aThe numbers correspond to site locations in Fig. 1 (sites 1-37) and Fig. 2 (sites 38-46).
aThe numbers correspond to site locations in Fig. 1 (sites 1-37) and Fig. 2 (sites 38-46).
er Lake at ca. 4700 B.P., probably by geomorphic means (diversion of the Walker River), yielded a late Holocene record of moisture variation, reconstructed primarily from diatom and ostracode data (Bradbury, 1987; Bradbury et al., 1989). The data indicate low lake levels and high salinity between 2400 and 2000 B.P., followed by subsequent refilling. Pore fluids in sediments from Pyramid Lake, another remnant of Lake Lahontan, suggest it remained in existence during the early Holocene when Walker Lake was dry (Benson, 1978). Dating of tufa deposits around Pyramid Lake (Benson and Thompson, 1987) suggests low lake levels between 2500 and 1800 B.P. Mono Lake in California, underwent a series of oscillations during the Holocene and reached its Holocene high stand at ca. 3400 B.P. By 1800 B.P., it had fallen to a low stand and subsequently underwent a series of century- and subcentury-scale oscillations (Stine 1990). These oscillations include two intervals of prolonged drought in the Medieval period (A.D. 9001300), both of which exceeded historic droughts in magnitude and duration (Stine, 1994).
For Arizona, lacustrine records indicate moist conditions in the early Holocene, followed by dry conditions after 8500 B.P. (Waters, 1989; Anderson, 1993; Blinn et al., 1994; Hasbargen, 1994). Within the dry mid-to late Holocene, two marl units in a core from Lake Cochise (Wilcox playa), dated at ca. 5400 and between 4000 and 3000 B.P., suggest mid-Holocene pluvial intervals when the basin filled (Waters, 1989). A continuous lacustrine diatom sequence for Montezuma Well suggests higher lake levels between 5000 and 3500 B.P., followed by a brief arid period to 3000 B.P. (Blinn et al. 1994). The lithology of a core from Potato Lake indicates that it refilled at ca. 3000 B.P., after mid-Holocene desiccation (Anderson, 1993). However, diatom, pollen, and macrofossil records for nearby Stoneman Lake suggest persistent aridity until 2000 B.P. (Hasbargen, 1994). In southern California, a short episode of increased moisture at ca. 3600 B.P., as well as a more recent event at 390 B.P., is indicated by interbedded lake sediments in a presently dry basin in the Mojave Desert (Enzel, 1992). Thus, there is considerable evidence for aridity throughout the Southwest between ca. 8500 and 5500 B.P. and of episodes of increased moisture between 5400 and 3000 B.P. However, late Holocene patterns of lake-level change are spatially variable, and thus, the nature of climate variation is unclear. Lithologic data for basins in the southern High Plains of Texas show lacustrine deposition prior to 8500 B.P., followed by desiccation in a number of locations and continuous deposition in others until ca. 5500 B.P. (Holliday, 1989). Deposits of eolian origin and widespread desiccation of lake basins between 6500 and 4500 B.P. suggest maximum aridity in this interval.
Although pollen data for the Pacific Northwest are abundant, there are virtually no paleolimnological studies of climate variation. The available lithologic data suggest that lake levels were very low in the early Holocene and rose sometime after 5000 B.P. (Barnosky et al., 1987).
In the northwestern part of the Canadian Plains (western and central Alberta), many lakes were dry at the onset of the Holocene, suggesting arid conditions (Schweger and Hickman, 1989). Flooding of these formerly dry basins began at various times in the mid-Holocene (7500-4000 B.P.), although low lake levels and high salinities persisted at most sites until 4000 B.P. (Schweger and Hickman, 1989; Vance et al., 1992, 1995). In contrast, for southern Saskatchewan (Ceylon Lake) and southern Manitoba (Lake Manitoba), mineralogical and lithologic data indicate that these lakes were fresh; therefore, conditions must have been wet until 8000 B.P., followed by high salinities and low lake levels to 4500 B.P. (Teller and Last, 1981; Last, 1990).
For the northern Great Plains of the United States, evidence from diatoms, ostracodes, and carbonate geochemistry indicates that lakes were fresh and lake levels were high at the onset of the Holocene. However, levels began to decline in the early Holocene, and many lakes became hydrologically closed. Once they were hydrologically closed, the lakes showed an increase in salinity in response to decreases in effective moisture, although there was regional variation in the magnitude and rate of salinity change and, thus, in the pattern of inferred climate (see summary in Laird et al., 1996a). For North Dakota, diatom and ostracode stratigraphies for several lakes indicate that the rate of salinity increase accelerated after 8000 B.P., and maximum salinity was reached at ca. 6500 B.P., although clearly conditions remained quite dry for several thousand years afterward (Laird et al., 1996a; Xia et al., 1997). Similarly, in a suite of lakes in western Minnesota, paleolim-nological proxies (diatoms, ostracodes, isotopes, and mineralogy) indicate gradually increasing aridity from the early Holocene onward, with an accelerated rate of increase after 8000 B.P. and maximum aridity in the interval between 7200 and 5200 B.P. (Forester et al., 1987; Bradbury and Dean, 1993; Digerfeldt et al., 1993; Schwalb et al., 1995; Smith et al., 1997). Further east in Wisconsin (Winkler et al., 1986), a sand lens in cores from Lake Mendota suggests lake-level lowering and aridity beginning at ca. 6900 B.P.
For the southern part of the northern Great Plains, poor dating control and the small number of sites pre clude a clear understanding of Holocene moisture patterns. In the meromictic Medicine Lake in South Dakota, maximum diatom-inferred salinity was reached by 9000 B.P., and high salinity persisted throughout the mid-Holocene (Radle et al., 1989). However, sedimen-tological and geochemical evidence suggests that the lowest lake levels may have occurred several thousand years later (Valero-Garces et al., 1995). Isotopical and geochemical data for Pickerel Lake in South Dakota similarly indicate the onset of dry conditions at ca. 9000 B.P., but a shift in ostracode species dominance at ca. 5500 B.P. suggests increased aridity at this time relative to earlier millennia (Smith, 1991; Schwalb and Dean, 1998). Thus, the timing of maximum aridity in this region is currently uncertain.
Many lakes and wetlands in the Nebraska Sandhills were formed just prior to the onset of the Holocene by blockage of drainage channels by eolian sand during an interval of extreme aridity. In the western Nebraska Sandhills during the middle Holocene (ca. 6000 B.P.), such a blockage by sand caused a 25-m rise in the regional water table (Loope et al., 1995; Mason et al., 1997). As a result, lakes formed in several interdune depressions and accumulated unusually thick sequences of organic sediments. Thin sand layers in several of these lakes represent fluctuations in lake levels during episodes of late Holocene (post-1500 B.P.) dune movement (Mason et al., 1997).
Although evidence for a mid-Holocene dry interval is widespread throughout the northern Great Plains and the Midwest, it is unclear whether the dry climate was cool or warm. For Elk Lake in Minnesota, ostra-code and oxygen isotope data suggest that, although conditions were dry beginning at ca. 6800 B.P., temperatures were cold until 5900 B.P. and then became warmer (Forester et al., 1987; Dean and Stuiver, 1993). Isotopic data for Pickerel Lake in South Dakota have been also interpreted as indicative of a cool early to mid-Holocene (Schwalb and Dean, 1998; see also Xia et al., 1997). In contrast, pollen analyses for Elk Lake suggest warmer July temperatures than now from ca. 7000 B.P. through the mid-Holocene (Bartlein and Whitlock, 1993). Unfortunately, there are presently insufficient data for other sites to confirm the nature of early to mid-Holocene regional temperature.
Sites throughout the northern Great Plains show a decrease in salinity and an increase in moisture beginning between 5000 and 4000 B.P. (reviewed in Laird et al., 1996a; see also Bradbury and Dean, 1993; Schwalb et al., 1995; Smith et al., 1997; Xia et al., 1997). In most locations, moist conditions have persisted to modern times, although for many sites, the data suggest large fluctuations between wet and very dry, especially after 2200 B.P. (Laird et al., 1996b; Schwalb and Dean, 1998).
In Wisconsin, a transition from sand to gyttja at 3200 B.P. marks higher lake levels and enhanced moisture at Lake Mendota (Winkler et al., 1986), slightly later than at most sites to the west, a pattern also suggested by vegetation data for Iowa (Baker et al., 1992).
At present, only a few high-resolution lacustrine records of decadal or subdecadal climate variation exist for North America, and these are restricted to north-central North America (Laird et al., 1996b; Dean, 1997; Campbell, 1998). Data for several lakes in North Dakota (Laird et al., 1996b; Yu and Ito, 1999) suggest that the climate of the last 2000 years was hydrologically complex, with frequent alternations between wet and dry intervals. The data indicate that severe drought was characteristic of Medieval times (A.D. 900-1300), but that this interval was not unusual relative to previous centuries. Similarly, intervals of major drought were also typical within the Little Ice Age (lIA) (A.D. 14001850), although extremely fresh conditions, suggesting high precipitation, are evident from A.D. 1330-1430 and in the early decades of the 1800s. The data suggest intervals of multiple decades to more than a century when drought was frequent and severe, indicating that the frequency of extreme events may have been greater at times during the last 2000 years than in the twentieth century. A record of eolian activity inferred from sedimentary aluminum and quartz concentrations in Elk Lake in Minnesota also indicates decadal- to century-scale cycles of drought during the last 1500 years (Dean,
1997), with the most distinct drought intervals corresponding in time to the LIA (see also Fritz et al., 1994). Similar cycles occur in magnetic susceptibility and organic carbon in Pickerel Lake in South Dakota (Schwalb and Dean, 1998).
A high-resolution, sediment grain size record for a lake in boreal parkland in central Alberta (Campbell,
1998) suggests several directional late Holocene shifts in streamflow magnitude into the lake. Median sediment grain size in a midlake deepwater core, which is argued to be directly related to stream inflow (coarse grain size indicates high flow), decreases markedly during Medieval times relative to the previous ca. 1000 years, suggesting an interval of very dry conditions. Inferred moisture increased during the last 600 years and was higher than at any other time in the 4000-year record.
Interpretations of lake records for the Great Lakes region of eastern North America disagree about Holocene moisture patterns. Most records suggest that lake levels were highest during the early Holocene and subsequently fell. Most authors agree that the climate was warm from 8000-2000 B.P. (Stuiver, 1970; Drum-mond et al., 1995; Krishnamurthy et al., 1995); some suggest that maximum aridity occurred from 8000 to 5500 B.P. (Colman et al., 1990; Edwards et al., 1996; Anderson et al., 1997), whereas others suggest it was driest between 5000 and 2000 B.P. (Dwyer et al., 1996; Yu et al., 1997). In Lake Michigan, increased carbonate deposition, bands of iron sulfide, and ostracode taxa characteristic of higher salinity in core sections dated at between 7000 and 5000 B.P. have been interpreted as evidence for a drier climate, and the data suggest that lake levels in the Great Lakes may have been as much as 50 m lower than they are now (Colman et al., 1990; Rea et al., 1994). Several records for the Finger Lakes region of New York, to the east, show the highest calcium carbonate deposition between 10,000 and 5000 B.P. (Dwyer et al., 1996; Anderson et al., 1997; Mullins, 1998), caused by enhanced carbonate precipitation as a result of higher water temperatures. Some of the climatic interpretations are based on oxygen isotopes in carbonates, which vary with changes in temperature, moisture source, and evaporative concentration. At Crawford Lake in southern Ontario, apparent depositional hiatuses in shallow-water cores between 5000 and 2000 B.P., coupled with lighter 818O values in marls, are interpreted as evidence of reduced input of summer Gulf moisture, which is relatively heavy (Yu et al., 1997). In contrast, a nearby site shows heavier values of 818O during this interval, attributed to increased flow of Gulf moisture (Edwards et al., 1996). The latter interpretation is broadly consistent with a compilation of lithologic data on lake-level fluctuations in eastern North America, which shows fewer basins with low lake levels beginning at 4000 B.P. (Webb et al., 1993). Intensified water vapor transport from the Gulf of México from 8000 until 2000 B.P. has been suggested from a record of 8D in lacustrine organic matter in western Michigan (Krish-namurthy et al., 1995), although here the relative roles of temperature versus moisture source cannot be determined. The inability to disentangle temperature versus precipitation influences from these lacustrine proxy records is a general problem, which may be the cause of some of the discordant interpretations mentioned previously. Clearly, more sites are needed from eastern North America, as well as a more developed understanding of lacustrine isotopic variations.
A single, detailed record of lake-level fluctuations at multidecadal resolution has been inferred from grain size analysis of the sediments of Herring Lake, where water levels are controlled by fluctuations in the adjacent Lake Michigan (Wolin, 1996). An increase in coarser grained sediments is used to infer low lake stages from A.D. 0-150, 450-780, 920-1010, 1220-1450, and after 1670, thus indicating significant decadal- to century-scale variations in atmospheric circulation patterns.
There are relatively few paleolimnological records of Holocene climate for northern latitudes. For northwestern Alaska, trace element analyses of ostracodes suggest that the climate was cold and dry in the early Holocene and remained dry until after 7000 B.P. (Hu et al., 1998). The data suggest that the temperature increased during the early Holocene (see also Bradbury et al., 2000), with peak temperatures between 8500 and 8000 B.P., followed by cooling until 7000 B.P. The climate probably warmed coincident with increasing effective moisture between 7000 and 6000 B.P., but became cooler between 4500 and 1500 B.P.
In central Canada, at sites just north of the modern tree line, a rapid change at 5000 B.P. in 818O of aquatic cellulose and diatom species composition, coincident with a shift from tundra to forest tundra, suggests warming and an increase in effective moisture, which persisted for ca. 1000 years (MacDonald et al., 1993; Edwards et al., 1996; Pienitz et al., 1999). Isotopic data also suggest a several-hundred-year interval of particularly wet climate centered at ca. 2700 B.P. (Wolfe et al., 1996).
For the Arctic, diatom studies have been used to reconstruct changes in lacustrine productivity, chemistry, and habitat associated with changes in temperature (Douglas and Smol, 1999). On Ellesmere Island, mid-Holocene increases in diatom species associated with aquatic mosses indicate the expansion of these mosses as a result of decreased ice cover during warm intervals (Smol, 1983). Early to mid-Holocene increases in diatom concentrations also probably reflect the warmer temperatures and longer open-water periods, which allowed enhanced diatom production (Williams, 1994; Wolfe, 1996; Wolfe and Hartling, 1996). For the High Arctic, the data suggest a colder climate after ca. 4000 B.P. (Smol, 1983). For other parts of the Arctic (Wolfe and Hartling, 1996), however, Holocene changes in diatom assemblages and sediment chemistry primarily reflect variations in lake water chemistry, driven by changes in catchment vegetation and alkalinity generation within the lake, rather than the direct influence of temperature.
Patterns of late Holocene climate change have been addressed with sedimentological and microfossil studies of Arctic lakes, including some with varved sediments. Interpretations based on varve thickness for a lake on Ellesmere Island suggest a cool climate from
3000-2400 B.P., warmer conditions at 2400-1200 B.P., subsequent cooling until ca. 700 B.P., and variable conditions during the LIA interval (Lamoureux and Bradley, 1996). Analyses for the last few centuries for several Arctic lakes suggest increased sediment deposition, increased diatom concentrations, and changes in diatom species composition indicative of warming, beginning at some sites in the mid-nineteenth century and at others in the twentieth century (Douglas et al., 1994; Gajewski et al., 1997; Overpeck et al., 1997).
A synthesis of lake records for North America (Figs. 3 and 4) suggests millennial-scale patterns of lacustrine and climate change driven by changes in major boundary conditions, particularly insolation and wastage of the Laurentide ice sheet. The evidence for lowered lake levels and severe aridity at the onset of the Holocene (10,000 B.P.) in the Pacific Northwest and northern Great Basin (Fig. 3) is consistent with the higher summer insolation and a resultant intensification of the eastern Pacific subtropical high, which blocked the flow of westerly moisture (Barnosky et al., 1987). These conditions subsequently diminished after 5000 B.P. as insolation decreased. The higher frequency moisture fluctuations apparent in the late Holocene records for the northern Great Basin are likely a result of winter precipitation variability associated with changes in the mean position of the winter jet and a consequent shift in the position of westerly storm tracks and high-pressure cells (Stine, 1994).
In the Southwestern United States, the modern hy-drologic regime is influenced by both winter snow and rain, as well as by summer monsoonal precipitation. A climate model for 9000 cal. B.P. suggests that there was an upper level anticyclone over the Colorado Plateau (Kutzbach, 1987), a configuration that is presently associated with wetter summers. Winter precipitation, although reduced from its late Pleistocene maximum, was also probably higher in the early Holocene than it is today (Van Devender et al., 1987) because the winter jet stream was still displaced south of its current position. After 9000 B.P., the northerly migration of the winter jet stream resulted in decreased winter precipitation across the southwestern United States and widespread desiccation of lake basins. Some sites show pluvial phases within the mid- to late Holocene arid interval, suggesting that there were intervals of intensified summer monsoons or of winter floods caused by a southerly displacement of the North Pacific storm tracks (En-zel, 1992). Late Holocene moisture patterns appear to be spatially variable, perhaps as a result of the complex
terrain and consequent variability in precipitation patterns (Mock, 1996).
In the northern continental interior of North America, between ca. 40° and 50oN latitude, the onset of arid conditions occurred just prior to the beginning of the Holocene (10,000 B.P.) at western sites (Alberta) and in the early to mid-Holocene (9000-6500 B.P.) eastward (southeastern Saskatchewan, Manitoba, North Dakota, Minnesota, and Wisconsin) (Fig. 3). Climate modeling (Hostetler et al., 2000; COHMAP [Climate of the Holocene—Mapping of Pollen Data] Members, 1988) suggests that this contrast between western and central North America in the onset of aridity is a result of early Holocene modulation of insolation-driven aridity by the persistent Laurentide ice sheet in eastern North America, which created steep climatic gradients at the margin of the ice sheet. For North Dakota and western Minnesota, the data suggest that the driest conditions within the mid-Holocene were between 7200 and 5200 B.P. The timing of maximum aridity within the mid-Holocene interval in Saskatchewan, Manitoba, and Wisconsin is not clear, either because of the small number of sites or because interpretations are based on lithology and mineralogy, which show thresholds rather than continuous responses to limnological and climatic change. In interior regions just to the south
(South Dakota and Nebraska), which were more distant from the ice sheet, severe aridity may have occurred earlier; however, there are currently too few sites to assert this with any confidence.
Moisture in the North America continental interior derives largely from the Gulf of México, with a precipitation maximum in summer. Seasonal dry intervals are the result of strong zonal (westerly) flow and the dominance of dry Pacific or Arctic air masses, which divert Gulf moisture to the east. Extending these observations to Holocene lacustrine records suggests that there was strong zonal flow beginning in the early Holocene, increasing in strength at ca. 7000 B.P., such that areas to the east were more frequently dry. The onset of increased moisture began at varied times in the mid-Holocene in central and western Alberta, between 4700 and 4000 B.P. throughout Saskatchewan, Manitoba, North Dakota, and Minnesota, but not until ca. 3200 B.P. in eastern Wisconsin (Fig. 3). This apparent west-to-east time-transgressive pattern suggests a series of shifts in the location and shape of the planetary Rossby waves, which changed the mean position of the summer jet stream and the position of high- and low-pressure cells across the North American continent.
The overall pattern of mid-Holocene aridity throughout western North America is consistent with a
Clh P3 b
Was this article helpful?