Expedition 364) recovered what is, to our knowledge, the first record of the few hundred thousand years immediately after the impact within the Chicxulub crater. Site M0077, which was drilled into the peak ring of the crater 7 (Extended Data Fig. 1), sampled an approximately 130-m-thick, generally upward-fining suevite (that is, melt-bearing impact breccia) overlying impact melt rocks and fractured granite 17 . The boundary between the suevite and overlying earliest-Palaeocene pelagic limestone is in core 40-1 (Fig. 1), and comprises a 76-cm-thick upward-fining, brown, fine-grained micritic limestone that we term the 'transitional unit' . The lower portion of the transitional unit is laminated below 54-cm core depth and contains no trace fossils (Fig. 1 and Extended Data Fig. 2). The laminations are thin, graded beds with sub-millimetre-scale cross-bedding that indicates bottom currents, and are likely due to the movement of wave energy-including tsunami and/or seiches-in the days after the impact. The fine grain size (primarily clay to silt, with some sand-sized grains concentrated in the graded beds) suggests that much of the material in the transitional unit was deposited from resuspension and settling. The transitional unit is overlain by a white pelagic limestone. The lowermost sample taken in this limestone (34 cm core depth) contains the planktic foraminifer Parvularugoglobigerina eugubina (which marks the base of Zone Pα), other foraminifer of the same genus (P. extensa, P. alabamensis) and Guembelitria cretacea. Because many other species that originate within Zone Pα first appear a few centimetres higher in the section (31-32 cm), we conclude that the base of the limestone lies very near the base of this zone, 30 kyr after the impact 18 .

Biostratigraphy and basic assumptions about depositional and crater processes indicate that the transitional unit was deposited between several years and 30 kyr after impact (Fig. 2). To better constrain this, we use the abundance of extraterrestrial 3 He to determine sediment accumulation rates (see Methods). This proxy provides a firm upper limit of 8 kyr for deposition, assuming none of the 3 He is reworked. If even a small amount of 3 He is reworked (which is very likely given the prevalence of reworked microfossils and impact debris), then the transitional unit was deposited in a period of time of less than about Letter reSeArCH 1 kyr, which is below the resolution of the method. With no sediment source other than settling of material suspended by the impact and subsequent tsunami and seiches, a more realistic estimate-based on Stokes' law-for the duration of this unit suggests about 6 years for the settling of a 2-μm grain of carbonate (an upper limit, as most grains are much larger; see Supplementary Information for further discussion). Fig. 2 | Constraints on the age of the transitional unit. Maximum durations of the transitional unit based on biostratigraphy (which suggests it was deposited in less than 30 kyr), extraterrestrial 3 He (which suggests it was deposited in approximately 8 kyr if there is no reworking, or less than 1 kyr if there is any reworking) and Stokes' law, which suggests it was deposited in less than 6 years.

The lower portion of the overlying limestone, which contains fossils that appear approximately 30 kyr after the impact, appears conformable with the transitional unit and must therefore be condensed owing to low pelagic sedimentation in the first few tens of thousands of years after the impact.

Clear, discrete trace fossils, including Planolites and Chondrites, characterize the upper 20 cm of the transitional unit (above 54 cm) (Fig. 1 and Extended Data Fig. 2), providing unequivocal evidence for benthic life in the crater within years of the impact. Flattening of the structures indicates that the traces were formed while the sediment was still soft, during or shortly after the deposition of the transitional unit. Infilling of the burrows with brown, fine-grained micrite also suggests traces were syndepositional and not derived from mixing of the Danian limestone above the transitional unit. Trace fossils produced during deposition of the limestone, as indicated by light infilling material, are distinct and occur only in the uppermost few centimetres of the transitional unit (Extended Data Fig. 2).

The transitional unit microfossils are dominated by clearly reworked Maastrichtian foraminifera and nannoplankton, known across the Gulf of Mexico and Caribbean as the Cretaceous/Palaeogene (K/Pg) boundary cocktail 19 (Extended Data Fig. 3 and Supplementary Table 1). Although overall foraminiferal abundance (plotted as the number of foraminifera per gram of sedimentary rock; Fig. 1) is high at the base of the unit, species known to range across the boundary ('survivor species') are rare in the lower transitional unit and become more common up-section even as total foraminifera decline (Fig. 1). Survivor species, here defined as G. cretacea, Muricohedbergella monmouthensis and Muricohedbergella holmdelensis 20 , dominate a depauperate assemblage in the upper 20 cm of the transitional unit, coinciding with the first appearance of trace fossils (Extended Data Figs. 4,5).

The nannofossil assemblage in the transitional unit contains reworked Cretaceous specimens, including a group of clearly overgrown species (such as Aspidolithus parcus (also known as Broinsonia parca) and Eiffellithus eximius) that became extinct near the Campanian/ Maastrichtian boundary. The remainder of the Cretaceous species, which dominate the assemblage, range to the top of or beyond the latest Maastrichtian age (Supplementary Table 2). Unusually small (less than 2 μm) and delicate specimens of Micula are observed throughout the transitional unit and increase in abundance up-section (Fig. 1), along with small Retecapsa (Extended Data Fig. 6). Taxa common at other sites of the earliest Danian stage are also present, including disaster genera (opportunistic groups that can tolerate high environmental stress) such as Thoracosphaera and Braarudosphaera. Unlike the foraminifera, there are no clear stratigraphic trends in overall nannoplankton abundance (Fig. 1).

Because survivor species lived both before and after the K/Pg mass extinction, it is impossible to determine for certain whether individual specimens in the transitional unit colonized the crater after the impact. However, the populations of foraminifera and nannoplankton are substantially different from those of the latest Cretaceous 12 (that is, the expected population if the whole assemblage was reworked), suggesting that these taxa were true survivors (Fig. 1 and Extended Data Fig. 6). G. cretacea, a common component of the survivor assemblage in the upper transitional unit, was restricted to marginal marine waters during the Maastrichtian and would not have been present at the pre-impact site, which was over 100 m deep 21 and over 500 km from shore 22 . The nannofossil assemblage in the transitional unit is considerably different from typical latest Maastrichtian assemblages, with some genera over-represented (Watznaueria and Retecapsa) and others underrepresented (Eiffellithus, not including E. eximius, Arkhangelskiella, Chiastozygus and Prediscosphaera) (Extended Data Fig. 6). Additionally, Micula-a robust taxon often used as a proxy for dissolutionis not as abundant as elsewhere, indicating that these unusual abundances are not due to poor or selective preservation (Extended Data Fig. 6).

This initial appearance of life is notably fast, especially because crater-specific factors do not seem to have had a negative effect on the local recovery of life. A vigorous, high-temperature hydrothermal system was established within the crater and may have persisted for millions of years after the impact 23 , especially across the peak ring where rocks exhumed from deep in the crust were extensively fractured 7 . Nevertheless, the appearance of burrowing organisms within years of the impact indicates that the hydrothermal system did not adversely affect seafloor life. Impact-generated hydrothermal systems are hypothesized to be potential habitats for early life on Earth 24 and on other planets, particularly below the surface. However, for marine impact craters in open ocean communication, such as Chicxulub (Extended Data Fig. 1), our data indicate that locally substantial but comparatively small volumes of hydrothermal fluids were overwhelmed by the 1.3 × 10 4 km 3 of well-mixed ocean water that filled the basin.

Likewise, the open connection with the Gulf of Mexico prevented the development of anoxia in the crater. Our analyses of I/Ca ratios suggest that local dissolved oxygen was high and stable in Zone Pα (Fig. 3). This is in contrast to the smaller (85-km wide) Eocene Chesapeake Bay impact crater, where anoxia due to restriction is attributed as the cause of delayed recovery of the benthic ecosystem on the crater floor 25 . This comparison suggests that the establishment of life within marine impact craters is controlled more by circulation (and thus crater geometry) than by the magnitude of the impact or global environmental effects.

The overlying pelagic limestone, which was deposited within Zone Pα (30-200 kyr after the impact) contains abundant evidence of high productivity in a thriving ecosystem. The assemblage of planktic foraminifera in Zone Pα is diverse and abundant (Fig. 3). Good preservation in the lowermost sample (34 cm core depth) enabled the identification of over 60 species of benthic foraminifera, and benthics make up 12% of the total foraminiferal assemblage at this level (Supplementary Table 1). This percentage of benthics 26 and the overall benthic assemblage 27 are both typical of a palaeo-water depth of about 600-700 m (around the boundary between the upper and middle bathyal zones) 10,27 . At the base of the white limestone, trace fossils increase in size, abundance and diversity relative to the underlying Letter reSeArCH transitional unit. The abundance and diversity of benthic organisms indicate that by about 30 kyr after the impact, seafloor conditions had returned to normal and sufficient organic matter flux existed to sustain a diverse, multilayer benthic community.

Conversely, the nannoplankton assemblage in the Danian limestone is dominated by Braarudosphaera and calcareous dinoflagellate cysts (for example, Thoracosphaera), which are common disaster taxa in the early recovery interval. Large, foraminifer-sized calcispheres appear after about 100 kyr. Calcareous phytoplankton in the earliest Danian clearly represent a low-diversity, high-productivity bloom. Genera such as Neobiscutum and Prinsius, which are common bloom taxa in the recovery interval at other Northern Hemisphere sites, do not become common until several metres higher in the section, over one million years after the impact. Organic microfossils are completely absent from the study interval, probably owing to poor preservation of organic material.

Geochemical palaeoproductivity proxies, particularly Ba/Ti and Ba/ Fe ratios, also indicate high productivity in the post-impact Danian limestone (Fig. 1). Ba/Ti ratios of about 1.0 at the base of the limestone (approximately 30 kyr after the impact) and about 2.0 above that (15 cm higher, or about 100 kyr after the impact) indicate relatively high and increasing productivity in the Chicxulub basin in the earliest Danian.

The recovery of productivity in the crater is faster than that at many sites, including those in the Gulf of Mexico, some of which took 300 kyr or more to recover to a similar extent 8,11 . Therefore, we find that proximity to the impact was not a control on recovery in marine ecosystems. The wide range of rates of recovery in the oceans show no relationship with geographic distance to the crater and so are best explained by natural ecological interactions, such as incumbency and competitive exclusion, between organisms within recovery ecosystems 8,14 . These trends can be used to understand the rates of recovery after other major extinction events and to predict the long-term recovery of modern ecosystems affected by pollution and climate change.

Any Methods, including any statements of data availability and Nature Research reporting summaries, along with any additional references and Source Data files, are available in the online version of the paper at https://doi.org/10.1038/s41586-018-0163-6.

Sample size was determined according to standard community practice (collecting approximately 300 specimens per sample, when possible). No statistical methods were used to predetermine sample size. The experiments were not randomized and investigators were not blinded to allocation during experiments and outcome assessment.

The IODP-ICDP Expedition 364 drilled the peak ring of the Chicxulub crater in the spring of 2016 (Extended Data Fig. 1). Samples were taken at the Bremen IODP core repository during the Expedition 364 sampling party. Core depth in centimetres-with zero at the top of the section (616.24 m below sea floor)-are reported throughout. Core material was indurated, and ~0.5-cm quarter-rounds were cut out with a rock saw. Owing to the need to reserve core material for rare earth element geochemistry (data not shown), the lowermost ~1.5 cm of the Danian limestone was not sampled. Individual samples were subdivided for foraminifera, calcareous nannoplankton and discrete geochemical analyses.

Forty-three samples were examined for planktic and benthic foraminifera from core 40 from 0-110 cm depth. Samples were weighed, crushed with a mortar and pestle, soaked overnight (or longer) in a 10% solution of hydrogen peroxide buffered with borax and washed over a 43-μm sieve to ensure capture of small Danian taxa. The sieve was soaked in methylene blue dye between samples to identify contaminated specimens. Samples were then dried in an oven, split to obtain a manageable volume of material, and examined for foraminifera, calcispheres, and other sand-sized particles. In the Danian limestone, at least 300 specimens were counted to establish a statistically robust population 28 and the rest of the residue was then examined for biostratigraphically important taxa. Low abundances in the transitional unit precluded 300-specimen counts. However, we demonstrate that our values are sufficient to reject the null hypothesis (that the observed enrichments in survivor taxa are the result of random noise) with binomial confidence limits. This calculation traditionally provides the basis for the 300-specimen 'rule': counting 300 specimens provides statistical confidence at a 95% confidence interval that a species that makes up 1% of the population is represented in the count 28 . As we show, fewer specimens are sufficient to demonstrate the presence of a survivor population in our samples. Binomial confidence limits for samples with fewer than 300 specimens are reported in Supplementary Table 1. Additionally, a single unusually well-preserved sample at the base of the post-impact limestone was examined for rare benthic species to determine the true diversity of benthic foraminifera at the base of the unit (Supplementary Table 1). Planktic foraminifer biozonation follows the P zones of Berggren and Pearson 29 as modified by Wade et al. 18 .

Ninety-seven samples were examined for nannofossils. Samples were disaggregated in water, and smear slides were made from the supernatant. Slides were observed in a transmitted light microscope at 1,600× until at least 100 specimens were observed (Supplementary Table 2). Standard taxonomy was applied (http:// www.mikrotax.org/Nannotax3/index.php?dir=Coccolithophores). The abundance of taxa at site M0077 was compared to a previous compilation of global K/ Pg nannoplankton 12 .

Ichnological analysis was conducted from 0-110 cm. Ichnological observations were conducted on core material and a detailed and continuous analysis of digital images. To improve visibility of ichnological features, images were treated by a digital image methodology, based on the modification of image adjustments as levels, brightness and vibrance 30,31 . Ichnotaxonomical classification of trace fossils was based on the overall shape and the presence of diagnostic criteria such as size and presence of branches 32 . Special attention was given to the infilling material of biogenic structures.

The measurement of I/(Ca + Mg) was carried out using a procedure similar to a previously described method 33 . For each sample and geostandard, approximately 3-4 mg of carbonate powder was weighed out, dissolved in ~0.45 M nitric solution and then diluted using 0.1 M nitric acid and 0.5% TMAH solution. All reported measurements are from samples that had a matrix of 50 ± 5 p.p.m. calcium solution to ensure the most precise iodine measurement. Dissolved samples had TMAH solution added within an hour to avoid any possible loss of volatilized iodine 33 . Samples were measured using an Agilent inductively coupled plasma mass spectrometer 7500 cs housed within the geochemistry group of the National High Magnetic Field Laboratory at Florida State University. A previously reported known sample, Key Largo (KL 1-1) was used to ensure reliable reproducibility. Our value of 5.51 μmol/mol was within error of the reported value of 5.55 μmol/ mol. A previous study 34 found that generally low oxygen conditions correspond to ~2.6 μmol/mol for I/(Ca + Mg). Values are reported in Supplementary Table 3.

Section 1 of core 40 was scanned with an AVAATECH XRF Core Scanner II at MARUM (Bremen, Germany) during the onshore phase of Expedition 364 (Fig. 1). The split core was covered with a 4-μm-thick SPEX CertiPrep Ultralene foil to avoid contamination. X-ray fluorescence data were acquired with a Canberra X-PIPS silicon drift detector with 1550 eV resolution, a Canberra DAS 1000 dig-ital spectrum analyser and an Oxford Instruments 50 W XTF011 X-ray tube with rhodium target material. X-ray spectra were processed with WIN AXIL software from Canberra Erisys at a resolution of 12 mm and a step of 10 mm. Scans were conducted at different voltages to determine a range of element concentrations: 50 kV, with a beam current of 1 mA (Ba and Sr; average dead time of 5%), and 10 kV with a beam current of 0.15 mA (Al, Si, K, Ca, Ti, Fe, Mn and S; average dead time of 11%). For each scan, sampling time was 20 s per spot. He is delivered to the Earth's surface by cosmic dust grains and over short time spans (about one million years) can be used as a constant flux proxy 35 . Previous work has shown that the K/Pg impactor was not associated with enhanced 3 He flux, and the mean extraterrestrial 3 He flux from cosmic dust accretion at the end of the Cretaceous (106 × 10 -15 cc (standard temperature and pressure) per g per cm 2 per kyr) was used to estimate the duration over which the K/Pg boundary clay was deposited at Gubbio and El Kef 36 . We use a similar approach here to establish the sedimentation rate of the transitional unit, which we use to develop an age model.

Helium isotope ratios and concentrations were measured on ~1-g aliquots of sediment following standard analytical procedures 31 . Extraterrestrial 3 He concentrations were computed from measured He isotopic compositions using an isotopic deconvolution model 36 . Results are shown in Extended Data Table 1. 3 He concentrations and 3 He/ 4 He ratios are generally low compared to typical marine sediments of similar age 37,38 . Nevertheless, with the exception of the lowest sample in the transitional unit (106.5 cm), the fraction of 3 He attributable to an extraterrestrial source is high, ranging from ~0.70 to 0.96. The deepest sample has a similar 3 He concentration to other samples in the transitional unit, but ~5 times more 4 He. This elevated 4 He probably arises from a higher concentration of terrigenous 4 He-bearing material deposited rapidly after the impact.

We see no evidence for extraterrestrial He carried in impactor fragments, such as highly elevated and/or highly variable 3 He and 3 He/ 4 He ratios. The absence of such a signal is consistent with either (a) the absence of impactor fragments in the material analysed or (b) the loss of extraterrestrial 3 He from the impactor via heating, vaporization or fusion. Note that, unlike many tracers of the impactor (such as Ir), deposition of fused or vaporized impactor will leave no trace in the sedimentary record because once He is lost into the atmosphere, it can no longer be retained in sediments. Reporting summary. Further information on experimental design is available in the Nature Research Reporting Summary linked to this paper.

Extended Data Fig. 1 | Location of site M0077 in the Chicxulub crater as seen using gravity data. Black dots are cenotes. Modified from Gulick et al. 21 . Life Sciences Reporting Summary Nature Research wishes to improve the reproducibility of the work that we publish. This form is intended for publication with all accepted life science papers and provides structure for consistency and transparency in reporting. Every life science submission will use this form; some list items might not apply to an individual manuscript, but all fields must be completed for clarity.

For further information on the points included in this form, see Reporting Life Sciences Research. For further information on Nature Research policies, including our data availability policy, see Authors & Referees and the Editorial Policy Checklist.

Describe how sample size was determined. Sample size was determined using standard population sizes for micropaleontological datasets (e.g., Buzas, 1990, J. of Paleo) except for samples which contained fewer than 300 specimens. These were all in the transitional unit, and were only analyzed for the proportion of 2 overall groups (survivors vs. non survivors; Guembelitria vs. everything else) and thus did not require the 300 specimen count needed for more advanced ecological analysis. We show this by reporting binomial confidence intervals for these data in Tables S2 andS3 2. Data exclusions Describe any data exclusions. We did not include benthic foraminifera in the %survivors foraminifer dataset (i.e., it is only %planktic survivors out of the total population of planktic foraminifera). Benthics did not experience an extinction at the KPg boundary, so *all* benthics are technically survivors and we felt this would bias the data.

Describe whether the experimental findings were reliably reproduced.

It is hard to judge the reproducibility of paleontological samples, as the act of picking foraminifera from sample changes the remaining population. However, we had several replicate samples from the same depth interval, and these produced very similar proportions of the groups measured (e.g., %benthics, % Parvularugoglobigerina, %Chilguembelina, etc.). Analytical runs for geochemical samples included standards of known composition to constrain error and ensure reproducibility. (see discussion in Methods section).

Describe how samples/organisms/participants were allocated into experimental groups.

Samples were taken at a regular interval throughout the core. Foraminifer samples were split with a microsplitter to obtain a representative subsample for population counts (when the population was >300); nannoplankton were counted in fields of view along a complete transect.

Describe whether the investigators were blinded to group allocation during data collection and/or analysis.

No blinding was utlized in this paleoecological/geochemical study.

Note: all studies involving animals and/or human research participants must disclose whether blinding and randomization were used.