Southern Ocean evidence for reduced export of North Atlantic Deep Water during Heinrich event 1
Abstract
Deep-sea corals form unique, high-resolution archives of ocean circulation that can be dated using the decay of uranium to thorium. They are abundant in the Southern Ocean, and can provide unprecedented insights into ocean circulation and ocean chemistry on sub-millennial time scales in areas where application of traditional paleoceanographic proxies is difficult. Here we present the first coupled neodymium (Nd) isotope and radiocarbon data from deep-sea corals in the Drake Passage (Southern Ocean) adding new constraints on ocean circulation during the last Heinrich event (H1; 16.7 ka). The modern-day Drake Passage water column is homogeneous with respect to Nd isotopes (expressed in epsilon units, ϵNd). Its isotopic value of close to −9 is largely controlled by the mixture of North Atlantic Deep Water and Pacific Deep Water. The aragonite of modern Drake Passage corals reflects this water-column value, whereas a fossil coral from H1 is significantly higher at −6.4 ± 0.4. We interpret this ~2.5 ϵ unit shift as a reduction in the influence of North Atlantic–sourced Nd in the Southern Ocean during H1. This interpretation is supported by a series of radiocarbon analyses on the same sample, and is consistent with a twofold or greater reduction in export of North Atlantic waters from the Atlantic Basin. Combining analyses of radiocarbon and Nd isotopes on U-series dated deep-sea coral skeletons holds great potential for quantification of past ocean ventilation rates.
INTRODUCTION—ATLANTIC CIRCULATION DURING HEINRICH EVENTS
A wealth of paleoclimate data from the Last Glacial Maximum (LGM) has established that there was a reorganization of Atlantic circulation coincident with cool temperatures and low CO2 (e.g., Curry and Oppo, 2005). Less is known about Atlantic circulation during transient Heinrich (H) events, which are characterized in the marine record by substantial accumulations of ice-rafted debris in the North Atlantic (e.g., Hemming, 2004). In general, existing studies show that circulation was indeed different during H1. For example, at the Bermuda Rise, foraminiferal Cd/Ca ratios show a distinct change in water-mass structure (Boyle and Keigwin, 1987) and sedimentary Pa/Th ratios exhibit their highest values (McManus et al., 2004). Both of these results are consistent with a reduced export of North Atlantic Deep Water (or its deglacial equivalent, hereafter also referred to as NADW) from the Atlantic, similar to the results of modeling studies (Rahmstorf, 1995). However, Cd/Ca is a passive tracer that does not record flux, and the nature of the relationship between Pa export and NADW circulation is not straightforward (Keigwin and Boyle, 2008). One way to test whether there was a reduction in the export of NADW is to examine sites that are outside the Atlantic Basin, where we would expect to see the impact of less NADW export.
One such location is the Drake Passage in the Southern Ocean. The Antarctic Circumpolar Current (ACC) incorporates waters derived from the Atlantic and the Pacific, so a reduction in NADW export from the Atlantic is likely to be associated with a decrease in its relative mass proportion in the ACC. In the Drake Passage the ACC is constricted and travels over rough topography, leading to strong mixing (Naveira Garabato et al., 2004), making it a prime location for reconstructing the representative geochemical characteristics of the ACC in the past.
DEEP-SEA CORALS—RECORDERS OF OCEAN CHEMISTRY
Constructing well-dated records of ocean chemistry and past conditions within the Drake Passage is challenging. Strong currents tend to scour away sediments before they can accumulate. Even where sediments exist, the corrosive nature of the water limits carbonate preservation so that many traditional paleoceanographic techniques, including radiocarbon age dating, become impractical. One way to alleviate these difficulties is to analyze the chemical composition of deep-sea coral skeletons such as Desmophyllum dianthus. This species is a cosmopolitan, solitary deep-sea scleractinian coral that thrives in strong currents, lives for about a century (Adkins et al., 2004), and can be dated precisely using the radioactive decay of uranium to thorium (Cheng et al., 2000). The chemical composition of its skeletal aragonite is controlled by the composition of seawater, but this signal can be complicated by environmental parameters and biologically mediated effects (Adkins et al., 2003; Gagnon et al., 2007; Sinclair et al., 2006; Smith et al., 2000). Two tracers that have been shown to record seawater reliably are radiocarbon and neodymium isotopes (Adkins et al., 2002; van de Flierdt et al., 2006): the former provides a dynamic tracer of the carbon cycle, while the latter has been used as a quasi-conservative water-mass tracer (Frank, 2002; Goldstein and Hemming, 2003).
Neodymium is not thought to be utilized during biological cycling, and its relatively short ocean residence time of ~500 yr (Tachikawa et al., 2003) is suitable for reconstructing millennial-scale circulation features (Piotrowski et al., 2008, and references therein). The dissolved Nd isotopic composition of seawater is primarily a function of the age and lithology of continental sources (Frank, 2002; Goldstein and Hemming, 2003). NADW has low values (ϵNd ~–13.5, where ϵNd is the deviation of a measured 143Nd/144Nd ratio from the bulk earth value of 0.512638 in parts/10,000; Jacobsen and Wasserburg, 1980), reflecting old continental crust surrounding the North Atlantic (Lacan and Jeandel, 2005; Piepgras and Wasserburg, 1987). In contrast, deep-water values in the North Pacific are higher (ϵNd ~–4), reflecting young Circum-Pacific volcanics (see summary in van de Flierdt et al., 2004). The Southern Ocean is characterized by ϵNd values that are between those observed in the North Atlantic and the North Pacific (Jeandel, 1993; Piepgras and Jacobsen, 1988). Besides water-mass mixing, local and regional inputs may influence the ϵNd of newly formed Southern Ocean waters. However, today Drake Passage seawater ϵNd is homogeneous at ~−9 at all measured water depths (Piepgras and Wasserburg, 1982), and similar ϵNd values have been observed in the Ross Sea (Tazoe et al., 2006) and in bottom waters of the South Atlantic (Jeandel, 1993).
We have measured the ϵNd of a modern cold-water coral skeleton in the Drake Passage (ϵNd = −9.2 ± 0.9, 400 m; Fig. 1; Table DR1 in the GSA Data Repository1) and it directly records the composition of local seawater (ϵNd= −9.2 ± 0.8, 650 m; Fig. 2, Piepgras and Wasserburg, 1982). This apparent one-to-one relationship between ϵNd in modern sea-water and deep-sea coral aragonite holds up on a global scale and for different species of deep-sea corals (van de Flierdt et al., 2006). Hence, deep-sea corals form an exciting new archive for seawater ϵNd, which is not affected by the complications of extracting authigenic phases from bulk marine sediments.
Map of Drake Passage. Subantarctic and Polar Fronts shown in white marking the main flow of the Antarctic Circumpolar Current (Orsi et al., 1995). Arrows showing Pacific Deep Water (PDW) and North Atlantic Deep Water (NADW) are schematic, and do not represent true path. Yellow squares show locations of water column profiles in which neodymium isotopes have been measured (Piepgras and Wasserburg, 1982); asterisk indicates site shown in Figure 2. Red dots show locations of corals.
Drake Passage depth profiles. A: Black diamonds show ϵNd of seawater in Drake Passage (Piepgras and Wasserburg, 1982). Blue triangle is ϵNd of the modern coral. Red triangle is ϵNd of the H1 coral. Black triangle is ϵNd of flakes of ferromanganese crust scraped off surface of the H1 coral. B: Top radio-carbon scale is for the modern, and the lower for H1. Scales have been offset so that modern and H1 atmospheres (shown by blue line) are aligned, allowing direct comparison of atmosphere-water column difference. The H1 atmosphere of 410‰ is average of Cariaco Basin Δ14C values from 17 to 16.4 ka based on the Hulu time scale (Hughen et al. 2006). Black diamonds are GEOSECS (Geochemical Ocean Section Study) water column Δ14C (Stuiver and Ostlund, 1980). Red triangles are H1-aged corals from Drake Passage: filled are from this study, and open are from Goldstein et al. (2001). Arrows are schematic and highlight that the H1 offset was larger than that of modern day.
ISOTOPIC COMPOSITION OF HEINRICH 1 DEEP-SEA CORAL SKELETON FROM THE DRAKE PASSAGE
Uranium series analysis of a fossil D. dianthus coral specimen from ~1100 m in the Drake Passage has established an H1 age of 16.7 ka (Table DR2). Its ϵNd of −6.4 ± 0.3 is more than 2ϵ units higher than modern seawater (Fig. 2) (Piepgras and Wasserburg, 1982). This change in ϵNd in the Drake Passage is contemporaneous with the least Pa being exported from the North Atlantic (Fig. 3). We analyzed flakes of the Fe-Mn coating from this particular coral and found that its value of −7.0 ± 0.3 is between modern and H1 values. This value is identical to the surface layer of a nearby ferromanganese nodule (Albarede et al., 1997), likely representing an integrated seawater signal from the glacial through to the Holocene. The coral results presented here are the first unambiguous, absolutely dated Nd isotopic values from the deglacial Southern Ocean.
Time series records of deglaciation. Upper panel: shown in blue is the δ18O of Greenland ice core GISP (Greenland Ice Sheet Project) 2 (Grootes et al., 1993) with Younger Dryas (YD), Heinrich event 1 (H1), and Last Glacial Maximum (LGM). Lower panel: green time series is 231Pa/230Th ratio measured in 4500-m-deep sediment core from Bermuda Rise (McManus et al., 2004). Red triangles are ϵNd values of two deep-sea corals analyzed in this study. Scaling is set so that modern Drake Passage ϵNd and North Atlantic Pa/Th plot together, and the highest Pa/Th values at H1 are aligned with the ϵNd value in the Drake Passage that we would expect if there was no input of Nd from the Atlantic (i.e., −4 ϵ units).
The main body of water in the Drake Passage to the south of the Subantarctic Front is composed of upper and lower circumpolar deep waters (Naveira Garabato et al., 2004), which show hydrographic features inherited from Pacific Deep Water (PDW) and NADW, respectively (Orsi et al., 1995). In the absence of additional inputs, a Pacific ϵNd of −4.0 (Piepgras and Jacobsen, 1988) and an Atlantic ϵNd of −13.5 (Piepgras and Wasserburg, 1987) imply that ~55% of the Nd in the Drake Passage comes from the Atlantic (Table DR3). The H1 ϵNd of −6.4 is appreciably closer to the typical Pacific value than the modern day, and allows only ~25% of the Nd to be sourced from the Atlantic. In this calculation the difference in Nd concentrations between different source regions is not considered, because we do not convert Nd fluxes to water-mass mixing ratios. Modern-day NADW has a lower concentration than PDW (e.g., Piepgras and Wasserburg, 1987, Piepgras and Jacobsen, 1988), so conversion to water-mass mixing would require a smaller PDW contribution.
There are several complications to this approach. First, deglacial changes in the ϵNd composition of NADW or PDW values would alter the isotopic composition of waters in the ACC. The data are scarce, but existing constraints on the ϵNd of both deglacial NADW and PDW show little or no change, so we discount this bias (Foster et al., 2007; van de Flierdt et al., 2006; Marchitto et al., 2005). Second, increased exposure of the South America and Antarctic shelves during H1 could result in an increased input of local Nd. Although these shelves do have a radiogenic signature (high ϵNd; Hegener et al., 2007), we consider increased shelf exchange an unlikely source for the observed H1 shift. The shelves are well developed today, and seawater values close to the coast display nonradiogenic values (Piepgras and Wasserburg, 1982). Furthermore, the H1 coral was located in the center of the Drake Passage, within a fast flowing current, where exchange times of seawater with particulates and the local seafloor are likely to be minimal.
We consider that a reduction in the flux of chemical components transported by NADW to the Drake Passage (at least twofold, from 55% to 25%) is the most likely explanation for the higher ϵNd of the Drake Passage during H1. Independent of our interpretation, the results have important implications for past and future studies seeking to use ϵNd records in the Atlantic Ocean to unmix Northern and Southern Component Waters (e.g., Piotrowski et al., 2008).
We also measured the radiocarbon content of the H1-aged coral in a four-point time series along the growth axis (Adkins et al., 1998; Eltgroth et al., 2006; Robinson et al., 2005). We determined that there was no change in Δ14C (the ratio of 14C/12C in the sample relative to a pre-industrial, prenuclear atmospheric standard in units of ‰) during its lifetime (Table DR4). The average value of 193‰ ± 12‰ is ~−220‰ depleted in radiocarbon compared to the contemporaneous atmosphere, which was ~410‰ (average of Cariaco Basin Δ14C values from 17 to 16.4 ka; Hughen et al., 2006; Fig. 2). A second H1 deep-sea coral from the same location has even less radiocarbon (Goldstein et al., 2001). Its age is slightly younger, but propagation of the quoted uncertainties allows for overlap in calendar age and Δ14C (Fig. 2). Results from both corals show that the H1 atmosphere-water column depletion was larger than the modern Δ14C offset of ~−160‰ (Stuiver and Ostlund, 1980; Fig. 2). The Δ14C of the ACC is set by the source waters, including 14C-enriched NADW and 14C-depleted Pacific waters, by ocean ventilation times, and by equilibration with the atmosphere. The difference between the modern and H1 water column to atmosphere Δ14C depletions is therefore suggestive of a reduced radiocarbon content of the source waters, a change in circulation (e.g., reduced ratio of NADW to PDW), or a change in local oceanographic conditions (e.g., increased ice cover or stratification) of the waters in the ACC. When coupled with our ϵNd results, it is likely that reduced NADW input played an important role in the radiocarbon depletion. Recent results from intermediate water depths in the Pacific Ocean suggest that reduced radiocarbon content of Pacific waters may also be an important factor (Marchitto et al., 2007).
The data presented here provide new support for the concept of reduced NADW export during one of the most extreme climate events of the last deglaciation, H1. The application of combined radiocarbon and Nd isotope measurements in radiometrically dated deep-sea corals has the potential to yield insights into dynamics of climate change far beyond the observations made in this paper. Applied to carefully chosen locations and time slices, this new tracer may allow a quantitative understanding of past ocean ventilation rates.
Acknowledgments
We acknowledge Stephen Cairns at the Smithsonian Museum of Natural History for supplying the corals used in this study, and Alex Gagnon for help with sampling them. We also acknowledge Maureen Auro and Vivien Cumming, who assisted in the preparation of the corals before geochemical analysis, Jess Adkins and Jerry McManus for invaluable discussions, and the Lamont-Doherty Earth Observatory isotope geochemistry group for maintaining the laboratories and mass spectrometers. Financial support was provided by the Woods Hole Oceanographic Institution's Ocean and Climate Change Institute, National Science Foundation (NSF) grant OCE-0623107 to van de Flierdt and Robinson, and NSF grant ANT-0636787 to Robinson.
Footnotes
-
↵GSA Data Repository item 2009055, supplementary methods, and Tables DR1–DR3, is available online at www.geosociety.org/pubs/ft2009.htm, or on request from editing{at}geosociety.org or Documents Secretary, GSA, P.O. Box 9140, Boulder, CO 80301, USA.
-
- Received 18 July 2008.
- Revision received 8 October 2008.
- Accepted 14 October 2008.
- © 2009 Geological Society of America
