Abrupt Climate Oscillations During the Last Deglaciation in Central North America

See allHide authors and affiliations

Science  18 Dec 1998:
Vol. 282, Issue 5397, pp. 2235-2238
DOI: 10.1126/science.282.5397.2235


Evidence from stable isotopes and a variety of proxies from two Ontario lakes demonstrate that many of the late glacial–to–early Holocene events that are well known from the North Atlantic seaboard, such as the Gerzensee-Killarney Oscillation (also known as the Intra-Allerød Cold Period), Younger Dryas, and Preboreal Oscillation, also occurred in central North America. These results thus imply that climatic forcing acted in the same manner in both regions and that atmospheric circulation played an important role in the propagation of these events.

The transition from the last glacial maximum to the present interglacial (Holocene) has great importance in understanding how Earth's climate system can abruptly switch from one state to another. The most detailed records of this transition—the late glacial period—are from the North Atlantic region, which appears to have acted as either a trigger or an amplifier of late glacial climate events. This transition [∼13,000 to 9000 14C years before the present (yr B.P.)] was characterized by several broad-scale climatic oscillations in the North Atlantic region (1–6). Isotopic records from Europe (4, 7,8) and Greenland ice cores (2, 3) reveal a late glacial climate sequence in which the warm Bølling-Allerød (BOA) was followed by the cold Younger Dryas (YD) at ∼11,000 to 10,000 14C yr B.P. and then by the warm Holocene. In continental North America, however, climate proxy data provide mixed and limited evidence for these climate oscillations (9), and this sequence was not seen in isotopic records (10). Establishing the geographic extent, sequence, and magnitude of these climate oscillations is essential for understanding the mechanisms and causes of abrupt short-term climatic changes.

Here we describe several sedimentary records from two small lakes (Fig. 1), which show the late glacial and early Holocene climate changes. The two sites from separate basins eliminate local hydrology as the cause of observed changes. The basins are predominately situated in Ordovician and Silurian dolomites covered by thin glacial deposits. Three cores of sediment from Crawford Lake (43°28′N, 79°57′W) and one from Twiss Marl Pond (informal name; 43°27′N, 79°57′W) were analyzed for carbonate stable isotopes (δ18O and δ13C), lithology, elemental geochemistry, pollen, plant macrofossils, and freshwater gastropods. The δ18O values from bulk marl and mollusk shells were used as proxy for temperature (11). Other proxy data were used to infer lake and watershed conditions during climatic changes (Fig. 2) (12). Chronology was controlled by four AMS 14C dates on terrestrial plant macrofossils and by correlation with dated regional pollen sequences (13).

Figure 1

Map showing locations of Crawford Lake and Twiss Marl Pond at the edge of Ontario's Niagara Escarpment, Canada (A), and coring locations (DC, SC, and BC) at Crawford Lake (B).

Figure 2

Combined diagrams of proxy climate data recorded in the late glacial and early Holocene sediments at Twiss Marl Pond (A) and in core DC (B), core SC (C), and core BC (D) at Crawford Lake. PB, Preboreal Oscillation; G/K/IACP, Gerzensee (7), Killarney (6), and Intra-Allerød Cold Period (1); V-PDB, Vienna Pee Dee belemnite; PAZ, pollen assemblage zones. At Twiss Marl Pond, the δ18O was measured on shells of two mollusk species, Pisidium ferrugineum (lower thick line) andValvata sincera (upper thin line). A more detailed discussion is given elsewhere (18). All data will be available at

The 18O/16O ratios (δ18O) in authigenic lake carbonates are a proxy of continental climate and depend on the isotopic composition of lake water and on water temperature. The isotopic fractionation between calcite and water varies by −0.24 per mil (‰) per degree celsius of temperature (14). A strong positive link between δ18O in atmospheric precipitation and mean annual surface temperature exists in the high and mid-latitude regions; the average coefficient is about 0.6‰ per degree celsius (15). The combination of these two factors and the assumption that water temperature closely tracks air temperature lead to an estimate of a coefficient of 0.36‰ per degree celsius between δ18O in carbonate and air temperature. In the late glacial and early Holocene, the climate in the Great Lakes region resembled that of a high-latitude region. A strong temperature gradient likely prevailed immediately south of the continental ice sheets (16). The climatic regime during that time may have imposed a strong link between δ18O and temperature, so the variation of temperatures along this strong gradient might be sensitively reflected in δ18O values of atmospheric precipitation.

The δ18O results from mollusk shells (Fig. 2A) and fromChara encrustations and bulk marl (17, 18) at Twiss Marl Pond show a negative shift of 1.3‰ at 10,92014C yr B.P. (at 490 cm) and a positive shift of up to 2‰ at ∼10,000 14C yr B.P. (at 390 cm). The more negative intervening interval indicates a cold period corresponding with the YD event. A minor negative excursion of 0.4‰ at 9600 14C yr B.P. (at 370 to 380 cm) may correlate with the Preboreal Oscillation (PB) (1–5, 7). This minor oscillation is also indicated by the recurrence of Picea pollen at Twiss Marl Pond and more clearly at Crawford Lake and by a distinct minerogenic layer (Fig. 2B). Before the YD interval, another slight negative excursion of δ18O at 500 to 510 cm in mollusk shells may correlate with the Gerzensee Oscillation (G) in central Europe (7, 4), the Killarney Oscillation (K) in Atlantic Canada (6), and the Intra-Allerød Cold Period (IACP) in Greenland (1–3). It is also indicated by a peak for elemental A1.

The δ18O profile of core SC (Fig. 2C) from the shallow basin at Crawford Lake shows the BOA warming at ∼12,50014C yr B.P. (minimum 1‰ positive shift in δ18O from 378.5 to 374 cm); the BOA warm period from ∼12,300 to 11,000 14C yr B.P. (from 374 to 361 cm); a pre-YD cooling event, corresponding with G/K/IACP oscillations, shortly before 11,000 14C yr B.P. (at 364 cm: ∼0.8‰ negative excursion); the YD cold period from 11,000 to 10,000 14C yr B.P. (from 361 to 350 cm: 1.5‰ negative excursion); the Holocene warming at 10,000 yr B.P. (at 351 cm: ∼1.5‰ positive shift); and the PB at 9650 14C yr B.P. (at 344 cm: 0.4‰ negative excursion). Cores DC (Fig. 2B) and BC (Fig. 2D) at Crawford Lake show similar patterns but with a thinner record and lower temporal resolution.

These observed patterns cannot be attributed to local hydrological effects, because (i) similar patterns are seen at the two sites, (ii) dissolution of dolomite bedrock (with a δ18O value of −6.46‰) as suggested by elemental geochemistry and lithology analysis (Fig. 2) (18) cannot account for the observed patterns, and (iii) the δ18O records from both sites correlate well with the δ18O records from Greenland ice cores (Fig. 3) and central European lake sediments. For the carbonate δ18O–air temperature correlation of 0.36‰ per degree celsius, the 1‰ decrease in δ18O at the beginning of the YD event could imply a 3°C decrease in mean annual air temperature, and the 2‰ increase at the beginning of the Holocene could imply a 6°C increase. This relation may, however, not be valid for long-term climatic change. For example, we have no independent data to evaluate the evaporative enrichment of δ18O (19). The δ18O values suggest that temperature or precipitation (or both) may have fluctuated during the YD event, as also seen in European (7) and Greenland isotopic records (Fig. 3). The YD event appears to have been cold at the beginning and slightly warmer in the middle. The variations during the YD interval are also indicated by elemental concentrations and sediment composition at two study sites. These new records are comparable in sequence and relative magnitude with records from ice cores (2, 3), and marine (1) and European lacustrine sediments (4, 7, 8).

Figure 3

Correlation of Greenland ice core [Greenland Ice Sheet Project 2 (GISP2)] δ18O (left bar-curve) (2) and Crawford Lake (core SC, heavy curve; core BC, dashed curve) carbonate δ18O profiles on calendar- and 14C-year time scales, respectively. The calendar ages from GISP2 are from annual ice layer counting, and the 14C ages from Crawford Lake are based on AMS 14C dates at the two study sites and on correlation with the dated regional pollen sequence (13). At Crawford Lake, the sequence of climatic oscillations includes the BOA warming, G/K/IACP, YD, PB, and HE-5 (22), and possibly OD (Older Dryas). HE-5 here is dated at approximately 7500 14C yr B.P., based on the majorPinus–to–Tsuga-Fagus pollen transition and is equivalent to the cooling event at Greenland Summit at 8200 calendar years B.P. (20). All these oscillations recorded at Crawford Lake are about one-third to one-half the amplitude of the Greenland record, and PB and HE-5 are half the amplitude of the YD and of G/K/IACP at both locations.

At Twiss Marl Pond and Crawford Lake, the YD cooling is indicated by a negative excursion in δ18O values together with a persistence of and slight increase in shrub and herb pollen, a slight decrease of pollen concentration, decreased carbonate, and increased erosion-derived elements, all of which suggest more openings in forests and accelerated soil erosion under a cold climate. The lack of strong upland vegetational evidence for the YD at the two study sites is attributed to an insensitive response of the nonecotonal vegetation at that time (18). A decrease in calcite precipitation or an increase in eroded minerogenic matter during the YD cold interval has been recorded at marl lakes in Switzerland (4).

Two cores (SC and BC) from Crawford Lake (Fig. 3) also show evidence for the short-term cooling event at approximately 8200 calendar yr B.P. (∼7500 14C yr B.P.). The δ18O values drop by ∼0.8‰ then, or about half the magnitude of the YD cooling. This prominent Holocene climatic event is widespread and has left a global record (20, 21). This event and the PB have been termed Holocene Events 5 and 8 (HE-5 and HE-8), respectively, from the North Atlantic marine sediments (22). At Crawford Lake after ∼6500 14C yr B.P., the δ18O profiles (23) are no longer correlative with Greenland isotopic records, which suggests that the climatic regime appears to have changed and that climatic change occurred more frequently on a regional basis.

The sequence and relative magnitude of late glacial–early Holocene climatic changes recorded at Crawford Lake and Twiss Marl Pond match in detail the records from the North Atlantic region and indicate that these oscillations are probably an expression of broad climatic changes. Our data show that, in addition to the YD, other minor oscillations (PB and G/K/IACP) affected the climate beyond the North Atlantic region. The close match of the δ18O records at the two sites presented in this study and at other sites in the North Atlantic region indicates that the climatic forcing acted in the same manner in both the North Atlantic region and interior North America and that the climatic signals were probably carried through the atmosphere over the Northern Hemisphere.

  • * To whom correspondence should be addressed at the Canadian Forest Service, 5320 122 Street, Edmonton, Alberta T6H 3S5, Canada. E-mail: zyu{at}


View Abstract

Navigate This Article