Time Scales of Critical Events Around the Cretaceous-Paleogene Boundary

See allHide authors and affiliations

Science  08 Feb 2013:
Vol. 339, Issue 6120, pp. 684-687
DOI: 10.1126/science.1230492


Mass extinctions manifest in Earth's geologic record were turning points in biotic evolution. We present 40Ar/39Ar data that establish synchrony between the Cretaceous-Paleogene boundary and associated mass extinctions with the Chicxulub bolide impact to within 32,000 years. Perturbation of the atmospheric carbon cycle at the boundary likely lasted less than 5000 years, exhibiting a recovery time scale two to three orders of magnitude shorter than that of the major ocean basins. Low-diversity mammalian fauna in the western Williston Basin persisted for as little as 20,000 years after the impact. The Chicxulub impact likely triggered a state shift of ecosystems already under near-critical stress.

The mass extinction at the boundary (KPB) between the Cretaceous and Paleogene periods, ~66 million years ago (Ma), likely involved the catastrophic effects of a bolide impact (1), although other factors may have played an important role (25). To a large extent, ambiguity between the possible causes stems from inadequate age resolution of relevant events near KPB time. Existing geochronologic data surrounding the linkage between the KPB and the Chicxulub structure in the northern Yucatán Peninsula of Mexico actually exclude synchrony, indicating that the Chicxulub impact and cogenetic impact melt droplets, termed "tektites" (68), postdated the KPB by 183 ± 65 (9) thousand years (ky) and 181 ± 71 ky, respectively (see supplementary materials). In contrast, some data suggest that the Chicxulub impact predated the KPB by several hundred thousand years, and that discrete tektite-bearing horizons in the Gulf of Mexico region were derived from multiple impact events (10).

We acquired high-precision 40Ar/39Ar data to clarify these temporal relationships and thereby facilitate a clearer sequencing of events associated with the KPB extinctions and subsequent ecosystem recovery. We analyzed multiple samples of the tektites to refine the age of the Chicxulub impact, and of bentonites (altered volcanic ashes; Fig. 1) clearly associated with the KPB to test for synchrony of the boundary with the impact. 40Ar/39Ar ages (Figs. 2 and 3) were determined (see supplementary materials) by incremental heating of 14 tektites from Beloc, Haiti, giving a weighted mean age of 66.032 ± 0.058/0.072 Ma (11) that is indistinguishable from that determined by previous studies (12, 13) when normalized to the same calibration. Combining all data yields an age of 66.038 ± 0.025/0.049 Ma for the tektites.

Fig. 1

Stratigraphic sections in the Hell Creek area of northeastern Montana (inset) showing positions of dated bentonites in relation to Ir anomalies (16, 41) and carbon isotope records (24) from the same sections. The two sections have different vertical scales; thin dotted line connects horizons at 1 m above the Ir anomaly in the two sections. Ages shown are from 40Ar/39Ar analysis of sanidine except one (U/Pb) from U-Pb analysis of zircon. Age uncertainties (in parentheses) refer to last significant figures shown and include analytical sources only. White dots labeled P and K on both sections show the lowest occurrence of Paleocene pollen and the highest occurrence of Cretaceous pollen, respectively (15).

Fig. 2

Summary of single-crystal geochronology results for volcanic ashes whose stratigraphic relations are shown in Fig. 1. Individual ages are shown in ranked order with 1σ analytical uncertainty limits. Samples interpreted as xenocrysts are shown in gray and are excluded from age calculations. Uncertainty limits for the weighted mean age for each sample are shown by red lines. U-Pb results for zircon crystals (A) and 40Ar/39Ar results for sanidine (B) from an ash in the Hauso Flats Z (HFZ) coal, 18 m above the KPB, yield indistinguishable results. Two ashes in the Z-coal in the Hell Creek marina road section (C and D) yield ages consistent with stratigraphic order although they are mutually indistinguishable at 68% confidence. Four independent data sets, from three irradiations and two labs, for the IrZ coal bentonite (E) yield consistent results.

Fig. 3

Summary of incremental heating 40Ar/39Ar age (lower panels) and Ca-K composition data (upper panels) for (A) fourteen tektites from Beloc, Haiti, and multgrained feldspar samples from the IrZ (B) and Z2 (C) bentonites shown in Fig. 1.

We also performed 40Ar/39Ar dating on sanidine separated from four bentonites in three distinct coal beds within two widely separated stratigraphic sections in the Hell Creek region of northeastern Montana. Extensive studies in this region have documented faunal, floral, and chemostratigraphic aspects of latest Cretaceous through early Paleogene terrestrial strata. Both sections contain well-documented Ir anomalies coincident with the biostratigraphically defined KPB (Fig. 1). In the Hauso Flats section, we analyzed samples from two localities ~200 m apart of a bentonite from the IrZ coal, located stratigraphically only a few centimeters above the horizon yielding the largest iridium anomaly [up to 11.7 parts per billion (ppb) at the nearby Herpijunk locality; (14)] reported from this area and 5 cm above the highest occurrence of Cretaceous pollen in the section (15). All of our data combined yield a weighted mean age of 66.043 ± 0.011/0.043 Ma. A bentonite from the Hauso Flats Z (HFZ) coal, 18 m stratigraphically above the IrZ coal, yielded an age of 65.990 ± 0.032/0.053 Ma. Isotope dilution–thermal ionization mass spectrometry (ID-TIMS) U-Pb analyses of 15 chemically abraded zircons from the same HFZ coal bentonite yielded a weighted mean age of 65.988 ± 0.074 Ma, in agreement with the 40Ar/39Ar results. In the Hell Creek Marina road section, we analyzed two bentonites within the Z coal, which lies 50 to 60 cm above a 0.57-ppb Ir anomaly (16). The two bentonites are separated stratigraphically by ~30 cm and yield results for the lower (Z2) and upper (Z1) bentonites of 66.019 ± 0.021/0.046 Ma and 66.003 ± 0.033/0.053 Ma, respectively.

The IrZ coal bentonite is much closer stratigraphically to both impact signals and the biostratigraphically defined KPB than the Z coal bentonites; thus, it should be regarded as the closest stratigraphic proxy for the KPB. Accordingly, a comparison with the pooled age for the Beloc tektites indicates a statistically insignificant age difference of 5 ± 27 ky between the two events. Thus, the hypothesis that the Chicxulub impact predated the KPB by ~300 ky (10) is unsupported by our data. Our preferred absolute age for the KPB, including propagated systematic uncertainties, is 66.043 ± 0.043 Ma. This age, which is intrinsically calibrated by both 40K and 238U decay constants (17), is sufficiently precise to discriminate between 100-ky orbital eccentricity cycles at 66 Ma, in principle allowing comparison with astronomical tuning approaches to dating the KPB. However, the uncertainty in the astronomical solution (~40,000 years ago) at 66 Ma (18) effectively limits this discrimination because the 100-ky cycle is not reliable in the solution due to chaotic behavior of the solar system. Our age for the KPB, if based on the Kuiper et al. (19) calibration, would be 65.836 ± 0.061 Ma, which would be sufficiently precise to discriminate between 405-ky but not 100-ky orbital cycles. Because circum-KPB marine records generally lack appropriate materials for high-resolution radioisotopic dating, astronomical tuning potentially represents the best means of temporally calibrating marine records and enabling their comparison with terrestrial records in this time interval. A fully calibrated astronomical solution can potentially enable deconvolution of orbital forcing from other causes of climate change.

The KPB age as determined by our data for the IrZ coal bentonite agrees with the astronomical age (66 ± 0.07 Ma; option 2) derived from deep-sea cores (20) and the preferred astronomical age (65.957 ± 0.040 Ma) inferred by Kuiper et al. (19) from the Zumaia section of Spain, but not with that (65.25 ± 0.06 Ma) of Westerhold et al. (21) (fig. S6). We conclude that the younger age inferred by Westerhold et al. (21) is a consequence of miscalibration by two 405-ky eccentricity cycles (see supplementary materials) and that the terrestrial and marine Ir anomalies are synchronous. This conclusion is strengthened by our concordant U-Pb and 40Ar/39Ar dates for the HFZ coal, which indicate that the KPB is older than 65.988 ± 0.074 Ma and 65.990 ± 0.053 Ma, respectively.

Our data indicate that the stratigraphic interval of ~18 m between the IrZ and HFZ coals in the Hauso Flats section corresponds to a duration of 53 ± 34 ky, whereas the previous 40Ar/39Ar data from the same section (15) suggested a duration of 390 ± 70 ky. This interval spans most of the Pu1 basal Puercan (North American Land Mammal Age) mammalian fauna known from northeastern Montana and adjacent areas in Canada, containing only 15 recognized species compared with 27 in the preceding (pre-KPB) Lancian stage (22). The brevity of the IrZ-HFZ interval of the basal Tullock Formation implied by our new data indicates that the depauperate Pu1 fauna persisted as briefly as 20 ky and supports the hypothesis (22) that much of the post-KPB vertebrate faunal recovery in the Hell Creek area occurred by immigration rather than evolutionary radiation, given that the duration of speciation events for mammals (at least, late Cenozoic ones) typically exceeds hundreds of thousands of years (23).

Our dating of the IrZ and HFZ bentonites constrains the terrestrial, hence atmospheric, –1.5 per mil (‰) δ13C isotope anomaly in the Hauso Flats section (24) to have occurred early within the first 53 ± 34 ky of the Paleogene. Scaling the sediment accumulation rate by linear interpolation between the two dated horizons and allowing the maximum possible stratigraphic extent of the anomaly (considering sample interval) yield a maximum duration of 5 ± 3 ky for the anomaly. Similarly, applying our date for the IrZ bentonite to the iridium anomaly and scaling to the Z2 bentonite in the Hell Creek Marina road section (24) yields a maximum duration of 13 ±13 ky for the –2.0‰ δ13C anomaly there.

The terrestrial δ13C anomaly is markedly consistent in magnitude, timing, and rapidity of onset with marine records, although the latter commonly show much longer recovery time scales. Some marine records [e.g., (1)] show an initial decrease of 1 to 2‰ followed by a rapid increase of ~1‰ on the time scale of several thousand years, succeeded by a much more gradual increase over several million years to pre-KPB values. Other cases show more rapid recovery of marine δ13C to pre-KPB values, as in the Agost section of southern Spain, where the duration of the anomaly and partial recovery are estimated to have occurred over 3 to 5 ky, and full recovery to have occurred over <100 ky (25). Differences between δ13C values in planktic versus benthic foraminifera from Atlantic and Pacific cores also show precipitous drops at the KPB, interpreted to reflect a major disruption in mixing between surface- and deep-water masses (26). Restoration of pre-KPB values of the differential, hence of normal ocean circulation, experienced a protracted recovery spanning several millions of years and was likely the rate-limiting determinant in recovery of general marine productivity (26).

Our results strengthen conclusions that the Chicxulub impact played an important role in the mass extinctions. However, global climate instability preceded the KPB (and thus, in view of our data, the Chicxulub impact) by ~1 million years (My) (2729). During this interval, six abrupt shifts of >2°C in continental mean annual temperatures have been inferred from paleoflora in North Dakota (29). The most dramatic of these temperature oscillations, a drop of 6° to 8°C, occurred <100 ky before the KPB (29) and was closely synchronous with notable mammalian turnover in the Hell Creek area (30). Several cycles of latest Cretaceous sea-level oscillations are recorded in the Williston basin with an overall regression peaking just before the KPB (31), possibly a glacio-eustatic response to climatic cooling. Cooling at this time is consistent with a global sea-level drop of ~40 m beginning in geomagnetic polarity chron 30n and ending in chron 28r (32), clearly spanning the KPB. This event followed closely on a sharp sea-level drop and subsequent rise of ~30 m, coincident with the highest δ18O values recorded for the 30 My before or afterward, which occurred in the middle of chron 30n (32), ~1 My before the KPB. Recognition of these and other relatively brief events led Miller et al. (32) to infer the existence of multiple ephemeral Antarctic ice sheets between 100 and 33 Ma.

We suggest that the brief cold snaps in the latest Cretaceous, though not necessarily of extraordinary magnitude, were particularly stressful to a global ecosystem that was well adapted to the long-lived preceding Cretaceous hothouse climate. The Chicxulub impact then provided a decisive blow to ecosystems thus already under critical stress, and in essence pushed the global ecosystem across a threshold that triggered a planetary state shift (33, 34). Although the atmospheric carbon cycle was disrupted only briefly, and initial mammalian faunal changes after the KPB may have been dominated by migration, some changes such as the disappearance of nonavian dinosaurs were permanent. Thus, whereas some paleoenvironments may have been restored relatively rapidly, terrestrial and marine ecosystems changed forever.

The cause of the precursory climate perturbations that pushed some ecosystems to the tipping point is unclear, but a leading candidate is volcanogenic volatile emissions (35) from early pulses of the episodically erupted Deccan Traps (36, 37). The magmatic event producing the Deccan Traps was clearly initiated prior to the KPB (38), and the most voluminous middle pulse of volcanism may be linked to either (i) the inception of a two-staged decline in marine 187Os/188Os beginning about 300 ky before the KPB (39) or (ii) the KPB itself (36, 37, 40). Existing geochronological data are insufficiently precise to constrain these relationships with age resolution comparable to that presented here for the KPB and the Chicxulub impact. Refining the timing and tempo of Deccan volcanism remains a considerable challenge whose resolution is key to evaluating the role of this event in the causes of biotic and environmental change at the KPB.

Supplementary Materials

Materials and Methods

Supplementary Text

Figs. S1 to S7

Tables S1 to S4

References (4266)

References and Notes

  1. Uncertainties here and throughout are stated at the 68% confidence level.
  2. Uncertainties given as ±X/Y refer to values excluding (X) and including (Y) systematic sources as defined in the supplementary materials.
  3. Acknowledgments: We thank the Ann and Gordon Getty Foundation, U.C. Berkeley's Esper S. Larsen Jr. Fund, and NSF (grants EAR 0844098 and EAR 0451802) for support of B.G.C.'s work; the Natural Environment Research Council for continued funding of the Argon Isotope Facility at the Scottish Universities Environmental Research Centre; the GTSNext project of the Marie Curie Foundation; the Marie Curie Fellowship program for support of L.E.M.; the Netherlands Organisation for Scientific Research (grant 863.07.009) for support of K.F.K.; F. Maurrasse for providing the tektites; W. Alvarez, A. Barnosky, W. Clemens, T. White, and G. Wilson for discussion and comments on the manuscript; and three anonymous reviewers for helpful suggestions. Data are available in the supplementary materials.
View Abstract

Stay Connected to Science

Navigate This Article