Reconciliation of the Devils Hole climate record with orbital forcing

See allHide authors and affiliations

Science  08 Jan 2016:
Vol. 351, Issue 6269, pp. 165-168
DOI: 10.1126/science.aad4132

The difference is all in the water

Glacial cycles are in part controlled by the pattern of incident solar energy determined by the geometry of Earth's orbit around the Sun. The classic record of the penultimate deglaciation from Devils Hole, Nevada, did not reconcile the presumption of so-called orbital forcing, however, suggesting that deglaciation began ~10,000 years too early. Moseley et al. present analyses of a new set of data from Devils Hole that show that the deglaciation indeed occurred at the time expected on the basis of orbital forcing. The age offset displayed by the older samples apparently was caused by interaction with groundwater, which preferentially affected the deeper original samples but not the new shallower samples.

Science, this issue p. 165


The driving force behind Quaternary glacial-interglacial cycles and much associated climate change is widely considered to be orbital forcing. However, previous versions of the iconic Devils Hole (Nevada) subaqueous calcite record exhibit shifts to interglacial values ~10,000 years before orbitally forced ice age terminations, and interglacial durations ~10,000 years longer than other estimates. Our measurements from Devils Hole 2 replicate virtually all aspects of the past 204,000 years of earlier records, except for the timing during terminations, and they lower the age of the record near Termination II by ~8000 years, removing both ~10,000-year anomalies. The shift to interglacial values now broadly coincides with the rise in boreal summer insolation, the marine termination, and the rise in atmospheric CO2, which is consistent with mechanisms ultimately tied to orbital forcing.

Changes to Earth’s orbital configuration relative to the Sun, known as the Milankovitch hypothesis, astronomical theory, or orbital forcing, have long been considered the leading theory for the primary mechanism driving Quaternary glacial-interglacial cycles (13) and associated climate change. The hypothesis is supported by a huge array of evidence from paleoclimate records across the globe, which show that major shifts in climate took place throughout the Quaternary on orbital time scales (2, 4, 5). One particular seminal paleoclimate record, however, the ~500-thousand-year (ky) Devils Hole record from Nevada (fig. S1) (36°25’N, 116°17’W; 719 m above sea level), has challenged this hypothesis for nearly three decades (68). Of note is the controversy (4, 911) surrounding the timing of events at the end of the penultimate glacial period [Termination II (TII)]. At issue is whether TII preceded the rise in boreal insolation, in which case TII could not result directly from orbital forcing. The current Devils Hole chronology places the TII shift to interglacial values ~10 ky before the rise in boreal summer insolation (Fig. 1) and the duration of the last interglacial period to almost double its equivalent in other records, both of which are inconsistent with a straightforward explanation in terms of orbital forcing.

Fig. 1 Comparison of Devils Hole records with other paleoclimate archives.

(A) 230Th ages with 2σ uncertainties for DHC2-8 (light blue) (23), DHC2-3 (light-mid blue) (23), DH-11 (light green) (7, 8, 23), DH-2 (dark green) (this study), DH2-E (mid-dark blue) (this study), and DH2-D (dark blue) (this study). VPBD, Vienna Pee Dee belemnite standard. (B) δ18O records for samples in (A) [same color codes and references as in (A)] and 65°N July insolation (dark gray) (32). (C) δ13C records for DH-2, DH2-E, and DH2-D [same color codes as in (A)]. (D) Growth rate of samples from this study with [same color codes as in (A)]. (E) Stacked global benthic δ18O records (5). (F) Antarctica CO2 composite curve (3840). (G) 230Th ages on mammillary calcite with 2σ uncertainty (24). Pre-folia, red; post-folia, black; pre-growth hiatus, blue. Arrows indicate the direction of water-table change.

The controversial findings generated discussion on (i) dating accuracy (1216) and resolution (17, 18); (ii) the possibility that the record represents a regional hydrological signal (1921); and (iii) phase leads and lags between proxy records (7, 18, 22). Deviation from an orbital pacing was highlighted in the original Devils Hole record (6), with a chronology based on alpha-counting 230Th and 234U/238U dating of sample DH-2. This record was later replicated with higher-precision thermal ionization mass spectrometric ages on sample DH-11 and extended deeper in time (7, 8). A mechanism was immediately suggested whereby water-sourced 230Th incorporated during calcite growth could lead to artificially old ages (12, 14). However, measurements on the outer surfaces of DH-11 and DH-2, both of which had been submerged for an extended period of time, showed that this process operated only at low efficiency, under the presumption that the adsorption of 230Th was irreversible (15, 16). Several years later, concordant 231Pa and 230Th ages were obtained on an ~80-ky-old subsample of DH-11 (13), consistent with an accurate chronology for this portion of the record. In the following two decades, the accuracy of the DH-11 chronology remained unchallenged.

More recently, marine cores off the west coast of the Americas recorded changes in temperature before marine δ18O terminations (21). These observations were interpreted in terms of changing surface currents resulting in changing temperature. If Devils Hole recorded these early changes in regional temperature, the difference between the Devils Hole record and the marine δ18O record would be explained.

The record was then extended to the Holocene, using samples DHC2-3 and DHC2-8 (23), both collected from Devils Hole 2 (24), a cave similar to Devils Hole and 200 m away from it. The portion near TI could also be explained in terms of changing surface currents in the eastern Pacific (21).

Despite the apparent resolution, the controversy has subsequently been revived. Great Basin dripstone (fig. S1) records exhibit shifts to interglacial values around the time of marine TII (25, 26) and TI (27) and therefore replicate neither of the early shifts documented at Devils Hole (7, 8, 23). Thus, the controversy currently centers on major chronological discrepancies between the subaerially formed Great Basin dripstone and subaqueously formed Devils Hole records.

To address this controversy, we tested the reproducibility of the Devils Hole record by reanalyzing DH-2 (fig. S2) (6) from Devils Hole proper and by analysis of four newly drilled cores (fig. S2) from Devils Hole 2. The previously studied samples were all collected at substantial depth below the current water table (DH-2 = –21 m; DH-11 = –30 m; DHC2-3 = –25 m; DHC2-8 = –60 m), whereas the new cores from Devils Hole 2 were collected from above the water table (DH2-D = +2.1 m; DH2-E = +0.8 m; DH2-G = +4.5 m; DH2-J = +5.5 m) (fig. S3).

Considering our data, the results confirm the overwhelming majority of the features of the original studies (68, 23), verifying the original analytical results and the reliability with which the groundwater system and the calcite precipitated from it record climate. Portions of the record have now been replicated four times with similar results. The general character and range of δ18O variations, and the U isotopic compositions and concentration, are all similar to those in the original reports (6, 8, 23) (Fig. 1 and table S1). Further, large portions of our chronology replicate the original chronologies (8, 23). Between 28 to 112 thousand years ago (ka), DH2-D replicates at significantly higher resolution the timing of the previous chronology (8, 23) (Fig. 1), while further adding support for the timing of the isotopic maximum that occurs at about the time of marine isotope stage 5a, with an age of 82.5 ± 0.7 ka, replicating the original 230Th age of 80.6 ± 2.5 ka (7, 8) and the 231Pa age 82.5 ± 2.8 ka (13).

Considering our data sample by sample, starting with our deepest sample (DH-2: –21 m), we find that within quoted uncertainties, the newly analyzed portion of DH-2 replicates both the original DH-2 (6) and DH-11 records (7, 8), thus verifying the original analytical results, including measurement of the U and Th isotopes used in calculating ages. The first hint of an age discrepancy occurs at about the time of TII, because the shift in δ18O values in our DH-2 record is nominally later than the shift in DH-11 (7, 8) (Figs. 1 and 2).

Fig. 2 Timing of TII and the last interglacial in speleothem records.

(A) DH-2 (dark green) (this study) and DH-11 (light green) (7, 8, 23). (B) DH2-D (dark blue) (this study) and DH2-E (mid-blue) (this study). (C) Great Basin composite dripstone record (“Leviathan chronology,” pink) (30); Lehman Cave (red) (26) (dark red) (25); and 65°N July insolation (gray) (32). The midpoints of transitions are indicated with x’s and vertical dashed lines.

Moreover, comparison of our two highest-elevation δ18O records (DH2-D: +2.1; DH2-E: +0.8 m) to those from deeper cores yields large discrepancies in chronology, well outside of analytical error (Figs. 1 and 3 and fig. S4). These discrepancies are largely confined to and clearly observed at times that correspond to each of the last two terminations, but are best resolved during TII because calcite deposition rates are higher than during TI (Fig. 1 and fig. S5). The 230Th chronology of the shift to interglacial values is systematically younger for cores collected at higher elevation (Figs. 1 and 2), with the shallowest core recording a time of 132.2 ± 1.5 ka (24) for the midpoint of the shift to interglacial values, as compared with a value of 142 ± 3 ka (18) for the deepest sample for which TII data have been reported. The shift toward glacial values at the end of the last interglacial is similar in all records (Figs. 1 and 2). Therefore, the duration of the last interglacial δ18O peak as recorded in the samples also shifts systematically with sample elevation, with a last interglacial duration of 16.1 ky (measured from the midpoint of rise to the midpoint of fall) recorded in DH2-D as compared to a DH-11 duration of ~22 ky (table S2) (7, 8, 18). Considering just the portion of the peak that records the highest δ18O values, DH2-D records a 6-ky duration (from 127 to 121 ka), which is close to half the duration recorded in DH-11 (7, 8, 18). Thus, the shallowest core records the latest shift to interglacial conditions and the shortest duration for the last interglacial δ18O peak (Figs. 1 and 2).

Fig. 3 TI in Devils Hole and Great Basin paleoclimate records.

(A) 230Th ages with 2σ uncertainties: DHC2-8 (light blue) (23), DHC2-3 (light-mid blue) (23), DH2-E (mid blue) (this study), and DH2-D (dark blue) (this study). (B) δ18O records for samples in (A) with same color codes and 65°N July insolation (dark gray) (32). (C) δ18O Great Basin composite dripstone record (“Leviathan chronology”) (30) and 65°N July insolation (dark gray) (32).

The explanation for the relationship between apparent age and sample elevation probably lies in 230Th incorporated in the growing calcite from the water, as raised previously (12, 14) and argued to be negligible (15, 16). The discussion at that time assumed irreversible adsorption of 230Th onto the walls (12, 1416). In light of our data and considering advanced understanding in the distribution of 230Th in seawater (28, 29), we consider irreversible exchange unlikely. Indeed, vertical profiles through the ocean water column show that both particulate and dissolved 230Th increase in concentration with depth, indicating reversible exchange of 230Th between water and particulate matter (28). Furthermore, sharp increases in dissolved 230Th in the upper 200 m of the ocean have been demonstrated (29). The systematic increase of 230Th ages with depth (Figs. 1, 2, and 4) in Devils Hole may thus be the result of increasing concentrations of 230Th down the water column. The strong correlation (coefficient of determination, R2 = 0.97) between the age of the TII midpoint in each sample versus depth (Fig. 4) (24) is consistent with a reversible exchange–generated increase of 230Th with depth in the TII Devils Hole water column.

Fig. 4 The age of the TII δ18O midpoint of –15.7 per mil versus depth below the water table [+5.5 m above present elevation (24)] for each sample.

DH-11 (light green) (7, 8, 23), DH-2 (dark green) (this study), DH2-E (mid blue) (this study), and DH2-D (dark blue) (this study). Uncertainties are 2σ. R2, coefficient of determination.

The appearance of age anomalies during terminations may result from several factors. First, the water table was higher than at present during both TI and TII, before it progressively declined during the respective following interglacials (Fig. 1), in agreement with progressively drier conditions during interglacials (30, 31). One would therefore expect that for a given elevation, the water 230Th concentration would have increased during the pluvial periods associated with terminations and then decreased over the course of the proceeding interglacial. Second, lower depositional rates during times of the anomalies (Fig. 1 and fig. S5) may play a role in the incorporation of excess 230Th onto the calcite (12). Finally, wetter conditions associated with higher water tables could be associated with increased groundwater flow rates and 230Th fluxes into the open fractures.

Regardless of the details of the mechanism, if progressively higher levels of 230Th at depth are responsible for the anomalies, then the shallowest core, which records the youngest ages, has the most accurate chronology. Thus, we consider the chronology of the DH2-D δ18O record to be the closest to the true age. Because the DH2-D record is in close agreement with the timing of δ18O shifts in Great Basin dripstone records [Figs. 2 and 3; particularly considering an up to 2-ky groundwater transit time in the Devils Hole system (23)], we conclude that DH2-D has negligible initial 230Th and provides an accurate chronology.

In contrast to DH-11 (7, 8), the timing of TII and the duration of the last interglacial δ18O peak (either midpoint to midpoint or peak value duration) in the DH2-D record are all consistent with orbital forcing. The DH2-D δ18O shift to interglacial values is broadly coincident with the rise in boreal summer insolation (32), the rise in atmospheric CO2 (33), the marine termination (34), and the TII Weak Monsoon Interval (4). The mechanism for 18O enrichment in precipitation could be increasing temperature (68, 21, 23), in which case the temperature rise could be caused by the rise in global atmospheric CO2, which has been closely tied to orbital forcing (4). However, it is also likely that changes in the proportion of summer (high-δ18O today) to winter (low-δ18O today) meteoric precipitation (i.e., changes in seasonality) played a role. If so, the low δ18O values before the glacial-interglacial transition could be partly explained by a larger proportion of low-δ18O cool season rainfall associated with pluvial conditions. The pluvial conditions in turn, could be caused by the splitting of the jet stream (35) around the Laurentide Ice Sheet to the north. In this case, the proportion of (low-δ18O) winter season moisture would diminish as the ice sheet melted, as a result of increased boreal summer insolation. The rise in insolation could also trigger (high-δ18O summer) North American monsoon-like rainfall in the region, increasing the mean annual δ18O of precipitation. We conclude that some combination of these orbitally based processes contributed to the observed shift in δ18O. Temperature changes of Devils Hole water have been small (36), so that the temperature-dependent calcite-water fractionation would have had little effect on calcite δ18O. Also, the rainfall seasonality mechanisms do not directly relate to regional temperature and could plausibly explain the absence of a Devils Hole “lead” that is analogous to the lead in temperature observed in the marine cores (21).

Finally, in our (DH2-D) chronology, the prominent low in δ13C at the end of TII is shifted to younger values by about 7 ky (Fig. 1) relative to the DH-2 and DH-11 (37) records, thus coinciding with the boreal summer insolation peak. Because this could be a time of regional warmth and relatively high warm-season rainfall, our timing supports the idea (37) that such δ13C lows are caused by more extensive vegetation cover and productivity in the source region for the aquifer.

Our chronologies from Devils Hole have demonstrated that there is a systematic offset in the age of calcite deposited at increasing depths in these open fractures across glacial terminations, thus helping to solve one of the great paleoclimate enigmas of the past three decades.

Supplementary Materials

Materials and Methods

Supplementary Text

Figs. S1 to S6

Tables S1 and S2

References (4159)


  1. Materials and methods are available as supplementary materials on Science Online.
Acknowledgments: This work was supported by the Austrian Science Fund (FWF) project no. FP263050 to C.S. and in part by NSF grants 1103403 to R.L.E. and H.C., 1103320 to R.L.E., and NSFC grant 41230524 to H.C. This research was conducted under research permit numbers DEVA-2010-SCI-0004 and DEVA-2015-SCI-0006 issued by Death Valley National Park. We thank I. J. Winograd for providing the study with sample DH-2, M. Wimmer for preparation and measurement of the stable isotopes, K. Wilson and R. Freeze for assistance in the field, and M. Cross for discussion of Great Basin paleoclimate. Data can be downloaded from the National Oceanic and Atmospheric Administration’s National Centers for Environmental Information (
View Abstract

Stay Connected to Science

Navigate This Article