Introduction

The impact event at the Cretaceous-Paleogene (K/Pg) boundary, ~66 million years ago, formed the ~200-km diameter Chicxulub impact structure on the Yucatán Platform of the Gulf of Mexico1,2,3,4, causing a mass extinction of over 70% of the fossil species5,6,7. The decline and subsequent recovery of productivity in the marine ecosystem has been extensively studied in the K/Pg boundary sediments on a global scale8,9,10,11. It has been proposed that the recovery of the global marine ecosystem, measured as primary productivity, was heterogeneous12,13,14. Recent micropaleontological and geochemical studies of the impact site show a rapid recovery of both the local population and export production15,16,17,18.

To understand post-impact changes in marine environments and ecosystems on a global scale, geochemical markers of the impact event, particularly highly siderophile element concentrations (HSEs: Os, Ir, Ru, Pt, Pd, and Re) and Os isotope ratios, can be used to accurately correlate K/Pg boundary sediments5,19,20. The anomalous concentration of HSEs provides a key temporal horizon that precisely links Chicxulub to K/Pg boundary sections worldwide7,21,22. Osmium isotope (187Os/188Os) records across the K/Pg boundary also constrain the time scale of the post-impact environmental recovery. Since the residence time of seawater Os (10–50 thousand years (kyr)23,24) is longer than the time scale of modern oceanic circulation, seawater 187Os/188Os values are dominated by changes in whole-ocean chemistry and relatively constant throughout the global oceans. Seawater 187Os/188Os ratios reflect contributions to the global ocean from weathered continental crust (187Os/188Os = ~1.4), mantle and extraterrestrial inputs (187Os/188Os = ~0.12–0.13)25. Detailed profiles of seawater 187Os/188Os ratios from the pelagic carbonate sections in the previous studies show similar Os isotope curves that drop sharply (187Os/188Os = ~0.17 to 0.2) at the K/Pg boundary and then recover to their original seawater values (187Os/188Os = ~0.4) over ~200 kyr19,20,26. A model for the 200 kyr recovery of seawater 187Os/188Os in the distal sites was directly compared with age models based on biostratigraphy, magnetostratigraphy, and cyclostratigraphy, allowing independent estimates of stratigraphic time based on Os isotope analysis20. The development of this Os isotope based geochemical clock20 has broad implications for paleoceanographic research, providing a tool for global correlation of pelagic carbonate sequences across the K/Pg boundary, which makes it possible to elucidate the onset of biotic crisis and subsequent recovery. However, the seawater 187Os/188Os curve in the Gulf of Mexico has not been reported, and correlations using the K/Pg Os isotope clock between the proximal and distal sites remain unclear.

Here, Os isotope ratios are measured in the early Paleocene limestone samples obtained from the IODP-ICDP Expedition 364 at Site M0077 in the Chicxulub impact basin in the Gulf of Mexico and early Paleocene limestone, marl, and clay samples from several sections in northeastern and southeastern Mexico (El Mulato, La Lajilla, Bochil, and Guayal sections) to reconstruct the Os isotope reference curve in the Gulf of Mexico (Figs. 1, Supplementary Note 1)22,27,28. Biostratigraphic correlations and calibrated ages between the Chicxulub site and the Mexican sections are calculated based on a biochronological scale relying on planktic foraminifera and astrochronological calibrations29,30. The obtained HSE concentrations confirm the contribution of extraterrestrial materials in proximal sediments all around the Gulf of Mexico region. For comparison, Os isotope ratios and HSE concentrations are also analyzed in suevite (polymict melt-rich impact breccia), impact melt rock, and granitic samples underlying the Paleocene limestones at Site M0077 (Supplementary Figs. 1–3).

Fig. 1: Location map.
figure 1

a Paleogeographic reconstruction for the K/Pg boundary (~66 Ma) showing the locations of the study site (IODP-ICDP Site M0077) within the Chicxulub impact basin and other pelagic sites reported for Os isotope profiles20. This map is redrawn after ref. 67 and ODSN generated at https://www.odsn.de/odsn/services/paleomap/paleomap.html. Note that this paleoreconstruction potentially represents the placement of tectonic blocks at ~66 Ma, but does not reflect sea level at the time of impact. b Geographic locations for the study site and sections within the Gulf of Mexico. Map modified after ref. 28.

Full size image

Results

Planktic foraminiferal biochronology

Planktic foraminiferal biostratigraphy of the Paleocene white limestone for Site M0077 is updated from that published by ref. 16, which was based on P zones31. The planktic foraminiferal age model for Site M0077 relies on Dan zones29. The lowest and highest occurrence data (LOD and HOD) of key taxa occur in the correct stratigraphic order (from Dan2 to Dan4c; Supplementary Figs. 4, Supplementary Data 1). However, the LOD of Eoglobigerina simplicissima, which defines the base of Dan3b, could not be recognized in Site M0077, and the base of Dan3b was inferred to be at 616.23 mbsf (Supplementary Figs. 4, Supplementary Data 1).

Calibrated ages are assigned to each datum in accordance with ref. 29. The age of each sample in the Paleocene white limestone is determined by average sedimentation rates, which are constrained by the stratigraphic thickness and the calibrated time duration of the planktic foraminiferal biozones (Supplementary Figs. 4, Supplementary Data 1)29. Calculated sedimentation rates from each planktic foraminiferal biozone from the lower to the upper part of the white limestone are 0.667 cm kyr–1, 2.6 cm kyr–1, 0.125 cm kyr–1, 0.669 cm kyr–1, 0.256 cm kyr–1, 0.068 cm kyr–1, and 0.198 cm kyr–1 in ascending order (Supplementary Figs. 4, Supplementary Data 1). The high sedimentation rate calculated for Biozone Dan3a (2.6 cm kyr–1) is anomalous, probably resulting from assuming that there is no hiatus affecting Biozone Dan3b.

Rhenium-osmium isotopes and highly siderophile elements

Age-corrected initial Os isotope ratios (187Os/188Osi) in Site M0077 are low (187Os/188Osi ~ 0.19) in the ~3-cm-thick green-gray marlstone that transitions into the lowermost Paleocene white limestone22 and then gradually increase in the Paleocene white limestone (Fig. 2a). 187Os/188Osi recover to pre-impact value (~0.4; ref. 20) at 614.32 mbsf in the lower part of Biozone Dan4c (Figs. 2a, Supplementary Data 2). In Biozone Dan4c, 187Os/188Osi remain constant at steady state and subsequently increase to ~0.45. Based on the sedimentation rates calculated by the planktic foraminiferal age model, the depth of 614.32 mbsf corresponds to ~65.265 Ma29, indicating that the recovery time of the Os isotope ratio at Site M0077 occurs more than ~700 kyr after the K/Pg impact. Measured 187Os/188Os ratios of suevite and granite underlying the Paleocene white limestone are generally high, ranging from 1.1 to 4.0 (Supplementary Figs. 5, Supplementary Data 2). In contrast, the measured 187Os/188Os ratios for the impact melt rocks display significant variations, ranging from unradiogenic values of 0.18–0.19 for the lower impact melt rocks found in the granite unit (917.18 to 917.25 mbsf) to radiogenic values of 0.9–1.5 for the upper impact melt rocks, which overlies shocked granitoid target rock of the peak ring (721.01 to 744.20 mbsf; Supplementary Figs. 1 and 5, Supplementary Data 2).

Fig. 2: Stratigraphic profiles of 187Os/188Os ratios, HSE concentrations, and Os/Ir ratios of the Paleocene white limestone at Site M0077.
figure 2

a 187Os/188Os ratios represent age-corrected (red circle) and measured (white circle) isotope ratios. b Os and Re concentrations. c Ir and Ru concentrations. d Pt and Pd concentrations. e Os/Ir ratios. Planktic foraminiferal zonations A21 and W11 are from ref. 29 and ref. 31, respectively. Os/Ir ratios for CI chondrite and upper continental crust (UCC) are from refs. 68,69,70.

Full size image

Osmium isotope ratios from marl samples in the La Lajilla section of northeastern Mexico show a minimum value (187Os/188Osi ~ 0.241) just above the K/Pg impact-induced Complex Clastic Unit (CCU)32 and recover to ~0.3 in about 70 kyr (Figs. 3d; Supplementary Data 3). The marl samples in the El Mulato section located in northeastern Mexico show relatively high 187Os/188Osi of ~0.320 at 1 cm above the CCU, and its lowest value about 15 cm above the CCU (Fig. 3c). Subsequently, the isotope ratios gradually increase to ~0.33 about 200 kyr after the impact. The Bochil section in southeastern Mexico shows the lowest 187Os/188Osi values (~0.167) in the clay sample about 3 cm above the CCU and then gradually increases in the limestone and marl samples, although the Os isotopic ratios are less than 0.3 at 200 kyr after the impact (Fig. 3a). The 187Os/188Osi ratios in the Guayal section located in southeastern Mexico show the lowest value just above the CCU ( ~ 0.192) in the clay sample and increase to ~0.27 in the uppermost marly limestone layer of the measured sample (~110 cm above the CCU; 80 kyr after the impact; Fig. 3b).

Fig. 3: Stratigraphic sections showing age, biozones, lithology, 187Os/188Os ratios, and Os and Ir concentrations in the Mexican sections.
figure 3

a Bochil, b Guayal, c El Mulato, and d La Lajilla sections. 187Os/188Os ratios represent age corrected (colored circle) and measured (white circle) isotope ratios. Planktic foraminiferal zonations A21 and W11 are from ref. 29 and ref. 31, respectively.

Full size image

The Os concentrations of the Paleocene white limestone at Site M0077 gradually decrease upward throughout the limestone (Fig. 2b). The high Ir concentrations at the base of the limestone (~0.49 ppb) decrease abruptly in the lowest part of the limestone, and then remain almost constant throughout the limestone (Fig. 2b). As a result, the Os/Ir ratios of the limestones in the lower part of the study section are high compared to the ratios of upper continental crust (UCC) and CI chondrite (Fig. 2e). The concentrations of HSE show high values for Os-Ir-Ru at the base of limestone (Fig. 2), and their CI chondrite-normalized HSE patterns exhibit a relatively flat pattern (Fig. 4a)22. Other limestone samples between 616.48-610.33 mbsf exhibit low HSE concentrations (Fig. 2). These samples are depleted in Os-Ir-Ru and enriched in Pt-Pd-Re and characterized by Os/Ir > 1 (Figs. 2 and 4a).

Fig. 4: CI chondrite-normalized HSE patterns of Paleocene samples.
figure 4

a Paleocene white limestones in Site M0077. The dashed line represents the sample from the base of limestone (616.52 mbsf). b Paleocene samples in the Mexican sections. c Paleocene samples from a few cm above the K/Pg impact-induced Complex Clastic Unit (CCU)32 in the Mexican sections. The HSE patterns of upper continental crust (UCC) are also shown in comparison. HSE concentrations for CI chondrite and UCC are from refs. 68,69,70.

Full size image

The Ir and Os concentrations of the impact melt rocks are generally low, similar to the average UCC values (Supplementary Fig. 6). Only two samples, 163R3_35.5–41.0 and 163R3_47.0–48.0 from the lower impact melt rocks show slight enrichments in Os and Ir concentrations (Supplementary Fig. 6). The suevite samples are relatively similar in Os and Ir concentrations to the upper impact melt rock samples. The granite sample shows significantly lower concentrations of HSE compared to UCC (Supplementary Fig. 6).

In other Gulf of Mexico sections, Os and Ir concentrations remain high up to a few cm above the CCU (Fig. 3). The Ir concentration decrease abruptly, followed by constant low values in each section (Fig. 3). In all four sections, Os/Ir ratios show high values (Os/Ir > 1) relative to CI chondrite and UCC similar to the concentration/pattern observed at Site M0077 (Supplementary Fig. 7). The CI chondrite-normalized HSE patterns show relatively flat patterns for samples immediately above the CCU, while samples at higher stratigraphic levels display coherent patterns characterized by depletion in Os-Ir-Ru relative to Pt-Pd-Re (Fig. 4b, c).

Discussion

Previously, a geochemical “clock” was proposed to determine the time elapsed since the K/Pg impact using the change from low to steady-state in the 187Os/188Os ratio20. This method assumes that the residence time for Os in early Paleocene seawater is ~40 kyr due to the Os one-box model after the K/Pg Chicxulub impact20, similar to values estimated for the modern ocean (10–50 kyr)23,24. This makes the Os clock useful for constraining the age of sediments deposited in global oceans after the K/Pg impact. The seawater 187Os/188Os recovery from low values (~0.17) to steady state (~0.4) in ~200 kyr after the K/Pg impact relies on data obtained at ODP Hole 1262B (Walvis Ridge, South Atlantic Ocean) and ODP Hole 690 C (Maud Rise, Southern Ocean), for which the ages were based on orbital tuning and magnetostratigraphy, respectively20. Recently, Krahl et al. 33 revised the age of each sample from ODP Hole 1262B by interpretation derived from the astrochronological framework and the planktic foraminifera high-resolution biostratigraphy from the well-known and also well-calibrated Zumaia section in Spain (Supplementary Data 4)30,34. The age model for ODP Hole 690 C was also recalibrated against the latest paleomagnetic data (Supplementary Data 5). Although these age calibrations result in slight shifts in the 187Os/188Os curves at both pelagic sites, the length of the impact-induced Os isotope excursion remains ~200 kyr (Fig. 5a), comparable to the time scale reported by ref. 20.

Fig. 5: Comparison of 187Os/188Osi profiles after the K/Pg impact.
figure 5

a 187Os/188Osi ratios in the Chicxulub impact basin (Site M0077) and distal pelagic sites20 for ~3.5 million years (myr) after the impact. b 187Os/188Osi ratios of Paleocene (Danian) samples in the Gulf of Mexico. Gray line represents the calculated recovery model of seawater 187Os/188Os by ref. 20.

Full size image

The Os isotope variation of the Paleocene samples from Site M0077 within the Chicxulub impact basin is consistent with the data from both Hole 1262B and Hole 690 C showing that the ratios recover from low (187Os/188Os ~ 0.17) to steady state values (187Os/188Os ~ 0.4) after the K/Pg impact19,20. However, the recovery time for the Chicxulub impact basin occurs around ~700 kyr after the impact, much longer than ~200 kyr observed at the distal pelagic sites (Fig. 5a). Similarly, the 187Os/188Osi recovery curves of the Bochil and El Mulato sections in the Gulf of Mexico deviate from the reference curve of the distal pelagic sites towards lower 187Os/188Osi, only reaching ~0.3 after 200 kyr (Fig. 5b).

A possible explanation for the difference between the Gulf of Mexico and pelagic 187Os/188Osi curves is that the reduced 187Os/188Os observed for the Gulf of Mexico sediments may be due to remobilization or reworking of the K/Pg deposits enriched in chondritic components. Although the occurrence of intense sedimentary disturbances such as bioturbation or upward reworking can largely be ruled out for the sections studied, the possibility remains that fine meteoritic dust particles from nearby sites could be re-supplied to the Gulf of Mexico. Goderis et al. 22 identified an upward contribution of meteoritic components to post-impact sediments based on very high Ir concentration and distinctive CI chondrite-normalized HSE pattern in the basal white limestone at Site M0077 ( ~ 0.5 ppb; ~616.52 mbsf; Figs. 2c and 4a). By contrast, the Paleocene white limestones above 616.47 mbsf show remarkably low and constant Ir concentrations (< 0.065 ppb) with the general UCC-like HSE patterns. This strongly suggests that the apparent contribution of extraterrestrial Ir ends shortly after the K/Pg impact (Figs. 2c and 4a). Similarly, sharp Ir decline after the K/Pg interval is commonly observed in the Mexican sites including the Arroyo el Mimbral35 and other four sections studied here (Fig. 3). The basal Paleocene sediments in a few cm above the CCU and the upper sections display contrasting HSE patterns as with the Site M0077 (Fig. 4b, c). Indeed, distinct declining trends between Os and Ir abundances throughout the Paleocene sediments are broadly consistent with their contrasting residence times estimated from the modern ocean of ~40 kyr for Os and ~2 kyr for Ir23,36 (Fig. 2 and 3), indicating that the hydrogenous component dominates the sedimentary budget. Thus, the stratigraphic variations in HSE concentrations for the Gulf of Mexico sites do not support the idea that the reduced Os isotope ratios during ~700 kyr are caused by reworking of chondritic components during Paleocene carbonate sedimentation.

From these observations, we invoke that the Paleocene sediments in the Gulf of Mexico record a continuous but unique temporal evolution of seawater Os isotopic composition after the K/Pg impact. Distinct 187Os/188Os evolutions for the global ocean and the Gulf of Mexico cannot be expected in the modern environment, since the water mass renewal rate of the Gulf of Mexico is estimated to be much faster than the Os residence time τ = 40 kyr. For example, all the seawater in the Gulf of Mexico (~2.4 × 1015 m3) can be renewed within 3 years if the very large transports observed in the Yucatan Channel and the Straits of Florida [~27.6 Sv (1 Sv = 106 m3 s–1); ref. 37] represent the entry and exit of the Gulf of Mexico throughflow. This in turn implies that the delayed recovery of seawater 187Os/188Os in the Gulf of Mexico could not be acquired unless the ventilation was significantly reduced shortly after the K/Pg impact.

The openness of the Gulf of Mexico to the global ocean since the Triassic-Jurassic basin opening associated with the breakup of Pangea has been discussed with paleogeographic reconstructions of the regional tectonic and depositional history38,39. At about 80 Ma, when sea level peaked in the Late Cretaceous40, most reconstruction depicts the Western Interior Seaway directly connected to the Pacific. However, this connection was withdrawn by 66 Ma, leaving only the oceanic passages to the Atlantic between the Yucatán and Bahamas platforms. Recent data compilation of sedimentary sections from many industrial wells distributed throughout the Gulf of Mexico has well documented the widespread deposition of deep-water chalk and shallow marine carbonate with limited siliciclastic inputs at the end of the Cretaceous, largely due to a small number of drainage systems and circulation of ocean and shelf currents connected to the global ocean39.

The high-energy Chicxulub impact at 66 Ma initiated earthquakes and tsunami waves, resulting in enormous redistribution of sediments into and out of the Gulf of Mexico41,42. For example, tsunami and mass wasting deposits recognized in the Circum-Gulf region such as Mexican and North American sites exhibit a wide range of thickness, from ~1 m to nearly 1000 m, typically containing upward fining sequences28. Large quantities of redistributed sediment are also documented from several sections in Cuba (< 700 m), which represent slope and basin deposits in the proto-Caribbean Sea outside the Gulf of Mexico43. Thus, we postulate that the impact-induced deposition generated by the tsunamigenic debris flow and platform collapse reshaped the Gulf of Mexico gateway between the Yucatán and Bahamas platforms and significantly reduced the deep-water connection to the global ocean, although there are no known estimates for the volume or thickness of deposits accumulated in the gulf entrance/exit just after the impact.

To address the delayed recovery of 187Os/188Os values in the Gulf of Mexico in this context, we carried out the Os mass balance calculation for the global ocean and the Gulf of Mexico separately using a simple box model (Supplementary Fig. 8, Supplementary Data 6). Assuming that the Gulf of Mexico is an Os pool affected by inputs of global seawater Os and unradiogenic Os sourced from chondritic impactor, the temporal evolution of seawater 187Os/188Os recorded in Site M0077 borehole is simulated if relative contributions of the two Os sources gradually change after the K/Pg asteroid impact. The model 187Os/188Os curve fitted to the Site M0077 data is reproduced under the following tuned parameters; 1) the renewal rate of water mass in the Gulf of Mexico increases from 0.015 to ~0.060 Sv between 66.001 and 63.75 Ma, and 2) meteoritic Os with unradiogenic composition is delivered to the Gulf of Mexico, decaying asymptotically from 81 mol/yr (at the K/Pg impact ~66.001 Ma) to 0 at 63.75 Ma (Supplementary Note 2, Supplementary Fig. 9, Supplementary Data 6). It should be noted that ~0.060 Sv at 63.75 Ma is close to the maximum value for allowing the box model calculation, for given the seawater volume of the Gulf of Mexico and the Os residence time in the seawater. Actually, nearly identical 187Os/188Os between the Gulf of Mexico and the global ocean at ~65.265 Ma would support that “vigorous” flow (> ~0.060 Sv) was rapidly achieved ~700 kyr after the Chicxulub impact.

Considering that Os is known to be highly mobile during alteration and weathering processes44,45, there are two possible mechanisms to acquire a large but continuous input of meteoritic Os with unradiogenic composition into the Gulf of Mexico: (1) through weathering of impact ejecta deposited on the nearby continents such as North America and Mexico, or (2) by venting of hydrothermal fluids reacting with the impact melt sheet underneath the Chicxulub structure46. Previous studies infer that a large fraction of meteoritic Os that resided in the chondritic projectile is likely transported globally as airborne microscopic dust and impact vapor condensates that partially dissolved into seawater immediately after the impact event22,47. In contrast, larger fragments of melt droplets or sheets enriched in projectile components are distributed and deposited concentrically around the impact site, possibly with larger amounts downrange direction to the southwest48,49. Although their compositions and constituent minerals/phases remain unknown, it is possible that the incongruent weathering of these materials could have released significant and negligible amounts of unradiogenic Os and Ir, respectively. If such Os were transported with some time lag by riverine flow to the Gulf of Mexico47, this could explain the delayed recovery of 187Os/188Os after the K/Pg impact observed in our studied sections. However, this mechanism may not be effective because the impact melt rocks enriched in HSE have rarely been observed on the continents surrounding the impact site, in contrast to their occurrence within the Chicxulub impact basin45,48. Furthermore, our data demonstrate that the Mexican sites distributed on the continental margin show smaller 187Os/188Os excursions than that of Site M0077, located in the impact basin (Fig. 5b). In other words, the largest 187Os/188Os excursion for Site M0077 among the study sites likely indicates that the crater hydrothermal mechanism is the dominant process for introducing the unradiogenic Os into the Gulf of Mexico seawater. Indeed, contemporaneous limestones from Site M0077 show excess concentrations of Mn, P, and Pb during reduced 187Os/188Os ratios16,27,46. Kring et al. 46 suggest that Mn enrichment in post-impact limestone sediments overlying suevite may be due to hydrothermal venting after the impact at Site M0077. Therefore, delayed recovery of seawater 187Os/188Os in the Gulf of Mexico suggest that it is influenced by a long-term supply of hydrothermal fluids activated in the Chicxulub structure after the impact event, as discussed below.

The Chicxulub impact uplifted crustal rocks from a depth of 8–10 km and produced a melt sheet of 104–105 km3 (refs. 50,51). Such a large impact event generates long-term heat sources that include shock-deposited heat, a large melt sheet, and a central uplift. The impact event created a porous and permeable structure in the Chicxulub impact structure that was an ideal host for the hydrothermal fluid system. Numerical simulations of the hydrothermal activity in the Chicxulub structure indicate that the system extended to a depth of 5–6 km and its lifetime (i.e., the time required for the system to cool below 90 °C within 1 km of the surface) ranges from 1.5 to 2.3 million years52. Mineralogical observations over almost the entire length of Site M0077 drill core support an extended duration thermal evolution model. The volume of area is estimated at ~1.4 × 105 km3 (ref. 46).

The hydrothermal activity at Chicxulub structure could play a significant role in the observed 187Os/188Os variations in the Gulf of Mexico seawater. This is inferred from the recent study demonstrating that the 5.8 m thick early Paleocene limestones in the M0077 core contain significant amounts of Mn released from post-impact hydrothermal plumes into the Gulf of Mexico seawater (Fig. 6b)46. Although early Paleocene limestones from Site M0077 do not show a significant change in color (Supplementary Fig. 2), a strong positive correlation between Os/Al and Mn/Al (R2 = 0.90) was observed for our data during the delayed recovery of 187Os/188Os values (614.32 to 616.57 mbsf). Other geochemical proxies for hydrothermal activity such as the P and Pb enrichments53,54 are also observed in the interval with unradiogenic 187Os/188Os (Fig. 6). This covariation strongly suggests that together with Mn, P, and Pb, the hydrothermal plumes supplied unradiogenic Os into the Gulf of Mexico for at least ~700 kyr after the Chicxulub impact, based on our calibrated age. Similar Mn, P, and Pb enrichments occurs in post-impact hydrothermal sediments of another drill core within the Chicxulub structure (ICDP Yaxcopoil-1)53. In the Sudbury structure, Mn, P, and Pb enrichments are also observed in the Onaping formation55.

Fig. 6: Osmium isotope profile and paleoenvironmental change after the impact event.
figure 6

a Age-corrected (red circle) and measured (white circle) 187Os/188Os ratio and Os concentration at Site M0077. b Mn/Al ratio. c P/Al ratio. d Pb/Al ratio. e Relative abundance plots for nannoplankton taxa. f Relative abundance of planktic foraminiferal trophic groups. g The percentage of benthic foraminifera relative to all foraminifera. b, c and d are XRF core scan data reported by ref. 16. Data of e, f, and g are from ref. 62 and ref. 16, respectively.

Full size image

At mid-ocean ridge vents, Mn oxidizes more slowly than other dissolved elements in hydrothermal plumes, and Mn oxide particles can be transported for several hundreds of km, before precipitating out of suspension to form sediments53,56. Studies of present and past marine sediments suggest that Os dissolved in seawater is adsorbed onto Mn (oxyhydr) oxide particles, preserving the seawater 187Os/188Os in the sediments25,57,58,59. Although mantle-derived unradiogenic Os is released from hydrothermal vents at mid-ocean ridge settings, it is difficult to presume the Os isotope ratios in the Chicxulub hydrothermal plumes. If the amount of meteoritic material introduced into the Chicxulub impact structure was low due to the assumed steeply-inclined trajectory of the impactor60,61, highly radiogenic Os derived from the pre-impact lithologies (continental granites) may have been dominated in hydrothermal fluids.

Feignon et al. 45 reported the 187Os/188Os values and HSE compositions of suevites and impact melt rocks as well as pre-impact lithologies recovered from Site M0077 on the peak ring. Since the majority of samples display low Ir contents (< 40 ppt) with HSE patterns similar to those for typical crustal compositions (e.g., UCC), they concluded that any meteoritic components within the crater are highly diluted. However, two impact melt rocks exhibit significant enrichments of Ir (250–324 ppt) and Os (125–344 ppt), which are likely attributed to meteoritic contamination. The most striking feature of the depth profile of M0077 drill core is that the 187Os/188Os ratios of the granitic basement and impact melt rocks in the lower unit (750-1250 mbsf) are consistently low (187Os/188Os < 0.2), indicative of the post-impact hydrothermal overprint on an extensive part of the basement (Supplementary Fig. 5). This is supported by the fact that the least altered amphibolite (80R2_61–63.5) and granite (95R1_91.0–93.0) preserve highly radiogenic 187Os/188Os ratios of 2.47 and 4.02, respectively. Assuming that these two samples represent pre-impact lithologies, mixing calculation traces the contamination of the meteoritic component. The majority of data from the basement broadly follows mixing areas between radiogenic amphibolite-granite and chondrite on a diagram of 187Os/188Os versus Os concentration, even though the data for Os/Ir ratios versus 187Os/188Os tend to deviate toward higher Os/Ir ratios from the mixing areas (Supplementary Fig. 10). These deviations can be attributed to the selective hydrothermal remobilization of Os but not for Ir. Some contributions from mafic target rock cannot be ruled out due to the occurrence of Os-rich dolerite dykes in the depth interval (854.6 mbsf: 140R2_5–8). Nevertheless, the data strongly suggest that hydrothermal fluids within the Chicxulub structure mobilized unradiogenic Os, which could then affect the seawater 187Os/188Os ratios of Gulf of Mexico seawater via prolonged hydrothermal venting.

If the above scenario is plausible, temporal relationship between hydrothermal supply of nutrients and marine ecosystem change should be examined carefully. Detailed analyzes of a record of calcareous nannoplankton and planktic foraminifera from Site M0077 reveal that life in the impact basin reappeared soon after the impact, and that a high-productivity ecosystem is established after only ~30 kyr15. Their post-extinction assemblages are dominated by calcareous dinoflagellate cysts (Cervisiella spp.), Braarudosphaera spp., and microperforate foraminifera, all of which are adapted to high-nutrient (eutrophic) environments (Fig. 6e, f)62. Furthermore, the abundance of benthic foraminifera relative to planktic foraminifera, which can be regarded as an indicator of organic matter fluxes from the surface ocean to the deep seafloor, increased sharply in Biozone Dan4a (Fig. 6g)16. It has been presumed that these recovery patterns are likely driven by ecological rather than environment processes16,62.

Interestingly, the time scale for complete recovery from the impact induced Os isotope excursion coincides with a transition in the marine ecosystem from eutrophic to oligotrophic conditions (Fig. 6e, f). Turnovers in the calcareous nannoplankton and planktic foraminiferal assemblages are marked in Biozone Dan4c, which coincide with the lower part of Biozone P1c (Fig. 6e, f). In this stratigraphic interval of 65.485–63.888 Ma, calibrated by biochronological scale (Supplementary Data 1), seawater 187Os/188Osi ratios in the Gulf of Mexico become identical to those of the steady-state recorded in the pelagic sites (~0.4)20. Also, at this interval, Mn, P, and Pb concentrations decrease to background levels inferred from overlying middle Paleocene sediments (Fig. 6). Indeed, the declining pattern of P concentration in early Paleocene limestone at Site M0077 at 614 mbsf and further at 613.2 mbsf coincides with major assemblage changes in the calcareous nannoplankton and planktic foraminifera trophic group (Fig. 6c, e, f). While this timing may be coincidental, these covariations suggest that the hydrothermal activity in the impact basin led to large-scale shifts in the dominant nannoplankton and planktic foraminiferal trophic groups, and that the hydrothermal supply of phosphorus, a limiting nutrient in primary productivity facilitated the early recovery of diverse life, as speculated by ref. 17.

Our study infers that there are some intimate relationships between the impact-induced hydrothermal activity and the biosphere recovery within the Chicxulub impact basin. However, in order to test the scenario that hydrothermal supply of nutrients and energy was a key driver for changes in the biota of the marine surface environment, it is necessary to clarify how these nutrients were supplied by the hydrothermal system and entered the euphotic zone to drive enhanced primary production. Because there are no large-scale hydrothermal systems in the oceans today comparable to that formed by the Chicxulub impact, sedimentary records from variable sites in nearby oceans are critical to examining the extents of hydrothermal effects on the marine surface environment. For example, there is the problem of why similar trends in export productivity (e.g., Ba/Ti ratio) are observed in the Caribbean Sea63, far from the hydrothermal system, which cannot be explained if we assume that the supply of nutrients from the hydrothermal system was limited to the Gulf of Mexico. Thus, further research on Os isotope and other geochemical profiles in different sites such as the Caribbean Sea and the Pacific Ocean are clearly required to confirm the versatility of the Os isotope clocks and to reconstruct the recovery patterns of the marine ecosystem.

Methods

Planktic foraminiferal analyzes

The age models for Site M0077 and Mexican sections (El Mulato, La Lajilla, Bochil, and Guayal) are based on planktic foraminifera high-resolution biostratigraphy (see refs. 16,28,29; and references herein). The studied rock samples were disaggregated using two techniques; 1) using a H2O2 solution for 6 to 24 h (to disaggregate marl and clay samples), and 2) using a solution with 80% acetic acid for 4 to 6 h (to disaggregate very lithified marly limestone and limestone samples). The residual fractions were then washed through a 63 μm sieve and dried at 50°C.

All correlations and ages were obtained following the same taxonomic and biostratigraphic criterion (Supplementary Data 1 and 3). The planktic foraminiferal biochronological scale used was magnetochronologically calibrated by refs. 29,64, and astrochronologically by ref. 30. This scale is based on the Dan biozonation of ref. 29, which used the following key-biohorizons: HOD of Abathomphalus mayaroensis (= K/Pg mass extinction event; 66.001 Ma; base of Dan1), LOD of Parvularugoglobigerina longiapertura (65.995 Ma; base of Dan2), LOD of Parvularugoglobigerina eugubina (65.983 Ma; base of Dan3a), LOD of Eoglobigerina simplicissima (65.973 Ma; base of Dan3b), LOD of Parasubbotina pseudobulloides (65.933 Ma; base of Dan4a), LOD of Subbotina triloculinoides (65.797 Ma; base of Dan4b), and LOD of Globanomalina compressa (65.485 Ma; base of Dan4c). The biostratigraphic study of Site M0077 also includes the following key-biohorizons: LOD of Acarinina trinidadensis (base of homonym biozone), and LOD of Acarinina uncinata (base of homonym biozone), magnetochronologically calibrated at 63.888 and 62.902 Ma by ref. 64.

Re-Os isotope and highly siderophile element analyzes

Rhenium-osmium (Re-Os) isotope compositions and highly siderophile element (HSE: Os, Ir, Ru, Pt, Pd, and Re) concentrations for 23 samples of Paleocene limestone and 16 samples (with 1 replicate) of suevite, impact melt rock, and granite within the Chicxulub impact basin, and 36 Paleocene samples from other K/Pg sites around the Gulf of Mexico (Bochil, Guayal, El Mulato, and La Lajilla sections) were determined by isotope dilution mass spectrometry (ID-MS) after quartz glass tube digestion. The methods used for sample digestion, chemical purification, and mass spectrometry are based on procedures outlined in refs. 59,65. Powdered samples (~0.25–0.5 g) and spike solutions enriched in 185Re and mixed 190Os-191Ir-99Ru-196Pt-105Pd were transferred into a quartz glass tube. After adding 3–4 mL of inverse aqua regia, each quartz glass tube was frozen in dry ice, sealed with an oxygen-propane torch, and placed in an oven at 240 °C for 72 h. Subsequently, 1–3 mL of 38% hydrofluoric acid (HF) was added to the residual solids after Os extraction for desilicification.

Isotope ratio measurements of HSE were conducted using two types of mass spectrometers. Osmium concentrations and isotopic compositions were measured by negative thermal ionization mass spectrometry (N-TIMS; Triton Plus, Thermo Fisher Scientific) at JAMSTEC, Japan. Osmium purified after CCl4 solvent extraction and microdistillation66 was loaded in HBr on baked 99.997% Alfa Aesar Pt wire and covered with a NaOH-Ba(OH)2 activator solution. The average total procedural blank for Os was 0.32 ± 0.46 pg (n = 9, 1 SD) with a 187Os/188Os ratio of 0.152 ± 0.015. Blank corrections were applied to all analyzes, which relies on the blank associated with each batch of digestion rather than using a long-term average. The uncertainties for 187Re/188Os and 187Os/188Os were estimated by error propagation of the blank uncertainties. Blank contributions for the measured Os concentrations and 187Os/188Os ratios of the samples from the Paleocene limestone in Site M007 were mostly insignificant: less than 1.4% and 0.9%, respectively. The blank contributions for the suevite, impact melt rocks and granite samples were variable depending on their Os concentrations (0.2–4.7% for Os and 0.03–4.8% for 187Os/188Os). The blank contributions for the measured Os concentrations and 187Os/188Os ratios for the samples from other Gulf of Mexico regions were less than 0.7% and 0.3%, respectively.

Purification of Ir, Pt, Ru, Pd, and Re was accomplished in three steps: (1) anion exchange chromatography; (2) cation exchange chromatography; and (3) solvent extraction using N-benzoyl-N-phenylhydroxylamine (BPHA). HSE concentrations were measured by high-resolution ICP-MS (HR-ICP-MS; Element XR, Thermo Fisher Scientific) housed in the University of Tokyo at Komaba. For sample introduction, a combination of 100 μ/min PFA self-aspirating nebulizer and dual cyclonic/Scott double-pass spraychamber was used during all measurements, and oxide level (HfO+/Hf) was set to 1%. Sample and standard solutions were interspersed throughout the analytical sessions to monitor and correct for instrumental fractionation. The monitored masses of analyzes and interferences are 89Y, 90Zr, 95Mo, 97Mo, 99Ru, 100Ru, 101Ru, 105Pd, 106Pd, 108Pd, 111Cd, 178Hf, 185Re, 187Re, 191Ir, 193Ir, 194Pt, 195Pt, 196Pt, and 202Hg. Although all raw signal intensities of samples were mathematically corrected for the measured isobaric oxide interferences, contributions of interferences to analyte signals are mostly insignificant (< 0.1%). The average total procedural blanks for the analyzed elements are 2.3 ± 1.7 pg for Ir, 1.4 ± 0.1 pg for Ru, 22 ± 7 pg for Pt, 4.0 ± 2.6 pg for Pd, and 1.9 ± 0.8 pg for Re (n = 9, 1 SD). As in the case of Os, all analyzes were blank-corrected. Blank contributions to samples from Site M0077 for Ir, Ru, Pt, Pd, and Re were < 37%, < 21%, < 22%, < 2.2%, and < 20%, respectively, except for granite sample 95R1_91.0-93.0 (42% for Ir, 22% for Pt, 52% for Pd, and 0.4% for Re). The blank contributions to samples from other Gulf of Mexico regions for Ir, Ru, Pt, Pd, and Re were < 3.0%, < 12%, < 32%, < 1.1%, and < 29%, respectively. The uncertainties on each sample were estimated by error propagation of the analytical uncertainties during ICP-MS measurement (2SE) and blank correction. The accuracy of the analytical methods was evaluated by measuring basaltic (BIR-1a) and sedimentary limestone (JLs-1) reference materials (Supplementary Data 2). Nine replicates of BIR-1a display good reproducibility (0.3% RSD for 187Os/188Os, 12% RSD for Os, 12% RSD for Ir, 4.2% RSD for Ru, 5.4% RSD for Pt, 4.5% RSD for Pd, and 1.3% RSD for Re in the ~0.5 g powdered samples; n = 9) and are in good agreement with data for larger powdered samples (~1 g; ref. 65). JLs-1 also displays good reproducibility for 187Os/188Os (0.5%) and Os (0.3%) and Re (1.5%) concentrations, whereas other HSE concentrations show larger variability (33% RSD for Ir, 53% RSD for Ru, 64% RSD for Pt, and 31% RSD for Pd in the ~0.5 g powdered samples; n = 4) due to their very low HSE concentrations.