fig. 5). As the high radiocarbon age and carbon isotope values of crystals indicate an old and ¹³C-enriched carbon source, we assume that ice-rafted calcite flour serves as a reservoir for authigenic calcite. Nørgaard-Pedersen reported authigenic carbonates in the coarse-size fraction of foraminifera-poor, clay to silt-sized ice-rafted Paleozoic Mg-calcite flour-rich (thereafter calcite flour) glacial sediments from Northern Greenland-Northern Canada, in sediment cores from the Gakkel Ridge (PS2166) and Morris Jesup Rise (PS2200)⁶⁶ (Fig. 1). He also considers increased calcite flour supply as causal for the formation of large authigenic calcites. Distributed by bioturbation throughout the sediments, or dissolved in porewater, calcite flour may support a persisting, pore/bottom water buffer capacity for more or less extended times of CO₂ storage^67,68,69. This process is comparable to the hard water effect, which results from the dissolution of limestone grounds in aquatic systems¹⁰. During times with corrosive bottom/porewater or by, in this oligotrophic environment, only facultative, organic matter degradation, calcite flour and variable amounts of biogenic carbonates may dissolve. At a later date, with well-ventilated bottom and porewater, high sedimentary clay contents^70,71 mediated by microbial communities^72,73, sulfate or iron⁶⁸ reduction eventually lead to increased porewater alkalinity and pH resulting in a precipitation of calcite⁷¹.

Fig. 9: Sediment characteristics core PS72/413-3.

a Linescan image with published chronostratigraphic interpretation⁵¹. b Lightness L* and green-red color space a*. c Relative abundance of carbonates quantified from X-ray diffraction analyses of bulk sediments⁹⁴. d Concentrations of planktonic and benthic foraminiferal and pteropod shells. MIS placement this study.

An alternative mechanism of authigenic carbonate formation has been recently described by Rogov et al.⁷⁴. Under favorable melting and freezing conditions combined with porewater chemistry changes, carbonates may grow by cryometamorphism in free pore spaces of permafrost on the shallow Siberian shelves. Such geochemical processes could also lead to the precipitation of calcites in the deep Arctic Ocean.

Especially, the oxygen and carbon isotope ratios of the crystals are extremely variable, both downcore and even within one crystal. Thus, besides calcite flour, regarded as a major source of authigenic calcite precipitation, variable other sources are needed to explain especially the sometimes extremely depleted δ¹⁸O values of crystals. The oldest radiocarbon age we could retrieve for PI_2 specimens was from core PS2185-6 at 25 cm; below this depth, sediments are barren in microfossils and coeval with a period when the Arctic Ocean was flooded with freshwater⁴⁹. In core PS72/410-1 (Fig. 1), which is adjacent to PS72/413-3, sedimentary authigenic neodymium isotopes also indicate increased freshwater inputs at the same time⁷⁵. At present neither models nor porewater data aid to understand this persisting precipitation process that, according to the existing data, can be explained neither by the degradation of labile organic material nor by increased methane concentrations. However, in the central Arctic Ocean, sedimentation rates are extremely low, and during glacial times periods with high sediment flux likely alternated with times of diminished sedimentation. Any porewater chemistry thus likely altered distinctly over time. Combined porewater and lithological investigations, potentially supported by experimental set-ups, will be needed to unravel the formation of the ubiquitous authigenic calcites in the central Arctic Ocean.

Can foraminiferal shells be cleaned from overgrowth?

Foraminifera shells are not solid calcite but they contain organic linings deposited either as sheets for the nucleation of calcite precipitation and shell protection³³, or surrounding single or groups of interlocking calcite mesocrystals⁷⁶ (Fig. 10).

Fig. 10: Diagenetic alteration of N. pachyderma shells.

a Pristine shell with secondary calcite crust and smooth shell interior. b, c Shell affected by initial dissolution, showing widened pores and especially at the border of previously existing organic sheets large cavities. d, e Shell with authigenic calcite on the exterior and interior (e) surface and penetrating the shell, hereby destroying the original shell structure (d), see also the rough interior shell surface (yellow line). PS2185-6. Core depth (cm): a–c 0, d, e 10 cm. Lines indicate the original position of the inner shell surface (yellow), the presumed border between primary shell and secondary crust (purple). Red arrows = growth direction of mesocrystals. Blue line highlights single mesocrystal.

Whereas the biologically precipitated pristine shell in cross-section appears rather solid, and especially in the secondary calcite crust mesocrystals are radially aligned (Fig. 10a), its porous structure and multiple organic/mesocrystal boundaries make it vulnerable to dissolution. Pores enlarge and mesocrystals corrode especially where in contact with organic sheets (Fig. 10b, c). When the pH rises and authigenic calcite precipitates, it also penetrates into these cavities, masking the original radial mesocrystal alignment (Fig. 10d, e).

The common way to remove authigenic overgrowth is to expose the shells to a weak acid (leaching) to hydrolyze the contamination prior to the AMS measurement of the remaining shells⁷⁷. Since in our samples the shells were exposed to a repeated interplay between dissolution and authigenic calcite precipitation, the porosity of the biogenic calcite far exceeds that of the authigenic calcite. Consequently, leaching would mainly affect biogenic but not authigenic calcite.

Impact on Arctic Ocean radiocarbon chronology and paleoceanographic reconstructions

At PS2185, one of the key sites in the Central Arctic Ocean, the PI_3 ages and stable oxygen and carbon isotopes are comparable to published records⁴ indicating that predominately PI_3 specimens were previously analyzed (Figs. 3c and 4). An almost instantaneous age shift on the order of 10–15 ka at this site in the PI_3 record, also observed at other Central Arctic Ocean sites (Fig. 7 and Supplementary Fig. 1), has been interpreted as an apparent absence of MIS 2 sediments^78,79, or a mixture of MIS1 and MIS 3 sediments⁷⁹. Our results suggest a more uniform age increase downcore derived from PI_2 specimens, indicating that previous interpretations might have suffered from age biases. We thus propose that the chronology for the past 30 ka should be critically re-evaluated (see methods). Measuring shells (foraminifera, ostracoda, pteropoda, bivalves, otoliths⁷⁹) with a distinct brownish discoloration, indicating the incorporation of exogenous components, should be avoided.

Consequently, previous interpretations of sediment core records need careful revisions. Apart from a different chronology of lithostratigraphic units that are useful for stratigraphic correlation at the Mendeleev Ridge (methods), our results may have a tremendous impact on paleoceanographic interpretations. If the presence of MIS 2 sediments is confirmed at many sites, the last glaciation may be studied at a low resolution in the Central Arctic Ocean. Since these sediments comprise a maximum of planktonic foraminifera and some biogenic calcite at the base of MIS 2 (Figs. 8d and 9), a permanent thick sea-ice cover during the entire MIS 2²⁷ can be excluded. The proposed high productive zone with seasonally open water conditions up to 84–85°N²⁷ may thus have extended at least to the Lomonosov Ridge. The relatively high abundances of N. pachyderma in core PS72/413-3 at the southern Mendeleev Ridge support a relatively high production during seasonally ice-free conditions in early MIS 2.

Final remarks and recommendations

The majority of overgrowth observed in our samples consists of a more or less thick homogenous coating of authigenic calcite which may easily be overlooked on the coarse secondary calcite crust of N. pachyderma (Fig. 2i, j). Only by careful inspection at high magnification under the stereomicroscope, such overgrowth can be noticed, but early tiny overgrowth may still be overlooked. Merely the investigation of foraminiferal shells from small sample splits under the stereomicroscope and SEM reveals information on the extent of overgrowth in the respective sample.

The unknown variability of sea-ice extent, upwelling and ocean-atmosphere exchange complicates an unambiguous calibration of radiocarbon ages, particularly in the Arctic Ocean⁸⁰. Overgrowth now adds an omnipresent additional complication to obtaining accurate radiocarbon chronologies. Besides authigenic calcite, initial dissolution is observed on some shells of PS2185-6 (Fig. 10d) and may here also contribute to the altered shell-based proxies^{10,12,13,14,25,43,44,45} of PI_3 specimen.

Furthermore, when assessing glacial radiocarbon ages, one has to keep in mind that planktonic foraminifera require primary production for nutrition. Under a much thicker ice cover, they could only reproduce and be ultimately deposited when ice thickness was episodically reduced. Coupled with very low sedimentation rates and intensive bioturbation, each MIS 2 radiocarbon age measured on PI_2 specimens will reflect one or several of such events. Deduced from the crystals, an old carbon source eventually admixed with variable small proportions of dissolved biogenic aragonite or calcite contributes to the precipitates. This makes the application of correction factors for overgrowth impossible.

This study highlights challenges afflicted with our present-day shell-based work. The commonly assumed chronology of Arctic Ocean sediments may deviate by more than 10,000 years from the true chronology in critical parts of the last deglaciation and thus impede the final resolution of ocean condition reconstruction controversies^49,81. The PI_2 specimen ages and isotope values reported here are reasonable for the Holocene and distinctively improved for the glacial. However, only specimens screened beforehand under the SEM, confirmed ideally to be free of overgrowth and dissolution, will enable credible radiocarbon and proxy measurements of glacial samples. These measurements are the cornerstones of Arctic chronology, paleoceanographic reconstructions and, ultimately, climate models.

Materials and methods

The surface samples and sediment cores analyzed in this study were retrieved with the RV Polarstern from sites and with devices as listed in Supplementary Table 1. X-ray images do not give evidence for hiatuses in sedimentary records. For both cores, numerous supplementary information to this paper can be retrieved from https://doi.org/10.1594/PANGAEA.938246.

Lithostratigraphy

The sedimentary sequence of core PS72/413-3 consists of alternating brown and gray beds interbedded with whitish to pink layers or clasts (Fig. 10a). Distinct mottling indicates that in particular brown beds are extensively bioturbated. As it has been previously proposed by Polyak et al.⁸², Stein et al.⁵¹ labeled the brown beds downcore and identified brown beds B1–B3 in core PS72/413. This lithostratigraphy has been used for stratigraphic correlation across the Mendeleev Ridge⁵¹. In the Mendeleev Ridge area, the brown beds are usually assigned to interstadials or interglacials, gray sediments to stadials or glacials, based predominately on the variability of calcareous foraminifera^6,83 Pink to white sediments are rich in detrital dolomite and reflect massive discharge of icebergs from the Laurentide ice sheet^{6,51,75,78,84}. The lithology of core PS72/413-3 is difficult to tie into the principal lithostratigraphy at the Mendeleev Ridge based on the new analytical data. The characteristic combination of low lightness L* and high green-red color space values a* in the brown beds⁶ is only seen in B1 and B2 (Fig. 10a, b). Calcite is only abundant in the topmost 10 cm (Fig. 10c), whereas calcareous microfossils (planktonic and benthic foraminifera, pteropods) are consistently abundant throughout the core (Fig. 10d, note logarithmic scale) being slightly more abundant at the top and base of the core. This pattern differs from those previously observed in cores from the Mendeleev Ridge area where brownish sediments contain considerably higher abundances of calcareous microfossils than grayish sediments⁶. Finally, the white layer W3 is not reflected in increased dolomite contents. Therefore, we generally adopt the lithostratigraphic interpretation of Stein et al.⁵¹ with slight revisions. Color reflectance data indicate a much thicker brown bed B1. The stratigraphic marker horizon W3 which is linked to brown bed B2 is difficult to identify because it thins out from the Amerasian Basin to the Makarov Basin⁵¹. Few pink-white clasts have been observed at ca. 14 cm but these are not associated with high dolomite contents. The peak abundance of dolomite at ca. 28 cm may rather be linked to W3 but a distinct layer is not observed. Since pink-white clasts may be enriched in a horizon at the base of B3 in sediment cores with relatively high sedimentation rates⁵¹, this peak is rather linked to a different white layer. Sediment layers with the highest Mg-rich calcite contents (Figs. 8c, d and 10c) are possibly coeval with those at sites located off northern Greenland⁸⁵. Here, grain size-specific X-ray diffraction (XRD) measurements in cores from site PS2200 show the highest concentrations of high Mg-calcite in the finest sediments, especially in the clay fraction⁸⁵. These sediment layers are derived from iceberg, sea-ice and ocean current transport of glacial rock flour eroded in northern Greenland based on evidence from sediment cores located near the northern mouth of Nares Strait^86,87. Therefore, we assume that most of the fine-grained calcareous detritus in core PS72/413-3 was also derived from glacial rock flour as is observed in neighboring cores⁸⁸.

Chronostratigraphy of sediment cores PS72/413-3 and PS2185-6

Stein et al.⁵¹ discussed published age models for sediment cores at the Mendeleev Ridge, which are rather tentative beyond the range of radiocarbon dating. Thus, brown beds B1 and B2 are consistently dated to the Holocene and late MIS 3, respectively^6,51,75, while B3 has been assigned either to MIS 4/5^6,75,78,83 or 3/4^75,89(Fig. 10a). These age models have been transferred by lithostratigraphic correlation to PS72/413-3 suggesting a much older age of ca. 40 ka at ca. 12 cm⁵¹ than the chronology based on PI_2 and PI_3 specimens. The new uncalibrated radiocarbon chronology confirms the age of B1 but arrives at much younger ages for the older brown beds. Both brown beds are dated to MIS 2 based on radiocarbon ages of PI_2 specimens, while the ages from PI_3 specimens support the presence of a hiatus, encompassing MIS 2. Comparable discrepancies are observed at site PS2185-6 when the new radiocarbon chronology of PS2185-6 is compared with the published record of the box corer PS2185-3^4,27,90 (Fig. 3). The stack of Spielhagen et al.⁹¹ was not used because we could not find any conclusive evidence for their correlation. Both cores were correlated one-to-one without any depth adjustments. The consistent offset between the chronology proposed by standard AMS¹⁴C ages and PI_3 specimens, and that based on selected PI_2 specimens call for a re-evaluation of Arctic chronostratigraphy in the range of radiocarbon chronologies.

X-ray diffraction

XRD pattern analyses have been done in the laboratories of the research group Crystallography and Geomaterials (University of Bremen, FB05 Geosciences). Dried bulk samples have been grounded to a fine powder (<20 µm particle size) and prepared with the Philips backloading system. A thorough preparation commonly increases reproducibility of the results; however, the standard deviation given by Moore and Reynolds⁹² of ±5% can be considered as a general guideline for mineral groups with >20% clay fraction. In addition, the determination of well-crystallized minerals like quartz, calcite or aragonite can be done with better standard deviations (±1–3%)^93,94. The X-ray diffractograms have been measured on a Philips X’Pert Pro multipurpose diffractometer equipped with a Cu-tube (k_α 1.541, 45 kV, 40 mA), a fixed divergence slit of ¼°, a 16 samples changer, a secondary Ni filter and the X’Celerator detector system. The measurements have been done as a continuous scan from 3 to 85° 2θ, with a calculated step size of 0.016° 2θ (calculated time per step was 50 s). Mineral identification has been done by means of the Philips software X’Pert HighScore™ and identification of sheet silicates has been done with the freely available Apple MacIntosh XRD interpretation software MacDiff 4.25⁹⁵. This was followed by full quantification of mineral assemblages of the bulk fraction via the QUAX full pattern method⁹⁴.

Micropaleontological work

One-centimeter-thick samples were taken continuously from core PS72/413-3, and at 5-cm intervals from core PS2185-6. All samples were freeze-dried, and their weight was determined (mean freeze-dried sample weight is 77 and 104 g for cores PS72/413-3 and PS2185-6, respectively) (Supplementary Tables 2 and 6). Surface sediments from multiple corer (MUC) casts were treated with a volume equivalent Rose Bengal/Ethanol (96%) mixture (1 g Rose Bengal per L Ethanol), not freeze-dried, and no weight was determined. All samples were washed with tap water over a 63 µm sieve and oven dried (50 °C). The residue was dry sieved over a 125 µm sieve. For core PS72/413-3 planktonic and benthic foraminiferal numbers were calculated from sample splits (microsplitter) (Supplementary Table 2). Isotope and AMS analyzes were performed on the size fraction >125 to <250 µm. Except for cores PS72/413-3 and PS2185-6, only the surface sample of MUCs was qualitatively inspected for overgrowth on foraminifera and sedimentary authigenic crystals. In plots, all data are plotted as sample mid depth (0–1 cm = 0.5 cm).

All stereomicroscope images are stacked digital images taken under a Zeiss Axio Zoom.V16 microscope equipped with ×16/16 Br. foc. oculars and the objectives PlanNeoFluar Z ×1.0 (numerical aperture 0.25, FWD 56 mm, magnification ×11…×179 with eyepieces PL ×16/16; object field in mm 23…1.4), and Apo Z ×1.5 (numerical aperture 0.37, magnification ×10.5…×168 with eyepieces PL ×16/16; object field in mm 15…0.95). The numerical aperture is 0.25, and the microscope uses the ZEISS ZEN 2.3 (blue edition) imaging software. Images were taken with a ZEISS Axiocam 506 color microscope camera. Each measured sample was documented beforehand by one to several images (see documentary of samples under https://www.pangaea.de/?q=doi:10.1594/PANGAEA.938246).

SEM images and EDS measurements are performed by means of a JEOL JSM-IT100 InTouchScope™ Scanning Electron Microscope. Samples mounted on aluminum stubs using conductive double-sided carbon tape are analyzed without conductive coating in low vacuum mode. Elemental concentrations of samples are the average of ten EDS point measurements, performed at slightly different positions on the sample with an integration time of 45 s per measurement. Due to the roughness of the sample surface, the angle between the sample surface and the EDS detector is not constant (like on a flat polished surface). Due to the topography of the sample, the elemental concentrations determined show a relatively high standard deviation. However, the qualitative differences between different types of samples (structures) are clearly revealed. The documentation of foraminiferal shells is performed on uncoated specimens, only in core PS72/413-3 the specimens from 1.5 to 10.5 cm depths were coated with charcoal in the 306 Vacuum Coater with Turbomolecular Pumping System of Edwards.

To document the degree of overgrowth in both cores downcore, from >125 µm to <2 mm size fraction all N. pachyderma specimens of sample splits containing 10–30 specimens of each sample were glued on SEM stubs and documented (see https://www.pangaea.de/?q=doi:10.1594/PANGAEA.938246). Hereby, a relative qualitative assessment of the overgrowth on N. pachyderma shells is provided by combined Axio Zoom and SEM images (see https://www.pangaea.de/?q=doi:10.1594/PANGAEA.938246).

Radiocarbon measurements

For PI_1 measurements, 20 and 80 specimens were picked from the whole sample >125 to 250 µm of core PS72/413-3, and split into subsamples of 20–30 specimens for replicate measurements. For PI_2 and PI_3 measurements, 60–80 specimens (min 20 if abundance was too low) collected from sample splits of 3.13–6.25%, were selected for each measurement. At site PS72/413-3, two depths containing >800 specimens of PI_2 (18.5 cm) and PI_3 (17.5 cm) were picked for comparison with measurements of smaller samples from the same depth (Supplementary Tables 4 and 5). For the ages of core PS2185-3 between 1000 and 2000 specimens were picked from the 125–250 µm fraction^4,93. From all N. pachyderma isolated, the best-preserved individuals were used for radiocarbon and spare individuals for isotope measurements. The majority of radiocarbon measurements were performed with the MICADAS at the Alfred Wegener Institute (AWI), unless otherwise noted. Equipped for gas measurements of samples containing 20–100 µg C^29,30 it is possible to obtain ¹⁴C ages on only 20–60 N. pachyderma specimens in contrast to approximately 1000 specimens required for standard AMS analyses. Radiocarbon analysis was conducted following standard operating procedures as detailed by Mollenhauer et al.⁹⁶. Briefly, 0.2–0.3 g of foraminifera shells were filled in septum sealed vials and processed using the carbonate handling system. The sample vials were flushed for 5 min with 70 mL/min ultra-pure helium (99.99% He) to remove atmospheric CO₂, and afterward 200 µL of phosphoric acid (≥85%, Fluka 30417) was added. The hydrolyzation of carbonates to CO₂ took place over ~30 min at 70 °C. Following complete hydrolyzation, CO₂ was flushed from the vials by a He flow. Residual moisture was removed from the gas stream over a phosphorus pentoxide trap and CO₂ was concentrated on the zeolite trap of the Gas Interface System. Following thermal expansion of the CO₂ from the zeolite trap, the gas was manometrically quantified and He was added to obtain a gas mixture of ~5% CO₂ in He. The gas mixture was subsequently fed to the MICADAS gas ion source. The radiocarbon data were normalized against Oxalic Acid II standard gas (CO₂ produced from NIST Oxalic Acid II, NIST SRM4990C) and blank correction was performed against size-matched blank foraminifera (pre-Eemian age carbonate fraction from GeoB3316-1, 521–541 cm, >150 µm) processed alongside the samples.

Oxygen and carbon isotope measurements

Stable isotopes were measured with the stable isotope ratio mass spectrometer Thermo Fisher Scientific MAT253 with an automated Kiel IV device at AWI. As this instrument performs best with 30–60 µg carbonate, 20–30 N. pachyderma or a respective number/weight of crystals were used for each measurement. External reproducibility of δ¹³C and δ¹⁸O measurements based on an internal laboratory standard (Jurassic limestone and Solnhofen limestone) measured over a 1-year period together with samples was better than 0.04‰ VPDB for ¹³C and 0.08‰ VPDB for ¹⁸O. Isotope data was converted to the delta notation⁹⁷. All carbonate isotope data are given in per mil (‰) relative to the VPDB standard.