Climatic effects of glacial Lake Agassiz in the midwestern United States during the last deglaciation

Stable isotope and pollen analyses of a sediment core from Deep Lake, Minnesota, provide new information on the climatic effects of glacial Lake Agassiz in Minnesota and insights into the cause of the prominent Picea recurrence during the Younger Dryas in the southern Great Lakes region. Bulk-carbonate δ18O exhibited large fluctuations between 12.0 and 9.1 ka (dates are in calendar years throughout this paper unless indicated otherwise), probably reflecting the effects of glacial Lake Agassiz superimposed on climatic warming related to large-scale climatic controls. In particular, a 3‰ decrease in δ18O 11.2–10.2 ka interrupted the δ18O enrichment of 1‰ from 12.0 to 11.2 ka and 3.5‰ from 10.2 to 9.1 ka. This δ18O decrease coincided with the expansion of Lake Agassiz. We interpret this decrease as a result of decreased summer temperature and increased precipitation derived from the cold and isotopically light meltwaters of Lake Agassiz. During this δ18O decline, Pinus pollen continued to increase at the expense of Picea pollen at Deep Lake, as at other Minnesota sites, providing evidence that climatic cooling induced by Lake Agassiz did not cause a reversal to a Picea-dominated vegetation. The absence of such a vegetational response implies that the prominent Picea recurrence during the Younger Dryas in the southern Great Lakes region was not caused solely by climatic cooling due to increased flux of meltwater from Lake Agassiz into the Great Lakes. Instead the Picea recurrence might have been driven primarily by the westward penetration of the Younger Dryas cooling. *Address correspondence to Hu. E-mail: huxxx018@ gold.tc.umn.edu. Figure 1. A: Map of Lake Agassiz II region (Teller, 1987). B: Bathymetry of Deep Lake. the penetration of the Younger Dryas cooling into the interior of North America (Shane, 1987; Wright, 1989; Shane and Anderson, 1993), or by the more local climatic cooling of the cold Lake Agassiz waters passing through the Great Lakes (Lewis and Anderson, 1989, 1992). To address these issues, we conducted stable isotope and pollen analyses of a sediment core from Deep Lake (47°41'N, 95°23'W), northwestern Minnesota (Fig. 1). Deep Lake is strategically selected for the purposes of this study. First, the carbonate-rich sediments from this lake offer an excellent opportunity for paleoclimatic reconstructions on the basis of stable isotope records. Second, the lake is suitable for reconstructing vegetational changes on the surrounding landscape because of its small surface area (~4 ha) and the relatively small areas of pollen sources (Jacobson and Bradshaw, 1981). Third, the lake sediments for the period of interest are annually laminated (varved), providing a means for precise chronological determination. Fourth, and most important, Deep Lake has the potential to record the environmental effects of Lake Agassiz because of its proximity to the proglacial lake. Deep Lake is located in a moraine 45 km east of Lake Agassiz at the time of its second expansion 11.0–10.0 ka (herein termed Lake Agassiz II), and its varved sediments are old enough to encompass this expansion. METHODS Three overlapping cores were recovered from the deepest part of Deep Lake with a square-rod piston corer. The cores, mostly consisting of varved sediments (Fig. 2), were correlated with distinctive markers in the varve sequence. We anchored the chronology on a calibrated date of 8.986 ka (accelerator mass spectrometry 14C date 8.090 ± 0.085 ka, Table 1) and extended it to 11.522 ka by varve counts easily made with the aid of a low-power (10X) microscope. Dates for the nonvarved section below were linearly interpolated between 11.522 ka and 12.610 ka (AMS 14C date 10.700 ± 0.107 ka, Table 1). Subsamples of 0.5 cm3 were taken with a calibrated sampler for the analyses of sediment composition, pollen, and stable isotopes of carbonates. Organic and carbonate contents (Fig. 2) were determined by loss on ignition at 550 °C and 950 °C, respectively. Pollen analysis followed standard procedures of Faegri et al. (1992), and at least 300 pollen grains were counted at each level. For the analyses of oxygen and carbon isotopes, bulk-carbonate samples were reacted at 25 °C for 1 hr with 104% H3PO4 made from ultra-pure P2O5 and triply distilled H2O (McCrea, 1950); this reaction should not result in the substantial dissolution of detrital dolomite, which could complicate paleoclimatic interpretation of the data. The isotopic composition of the released CO2 gas was analyzed with a Finnigan MAT delta E triple-collector mass spectrometer. PALEOCLIMATIC INTERPRETATIONS OF OXYGEN ISOTOPE RECORDS Because δ18O records from lake sediments may have a number of controls, their climatic interpretations are complicated. Here we discuss the possible controls of δ18O to facilitate the interpretation of the Deep Lake data in the following section. The controls of δ13C are not discussed in detail because we use our δ13C data only as supportive information for climatic inferences from the δ18O record. The δ18O of authigenic calcite is a function of water temperature and lake water δ18O (Stuiver, 1970), assuming isotopic equilibrium for carbonate precipitation from lake water. Changes in water temperature are relatively unimportant, because the coefficient is small (–0.24‰ ⁄°C) for the temperature-dependent fractionation between lake water and carbonate and because this effect can be easily compensated by evaporative changes in a closed basin (Talbot, 1990) such as Deep Lake. Before evaporative enrichment, the δ18O of lake water is determined by the δ18O of the source water modified by progressive temperaturedependent isotopic fractionation as the air mass passes over the continent (Dansgaard, 1964). The δ18O in precipitation increases with air temperature at the rate of 0.6‰ ⁄°C for the midwestern United States (Rozanski et al., 1993). Shifting moisture sources related to the interplay of the air masses from the Arctic, Pacific, and Caribbean could be important, because Deep Lake is located near the junction of these air masses. However, both the Arctic and Pacific air masses are extremely dry, so the Caribbean is essentially the sole moisture source. We assume that this was the case for the period of interest. Nevertheless, changes in the more local moisture source through the expansion and reduction of Lake Agassiz could be an important driver of our isotopic trends, given the proximity of Deep Lake to this vast proglacial lake. Therefore we consider evaporation, air temperature, and regional moisture source as the major climatic controls of the δ18O trends in the Deep Lake record. Carbonate mineralogy is a nonclimatic or indirect climatic factor affecting δ18O of sediments. Microscopic examinations show that the carbonates in the Deep Lake sediments are dominated by fine-grained calcite that probably precipitated from surface water, presumably as a result of algal photosynthesis during the summer. In addition, X-ray diffraction analysis of the sediment samples shows no correlation of mineralogical composition with our isotopic trends. Furthermore, δ18O analysis of ostracode valves, which are present at several levels (A. Schwalb, 1996, personal commun.), shows a trend consistent with that of bulk sediment samples. We therefore assume that the carbonates in the Deep Lake sediments are primarily authigenic calcite for the period of interest. CLIMATIC AND VEGETATIONAL RECONSTRUCTIONS FROM DEEP LAKE: EFFECTS OF GLACIAL LAKE AGASSIZ Between 12.0 and 11.2 ka, δ18O increased from –7.5‰ to –6.5‰ in the Deep Lake record (Fig. 2), probably as a result of climatic warming associated with increased insolation and decreased ice-sheet size (COHMAP Members, 1988). This interpretation is consistent with other climatic reconstructions from the midwestern United States. For example, many pollen records show a transition from Picea parkland to Pinus (pine) forests during this period (Webb et al., 1983; Wright, 1992). Using several statistical ap208 GEOLOGY, March 1997 Figure 2. Stratigraphy of oxygen and carbon isotopes, pollen assemblages, sediment composition (OM: organic matter), and gross lithology, Deep Lake. Open circles and triangles in isotope and pollen curves represent sampling points.The 14C age scale, which is provided for reference, is a straight conversion from the calendar age scale on the basis of Stuiver and Reimer (1993). proaches, Bartlein and Whitlock (1993) estimated that increases in both July and January temperatures of ~7 °C were associated with this vegetation change at Elk Lake, northwestern Minnesota. A similar vegetational change occurred in the Deep Lake region, as evidenced by the decreasing Picea and increasing Pinus percentages in our pollen record (Fig. 2). δ18O decreased from –6.5‰ to –9.5‰ between 11.2 and 10.2 ka (Fig. 2), suggesting marked decreases in air temperature and evaporation, and/or a shift in moisture source. A temperature decrease resulting from large-scale climatic controls is unlikely, because at this time insolation increased greatly and caused the Laurentide ice sheet to retreat rapidly (COHMAP Members, 1988), and the pollen records throughout the midwestern United States indicate rapid climatic amelioration. The Younger Dryas cooling event, recorded especially in the North Atlantic area, occurred before Lake Agassiz II. The most plausible cause for the 3‰ reversal in δ18O is the climatic effects of Lake Agassiz II, which is dated at 11.0–10.0 ka. Lake Agassiz expanded at this time when the Marquette glacial advance crossed the Superior basin to the Upper Peninsula of Michigan and blocked the eastern outlets (Teller and Clayton, 1983). The lake was fed by glacial meltwater in summer and was dotted by icebergs, which left prominent tracks on the clayey bottom (Clayton et al., 1965). This expansion probably resulted in cooler summers and greater precipitation in the downwind region in which Deep Lake was located, just as the North American Great Lakes (Norton and Boisenga, 1993; Gat et al., 1994) and Hudson Bay (Rouse, 1991) alter the climates to their lee today. Both cooler summers and greater precipitation would have contributed to the 18O depletion in our record. In addition, the isotopically light glacial meltwaters in Lake Agassiz (–15‰ to –17‰ based on the valves of ostracode Candona subtriangulata, Last et al., 1994) as an important moisture source likely contributed to the 18O depletion. A major decrease in lakesurface evaporation in summer probably occurred as a result of these climatic changes, as suggested by the highly significant covariance (R2 = 0.843; p < 0.001) in the decreases of δ18O and δ13C (Fig. 2) in this closed basin (Talbot, 1990). Winter changes might not have been large, because Lake Agassiz was probably frozen in the winter as a result of the constant flux of cold meltwater from a broad expanse of the ice front in the summer, restricted heat storage in this relatively shallow lake (Teller and Clayton, 1983), and cool periglacial atmospheric temperatures. The relative importance of the contributing factors for the 3‰ δ18O decrease cannot be separated with our data. An alternative interpretation for the δ18O decline 11.2–10.2 ka at Deep Lake is a cooling event immediately following the Younger Dryas, as hypothesized by Anderson et al. (1997). In a study at Seneca Lake, one of the Finger Lakes in New York, Anderson et al. (1997) suggested that a 1‰ decline in δ18O values 11.7–9.2 ka (10.1–8.2 14C ka) was correlated with meltwater pulse IB of Fairbanks et al. (1992), and that this represented extensive melting of the ice sheets and influx of meltwater to the Great Lakes, which reduced summer temperatures in the area. However, the duration of the δ18O reversal at Deep Lake was much shorter; here δ18O resumed its increase 10.2 ka and became enriched by 3.5‰ by 9.1 ka (see below). Considering the temporal coincidence of the δ18O decline at Deep Lake with Lake Agassiz II and the potential climatic effects of this huge proglacial lake, we prefer to interpret the Deep Lake record as a manifestation of the more local Lake Agassiz expansion. The pollen assemblages at the onset of Lake Agassiz II (Fig. 2) suggest that the vegetation of the Deep Lake region was at a Picea-Pinus forest ecotone, which should be susceptible to climatic forcing. However, during the first phase (11.2 and 10.4 ka) of the isotopic decrease, Pinus pollen percentages continued to increase at the expense of Picea pollen, as they do elsewhere in Minnesota. This suggests that climatic changes related to Lake Agassiz II did not alter the direction of vegetational transition from Picea to Pinus dominance around Deep Lake. However, as δ18O dropped to its minimum values between 10.4 and 10.0 ka, Pinus pollen percentages decreased gradually, mostly in favor of hardwood taxa Betula (birch) and Ulmus (elm). Starting at 10.2 ka, δ18O gradually increased to –6.0‰ by 9.1 ka (Fig. 2). This 3.5‰ enrichment probably reflected the cessation of Lake Agassiz effects and resumption of large-scale insolation controls (COHMAP Members, 1988) over the climate of the Deep Lake region, which would have resulted in increased summer temperature, decreased annual precipitation, and the resulting increase in evaporation. Such a climatic trend was probably responsible for the initial expansion of Pinus, the population density of which reached a maximum between 10.0 and 9.5 ka, as inferred from its maximum pollen percentages. In the following 400 yr, δ18O continued to increase, and it reached a maximum of –6.0‰ at 9.1 ka. However, Pinus pollen percentages declined markedly as the pollen percentages of herbaceous taxa increased. These isotopic and pollen changes were probably caused by increased evaporation and the consequent effective-moisture deficit, which led to the decline of Pinus forests and the establishment of prairie. These changes are typical of the vegetational and climatic histories (Webb et al., 1983; Bartlein and Whitlock, 1993) of this region. The effects of the expansion and reduction of Lake Agassiz on the climate and vegetation of the adjacent regions have not been documented by previous studies. Despite the existence of numerous pollen records, only the pollen record from Elk Lake, which was substantially farther away from Lake Agassiz II (~75 km) than Deep Lake, has sufficient temporal resolution to assess the effects of Lake Agassiz. On the basis of numerical climatic analyses of the pollen record from Elk Lake, Bartlein and Whitlock (1993) speculated that Lake Agassiz II might have been the cause of increased annual precipitation and decreased July and January temperatures. However, as these authors acknowledged, this interpretation is problematic because the chronology is insecure and because pollen changes responsible for the reconstructed climatic changes occurred throughout much of the midwestern United States. Stable-isotope analyses at this site (Dean and Stuiver, 1993) focused on the environmental reconstructions of the Holocene, the late glacial being represented by a few samples at coarse temporal resolution. IMPLICATIONS FOR THE CAUSE OF THE LATE GLACIAL CLIMATIC FLUCTUATIONS IN THE GREAT LAKES REGION Palynological studies in the southern Great Lakes area provide evidence for vegetational changes contemporaneous with the Younger Dryas episode of cooling. In particular, the recurrence of Picea was a striking feature in the pollen diagrams from the till plains of northern Ohio (Shane, 1987; Shane and Anderson, 1993). Here Picea pollen percentages declined a great amount ca. 15.4–12.9 ka, in association with increases in the pollen percentages of Fraxinus (ash), Quercus (oak), and other temperate hardwood species as a result of climatic warming. Picea pollen then increased along with Abies (fir) and Larix (larch) and then Pinus during the Younger Dryas, which is interpreted as a result of a climatic reversal from the previous warming. Two alternative causal mechanisms have been hypothesized: GEOLOGY, March 1997 209 (1) a westward extension of the Younger Dryas cooling from the North Atlantic (Shane, 1987; Wright, 1989), and (2) local “lake-effect” cooling due to increased flow of meltwater through the Great Lakes basins at this time (Lewis and Anderson, 1989, 1992), following the diversion of glacial Lake Agassiz drainage from the Gulf of Mexico to the St. Lawrence Valley. These hypotheses can be evaluated in the context of our conclusion about the climatic effects of Lake Agassiz II and the vegetation response in the Deep Lake region, because similar processes were involved in these two cases, even though the Picea recurrence in the southern Great Lakes area occurred 1000 yr before Lake Agassiz II. The magnitude of climatic cooling caused by Lake Agassiz II in adjacent Minnesota should be at least as great as that induced by increased meltwater influx to the Great Lakes for several reasons. First, Lake Agassiz II was far larger than all the Great Lakes combined (Teller and Clayton, 1983). Second, Deep Lake was much closer to Lake Agassiz II (~45 km) than the till plains sites were to the Great Lakes (up to several hundred kilometres). Third, Lake Agassiz meltwater, when passing through the Great Lakes, should be warmer, or at least not colder, than that in Lake Agassiz II itself because of surface warming during transit. Although the vegetation around Deep Lake was somewhat different from that of northern Ohio and Indiana for the respective periods of concern, it was at a Picea-Pinus ecotone that should be suitable for a Picea recurrence similar to that in the southern Great Lakes region. However, our results show that climatic changes in the Deep Lake region due to the expansion of Lake Agassiz, although enough to affect the stable isotope record, did not reverse the trend toward the replacement of Picea by Pinus. This implies that climatic cooling south of Lake Michigan due to the increased influx of meltwaters to the Great Lakes alone did not cause the Picea recurrence in the downwind region to the south. This striking palynological anomaly probably resulted from the westward penetration of the Younger Dryas cooling from the North Atlantic superimposed on the more local cooling that might be attributed to lake effects. ACKNOWLEDGMENTS Supported by National Science Foundation Research Training Group on Paleorecords of Global Change (BIR-9014277–0) through a postdoctoral fellowship to Hu and by NSF grant EAR-9508216. We thank P. Glaser, L. Shane, B. Ammann, W. Dean, K. Kelts, A. Leventer, and A. Schwalb for comments, and J. P. Bradbury and H. Mullins for reviews. R. McEwan, J. J. Xia, N. Hageman, S. Elkan, A. Schwalb, and D. Slawinski provided invaluable laboratory assistance. Limnological Research Center contribution 492. REFERENCES CITED Anderson, W. T., Mullins, H. T., and Ito, E., 1997, Stable isotope record from Seneca Lake, New York: Evidence for a cold paleoclimate following the Younger Dryas: Geology, v. 25, p, 127–130. Bartlein, P. J., and Whitlock, C., 1993, Paleoclimatic interpretation of the Elk Lake pollen record, in Bradbury, J. P., and Dean, W. E., eds., Elk Lake, Minnesota: Evidence for rapid climate change in the north-central United States: Geological Society of America Special Paper 276, p. 275–293. Broecker, W. S., Andree, M., Wolfli, W., Oeschger, H., Bonani, G., Kennett, J., and Peteet, D., 1988, The chronology of the last deglaciation: Implications to the cause of the Younger Dryas event: Paleoceanography, v. 3, p. 1–19. Clayton, L., Moran, S. R., and Bluemle, J. P., 1965, Intersecting minor lineations on Lake Agassiz Plain: Journal of Geology, v. 73, p. 652–656. COHMAP Members, 1988, Climatic changes of the last 18,000 years: Observations and model simulations: Science, v. 241, p. 1043–1052. Dansgaard, W., 1964, Stable isotopes in precipitation: Tellus, v. 16, p. 436–438. Dean, W. E., and Stuiver, M., 1993, Stable carbon and oxygen isotope studies of the sediments of Elk Lake, Minnesota, in Bradbury, J. P., and Dean, W. E., eds., Elk Lake, Minnesota: Evidence for rapid climate change in the north-central United States: Geological Society of America Special Paper 276, p. 163–180. Faegri, K., Iversen, J., Kaland, P. E., and Krzywinski, K., 1992, Textbook of pollen analysis: New York, Hafner, 328 p. Fairbanks, R. G., Charles, C. D., and Wright, J. D., 1992, Origin of global meltwater pulses, in Taylor, R. E., ed., Radiocarbon after four decades: New York, Springer-Verlag, p. 473–500. Flower, B. P., and Kennett, J. P., 1990, The Younger Dryas cool episode in the Gulf of Mexico: Paleoceanography, v. 5, p. 949–961. Gat, J. R., Bowser, C. J., and Kendall, C., 1994, The contribution of evaporation from the Great Lakes to the continental atmosphere: Estimate based on stable isotope data: Geophysical Research Letters, v. 21, p. 557–561. Jacobson, G. L., and Bradshaw, R. H. W., 1981, The selection of sites for paleovegetational studies: Quaternary Research, v. 16, p. 80–96. Last, W. M., Teller, J. T., and Forester, R. M., 1994, Paleohydrology and paleochemistry of Lake Manitoba, Canada: The isotope and ostracode records: Journal of Paleolimnology, v. 12, p. 268–282. Lewis, C. F. M., and Anderson, T. W., 1989, Oscillations of levels and cool phases of the Laurentian Great Lakes caused by inflows from glacial lakes Agassiz and Barlow-Ojibway: Journal of Paleolimnology, v. 2, p. 99–146. Lewis, C. F. M., and Anderson, T. W., 1992, Stable isotope (O and C) and pollen trends in eastern Lake Erie, evidence for locally-induced climatic reversal of Younger Dryas age in the Great Lakes basin: Climate Dynamics, v. 6, p. 241–250. Lowell, T. V., and Teller, J. T., 1994, Radiocarbon vs. calendar ages of major lateglacial hydrological events in North America: Quaternary Science Reviews, v. 13, p. 801–803. McCrea, J. M., 1950, On the isotopic chemistry of carbonates and paleo-temperature scale: Journal of Chemical Physics, v. 18, p. 849–857. Norton, D. C., and Boisenga, S. J., 1993, Spatiotemporal trends in lake effect and continental snowfall in the Laurentian Great Lakes, 1951–1980: Journal of Climate, v. 6, p. 1943–1956. Overpeck, J. T., Peterson, L. C., Kipp, N., Imbrie, J., and Rind, D., 1989, Climate change in the circum–North Atlantic region during the last deglaciation: Nature, v. 338, p. 553–557. Rouse, W. R., 1991, Impacts of Hudson Bay on the terrestrial climate of the Hudson Bay Lowlands: Arctic and Alpine Research, v. 23, p. 24–30. Rozanski, K., Araguás-Araguás, L., and Gonfiantini, R., 1993, Isotopic patterns in modern global precipitation, in Swart, P. K., et al., eds., Climate change in continental isotopic records: Washington, D.C.,American Geophysical Union, p. 1–36. Shane, L. C. K., 1987, Late-glacial vegetational and climatic history of the Allegheny Plateau and the till plains of Ohio and Indiana: Boreas, v. 16, p. 1–20. Shane, L. C. K., and Anderson, K. H., 1993, Intensity, gradients and reversals in late glacial environmental change in east-central North America: Quaternary Science Reviews, v. 12, p. 307–320. Stuiver, M., 1970, Oxygen and carbon isotope ratios of fresh water carbonates as climatic indicators: Journal of Geophysical Research, v. 75, p. 5247–5257. Stuiver, M., and Reimer, P. J., 1993, Extended 14C age calibration program: Radiocarbon, v. 35, p. 215–230. Talbot, M. R., 1990, A review of the palaeohydrological interpretation of carbon and oxygen isotopic ratios in primary lacustrine carbonates: Chemical Geology (Isotope Geoscience Section), v. 80, p. 261–279. Teller, J. T., 1987, Proglacial lakes and the southern margin of the Laurentide ice sheet, in Ruddiman, W. F., and Wright, H. E., Jr., eds., North America and adjacent oceans during the last deglaciation: Boulder, Colorado, Geological Society of America, Geology of North America, v. K-3, p. 39–70. Teller, J. T., and Clayton, L., 1983, Glacial Lake Agassiz: Geological Association of Canada Special Paper 26, 451 p. Webb, T., III, Cushing, E. J., and Wright, H. E., Jr., 1983, Holocene changes in vegetation of the Midwest, in Wright, H. E., Jr., ed., The Holocene: Minneapolis, University of Minnesota Press, p. 142–165. Wright, H. E., Jr., 1989, The amphi-Atlantic distribution of the Younger Dryas paleoclimatic oscillation: Quaternary Science Reviews, v. 8, p. 295–306. Wright, H. E., Jr., 1992, Patterns of Holocene climatic change in the midwestern United States: Quaternary Research, v. 38, p. 129–134. Manuscript received September 10, 1996 Revised manuscript received November 20, 1996 Manuscript accepted December 2, 1996 210 Printed in U.S.A. GEOLOGY, March 1997