Contrasting paired benthic and planktonic foraminifera radiocarbon ages from Bermuda Rise ODP Site 1063 during Heinrich Stadials 1 and 2

Abstract

We report 27 planktonic and 21 benthic radiocarbon ages from the subtropical marine sediment core ODP Site 1063 (Bermuda Rise) for the time range between 30 and 14 ka before present. Despite low abundances of benthic specimens, it was possible to measure radiocarbon ages down to ∼10 µg carbon using a MICADAS and the gas ion source developed at ETH Zurich. Based on a tentative radiocarbon–independent age-model we found that the radiocarbon reservoir of the bottom water varied moderately relative to the analytical and age-model related uncertainties throughout the examined time-period, but larger differences in the radiocarbon reservoir appear to have affected the upper ocean layer. In particular, radiocarbon levels around Heinrich Stadial 2 reveal surface radiocarbon content similar to that of the atmosphere, while during Heinrich Stadial 1 surface waters were significantly depleted in ¹⁴C.

Introduction

Ocean Drilling Program (ODP) Site 1063 (33°41´N, 57°37´W) was drilled during ODP Leg 172 at a water depth of 4,584 m (Keigwin et al. 1998) at the Bermuda Rise. The sediment drifts of the Bermuda Rise represent a high-resolution archive of paleoclimatic and paleoceanographic information. Due to sedimentation rates partly exceeding 150 cm/kyr during glacials (Channell et al. 2012) and because the site is bathed by the deepest water masses of the North Atlantic, this region has been intensively studied in particular in terms of reconstructing late Pleistocene deep-water mass geometry and the strength of the Atlantic Meridional Overturning Circulation (Böhm et al. 2015; Deaney et al. 2017; Gutjahr and Lippold, 2011; Henry et al. 2016; Jaume-Seguí et al. 2020; Keigwin and Boyle, 2008; Keigwin and Boyle, 2000; Lippold et al. 2009; Lippold et al. 2019; McManus et al. 2004; Roberts et al. 2010), while paleoceanographic reconstructions from this core location are repeatedly under discussion given the presence of benthic nepheloid layers (Lerner et al. 2019) and indications for high levels of bioturbation at this drift deposit site (Çağatay et al. 2001).

We carried out ¹⁴C measurements from co-existing planktonic and benthic foraminifera covering the time period between 30 and 14 ka before present (BP) to examine radiocarbon surface and bottom reservoir effects and to derive the ventilation age of the deep-water masses in the NW Atlantic since the Marine isotopic Stage 3 (MIS3).

Methods

Sediment samples

The examined 27 samples from ODP Site 1063 (Holes B and D) cover the interval from 2.13 to 14.62 meters of composite depth, which previous age models indicated to encompass the time period across Heinrich Events 2 and 1 (Keigwin et al. 1998, 2005). Prior to sampling, depth correlation between Hole D (22 samples) and Hole B (5 samples) has been refined by aligning high resolution magnetic susceptibility records from both holes (Keigwin et al. 2005), such that the selected samples covered the target interval evenly in age. Due to the high sedimentation rate, where accumulated material predominantly contains clays, silts, and nannofossils, and the water depth close to the regional lysocline, only small amounts of carbonate material were available in each sample and we therefore requested large samples, covering 2-cm core slices for analyses. Foraminifera were extracted from the sediment using standard methods (disintegration of sediment in de-ionized water followed by washing over a 0.063 mm sieve and manual picking from the dry-sieved 0.150 mm residue) at the Department for Geosciences, University of Tübingen. For the analysis of surface water radiocarbon ages, monospecific aliquots of the only sufficiently abundant planktonic foraminifera (Globorotalia inflata) were obtained. Other surface dwellers, such as Globigerinoides ruber or Trilobatus sacculifer, occurred in too low abundance only. For the analysis of deep-water radiocarbon ages, assemblages of mixed benthic species (including epifaunal and infaunal species) were sampled and separated. Sample masses of the planktonic foraminifera were between 4 and 37 mg, while benthic samples always yielded less than 1 mg (Table 1).

Table 1. Samples, core depth, sample weight, radiocarbon results and age of foraminifera from ODP Site 1063

AMS

All planktonic foraminifera samples with more than 5 mg were prepared for conventional radiocarbon AMS measurements with solid C targets. The samples were dissolved in 3N HCl under vacuum and dried in a water trap using a dry-ice acetone mixture. The remaining pure CO₂ gas is captured with liquid nitrogen and attached to the graphitisation line. Here, the CO₂ reacts with H₂ to C and H₂O at 575°C using an iron catalyst (Alfa Aesar, -325 mesh). The solid C samples were measured at the Klaus Tschira Laboratory for scientific dating (Kromer et al. 2013) against the Oxalic Acid II standard and blank values derived from approximately ∼200 ka old foraminifera from the same marine sediment core. Benthic foraminifera were much less abundant yielding amounts of less than 1 mg CaCO₃. Therefore, to determine the radiocarbon content in the scarce benthic species in addition to the contemporaneous planktonic samples, we measured the ¹⁴C content of CO₂ evolved from the carbonate and directly introduced it to the gas ion source as developed at the AMS lab in Zurich (Wacker et al. 2013a). This method allows analyses using as little as 100–600 μg of carbonate (about 10–60 μg C). A sample of 400 μg carbonate (40 μg C) gives typically a ¹²C- current of 10–15 μA over 20 minutes and yields a measurement precision of less than 7 ‰ for a modern sample (Wacker et al. 2013b). All samples were measured against the Oxalic Acid II standard accompanied by measurements of ¹⁴C-free ∼200 ka old foraminifera from the same sediment core for blank corrections. Using the gas ion source allows to measure radiocarbon in the CO₂ phase and reduces the risk of contamination during sample handling in the conventional AMS method. Nevertheless, measurement errors are generally larger due to the limited sample mass and the reduced ionization probability using the gas ion source compared to the solid target technique. The applicability of the gas ion source has been well addressed in several studies, which compared results of solid targets and the gas ion source of identical carbonate samples (Bard et al. 2015; Gottschalk et al. 2018; Molnár et al. 2021; Wacker et al. 2013a).

Tentative ¹⁴ C-independent age model

We refined the available age model for ODP Site 1063 for the examined time period (Böhm et al. 2015; Grützner et al. 2002) by tuning XRF derived CaCO₃ content of the sediment to the δ¹⁸O signal of Greenland NGRIP ice core (Andersen et al. 2004). As both records reflect a Daansgard/Oeschger event (DO) like behaviour in their signals with the ice core being well dated in the period of interest by annual layer counting, we use the prominent DO pattern to derive an absolute age for ODP Site 1063. We identified DO events between 40 and 15 ka BP in ODP Site 1063 sediment properties, allowing a millennial-resolution age tuning. For the time interval between 36 and 10 ka we identified 12 tuning points, hence one tuning point per approximately 2200 a (Figure 1, Table 2). We set the tuning points preferentially to the start and end-points of DO-events, and thus mark identical time stamps in both records, on which the tuning is based. Here, we assume, that there is no significant lag in the CaCO₃ content of ODP Site 1063 compared to the δ¹⁸O signal from NGRIP (Buizert et al. 2015), as fine scale variations (< 200 a) are recorded consistently in both archives. In order to improve our decision in setting the tuning points, we applied a Monte Carlo algorithm (Fohlmeister, 2012), which varies all tuning points within its given uncertainties to maximize the correlation between the ice core data and the elemental composition of the sediment core. Thus, we assume an age error of 0.2 ka for the absolute ages, i.e. the subjectively chosen tie points where optimised by an objective statistical method. The assumed age error of 0.2 ka is accounted for in the resulting radiocarbon reservoir ages (in addition to the ¹⁴C analytical uncertainties). Even if the variations in NGRIP δ¹⁸O and ODP Site 1063 CaCO₃ content did not occur simultaneously, a potential lagging mechanism must have been rather constant and small as the shape of the DO events is maintained during the sedimentation process, potentially resulting only in minor reservoir age offsets. Nonetheless, we note that the alignment of the variations in sedimentary CaCO₃ to the δ¹⁸O of NGRIP is based on the assignment of wiggles and may therefore represent only one out of other possible age-depth realisations.

Table 2. List of used tie points for the age model construction

Figure 1. Alignment of the CaCO₃ profile from ODP site 1063 (Keigwin et al. 2005) to the δ¹⁸O profile of the NGRIP ice core (Andersen et al. 2004). Red triangles denote position of tie points used for the age-model tuning. Blue symbols indicate positions of ¹⁴C-measurements of benthic (crosses) and planktonic (circles) foraminifera for this study.

Results

We report 27 planktonic and 21 benthic radiocarbon ages and thus 21 benthic-planktonic (B-P) ventilation ages from ODP Site 1063 in the time range between 30 and 14 ka BP (Table 1).

The B-P ages in our study vary from a range between –120 to 5000 ¹⁴C-years with an average of 1500. The bulk of the data falls within a broad band of values between 330 and 1990 ¹⁴C-years. The temporal variations within this band might not be subject to fine-scale interpretation given the analytical uncertainties introduced largely by the benthic measurements and the relatively coarse resolution of our ¹⁴C samples.

Discussion and conclusions

Deviating from the majority of our data three values stand out with very large (5000 and 3600 ¹⁴C-years) or very small (–120 ¹⁴C-years) B-P ages. Both the small negative B-P age around 17.4 ka BP (close to the onset of Heinrich Stadial 1), which would suggest the same water mass origin for the near-surface and the bottom water at this site, and the old B-P ages at 26.1 and 27.4 ka BP (before Heinrich Stadial 2), potentially indicating near-surface and bottom water masses of very different origin, are not reflected by the surrounding B-P ages of our record. We cannot exclude the possibility that such features might have been introduced by bioturbation or transport of allochthonous tests albeit the high sedimentation rate (on average ∼90 cm/ka) suggests this process to have a limited effect throughout the core. In particular, the three observed very high or very low B-P ages are not exclusively associated with high or low sedimentation rates. In support of these more extreme values representing real B-P differences, there is ¹⁴C-based evidence for similar fast and strong ventilation changes from various other Atlantic bottom water locations (e.g., data compilation by Rafter et al. 2022) within the time range between 25 and 15 ka BP.

Based on our tentative ¹⁴C-independent age model we calculated the radiocarbon surface reservoir age by comparing the ¹⁴C results of the planktonic species with the contemporaneous atmospheric value for the (¹⁴C-independent) site-specific chronology. The radiocarbon surface reservoir is a measure for the interaction of old deep-water masses with water at the surface or the intensification of exchange between deep and surface waters. In addition, we calculated the radiocarbon difference between the benthic foraminifera and atmospheric values. The benthic radiocarbon offset gives an information on how long the deep water mass was separated from the atmosphere (Skinner and Bard 2022). When considering the analytical uncertainties and assuming that our ¹⁴C-independent age model is correct, we found that based on our benthic results the radiocarbon reservoir of the bottom water was on average 2400 +/–800 ¹⁴C-years and varied little throughout the examined time-period. In contrast, derived from the planktonic data we found larger relative differences in the radiocarbon reservoir at the surface (1040 +/–910 ¹⁴C-years, Table 2, Figure 2). In particular, surface water radiocarbon levels around Heinrich Stadial 2 nearly reach those of the atmosphere, whereas during Heinrich Stadial 1 the surface waters were significantly depleted in ¹⁴C. These differences between both prominent Northern Hemisphere cold events require further research (Bauska et al. 2021; Shuttleworth et al. 2021), since deep ocean circulation proxies (e.g. ²³¹Pa/²³⁰Th, εNd) indicate similar conditions for both cold events (Gutjahr and Lippold 2011; Lippold et al. 2009; McManus et al. 2004; Roberts et al. 2010). Surface water radiocarbon content around Heinrich Stadial 2 indicates a good ventilation and equilibration of surface waters with the atmosphere. This implies a relatively strong stratification of surface waters preventing old, carbon rich bottom waters from upwelling. In contrast to Heinrich Stadial 2, Heinrich Stadial 1 surface waters seem not to have experienced such stratification. The high variability of the B-P ages at ODP Site 1063 and the deviating observations for Heinrich Stadial 2 and 1 are not in agreement with interpretation of other paleocirculation proxies from the Bermuda Rise (Gutjahr and Lippold 2011; Lippold et al. 2009; McManus et al. 2004; Roberts et al. 2010) and might be related not purely to oceanographic explanations. It is possible that the planktonic signal may not represent the surface mixed layer, because the analysed planktonic species G. inflata is known to add a significant portion of the shell calcite within the thermocline (van Raden et al. 2011). However, it is unlikely that the depth habitat of the species changed significantly between the two Heinrich Events (Jonkers et al. 2021). Another process potentially capable of influencing the apparent ¹⁴C age of a sediment horizon is bioturbation, in particular during times of variability in the abundances of different species. Bioturbational mixing of tests out of abundance maxima into older or younger sediment layers can result in apparent isotopic gradients between different species (Löwemark et al. 2008). However, the high sedimentation rate at this site and the clear manifestation of DO-cycles in the sedimentary record argue against a crucial role of bioturbation.

Figure 2. Decay-corrected (Δ¹⁴C) for the atmosphere (black solid line, IntCal2020; Reimer et al. 2020) in comparison with the radiocarbon content measured in planktonic (red ellipses) and benthic (blue ellipses) foraminifera. Error ellipses denote 95% confidence intervals. Cyan rectangles denote the approximate timing of Heinrich Stadials 1 and 2 as derived from peaks in sedimentary ²³¹Pa/²³⁰Th from the Bermuda Rise (Lippold et al. 2009; McManus et al. 2004).

Nevertheless, this data set reports high variability in radiocarbon levels during the coldest period of the last glacial cycle. This observation may provide helpful information for future studies on past ventilation changes of the Atlantic Ocean.