Deep-Sea Coral Evidence for Rapid Change in Ventilation of the Deep North Atlantic 15,400 Years Ago

Coupled radiocarbon and thorium-230 dates from benthic coral species reveal that the ventilation rate of the North Atlantic upper deep water varied greatly during the last deglaciation. Radiocarbon ages in several corals ofthe same age, 15.41 ::!: 0.17 thousand years, and nearly the same depth, 1800 meters, in the western North Atlantic Ocean increased by as much as 670 years during the 30- to 160-year life spans of the samples. Cadmium/calcium ratios in one coral imply that the nutrient content of these deep waters also increased. Our data show that the deep ocean changed on decadal-centennial time scales during rapid changes in the surface ocean and the atmosphere. R ecords from Greenland ice cores have revealed that the glacial polar climate shift ed extremely rapidly several times. From 18,000 to 40,000 years ago (ka), glacial climates were periodically punctuated by rapid returns to milder conditions, called interstadials, that lasted for hundreds of years ( 1). An abrupt change, fewer than several decades, marked the end of the Younger Oryas (YO) cooling event (11.5 ka) (2, 3). Analogous changes in sea surface properties have been found in sediment cores from the North Atlantic, Cariaco Ba sin, and Santa Barbara Basin (4). The glob al extent of these surface ocean and atmo sphere correlations shows that these reser voirs change coherencly and abruptly dur ing major climate transitions. rapid deep circulation change at the begin ning of the last deglaciation. Today the

R ecords from Greenla nd ice cores have revealed that the glacial polar climate shifted extremely rapidly several times. From 18,000 to 40,000 years ago (ka), glacial climates were periodically punctuated by rapid returns to milder conditions, called interstadials, that lasted for hundreds of years ( 1). An abrupt change, fewer than several decades, marked the end of the Younger Oryas (YO) cooling event (11.5 ka) (2,3). Analogous ch an ges in sea surface properties have been found in sediment cores from the North Atlantic, Cariaco Basin, and Santa Barbara Basin (4). The global extent of these surface ocean and atmosphere correlations sh ows that these reservoirs change coherencly and abruptly during major climate transitions. rapid deep circulation change at the beginning of the last deglaciation.
Today the deep North Atlantic Ocean is partially ventilated by North Atlantic Deep Water (NADW), a low-nutrient water mass  (19). Gray line is the E ·34 first derivative of a seventhorder polynomial fit to the ! ·36 surface coral sea-level data 0 (16,21). No attempt was co ·38 made to account for differ-~ ent depth habitats of partic-o.. ·40 ular coral species. Ages of ~ data used to construct the -42 line are shown as gray triangles (16) and circles (21). (B) Oxygen isotopic data from the GISP 2 ice core at 2 m

REPORTS
that is formed in the Nordic and the Labrador Seas. NADW reaches to bottom depths in the modern western Atlantic Ocean as far south as about 40°N and then spreads southward with a core at -3000 m (6). However, the large-scale circulation at the last glacial maximum (LGM) was different (7,8). Nutrient-rich bottom waters of a southern origin spread to 60°N (9-J I) and NADW shoaled co form its glacial analog, glacial N orth Atlantic intermediate/deep water (GNAI/DW). The vertica l boundary between low-nutrient GNAI/DW and southern source waters lay between 1500 and 2000 m (measured between 50° and 60°N) ( 11 ) . Our deep-sea coral samples are from 40°N in the western basin of the North Atlantic and a water depth around 1800 m. They lie n ear the boundary between glacial water masses and are a sensitive indicator of the tim ing of deglacial deep circulation shifts.
Similar to surface reef-building corals, some deep, aragonitic, solitary corals have paired light and dark density bands. This

*
The ro le of the deep ocean in these events has been uncertain. As the deep ocean is the largest reservoir of heat and mass in the climare system, it plays an important role in the mechanism of rapid climate ch ange. Shifts in deep circulation patterns, recorded by the stable carbon isotope composition ofbenthic foraminifera in deep-sea sediments, have been shown co correspond to intersradials of longer duration (5). One problem has been that the time resolutio n of sediments is limited by bioturbation of the upper few centimeters.
We present coupled 23°T h and 14 C dates and Cd/Ca ratios from deep-sea corals and show that this new archive records a very resolution is a proxy for atmospheric temperature at the Greenland Summit (1). (C) Relative abundance of the cold-dwelling planktonic foraminifera N. pachyderma in core Troll 3.1 (29), a proxy for SST. Because the end of the YD cooling occurs during a 14   t To whom correspondence should be addressed. (30) to match the ice core age for the termination of the YD, 11.64 ka (2), and the other control points are left the same (29). (D) Cd/Ca ratios from benthic foraminifera in core EN120-GGC1 at the Bermuda Rise, a proxy for deep ventilation (8,31). Higher values represent a larger content of nutrient-rich waters, of a southern origin in the Atlantic Ocean, and lower values correspond to a higher content of nutrient-poor waters, of a northern origin in the Atlantic Ocean. The coral data from 15.4 ka correspond to rapid events in all the other tracers from the atmosphere to the deep ocean. banding structure and growth rates of about 0.2 to 1 mm per year provide the potential for annual to decadal records of deep-ocean change (12,13). Deep-sea corals can live at depths of 60 to 6000 m, but most live between 500 and 2000 m (14). Corals have been dredged from the ocean floor since the days of the Challenger expedition ( 1872-1876) and thousands of samples exist in collections. Because most of these samples had not been dated, we developed an age screening method to sort samples (15). T he deep-sea scleracrinia Desmophyllum cristagalli and Solenosmilia sp. are two of several cosmopolitan species that are abundant in dredge collections. We measu red paired thermal ionization mass spectrometry (TIMS) uranium series dates ( 13,16) and accelerator mass spectrometry (AMS) radiocarbon dates (17) on six corals (T able 1). From these paired dates, we calculated the initial coral 14 C/ 12 C ratio, denoted 6 14 C (18). which is also the 6 14 C value of the water in which the coral grew. In the modem ocean, we know the initial 6 14 C value for deep-water masses and therefore can calculate a radiocarbon deficiency from a deep 6 14 C value measured downstream. However, the initial 6 14 C of surface waters at deep-water fonnation zones has changed with time. This initial water 6 14 C value is determined largely by the atmospheric 6 14 C history, which depends on the production rate of 14 C in the upper atmosphere and the exchange of 14 C berween active carbon reservoirs. To avoid this problem of variable initial 6 14 C, we calculated a 14 C projection age to remove the effects of changing atmospheric radiocarbon contents on the measured deep 6 14 C for each of our samples ( 19). T h is procedure extrapolates back from the deep value along a closed system 14 C decay path to the intersection with the known atmospheric 6 14 C record. The d ifference between the age of the intersection and the calendar age of the coral is the 14 C projection age. The more conventional ventilation age, measured re lative to the sea surface, is calculated by subtracting the surface ocean reservoir age from the 14 C projection age of the sample (Fig. IA). We subtracted the modern North Atlantic reservoir age of 400 years from the 14 C projection ages to obtain the ventilation age value relative to the tropical sL1rface ocean for all our data. During the YD the high-latitude North Atlantic reservoir age was 600 to 800 years versus the modern value of 400 years (20). Subtracting much more than 400 years from the projection age of our data at 13.7 ka would imply a negative ventilation age. Therefore, during the B¢1ling-Aller¢d warm period, the reservoir age must have been close to the modem value, with an upper limit of 700 years. A lthough our data are not from the YD, they provide an indication of how much the reservoir age can change.
During deglaciation sea level increased abruptly (16,21) (Fig. lA). Two major melt-water pulses from the Northern Hemisphere ice sheets influenced thermohaline circulation during this deglacial time period. u Direction ot Growth 2 30Th dating in five modern D. cristaga/11 constrains the extension rate to be no less than about 0.2 mm/year and therefore the life span of this sample is no more than about 160years (13). This evidence for a rapid change in deep ocean circulation corresponds to the first major deglacial event obseNed in many climate records (see Fig. 1).
Large inputs of fresh water to the North Atlantic may have reduced or halted NADW formation and caused abyssal circulation tO stagnate (22). Our deep-sea coral data show that ventilation at I680 to 1830 m was relatively rapid during meltwater pulse la (at about 14.0 ka) and somewhat more sluggish during the period of reduced melting, about 1000 years afterward (Fig. lA). One possible explanation of these data is that during rapid ice-sheet melting into the North Aclantic, the well-ventilated NADW shoaled. This signature is evident in the radiocarbon age of the 13.7-ka coral. By 12.9 ka, when melting was reduced, the NADW sank to deeper depths. As a result, the coral site was bathed by an older water mass. This scenario also agrees with the glacial nutrient data discussed earlier.
All four of the corals at 15.4 ka have the same 23°T h age but they have different radiocarbon ages. The calculated ventilation ages vary by a factor of 2 (legend to Fig. 1). Separate AMS 14 C dates for the tops and bottoms of three of these corals show an age reversal in each sample ( Table 2). In terms of radiocarbon ages, the biologically younger part of each coral (the top), is older than the biologically older portion (the bottom). We interpret this as a change in the ventilation age of the water during the lifetimes of the corals. From 23°T h dates on modem D. cris-tagalU, we expect that these corals live for no longer than about 160 years ( 14). The largest 14 C age difference, 670 :t 60 years for )FA 24.8, implies a ti. 14 C difference of about 80 per mil. This is nearly the range of prebomb ti. 14 C values in the entire water column of the modem Atlantic Ocean (23).
To check for the large water mass switch implied by these data, we also measured the Cd/Ca ratios in JFA 24.8. The Cd/Ca ratios change by nearly a factor of 2 through the coral (Fig. 2). These data are consistent with the 14 C evidence for a large change in circulation in less than 160 years at this site (24) . At 15.4 ka North Atlantic sea surface temperatures (SSTs) increased dramatically (Fig. IC). B¢lling-Aller¢d warming began over Greenland at about 14.6 ka (Fig. lB). Benthic Cd/Ca values at the Bermuda Rise also increased around 15.4 ka (Fig. 10). In addition, benthic 8 13 C values (25) from many North Atlantic sediment cores decreased at the same time, implying that bottom waters became more nutrient rich ( 10). Our data thus imply that the deep ocean can change during rapid climate events at a rate comparable to that of the atmosphere and the sU1face ocean. Whether the deep ocean is merely responding to or is actively modulating these rapid climate changes is not clear.
On   ID) show that the abyssal Atlantic Ocean was in the process of becoming better ventilated when the I800-m corals show more poorly ventilated waters. Presumably the nutricline at 2000 m shoaled because GNAI/ OW became denser than the underlying southern source water (I l). However, this is not a firm conclusion because the uncertainty between the two time scales cannot constrain the phasing of events.
The coupled 14 C and Cd/Ca data allow us to deconvolve the radiocarbon data and calculate a radiocarbon transit time of this southern source water mass to 40°N. The top to bottom 14 C age difference in JFA 24.8 is equal to the radiocarbon age of the southern source water mass (SO) minus a 50:50 mixture of this water mass and the radiocarbon age of the shallow, northern source water mass (NA): SO -(0.5(SO) + 0.5(NA)) = 670 The 14 C age of the deeper water mass (SO) is expanded into an initial value at its southern source (initial south) and an average transit time for the water mass to reach the coral location (t). This location is close enough to the northern source zone that NA is essentially the northern initial value (initial north): (initial south -initial north) + t = 1340 In the modern western Atlantic Ocean the in itial ages of the northern and southern waters are about 560 and 1400 years, respectively (27). If we assume the modern in itial Atlantic age difference (840 years) between northern and southern waters for the period around 15.4 ka, then the coral data imply an average transit time of 500 years for the southern source water mass. This is significantly longer than today's transit time. However, IS.4 ka may be a time of rapid variation in atmospheric ti. 14 C ( 28) and a t ime of older Pacific deep waters than today (19). Either of these differences from the modern climate cou ld have led to an increase in the age contrast between northern and southern source waters at their formation zones. Consequently, we cannot calculate precisely how much of the 500-year residual is due to more sluggish deep circulation and how much is due to variability in the initial radiocarbon ages at deep-water formation zones.