In search of differential sources of DSO waters
Today, DSO source waters are intermediate and deep waters of the Icelandic and Greenland Seas, that means, convection-induced waters of North Atlantic sources (NAC), Arctic Ocean waters, and the subsurface North Iceland Jet (NIJ) (Fig. 1). The isotope composition and ventilation age of waters feeding the DSO as recorded at Site PS2644 (and by fragmentary records from the southern exit of the Denmark Strait; Millo et al., 2006) document a number of short-term millennial-scale changes in circulation geometry (Fig. 4) to be compared with the modern regime of the overflow. Differential ventilation ages suggest a distinction of three different modes of DSO water: (1) Extremely low ages mark the DSO mode 1, both at the very end of GI 2 at 21.8 cal. ka (GICC05 record of Wolff et al., 2010) and at the end of the LGM between ~19.5 and 18.7 cal. ka (Mix et al., 2001), the center of time segment II (Fig. 4). (2) Most of LGM time segment I displays DSO mode 2 marked by a stable high ventilation age of 2200-2500 years, almost recurring near 18.7–18.4 cal. ka. (3) Finally, time segments III and IV show DSO mode 3 with an intermediate ventilation age of 1100-1600 years. Near 17.6–17.3 cal. ka, mode 3 starts with an age minimum of ~700-960 years similar to DSO mode 1. Finally, this mode continued until about 15.1 cal. ka (Fig. 2a), the base of a sedimentation gap (Voelker 1999).
DSO modes 1 and 2 are paired with two separate, but coeval populations of epibenthic δ13C values measured on single foraminifera specimens in each sediment sample, one with 0.8–1.4 ‰, one with 1.4–1.7 ‰ (encircled in Fig. 4), an absolute extreme in the estimates of global ocean ventilation (Millo et al., 2006). In part, these extreme populations may just represent the product of seasonal to decadal variability. The highest values suggest that DSO mode-2 waters then have been entrained from polar regions (Brakstad et al., 2023; in harmony with a model of Haine, 2021) that are largely bare of any organic carbon production. Based on data of single foraminifera specimens, maximum δ18O values of 5.5 ‰ likewise record extremely low intermediate-water temperatures typical of an Arctic source (Waelbroeck et al., 2011). Conversely, briefly intercalated ventilation age minima of DSO mode 1 indicate that a major portion of these waters then may have originated nearby, likely from convection in the eastern Nordic Seas, near Site GIK23074, as discussed below (Fig. 3).
Nd and Pb radiogenic isotope signatures suggest the sediment related to DSO modes 1 and 2 was largely sourced from northern Europe, Iceland, and probably, basaltic sediments on the Iceland-Scotland Ridge overflown by the North Iceland Jet (Figs. 1, 4, and 5). Altogether, DSO modes 1 and 2 may have come close to patterns of modern circulation geometry. The flow strength of mode 2, however, was probably reduced as suggested by medium high sedimentation rates at PS2644 (Fig. 4). Conversely, the flow of mode 1 was probably much stronger. At >21.8 cal. ka, it resulted in a sedimentation gap and erosional hiatus at Site PS2644, where the inflow to the DK Strait was constricted like in a funnel. The enhanced current regime is also documented at the end of GI 2, during the B/A, and major parts of the Holocene, by contrast to today, where a thin layer of few cm modern sediment is still preserved and14C dated (Voelker, 1999).
During time segment 4, features of DSO mode 3 were fundamentally different from LGM DSO modes 1 and 2. Apart from just medium high ventilation ages mode 3 shows a broad range of intermediate epibenthic δ13C values (0.6–1.4 ‰) that record an attenuated source water ventilation, possibly due to a modest HS-1 input of remineralized organic carbon. DSO mode 3 starts with a significant, almost abrupt 0.8 ‰ drop of maximum δ18O values, equivalent to a 3.4°C rise in minimum bottom water temperature, that occurred near the actual end of a last high in bottom water ventilation age (Fig. 4). The diverse range of δ18O data of epibenthic single species could reflect a broad array of bottom water temperatures for each sediment sample extending over 1.5–2.0 ‰, equal to about 8°C. Since this range appears unlikely, it probably includes an imprint of δ18O signatures of waters from different seasons and short-term changing source waters.
Per analogy to the records of PS2644, Sessford et al. (2018) report on Mg/Ca-based bottom water temperatures for stadial and interstadial periods over MIS3, analyzed in twin sediment core GS15-198-36CC. Here, minimum values of 1° to -1.5°C mark Greenland Inter­stadials (GI) 5, 6, 7, and GI 8, when bottom waters were formed by convection in the Nordic Seas, whereas warm temperatures of 2°–4°/ 5°C marked Greenland Stadials (GS) 4, 5, 7, 8, and the very onset of GI 8. Temperature anomalies reached up to 6°C and provide crucial evidence for short-term substantial changes in the origin and circulation geometry of DSO waters during MIS3, similar to those deduced in our study.
Last but not least, DSO mode 3 goes along with a significant rise in εNd and drop in 206Pb/204Pb isotope ratios, in particular during time segment IV (Figs. 4 and 5). The proxies form an important tracer of basaltic sediments from Iceland and/or the Iceland-Scotland Ridge, predominantly passed by mode-3 waters prior to reaching Site PS2644 (~68°N) at the northern entrance of the Denmark Strait, similar to the modern track of the ”North Iceland Jet” (Våge et al., 2011). Less likely, the isotope ratios may trace East Greenland basalts (S of 70°N) (Peate and Stecher, 2003), since the EGC acts as interjacent barrier (Fig. 1).
During time segment III, the flow strength of DSO mode 3 may have been similar to that of the preceding mode 2. During time segment IV, however, it may have decreased significantly or may have even been negligible, when meltwater advection reached a final maximum and hemipelagic bulk sedimentation rates at Site PS2644 displayed an abrupt rise by a factor of 3 to 5 (Fig. 4). The high sedimentation rates were paired with abundant ice-rafted debris (IRD) rich in hematite-stained red quartz grains that depict an ongoing flux of Devonian ’Old Red’ sediments picked up by icebergs along the N.E. Greenland margin (N of 72°30’N), hence reflect a continued afflux of the EGC (Voelker, 1999).
In search for a hypothetical ultimate source of DSO mode 3 waters during HS-1, sediment proxies provide several independent lines of evidence for a link to the topmost intermediate and/or subsurface waters in the (then possibly sea-ice covered) Atlantic south and southeast of Iceland: (i) Their dominant minimum epibenthic δ13C values of 0.6-0.9 ‰ reflect bottom waters far less ventilated during HS-1 than the output of peak glacial DSO modes 1 and 2. However, they largely resemble the low δ13C values of 0.2 - 0.9 ‰ for the coeval DSO output south of the Denmark Strait (at ~2000 m w. d. during late HS-1, in addition to some rare δ13C outliers of single specimens that record local brine water formation in the Irminger Sea (-0.2 to -1.5 ‰; Millo et al., 2006; Hagen and Hald, 2002; Sarnthein et al., 1994). (ii) The ventilation age range recorded for DSO mode 3 during time segments III and IV (Fig. 4) matches closely that reported for subsurface and intermediate waters south of Iceland (Thornalley et al., 2011). (iii) Though per se largely unknown, the minimum temperatures of topmost Atlantic Intermediate Waters (that, as we surmise, crossed the western Iceland-Scotland Ridge with the NIJ at depths of <200 m w. d.) have definitely been much higher than those of polar waters probably entrained at Site PS2644 along with LGM DSO modes 1 and 2 (e.g., coeval abrupt 1-‰ drop of HS-1 benthic δ18O values at Site SO82-5 south of Iceland, ~1400 m w.d.; van Kreveld et al., 2000) and per analogy to today. (iv) As outlined above, the highly radiogenic εNd values of DSO mode 3 could mean an antecedent passage over basaltic sediments as found on the Iceland-Faeroe Ridge and northern slope of Iceland, defined as ”North Iceland Jet” by Våge et al. (2011), which today supplies almost half of the total DSO flow rate (Figs. 1 and 6).