Low Water in Lake Erie: Evidence for Early Holocene Closed-Basin Conditions
Lake Erie today is fourth in the chain of Laurentian Great Lakes that overflow and drain by connecting rivers to the St. Lawrence River and Atlantic Ocean. Ninety percent of the water supply of Lake Erie comes by overflow from upstream Lake Huron. This supply was diverted from the Erie basin when the Huron basin drained directly to the Ontario basin about 12,500 to 11,000 14C (14,640 to 12,920 cal) years BP, and later, when it drained to the Ottawa River valley from 10,500 to 5500 14C (12,510 to 6320 cal) BP. Thus low water levels are expected to have existed in Erie basin during these periods of reduced inflow, in combination with an early Holocene drier-than-present climate. Evidence of low water levels was revealed by acoustic profile and sediment core data in the form of a beach and erosional shoreface 21 m below eastern Lake Erie 10 km south of Nanticoke, Ontario. Similarly, an erosion surface, interpreted as a wave-cut platform, extends southward from Rondeau, Ontario, and along the whole northern shore of the central Erie basin beneath postglacial lacustrine mud to depths of 30 m below present lake level. Cores dated by pollen- and PSV (paleomagnetic secular variation)-correlation to radiocarbon-dated onshore organic sediment sequences show that mud accumulation over the wave-eroded surface only began after 7600 14C (8400 cal) BP while a significant reduction in the mud accumulation rate occurred offshore from 10,500 to 7500 14C (12,510 to 8300 cal) BP. In the relatively shallow western Erie basin, sediment sequences contain plant detritus concentrations indicative of marsh environments between 12,500 and 11,000 14C (14,640 and 12,920) BP, and after 10,400 14C (12,270 cal) BP. After removing the effects of ongoing differential glacio-isostatic crustal warping, the above data revealed that lake levels were low (below the basin overflow outlet) when Huron basin drainage was diverted from the Erie basin. Throughout the early Holocene, the lake in the central and eastern Erie basins was hydrologically closed below the overflow sill, as indicated by the Nanticoke and Rondeau paleo-shore zones. After 7600 14C (8400 cal) BP the lake level rose about 7 m, then more slowly, still in hydrological closure, while geomorphic coastal zone indicators formed throughout the basin, identified by Holcombe et al. (2003; J. Great Lakes Research 29, 681-704). Finally, the lake level rose to overflow the Niagara River sills about 5,500 14C (6320 cal) BP with the resumption of Huron basin overflow, as at present.
Tracking Agassiz Floodwaters Beyond Hudson Strait: Correlation to the Onset of the 8.2 cal ka Cold Event
Barber et al. (1999, Nature 400: 344-348) originally hypothesized that floodwaters from glacial Lake Agassiz flowed from Hudson Strait directly into the Labrador Sea about 8.47 cal ka and suppressed thermohaline circulation there to initiate the 8.2 cal ka cold event. Keigwin et al. (2005, Paleoocean. 20: PA2003, doi:10.1029/2004PA001074) and Hillaire-Marcel et al. (2007, Geophys. Res. Let. 34: doi:1029/2007GL030396) did not find evidence for the floodwaters in the Labrador Sea. It is now known that Labrador Sea convection began about 1000 years later. Also, the original radiocarbon chronology offset the Agassiz outflow from the cold event by about 200-300 years. Energetic drainages of glacial lakes from the limestone- and dolomite-rich Hudson Bay and Hudson Strait regions carried suspended sediment which subsequently rained out over the floodwater trajectory to produce distinct sedimentary beds of enhanced detrital carbonate (DC) content. Using the DC beds as a proxy for floods emanating from Hudson Strait, our studies of sediment cores show that the Lake Agassiz and most other floods turned south after leaving the Strait and flowed over the Labrador and Newfoundland shelves and upper continental slopes, rather than into the Labrador Sea. DC beds in cores collected south of the Grand Banks of Newfoundland show that the Agassiz waters reached the Gulf Stream and were likely transported northeastward by the Atlantic Current into the Nordic seas, where they could have suppressed North Atlantic deepwater production. Bard et al. (1994, Earth and Planetary Sci. Let. 126: 275-287) have shown that corrections to North Atlantic 14C dates on biogenic carbonate depend mainly on the presence of Gulf Stream subtropical water and the duration of annual sea-ice cover that suppresses air-water CO2 exchange. Along the Labrador and Newfoundland margins Gulf Stream water is not a factor, but sea ice has a significant presence. Reservoir corrections applied to 14C dates on biogenic carbonate are based on the age of modern (pre-bomb) shells, and incorporate the effects of current sea-ice cover duration (5-6 months), but not necessarily the duration of former ice cover. Transfer function analysis of dinoflagellate assemblage data show that sea-ice durations at the time of the Agassiz floods ranged up to 11 months; this difference translates to increased corrections of up to -200 years for radiocarbon dates on foraminifera and mollusk shells. An additional -100 year correction allows for the likely presence of dissolved inorganic old' carbon in oceanic waters, indicated by relatively high (5 to 50 %) contents of Paleozoic-aged detrital carbonate in early Holocene sediments. These adjustments, when applied to new and previous Labrador and Newfoundland offshore 14C dates, show that the Agassiz floods were coeval with the onset of the 8.2 cal ka event oxygen isotope excursion in Greenland ice records. These findings raise confidence in the conclusion that ice-dam failure and rapid flooding of glacial Lake Agassiz into the North Atlantic Ocean played a significant role in causing abrupt climate change at 8.2 cal ka.
Characterizing the Discharge Features of Glacial Lake Agassiz during the Post-Marquette Period Using Marine Seismic-Reflection Methods
Glacial Lake Agassiz was the largest of the North American glacial margin lakes. Over its 4,000 year existence, Lake Agassiz varied substantially in aerial extent and volume. This variability was a function of the fluctuating retreat pattern of the Laurentide Ice Sheet's southwestern margin, differential isostatic rebound of the North American crust, the topography of the land exposed by the retreating ice, and erosion of the various outlet channels draining the lake. These factors combined to form a history of Lake Agassiz punctuated by sudden and sometimes catastrophic rerouting of its drainage from one outlet channel to another. The amount and routing of Lake Agassiz discharge has become controversial. However, extensive onshore observations of Glacial Lake Agassiz discharge features have firmly established that northwestern Lake Superior was a major drainage route following the retreat of the Marquette glacial advance ca. 9,500 years 14C BP. We describe a high-resolution single channel seismic reflection dataset collected with a small airgun that we acquired to test our hypothesis that this drainage event (corresponding to the Nipigon Phase of Lake Agassiz) left diagnostic stratigraphic and geomorphic signatures beneath Lake Superior. The unique bathymetry of northwestern Lake Superior, where water depth plunges off Nipigon and Black Bays, makes this location ideal for the identification and characterization of the Post-Marquette depositional features. The steep and sudden drop-off from the shallow water bays into the deep offshore waters of the lake would have caused the high-velocity floods to slow and drop much of the sediment they were carrying. Our results confirm the existence of these sediment packages, which are now buried below a thin blanket of Holocene sediment. They form wedges of sediment that are thickest (some over 70 m thick) in the deep water area adjacent to the flood outlet. The apron of sediment thins lakeward and shore-parallel away from the outlet. The seismic character of the basal units of the apron, proximal to the outlet, is chaotic and only very weakly stratified suggesting that these deposits represent coarse sediment laid down during the initial stages of the flood when flow was presumably at its peak. These sediments are overlain and draped by a weakly stratified package that is more widely developed (extending lakeward beyond the bounds of our survey). We interpret this unit, which becomes more stratified and thinner lakeward, to represent the fine grained sediment associated with the latter stages of the flood when flow had eased.