Research Article

Out of the Tropics: The Pacific, Great Basin Lakes, and Late Pleistocene Water Cycle in the Western United States

See allHide authors and affiliations

Science  28 Sep 2012:
Vol. 337, Issue 6102, pp. 1629-1633
DOI: 10.1126/science.1218390

Abstract

The water cycle in the western United States changed dramatically over glacial cycles. In the past 20,000 years, higher precipitation caused desert lakes to form which have since dried out. Higher glacial precipitation has been hypothesized to result from a southward shift of Pacific winter storm tracks. We compared Pacific Ocean data to lake levels from the interior west and found that Great Basin lake high stands are older than coastal wet periods at the same latitude. Westerly storms were not the source of high precipitation. Instead, air masses from the tropical Pacific were transported northward, bringing more precipitation into the Great Basin when coastal California was still dry. The changing climate during the deglaciation altered precipitation source regions and strongly affected the regional water cycle.

The Great Basin and southwestern United States were wetter during the last glacial period than today. From Oregon to West Texas, many modern desert valleys were once extensive lakes or wetlands. Largest and best known are Lake Bonneville, whose remnant is the Great Salt Lake, and Lake Lahontan in Nevada. Many Great Basin valleys have multiple ancient shorelines on valley slopes and exposed subaqueous sedimentary structures that together indicate a complex history of changes in precipitation-evaporation.

After radiocarbon dating became established, lake elevations were used to construct a regional late-Pleistocene climate history (1) for Lakes Lahontan and Bonneville and for many of the other Pleistocene lakes in the western United States (supplementary text). Lake level reconstructions combined with modeling and paleotemperatures could then be used to study how the water cycle changed. Lake levels in closed Great Basin drainages reflect the net water balance between precipitation and temperature-driven evaporation from the lake’s surface. The high lake levels of Lake Lahontan were mostly caused by higher precipitation and runoff, with lesser change caused by cooler temperatures (2, 3). The change in evaporation associated with 7°C cooling was sufficient to raise Lake Lahontan level by ~30 m out of the 180-m peak highstand. Higher precipitation (~140%) and runoff (~280%) are needed to cause peak lake levels. Once formed, the lakes themselves helped to maintain higher water levels by being a source of precipitation to their own watershed (4).

There was major water cycle variability during the deglaciation in the Great Basin (5). A major dry interval between 17.5 and 16 thousand years ago (ka) is proposed to separate an early wet interval from a major wet interval between ~16 and 14.5 ka (5). Our study examines both the timing and geographic distribution of wet intervals to determine the linkage between the Pacific Ocean sources of atmospheric water and the change in the water cycle.

New insights about the water cycle are possible because of a decade’s worth of study of sediments from drill sites along the California margin (Fig. 1) (6, 7). Sea surface temperature (SST) reconstructions and pollen data from multiple marine localities record changes in both coastal climate conditions and changes in the North Pacific subtropical gyre (8). Comparisons between climates in the coastal region and in the Great Basin give a geographic perspective about how the water cycle has changed during this major climate change.

Fig. 1

Alkenone estimates of mean annual SST for California margin time slices (8, 1520, 39), plus one radiolarian SST estimate at Site 1019 for the 15- to 17-ka time slice (22). (A) Modern era, 0 to 1 ka. (B) North Coastal wer period, 10 to 12 ka. (C) Lahontan highstand, 15 to 17 ka. (D) LGM, 19 to 21 ka. Sites marked with red open circles also have pollen assemblage data. Sites marked with plus signs are from gravity and piston cores (16); other data are from ODP drill sites, as marked in (A). Warmest SST off southern California and strongest coastal SST gradient occurred in the 15- to 17-ka period.

The “shift of the westerlies” hypothesis. It has been proposed that elevated lake levels within the Great Basin mark southward shifts of Pacific winter storm tracks during glacials, causing higher regional precipitation (9). According to this hypothesis, the jet stream split because of Cordilleran-Laurentide Ice Sheets, with a weak branch going north of the ice sheet and a strong branch going across southern California. The southern jet carried storms inland to fill the Great Basin lakes (9).

The modeling that helped define this hypothesis has shortcomings. The resolution of the 1980s vintage CCM0 model (9) is 4.4° in latitude and 7.5° in longitude, so shifts in storm tracks are not well constrained. SST was based on the early Climate: Long range Investigation, Mapping, and Prediction (CLIMAP) last glacial maximum (LGM) reconstruction (10). The SST fields during deglaciation were were not based on data but were interpolated between the LGM CLIMAP SST (10) and modern SST. New high-resolution model results based on the LGM CLIMAP SST (10) show LGM results similar to Cooperative Holocene Mapping Project (COHMAP), but with important differences (9). The new studies have higher numbers of Pacific storms than do modern conditions such as those of COHMAP (9), but with a weak “southern” winter jet stream and storm tracks that tend to be diverted northward into the Alaska Gyre (11, 12). New multiproxy LGM SST reconstructions have identified errors in the CLIMAP SST reconstruction, which may cause further revisions to LGM climate patterns (13). In the northeast Pacific, alkenone estimates of LGM SST along the California margin (Fig. 1) delineate an 800-km southward shift of the 8°C isotherm compared with CLIMAP. CLIMAP tropical SST appears to be too warm compared with the Multiproxy Approach for the Reconstruction of the Glacial Ocean Surface (MARGO) reconstruction (13). When cooler LGM tropical SST is input to the National Center for Atmospheric Research (NCAR) CCM3 model, temperate Pacific storm energy is weakened compared with simulations that use CLIMAP-based SST (14).

Because Pacific SST maps for each of the 3-thousand-year (ky) time slices in the COHMAP (9) reconstruction are interpolations, better information is needed to understand the deglaciation in the western United States. Better lake hydrographs in the Great Basin and better information about both eastern Pacific SST (8, 1520) and coastal vegetation (8, 2125) sharpen the temporal and spatial pattern of wet intervals to compare with the model of shifting westerlies. The new data suggest that a strong tropical connection at the LGM and during the deglaciation was important to western U.S. precipitation, rather than southward shifts of North Pacific storms.

Pleistocene wet intervals along the U.S. Pacific margin. Timing of wet periods for coastal California are based on the pollen time series. Distinctive pollen assemblages in coastal cores are associated with changes in temperature and/or precipitation in coastal river drainages and can be correlated directly to marine variability (8, 26). As climate cooled, cool temperate coastal forests typically expanded southward. Higher relative precipitation typically causes an increase in forest cover relative to grassland or shrubland, or an increase in wet-adapted versus dry-adapted tree varieties (21, 27). Southern California near Santa Barbara, for example, remained arid throughout the past 160 ky, but wetter intervals are marked by expansion of an upland pine forest (Fig. 2) (21, 28). In central California, decreases in pollen from herbs and shrubs and increases in tree pollen—alder, coastal redwood, and oak—mark wet intervals (8). In northern California, both temperature and precipitation affected the record. Alder, oak, and coastal redwood retreated southward at the expense of cool-adapted Sitka spruce and western hemlock as temperatures cooled (22, 23). We used the peak in alder pollen found during the deglaciation to define maximum precipitation because alder responds rapidly to changing conditions and was well within its temperature tolerance, even at the glacial maximum (19).

Fig. 2

Pollen records from ODP sites along the California margin used to measure SST and change in vegetation. Green bands denote coastal wet periods and show the early wet period in the south. The blue time series is of pollen from more wet-adapted vegetation, and brown is that from dry-adapted vegetation. (Top) ODP Site 1019, near Crescent City California (42°N) joining two age models: top 16 ka (19), and 16 to 75 ka (40). (Middle) Site 1018, west of Santa Cruz (37°N) (8). (Bottom) Site 893 in the Santa Barbara Basin (34°N) (24).

The southern California wet period [Ocean Drilling Program (ODP) Site 893, 34.3°N] occurs ~5 ky earlier than those in central and northern California (Fig. 2). Maximum wet conditions occurred between 17.4 and 14.2 ka, with much of the glacial interval wetter than modern conditions (21, 24). Central California (ODP Site 1018, 37°N) had a wet interval that was confined to the interval between 12.5 and 4.5 ka, ~5 ky after the Santa Barbara wet interval (8). Similar timing was found in a speleothem record from Moaning Cave at ~38°N in the western foothills of the Sierra Nevada (29). Brief wet intervals were found at about 14.1 and 13.3 ka, but an extended wet interval began ~12.5 ka and lasted until 9.4 ka.

In northern California (ODP Site 1019, 41.7°N), the wet interval began ~14 ka, peaked ~12 ka—coincident with rapid SST warming along the California-Oregon margin (15, 19)—and ended ~10 ka (19), slightly earlier than the wet interval in central California. The beginning of the wet interval is marked by an abrupt switch from dry, cool conditions inferred by high levels of artemisia and pine pollen to wetter conditions inferred by high levels of alder pollen.

Timing of lake level change in the Great Basin. If westerly storm tracks first moved southward then subsequently returned to the north, latitude bands should have the same timing for the coastal wet intervals and lake level highstands. By this hypothesis, as the westerlies shifted southward, there should have been pre-LGM drying in the northern Great Basin >40° N, concurrent with high lake levels south of 35° N. As glaciers waned, storms should have swept northward, briefly causing coastal wet intervals and Great Basin lakes in the central latitudes to be high.

We group high lake stands into three major intervals since the LGM: >17 ka, 14 to 17 ka, and <14 ka (Figs. 3 and 4 and table S1). Highest highstands in the 17- to 14-ka range are characteristic of the central Great Basin from ~37°N to 42°N, including Lakes Lahontan and Bonneville. The southern Great Basin, south of 37°N, typically had early lake highstands before 17 ka, or roughly similar highstands in the periods between 17 to 14 ka and >17 ka. The northern Great Basin, north of 42°N, had early maximum highstands ~21 ka (30) but with additional highstands ~13 to 10 ka. Cold temperatures, especially ice formation, may have been the primary factor for older lake highstands in the northern Great Basin, with moderately higher precipitation being a secondary factor (30).

Fig. 3

Lake level records in thousands of calendar years before present from (A) Lake Lahontan, Nevada (5, 41, 42) and (B) Lake Estancia, New Mexico (43) compared with the Site 1018 wet period (gray shading). Site 1018, being west-southwest of Lake Lahontan, should respond to any substantial precipitation that came directly across the Sierra Nevada mountains to the lake. The two lake level records show that the early wet interval was found in both the central Great Basin and New Mexico, and that the 17- to 14-ka wet interval was magnified in the central Great Basin.

Fig. 4

Timing of lake level changes and coastal wet periods in western North America (table S1). Multiple wet periods are found over the LGM and deglacial intervals. The largest events typically occur earliest in the southwest and later to the north. For a given latitude, the coastal region has wet periods persistently later than the interior western United States. Arrows represent possible storm tracks.

Tests of the westerly storm track hypothesis. There is little observational evidence to indicate that westerly storm tracks shifted southward at the LGM. First, the timing of wet periods is not the same along latitudinal bands. The highest lake levels in Lakes Bonneville and Lahontan (between 37° and 42°N) occurred between 17 and 14.5 ka, but in the coastal region west of the Sierra and north of 36°N, there were no wet periods then (Fig. 4). Vegetation associated with higher precipitation at Sites 1018 and 1019 (37° and 41.7°N) appears after 12.5 ka (Fig. 2) (8, 19, 22). Only the wet period in the Santa Barbara region to the south (Site 893, 34.1°N) matches with the 17- to 14.5-ka Lake Lahontan-Bonneville highstand.

Two wet periods should be found in central California if westerly storm tracks were displaced: one before the LGM as the Northern Hemisphere glaciers advanced and the storm tracks moved to the south, and a second one as the storm tracks moved northward again. However, at Site 1018 arid conditions persisted for the entire period between ~35 and 12.5 ka (8). Only in southern California (Site 893) is there evidence for moderately higher precipitation before the LGM, between 45 and 25 ka (24).

In the western Pacific, the westerly belt did not shift dramatically southward, implying little or no shift in the east near North America and little southward compression of the North Pacific subtropical gyre. Rea and Leinen (31) found that the amount of Asian dust carried to the oceans changed by a factor >1.5 over the past 30 ky, but the position of maximum dust flux marking the westerlies remained between 38 and 40°N. Yamamoto et al. (32) delineated the northern margin of the Kuroshio (the western Pacific boundary current) for 0 to 25 ka by SST and estimated that the maximum glacial southward compression of the gyre to be ~1.4°.

California coastal SST patterns (Fig. 1) do not fit with south-displaced winter storm tracks at the LGM. An enlarged subarctic gyre in association with a southward displacement of the subtropical gyre should cool SST to the south along the coast of Baja California. However, between 25 and 12.5 ka, a steep SST gradient was present between San Francisco and Santa Barbara [temperature change (ΔT) ≈ 5°] (18), compared with 0.9° during the Holocene (Fig. 1). Southern California SST warmed earlier than SST to the north (~25 ka), which is indicative of weak flow in the California Current (18). The warm Pleistocene SST pattern off southern California and large SST gradient off central California implies southerly winds to southern California, not westerlies.

Last, the LGM vegetation assemblage in southern California was not one that should have appeared with increased westerly storms. In the Holocene, coastal redwood (Sequoia) can only be found to the north of 36° N, where temperatures are cooler than in southern California and water is more available. At the LGM, coastal cooling was sufficient for redwoods to expand southward, but they did not because Southern California was still arid, just not as arid as in the Holocene (21).

“Out of the tropics.” The different timing of wet intervals on either side of the Sierra Nevada Mountains implies major water transport into the Great Basin from the south during the deglaciation (Fig. 4). North of 36°N, coastal climates were cooler and drier when Lake Lahontan and Bonneville reached their highest levels. The Lake Lahontan/Bonneville highstands between 17 and 14.5 ka were unsupported by additional precipitation travelling east across the Sierra Nevada. Even in high-resolution climate model experiments that show a “split” jet-stream and stronger Pacific cyclogenesis in response to LGM boundary conditions and CLIMAP-based SSTs, the enhancement of westerly storm tracks is confined to the subarctic North Pacific (12).

The temporal and spatial distribution of lake highstands fits with warm wet air masses moving north into the Great Basin from the eastern tropical Pacific. In the Holocene, coastal southern California and northern Baja California have persistent upwelling associated with northerly winds around the northeast Pacific subtropical high-pressure regime that cools SST and blocks tropical air masses (33). The warm ocean conditions found south of 34°N from ~25 ka until ~10 ka (Fig. 1) (18) are an indication that the conditions that drove upwelling were suppressed and imply that the northeast Pacific high was also less effective at blocking southern storms. Evidence for this scenario can be found by comparing the Devils Hole 18O time series to coastal SST records. Maximum tropical source water, as indicated by “warm” δ18O, is found at 17 ka (34), when all SST paleotemperatures in the California Borderlands are as warm or warmer than modern values (Fig. 1).

Seasonality of deglacial wet intervals. During the Holocene, southern-sourced precipitation annually appears with the summer North American monsoon, which is strongest in northwest Mexico, New Mexico, and Arizona (35). The deglacial tropical water source may well result from an enhancement of this summer precipitation pattern. Coastal precipitation and SST patterns imply that the northeast Pacific high may have been displaced or weakened relative to modern conditions, allowing wet Pacific tropical air more access into the Great Basin at its southern end. A weaker northeast Pacific high should have helped to enhance the summer precipitation by weakening winds that might block access of tropical Pacific air masses to the Great Basin. Enhanced contribution of moisture from the Gulf of Mexico associated with the summer monsoon could also increase moisture in the Great Basin without triggering a wet interval along the Pacific coast, and is therefore also consistent with the observed proxy data.

Wet conditions peaked earlier near Mexico relative to the rest of the Great Basin. Somewhat higher lake levels before 17 ka were found throughout much of the Great Basin, but highest lake levels in the south typically occurred before the 17- to 14-ka period of highest lake levels in the central Great Basin (see supplementary text). In the 17- to 14-ka time interval, coastal California did not receive higher levels of precipitation. Better penetration of tropical air into the central Great Basin through southern California and the Gulf of California, perhaps drawn by a stronger summer monsoon, would create a precipitation pattern such as the one observed. The time of maximum lake level in Lake Lahontan occurs when there is evidence of strong easterly winds in Oregon (36, 37) peaking at ~15 ka. The easterlies are evidence of the regular development of a low-pressure regime to the south of Oregon, perhaps associated with the summer monsoon. In support of the summer monsoon hypothesis, McClymont et al. (38) proposed that 15 to 17 ka represented a period of higher precipitation around the Gulf of California and used modeling to suggest that there was a strengthened summer monsoon at this time.

With the data now in existence, it is impossible to determine whether summer precipitation was more enhanced than winter precipitation between 17 and 14 ka. However, if winter storms were the major precipitation source, it is difficult to understand why coastal California remained dry.

Conclusions. Climate change associated with the deglaciation caused profound alterations of the western U.S. water cycle. Large lakes formed in the central Great Basin, mainly caused by increased precipitation. The timing of highest lake levels in the Great Basin are not synchronous, but have a progression from south to north that does not coincide with the northward progression of wet intervals along the U.S. west coast. Instead, highest lake levels for Lakes Lahontan and Bonneville match the timing of a wet interval found in southern California, but not with wet intervals on similar latitude bands. The evidence suggests that precipitation in the glacial western United States originated from the tropical eastern Pacific, perhaps via stronger spring/summer precipitation fed by tropical air masses rather than higher numbers of westerly winter storms.

Supplementary Materials

www.sciencemag.org/cgi/content/full/337/6102/1629/DC1

Supplementary Text

Table S1

References (4480)

References and Notes

  1. Acknowledgments: We thank the Ocean Drilling Program and its successor program, IODP, for collecting and archiving the drill cores used. We also thank M. Reheis for her insights in a thorough internal U.S. Geological Surbey review. The U.S. National Science Foundation supported this research through grants EAR-0450211 and OCE-9809438.
View Abstract

Navigate This Article