Evidence for an extraterrestrial impact 12,900 years

ago that contributed to the megafaunal extinctions

and the Younger Dryas cooling
R. B. Firestonea,b, A. Westc, J. P. Kennettd, L. Beckere, T. E. Bunchf, Z. S. Revayg, P. H. Schultzh, T. Belgyag, D. J. Kennetti,
J. M. Erlandsoni, O. J. Dickensonj, A. C. Goodyeark, R. S. Harrish, G. A. Howardl, J. B. Kloostermanm, P. Lechlern,
P. A. Mayewskio, J. Montgomeryj, R. Poredap, T. Darrahp, S. S. Que Heeq, A. R. Smitha, A. Stichr, W. Toppings,
J. H. Wittkef, and W. S. Wolbachr
aLawrence Berkeley National Laboratory, Berkeley, CA 94720; cGeoScience Consulting, Dewey, AZ 86327; dDepartment of Earth Sciences and eInstitute of
Crustal Studies, University of California, Santa Barbara, CA 93106; fNorthern Arizona University, Flagstaff, AZ 86011; gInstitute for Isotope and Surface
Chemistry, H-1525, Budapest, Hungary; hDepartment of Geological Sciences, Brown University, Providence, RI 02912; iDepartment of Anthropology and
Museum of Natural and Cultural History, University of Oregon, Eugene, OR 97403; jEastern New Mexico University, Portales, NM 88130; kSouth Carolina
Institute of Archaeology and Anthropology, University of South Carolina, Columbia, SC 29208; lRestoration Systems, LLC, Raleigh, NC 27604; mRozenstraat
85, 1018 NN, Amsterdam, The Netherlands; nBureau of Mines and Geology, University of Nevada, Reno, NV 89557; oClimate Change Institute, University
of Maine, Orono, ME 04469; pUniversity of Rochester, Rochester, NY 14627; qDepartment of Environmental Health Sciences, University of California,
Los Angeles, CA 90095; sP.O. Box 141, Irons, MI 49644; and rDepartment of Chemistry, DePaul University, Chicago, IL 60614

Communicated by Steven M. Stanley, University of Hawaii at Manoa, Honolulu, HI, July 26, 2007 (received for review March 13, 2007)

A carbon-rich black layer, dating to ⬇12.9 ka, has been previously be dramatically revealed by explaining that all were gone an instant
identified at ⬇50 Clovis-age sites across North America and appears before the black mat was deposited.’’
contemporaneous with the abrupt onset of Younger Dryas (YD) The cause of this extinction has long been debated and remains
cooling. The in situ bones of extinct Pleistocene megafauna, along highly controversial due, in part, to the limitations of available data
with Clovis tool assemblages, occur below this black layer but not but also because the two major competing hypotheses, human
within or above it. Causes for the extinctions, YD cooling, and overkill (5) and abrupt cooling (6), fall short of explaining many
termination of Clovis culture have long been controversial. In this observations. For example, Grayson and Meltzer (7) summarized
paper, we provide evidence for an extraterrestrial (ET) impact event serious problems with the overkill hypothesis, such as the absence
at ⬵12.9 ka, which we hypothesize caused abrupt environmental of kill sites for 33 genera of extinct mammals, including camels and
changes that contributed to YD cooling, major ecological reorgani- sloths. In addition, although abrupt cooling episodes of magnitudes
zation, broad-scale extinctions, and rapid human behavioral shifts at similar to the YD occurred often during the past 80 ka, none are
the end of the Clovis Period. Clovis-age sites in North American are
known to be associated with major extinctions. The possibility of
overlain by a thin, discrete layer with varying peak abundances of (i)
pandemic disease also has been suggested (8), but there is no
evidence for that in the Pleistocene record. Thus, the end-
magnetic grains with iridium, (ii) magnetic microspherules, (iii) char-
Pleistocene extinction event is unique within the late Quaternary
coal, (iv) soot, (v) carbon spherules, (vi) glass-like carbon containing
and is unlikely to have resulted only from climatic cooling and
nanodiamonds, and (vii) fullerenes with ET helium, all of which are
human overkill. The extinctions were too broad and ecologically
evidence for an ET impact and associated biomass burning at ⬇12.9 deep to support those hypotheses.
ka. This layer also extends throughout at least 15 Carolina Bays, which Extraterrestrial (ET) catastrophes also have been proposed. For
are unique, elliptical depressions, oriented to the northwest across example, LaViolette (9) suggested that a large explosion in our
the Atlantic Coastal Plain. We propose that one or more large, galactic core led to the extinctions. Brakenridge (10) postulated that
low-density ET objects exploded over northern North America, par- a supernova killed the megafauna and caused the worldwide
tially destabilizing the Laurentide Ice Sheet and triggering YD cooling. deposition of the black layer. Clube and Napier (11) proposed
The shock wave, thermal pulse, and event-related environmental multiple encounters with remnants of the mega comet progenitor
effects (e.g., extensive biomass burning and food limitations) con- of the Taurid meteor stream and Comet Encke. Although ET
tributed to end-Pleistocene megafaunal extinctions and adaptive events have long been proposed as a trigger for mass extinctions,
shifts among PaleoAmericans in North America. such as at the K/T (⬇65 Ma) (12) and P/T (⬇250 Ma) (13), there
has been no compelling evidence linking impacts to the late
comet 兩 iridium 兩 micrometeorites 兩 nanodiamonds 兩 spherules Pleistocene megafaunal extinctions and YD cooling.
In the 1990s, W. Topping (14) discovered magnetic microspher-
carbon-rich black layer, dating to ⬵12.9 ka (12,900 calendar ules and other possible ET evidence in sediment at the Gainey
A years B.P.) (1), has been identified by C. V. Haynes, Jr. (2), at
⬎50 sites across North America as black mats, carbonaceous silts, Author contributions: R.B.F., A.W., J.P.K., L.B., and W.T. designed research; R.B.F., A.W.,
or dark organic clays [supporting information (SI) Fig. 5]. The age J.P.K., L.B., T.E.B., Z.S.R., P.H.S., D.J.K., J.M.E., O.J.D., A.C.G., R.S.H., G.A.H., J.B.K., P.L.,
of the base of this black layer coincides with the abrupt onset of P.A.M., J.M., R.P., T.D., S.S.Q.H., A.R.S., A.S., W.T., J.H.W., and W.S.W. performed research;
R.B.F., A.W., J.P.K., L.B., T.E.B., Z.S.R., T.B., D.J.K., O.J.D., A.C.G., G.A.H., J.B.K., P.L., J.M., R.P.,
Younger Dryas (YD) cooling, after which there is no evidence for
S.S.Q.H., W.T., J.H.W., and W.S.W. contributed new reagents/analytic tools; R.B.F., A.W.,
either in situ extinct megafaunal remains or Clovis artifacts. In- J.P.K., L.B., T.E.B., Z.S.R., P.H.S., D.J.K., J.M.E., R.S.H., G.A.H., P.A.M., R.P., T.D., S.S.Q.H.,
creasing evidence suggests that the extinction of many mammalian A.R.S., A.S., W.T., J.H.W., and W.S.W. analyzed data; and R.B.F., A.W., J.P.K., and P.H.S.
and avian taxa occurred abruptly and perhaps catastrophically at wrote the paper.

the onset of the YD, and this extinction was pronounced in North The authors declare no conflict of interest.
America where at least 35 mammal genera disappeared (3), in- Freely available online through the PNAS open access option.
cluding mammoths, mastodons, ground sloths, horses, and camels, Abbreviations: YD, Younger Dryas; YDB, YD boundary; ET, extraterrestrial.
along with birds and smaller mammals. At Murray Springs, AZ, a bTo whom correspondence should be addressed. E-mail.com rbfirestone@lbl.gov.
well known Clovis site, mammoth bones and Clovis-age stone tools This article contains supporting information online at www.pnas.org/cgi/content/full/
lie directly beneath the black layer where, as described by Haynes 0706977104/DC1.
(4): ‘‘[T]he sudden extinction of the Pleistocene megafauna would © 2007 by The National Academy of Sciences of the USA

16016 –16021 兩 PNAS 兩 October 9, 2007 兩 vol. 104 兩 no. 41 www.pnas.org兾cgi兾doi兾10.1073兾pnas.0706977104

Table 1. Information about the YDB research sites, along with concentrations of selected YDB markers
Carbon Magnetic microspherules Magnetic grains Bulk

Clovis-age Date, Spherules Ni, IrM, IrB,

YDB Sites ka Misc. markers #/kg #/kg FeO, % TiO2, % g/kg H 2O FeO, % TiO2 ppm ppb ppb
Gainey, MI ⬇12.4 AGC 1,232 2,144 41 25 3.2 3.2 14 1.6 54 ⬍2 ⬍0.5
Murray Springs, AZ 12.99 AKGCFPSB 0 109 2.6 5.1 21 16 40 ⬍1 2.3
Blackwater Draw, NM 12.98 AKGCFPB 0 768 56 33 2.1 1.5 27 8.1 256 24 2.3
Chobot, AB ⬇13 AGCB 11 578 1.9 5.0 14 0.9
Wally’s Beach, AB 12.97 AK — 6 7.8 1.6 41 8.3 190 51 ⬍1
Topper, SC ⬍13.5 AG 2 97 1.1 0.7 25 49 440 2 ⬍1
Lommel, Belgium 12.94 ACB 0 16 14 67 0.8 0.8 23 21 23 117 ⬍1
Morley Drumlin, AB ⬇13 GCB 16 1,020 60 29 9.9 3.7 14 1.4 240 ⬍0.1
Daisy Cave, CA 13.09 GCFPB ⬎0 ⬎0 ⬎0 ⬍1
Lake Hind, MB 12.76 GCB 184 0 0.3 3.8
Carolina Bays, Min GCFS 142 20 0.5 0.3 18 21 ⬍1 0.5
Carolina Bays, Max 1,458 205 17 1.3 26 34 ⬍200 15 3.8

Radiocarbon ages are calibrated. More site age information is in SI Table 2. Percentages are by weight. A, artifacts from Clovis or contemporaries; K,
megafaunal kill-site; G, glass-like carbon; C, charcoal; F, fullerenes with ET He-3; S, soot; P, polycylic aromatic hydrocarbons; B, black mat; Ni, nickel in magnetic
fraction; IrM, Ir in magnetic fraction; IrB, Ir in bulk sediment. No measurable Ir was found outside the YDB. Ir uncertainties are ⫾10% at 117 ppb and ⫾90% at
0.5 ppb. Geochemical values are less than ⫾20%.

PaleoAmerican site in Michigan (see also ref. 15), and Lougheed group of ⬇500,000 elliptical lakes, wetlands, and depressions that
(16) and Bi (17) reported that late Pleistocene glacial drift con- are up to ⬇10 km long and located on the Atlantic Coastal Plain (SI
tained similar cosmic spherules. We now report substantial addi- Fig. 7). We sampled these sites because Melton, Schriever (20), and
tional data from multiple, well dated stratigraphic sections across Prouty (21) proposed linking them to an ET impact in northern
North America supporting a major ET airburst or collision near North America. However, some Bay dates are reported to be ⬎38
12.9 ka. Directly beneath the black mat, where present, we found ka (22), older than the proposed date for the YD event.
a thin, sedimentary layer (usually ⬍5 cm) containing high concen- Each of the 10 Clovis-age sites displays a YDB layer (average
trations of magnetic microspherules and grains, nanodiamonds, thickness of 3 cm) that contains a diversity of markers (magnetic
iridium (Ir) at above background levels, and fullerenes containing microspherules and grains, charcoal, soot, carbon spherules, glass-
ET helium. These indicators are associated with charcoal, soot, like carbon, nanodiamonds, and fullerenes with ET helium). The Ir
carbon spherules, and glass-like carbon, all of which suggest intense levels are above background in both bulk sediment and magnetic
wildfires. Most of these markers are associated with previously fractions at up to 117 parts per billion (ppb), which is 25% of levels
recorded impacts, but a few are atypical of impact events. We in CI (Ivuna type) chondritic meteorites (23). The YDB also
identify this layer as the YD boundary (YDB), and we refer to this exhibits uranium (U) and thorium (Th) in high concentrations that
incident as the YD event. are up to 25⫻ crustal abundance. At the 15 Bay sites examined,
At the sites studied, independent radiocarbon (1) and optically basal sediments and rim sands contain peaks in the same ET
stimulated luminescence dates that tend to cluster near 13 ka were assemblage found in the YDB at Clovis sites elsewhere.
used to establish the age of the YDB. For example, the end-Clovis
stratum (the YDB) is well dated at Murray Springs, AZ, (eight dates YD Event Markers. The various markers are summarized in Table 1
averaging 10,890 14C yr or calendar 12.92 ka) and the nearby Lehner and described in SI Text, ‘‘Research Sites.’’ Seven representative
site (12 dates averaging 10,940 14C yr or 12.93 calendar ka). Haynes North American sediment profiles are shown in Fig. 1.
(2) correlated the base of the black mat (the YDB) with the onset Magnetic microspherules. Magnetic microspherules measuring 10–
of YD cooling, dated to 12.9 ka in the GISP2 ice core, Greenland 250 ␮m peaked in or near the YDB at eight of nine Clovis-age sites
(see GISP2 chronology in SI Fig. 6) (18). Therefore, we have and in sediments from five of five Bays tested. Fig. 2 shows
adopted a calendar age of 12.9 ⫾ 0.1 ka for the YD event. representative microspherules from Canada, New Mexico, Michi-
We propose that the YD event resulted from multiple ET gan, and North Carolina. Several sites also yielded microspherules
airbursts along with surface impacts. We further suggest that the that appear to be silicates, requiring further analysis. Microspherule
catastrophic effects of this ET event and associated biomass burn- abundances average 390 per kilogram and are highest in the north,
ing led to abrupt YD cooling, contributed to the late Pleistocene ranging up to 2,144 per kilogram at Gainey. Analyses from Gainey,
megafaunal extinction, promoted human cultural changes, and led the Morley drumlin, and Blackwater Draw found the microspher-
to immediate decline in some post-Clovis human populations (19). ules to be enriched in titanomagnetite.
Magnetic grains. Magnetic grains measuring 1–500 ␮m, irregularly
Results
GEOPHYSICS

shaped and often subrounded, are more abundant than micro-

Research Sites. Ten Clovis and equivalent-age sites were selected spherules, and they show a distinct peak in the YDB at all 10
because of their long-established archeological and paleontological Clovis-age sites and are in all 15 Bays, reaching peaks above the
significance, and, hence, most are well documented and dated by pre-Bay paleosols at four sites. All had lower abundances at other
previous researchers (see SI Table 2). Two are type-sites where stratigraphic levels. Magnetic grains are mostly dark brown or black,
unique PaleoAmerican projectile point styles were first named: the although the magnetic fraction often contains terrestrial silicates
Clovis-point style at Blackwater Draw, NM, and the Gainey-point with magnetite inclusions. Concentrations of magnetic grains and
style at Gainey, MI. Three of the sites are confirmed megafaunal microspherules vary greatly between YDB sites, averaging 3.4 g/kg,
kill sites, and six of 10 have a black mat overlying the YDB. At with higher abundances at northern sites, such as Gainey, Chobot,
Blackwater Draw and Murray Springs, the YDB is found directly and the Morley drumlin. Lower abundances were found in the
beneath the black mat and overlying Clovis artifacts with extinct Carolinas and the southwestern U.S. Magnetic grains from south-
megafaunal remains. ern sites and Lommel, along with some YDB microspherules, are
The other sample sites were in and around 15 Carolina Bays, a enriched in titanomagnetite.

Firestone et al. PNAS 兩 October 9, 2007 兩 vol. 104 兩 no. 41 兩 16017

 Fig. 1. Sediment profiles for seven
sites. Concentrations are shown for
magnetic grains, microspherules,
charcoal, soot, glass-like carbon,
carbon spherules, Ir, Cr, and Ni,
which peak mostly in a narrow
stratigraphic section spanning only
a few hundred years. Ir open circles
indicate values below detection,
typically ⬍0.5–1 ppb. Ir uncertain-
ties are ⫾10% at 117 ppb and ⫾90%
at 2 ppb. Cr and Ni are less than
⫾20%. Keys are color-coded to
match the respective curves, and
graph points correspond to sam-
pling locations on the photograph.
The depth is in centimeters above or
below the YDB. The Blackwater
Draw image is a composite of three
photos. There is no photo for
Gainey. A profile for the Belgian site
at Lommel is shown in SI Fig. 8. The
locations of all sites that were sam-
pled are shown in SI Fig. 9.

Iridium and nickel. YDB sediments, but not the magnetic fractions, Of eight sites examined, soot was observed only in the YDB at two
are modestly enriched in Ni. For Ir, YDB magnetic grains from sites, Murray Springs (21 ⫾ 7 ppm) and Bay T13 (1,969 ⫾ 167 ppm),
seven of 12 sites exhibited a range of 2 (⫾90%) to 117 (⫾10%) ppb, where preservation possibly resulted from anoxic burial conditions.
and of those seven sites, three also had detectable Ir in the YDB In addition, the combustion of wood at very high temperatures
bulk sediment. The highest Ir value is ⬇25% that of typical produces diagnostic PAHs. High-temperature PAHs, which were
chondrites (455–480 ppb) (24) and ⬎5,000⫻ crustal abundance found at the K/T boundary (28), are present in the YDB, but not
(0.02 ppb) (25). In 17 measurements at these sites, no Ir was above or below it at each of three sites analyzed (Daisy Cave,
detected in magnetic grains above or below the YDB. For bulk Murray Springs, and Blackwater Draw), suggesting that intense fires
sediment, YDB Ir abundances at five of 12 sites range from 0.5 occurred at these locations.
(⫾90%) to 3.75 (⫾50%) ppb. However, the bulk sediment results Carbon spherules. Carbon spherules (0.15–2.5 mm) are black,
are near the detection limits of neutron activation analysis, and highly vesicular, subspherical-to-spherical objects (Fig. 3). SEM
further testing is required. analyses show them to have cracked and patterned surfaces, a
Upon retesting aliquots of high-Ir samples, five from nine sites thin rind, and honeycombed (spongy) interiors. SEM/energy
were confirmed, but Ir abundances were below detection in four dispersive spectrometer and microprobe analyses show that the
retests. Sample sizes were small, and variations are likely due to the spherules are dominantly carbon (⬎75%), with no evidence of
‘‘nugget effect.’’ In summary, no detectable Ir was found above or seed-like morphology or cellular plant structure, as in charcoal.
below the YDB and black mat at seven sites in 62 samples of both They were found in 13 of 15 Bays and only in the YDB at six of
bulk sediments and magnetic grains. Elevated Ir concentrations nine Clovis-age sites in concentrations up to ⬇1,500 per kilo-
were found only in the YDB and black mat at nine of 14 widely gram. In addition, we recovered them from one of four modern
separated sites (see Fig. 1, Table 1, and SI Table 3). forest fires (see SI Text, ‘‘Research Sites’’), confirming that they
Charcoal. Charcoal displays peaks in the YDB at eight of nine can be produced by intense heat in high-stand wildfires. At the
Clovis-age sites and is present in 15 of 15 Bays, reaching peaks in P/T boundary, Miurat discovered carbon spherules up to 90 wt%
four Bays with paleosols. The charcoal was identified optically and C and up to 20 ␮m in size, which he attributes to a controversial
by SEM based on its distinctive cellular structure and was found in cosmic impact ⬇250 Ma. More recently, Rösler et al.u reported
concentrations ranging from 0.06 to 11.63 g/kg.
finding carbon spherules from undated sediment across Europe,
Soot and polycyclic aromatic hydrocarbons (PAHs). Observed at the K/T
and these appear identical to spherules from the YDB layer. The
boundary (26) and distinguished by its aciniform morphology (see
authors report that they contain fullerenes and nanodiamonds,
SI Fig. 10) (27), soot forms only in flames through direct conden-
the latter of which are extraordinarily rare on Earth but are
sation of carbon from the gas phase. Soot was identified by using
found in meteorites and at ET impact sites (29), leading those
SEM imaging and quantified by particle size analysis and weighing.
authors to propose an ET association for the carbon spherules.
Fullerenes and ET helium. Of four sites analyzed, fullerenes with ET
helium, which are associated with meteorites and ET impacts
(30), were present in YDB sediments at three Clovis-age sites
(Blackwater, Murray Springs, and Daisy Cave). In Bay M33, they

tMiura,
Y., 37th Annual Lunar and Planetary Science Conference, March 13–17, 2006,
League City, TX, Vol. 2441, pp. 1–2 (abstr.).
Fig. 2. High-titanomagnetite microspherules from Blackwater Draw, NM uRösler, W., Hoffmann, V., Raeymaekers, B., Yang, Z. Q., Schryvers, D., Tarcea, N. (2006) First
(120 ␮m) (a); Chobot, AB, Canada (150 ␮m) (b), Gainey, MI (90 ␮m) (c), and International Conference on Impact Cratering in the Solar System, May 7–12, 2006,
Howard Bay, NC (100 ␮m) (d). Noordwijk, The Netherlands, abstr. 295464.

16018 兩 www.pnas.org兾cgi兾doi兾10.1073兾pnas.0706977104 Firestone et al.

 Fig. 4. Examples of glass-like carbon from Gainey, Bay M31, and Topper.

Fig. 3. Low-density carbon spherules are shown whole from the Chobot site
(a), sectioned and by SEM from Bay T13 (b), and at high magnification by SEM
from Bay B14 (c).
increasing distance from the Great Lakes region (see SI Fig. 12).
This nonrandom distribution is unlikely to be due to terrestrial
factors or interplanetary dust storms, but it is consistent with
also were found in glass-like carbon with an ET helium ratio that airburst/impacts over northern North America.
is 84 times that of air. By comparison, the ratio of the Tagish
Lake meteorite was 90 times that of air. Magnetic Microspherules and Grains. High concentrations of micro-
spherules (glass, clinopyroxene, spinel, or metallic) are accepted as
Glass-Like Carbon. Pieces up to several cm in diameter (Fig. 4) were evidence for at least 11 older ET impact events (33). Alternately,
found associated with the YDB and Bays, and their glassy texture microspherules are sometimes associated with volcanism, but when
suggests melting during formation, with some fragments grading YDB microspherules were analyzed by SEM/x-ray fluorescence
into charcoal. Continuous flow isotope ratio MS analysis of the and compared with known cosmic and volcanic microspherules (34,
glass-like carbon from Carolina Bay M33 reveals a composition 35), they appear to be nonvolcanic in origin. Analysis suggests an
mainly of C (71%) and O (14%). Analysis by 13C NMR of the ET origin, but because of high titanium (Ti) concentrations, the
glass-like carbon from Bay M33 finds it to be 87 at.% (atomic microspherules differ from typical meteoritic ones.
percent) aromatic, 9 at.% aliphatic, 2 at.% carboxyl, and 2 at.% The magnetic grains and microspherules are anomalously en-
ether, and the same sample contains nanodiamonds, which are riched in Ir and Ti (see Table 1 and SI Table 5) and are enriched
inferred to be impact-related material (see SI Fig. 11). Concentra- in water (up to 28 at.%), especially at northern sites. TiO2/FeO
tions range from 0.01 to 16 g/kg in 15 of 15 Bays and at nine of nine ratios of microspherules (0.48 ratio) and magnetic grains (0.76) are
Clovis-age sites in the YDB, as well as sometimes in the black mat, 4- to 250-fold higher than Alaskan terrestrial magnetite (⬍0.12
presumably as reworked material. Somewhat similar pieces were ratio in 347 samples) (36), crustal abundance (0.13) (25), CI
found in four modern forest fires studied (see SI Text, ‘‘Research chondrites (0.003) (23), and K/T impact layers (0.07) (12). These
Sites’’). ratios and the similarity in composition of YDB magnetic micro-
Quantities for selected markers are shown in Table 1, and spherules and magnetic grains (e.g., high Ti) from many sites across
abundances of all markers are given in SI Table 4. North America cannot be explained at this time, but the YDB
abundance of microspherules and magnetic grains most likely
Discussion resulted from the influx of ejecta from an unidentified, unusually
Age of the YDB. The YDB at the 10 Clovis- and equivalent-age sites Ti-rich, terrestrial source region and/or from a new and unknown
has been well dated to ⬇12.9 ka, but the reported ages of the type of impactor.
Carolina Bays vary. However, the sediment from 15 Carolina Bays
studied contain peaks in the same markers (magnetic grains, Carbon-Rich Markers. At Murray Springs, Haynes (37) first reported
microspherules, Ir, charcoal, carbon spherules, and glass-like car- the presence of glass-like or ‘‘vitreous’’ carbon in the black mat. In
bon) as in the YDB at the nearby Topper Clovis site, where this addition, he chemically analyzed the black mat layer, concluding
assemblage was observed only in the YDB in sediments dating back that it most likely resulted from the decomposition of charred wood
⬎55 ka. Therefore, it appears that the Bay markers are identical to and/or a prolonged algal bloom, both of which could result from
those found elsewhere in the YDB layers that date to 12.9 ka. event-related processes (e.g., climate change and biomass burning).
Although the Bays have long been proposed as impact features, Some black mats have no algal component, only charcoal. The
they have remained controversial, in part because of a perceived widespread peaks of charcoal in or near the YDB, and their
absence of ET-related materials. Although we now report that Bay association with soot and polycyclic aromatic hydrocarbons at
sediments contain impact-related markers, we cannot yet deter- specific sites, provide strong evidence for extensive wildfires. We
mine whether any Bays were or were not formed by the YD event. propose that glass-like carbon, carbon spherules, and nanodia-
monds were produced in the YDB by high temperatures resulting
Peaks in Markers. We investigated whether peaks in YDB markers from the impact and associated biomass burning.
might be attributed to terrestrial processes. The 25 sites examined
represent a wide range of depositional environments (fluvial, Ir Anomaly. Ir concentrations in sediments and ocean cores are high
lacustrine, eolian, alluvial, colluvial, and glacial), soil conditions for many accepted impact events, such as for the K/T and Chesa-
(aerobic/well drained to anaerobic/saturated), sediment composi- peake Bay (⬇36 Ma) (38). However, Ir values in the YDB bulk
GEOPHYSICS

tion (dense clay to gravelly sand), climatic regimes (semiarid to sediment are lower than at many K/T sites (e.g., 9.1 ppb at Gubbio,
periglacial), and biomes (grasslands to forests). The presence of Italy) (12), suggesting much less Ir in the YD impactor. The
identical markers found under such a wide range of conditions evidence indicates an Ir anomaly in both the YDB bulk sediment
argues against formation by terrestrial processes and is consistent and the magnetic fraction; however, for Ir in the bulk sediment, the
with an impact origin. We also examined whether the YDB might level of uncertainty remains high (⫾50–90%), in contrast to the
represent an interval of reduced deposition, allowing the accretion magnetic fraction, where values have higher certainty (up to
of interplanetary dust particles enriched in ET markers, such as Ir, ⫾10%), and are, therefore, more compelling. In 169 measurements
Ni, and ET helium. At Blackwater Draw, based on 24 calibrated 14C at 14 sites up to ⬇9,200 km apart, Ir was detected only in the YDB
dates from 13.30 to 10.99 ka, Haynes (31, 32) suggested that any sediments, YDB magnetic fraction, and the black mat. Ir never was
hiatus at the level representing the YDB most likely lasted less than detected above or below these layers, lessening concerns about the
a decade, which is insufficient to have produced a local Ir bulk high uncertainties, while providing strong evidence that Ir concen-
sediment level that is ⬎100⫻ crustal abundance. Furthermore, trations are above background in the YDB or black mat. The
abundances of microspherules and magnetic grains decrease with relatively low Ir and Ni peaks associated with the YDB are more

Firestone et al. PNAS 兩 October 9, 2007 兩 vol. 104 兩 no. 41 兩 16019

consistent with the generally proposed composition of comets and insufficient to fully understand impact dynamics, we can offer
inconsistent with the high-Ir content typical of most stony, nickel– speculation for future work.
iron, or chondritic meteorites. Toon et al. (46) suggest that an impact capable of continent-wide
Alternately, Ir peaks are found at major geologic boundary layers damage requires energy of 107 megatons, equivalent to an impact
with no confirmed impacts, and at least some of those Ir concen- by a ⬎4-km-wide comet (figure 1 in ref. 46). Although an impactor
trations may have resulted from volcanism. However, no major that size typically leaves an obvious large crater, no such late
North American volcanic episode is known at 12.9 ka, and, accord- Pleistocene crater has been identified. The lack of a crater may be
ing to Koeberl (39), such events produce Ir abundances of ⬍0.5 ppb, due to prior fragmentation of a large impactor, thereby producing
much less than we find in the YDB. Therefore, the high concen- multiple airbursts or craters. Hypervelocity oblique impact exper-
trations of Ir do not appear to be of volcanic origin. iments (P.H.S., unpublished data) indicate that a low-impedance
We also considered microbial concentration from Ir-rich adja- surface layer, such as an ice sheet, can markedly reduce modifica-
cent sediment, such as occurred in experiments by Dyer et al. (40), tion of the underlying substrate if the layer is equal to the projectile’s
who cultured microbes in Ir-rich igneous rocks and meteoritic diameter. These results suggest that if multiple 2-km objects struck
material. However, at all sites analyzed, non-YDB sediment levels the 2-km-thick Laurentide Ice Sheet at ⬍30°, they may have left
of Ir are very low (⬍0.1 ppb and possibly ⬍0.02 ppb) and are negligible traces after deglaciation. Thus, lasting evidence may have
insufficient to account for Ir levels up to 5,000⫻ crustal abundance. been limited to enigmatic depressions or disturbances in the
Given the association of high Ir with a suite of other event-related Canadian Shield (e.g., under the Great Lakes or Hudson Bay),
markers, an ET connection is more plausible. while producing marginal or no shock effects and dispersing fine
debris composed of the impactor, ice-sheet detritus, and the
Ice Core Evidence. Large increases in Ir and Pt occurred during the underlying crust.
Younger Dryas as recorded in the GRIP (Greenland) ice core by Toon et al. (46) also noted that if airbursts explode with energy
Gabrielli et al. (41), who attributed these increases to increased of 107 megatons at optimum height, they will cause blast damage
cosmic input. Although sample resolution in the ice core was too over an area the size of North America that is equivalent to a ground
low to permit us to specifically link the onset of these increased impact of 109 megatons (figure 5 in ref. 46). Such airbursts
fluxes with the timing of the YD event, the evidence is consistent effectively couple the impactor’s kinetic energy with the atmo-
with the YD event. sphere or surface (47, 48), producing devastating blast waves well
As evidence for biomass burning, Mayewski et al. (42, 43) above hurricane force (70 m䡠s⫺1) (46). In 1908, at Tunguska,
reported large ammonium and nitrate spikes in the Greenland Siberia, a object ⬍150 m in diameter, either a carbonaceous
GISP2 ice core at the onset of the YD. These GISP2 data are asteroid or a small, burned-out comet, produced a ⬍15-megaton
consistent with strong geochemical evidence in the GRIP ice core airburst with an intense fireball (107 °C) that scorched ⬇200 km2 of
trees and leveled ⬇2,000 km2 of forest yet produced no crater or
for massive biomass burning at the YD onset, especially a major
shock metamorphism (49). A debris shower from a heavily frag-
ammonium spike, in association with peaks in nitrate, nitrite,
mented comet (11) would have produced an airburst barrage that
formate, oxalate, and acetate (44). Altogether, the YD onset was
was similar to, although exponentially larger than Tunguska, while
one of the most robust intervals of biomass burning inferred from
causing continent-wide biomass burning and ice-sheet disruption,
the Greenland ice cores, although the source of this burning signal
but again possibly, without typical cratering.
must have been far more remote than sources today, because much
of the modern forested Arctic region was then covered by ice. The Environmental Effects. The YD event would have created a devas-
cause of this biomass burning is consistent with the YD event. tating, high-temperature shock wave with extreme overpressure,
followed by underpressure, resulting in intense winds traveling
Radioactive Elements. Some megafaunal bones in the YDB are
across North America at hundreds of kilometers per hour, accom-
highly radioactive relative to other stratigraphic intervals, as oc- panied by powerful, impact-generated vortices (50–52). In addition,
curred for some bones at the K/T boundary (see SI Figs. 13 and 14). whether single or multiple objects collided with Earth, a hot fireball
In addition, high concentrations of U and Th were found in the would have immersed the region near the impacts and would have
YDB sediment at six of six Clovis-age sites analyzed and in four of been accentuated if the impact angles were oblique (46, 53). For
four Bays with a paleosol, just as were found in the impact layers comparison, Svetsov (48) calculated that a Tunguska-sized airburst
at Chesapeake Bay (38) and the K/T (see SI Fig. 15) (45). Because would immerse the ground with a radiation flux severe enough to
the heavy minerals, zircon, monazite, and garnet, along with Ti-rich ignite 200 km2 of forest within seconds. Thus, multiple, larger
minerals, such as titanite, ilmenite, and rutile, sometimes contain airbursts would have ignited many thousands of square kilometers.
high concentrations of U and Th, we investigated whether lag At greater distances, the reentry of high-speed, superheated ejecta
deposits of those minerals might be the source of high radioactivity. would have induced extreme wildfires (53), which would have
We conclude that lag deposits may explain the high YDB radio- decimated forests and grasslands, destroying the food supplies of
activity at some sites but not at others. Ilmenite, rutile, and titanite herbivores and producing charcoal, soot, toxic fumes, and ash. The
are possible carriers given that they comprise up to ⬇2% of all number of ET airbursts or impacts necessary to induce the conti-
sediments, but zircon, monazite, and garnet are unlikely, because nent-wide environmental collapse at 12.9 ka is unknown.
they represent ⬍0.1% each (see SI Figs. 16 and 17). The elevated
levels of U and Th may result from multiple processes related to the Climate. A number of impact-related effects most likely contributed
impacts/airbursts, including formation of lag deposits, as well as the to the abrupt, major cooling at the onset of the YD and its
dispersal of ejecta from the impactor and/or the target area. maintenance for ⬎1,000 years. Cooling mechanisms operating on
shorter time scales may have included (i) ozone depletion, causing
Nature of the Event. The evidence points to an ET event with shifts in atmospheric systems in response to cooling, with the
continent-wide effects, especially biomass burning, but the size, side-effect of allowing increased deadly UV radiation to reach
density, and composition of the impactor are poorly understood. survivors on the surface (46); (ii) atmospheric injection of nitrogen
Even so, current data suggest that this impactor was very different compounds (NOx), sulfates, dust, soot, and other toxic chemicals
from well studied iron, stony, or chondritic impactors (e.g., at the from the impact and widespread wildfires (46), all of which may
K/T boundary). The evidence is more consistent with an impactor have led to cooling by blockage of sunlight, with the side-effect of
that was carbon-rich, nickel–iron-poor, and therefore, most likely a diminished photosynthesis for plants and increased chemical tox-
comet. Although the current geologic and geochemical evidence is icity for animals and plants (46); and (iii) injection of large amounts

16020 兩 www.pnas.org兾cgi兾doi兾10.1073兾pnas.0706977104 Firestone et al.

of water vapor and ice into the upper atmosphere to form persistent American fauna, and major adaptations and population declines
cloudiness and noctilucent clouds, leading to reduced sunlight and among PaleoAmericans. The unique, carbon-rich, YDB layer,
surface cooling (46). Although these cooling mechanisms tend to be coupled with a distinct assemblage of impact tracers, implies
short-lived, they can trigger longer-term consequences through isochroneity of the YDB datum layer and thus highlights its utility
feedback mechanisms. For example, noctilucent clouds can reduce for correlation and dating of the North American late Pleistocene.
solar insolation at high latitudes, increasing snow accumulation and These associations, if confirmed, offer the most complete and
causing further cooling in a feedback loop. The largest potential recent geological record for an ET impact and its effects, such as
effect would have been impact-related partial destabilization and/or global climate change and faunal extinction. This evidence also
melting of the ice sheet. In the short term, this would have suddenly would represent a record of a major ET event having serious,
released meltwater and rafts of icebergs into the North Atlantic and widespread consequences for anatomically modern humans.
Arctic Oceans, lowering surface-ocean salinity with consequent
surface cooling. The longer-term cooling effects largely would have Methods
resulted from the consequent weakening of thermohaline circula- Elemental analyses were performed by using prompt gamma-ray
tion in the northern Atlantic (54), sustaining YD cooling for ⬎1,000 activation analysis, neutron activation analysis, and inductively
years until the feedback mechanisms restored ocean circulation. coupled plasma MS. Microspherules, glass-like carbon, and
carbon spherules were analyzed by SEM/x-ray fluorescence.
Clovis and Megafauna. The impact-related effects would have been These methods are very standard and discussed further in SI
devastating for animals and plants. For humans, major adaptive Text, “Methods.”
shifts are evident at 12.9 ka, along with an inferred population
We thank C. V. Haynes, Jr., and J. Ballenger (University of Arizona,
decline, as subsistence strategies changed because of dramatic
Tucson, AZ), B. Kooyman and colleagues (University of Calgary, Calgary,
ecological change and the extinction, reduction, and displacement BC, Canada), B. Childress and J. Pike-Childress (Bureau of Land Man-
of key prey species (55, 56). Many sites indicate that both Clovis agement, Phoenix, AZ), M. Boyd and colleagues (Lakehead University,
people and extinct megafauna were present immediately before the Thunder Bay, ON, Canada), A. Huston, (Channel Islands National Park,
YD event, but, except in rare cases, neither appears in the geologic Ventura, CA), J. Johnson (Santa Barbara Museum of Natural History,
record afterward. At Murray Springs, butchered, still-articulated Santa Barbara, CA), M. Waters and A. Rodriguez (University of North
mammoth bones, Clovis tools, and a hearth were found buried Carolina, Chapel Hill), D. Kimbel (Restoration Systems, LLC), and W.
directly beneath the black mat, indicating that it buried them rapidly Newell (Restoration Systems, LLC) for providing/collecting samples and/or
for access to sites; M. Gifford for help with inductively coupled plasma MS
(37). YDB markers, including Ir at 51 ppb, occur inside an extinct analyses; Henry Wright (University of Michigan, Ann Arbor, MI) and
horse skull at the Wally’s Beach Clovis kill-site (57), again suggest- Denise Henry and Terrence Rettig (National Science Foundation, Arling-
ing rapid burial following the YD event. It is likely that some ton, VA) for early help and encouragement; J. Talbot (K/T GeoServices,
now-extinct animals survived in protected niches, only later to Argyle, TX), B. Cannon (Cannon Microprobe, Seattle, WA), C. J. Eastoe
become extinct because of insufficient food resources, overhunting, (University of Arizona), J. Edwards (Process NMR Associates, Danbury,
climate change, disease, flooding, and other effects, all triggered or CT), E. Hoffman (Activation Laboratories, Ancaster, ON, Canada), J.
amplified by the YD event. Feathers and J. Johnson (University of Washington, Seattle, WA), S.
Simpson (Becquerel Laboratories, Mississauga, ON, Canada), Philippe
Conclusions Claeys (Vrije Universiteit Brussel, Brussels, Belgium), and J. Southon
(University of California, Irvine, CA) for their careful analyses of samples.
Our primary aim is to present evidence supporting the YD impact This work was supported, in part, by U.S. Department of Energy Contract
event, a major ET collision over North America at 12.9 ka, which DE-AC02-05CH11231 and National Science Foundation Grants 9986999,
contributed to the YD cooling, the massive extinction of the North OCD-0244201, and ATM-0713769.

1. Taylor R, Haynes CV, Jr, Stuiver M (1996) Antiquity 70:515–525. 32. Haynes CV, Jr, Stanford DJ, Jodry M, Dickenson J, Montgomery JL, Shelley PH, Rovner
2. Haynes, CV, Jr (2005) in Paleoamerican Origins: Beyond Clovis, eds Bonnichsen R, Lepper I, Agogino GA (1999) Geoarchaeology 14(5):455–470.
BT, Stanford D, Waters MR (Texas A&M Univ Press, College Station, TX), pp 113–132. 33. Simonson B, Glass B (2004) Annu Rev Earth Planet Sci 32:329–361.
3. Grayson D, Meltzer D (2003) J Arch Sci 30:585–593. 34. Iyer SD, Prasad MS, Gupta SM, Charan SN, Mukherjee AD (1997) J Volcanol Geotherm Res
4. Haynes CV (1998) Mammoth Trumpet 13(2):2–6. 78:209–220.
5. Mosimann J, Martin P (1975) Am Scientist 63:304–313. 35. Wright F, Hodge P, Allen R (1966) Smithson Astrophys Obs Spec Rep 228:1–9.
6. Guthrie R (2006) Science 441:207–209. 36. Pan K-L, Overstreet WC, Robinson K, Hubert AE, Crenshaw GL (1980) USGS Professional
7. Grayson D, Meltzer D (2003) J Arch Sci 30:585–593. Publications 1135:3–22.
8. MacPhee R, Marx P (1997) in Natural Change and Human Impact in Madagascar, eds 37. Haynes CV, Jr (2007) in Murray Springs: A Clovis Site with Multiple Activity Areas in the San
Goodman S, Patterson B (Smithsonian Inst Press, Washington, DC), pp 169–217. Pedro Valley, Arizona, eds Haynes CV, Jr, Huckell BB (Univ of Arizona Press, Tucson, AZ),
9. LaViollete P (1987) Earth, Moon, Planets 37:241–286. pp 240–249.
10. Brakenridge G (1981) Icarus 46:81–93. 38. Bodiselitsch B, Montanari A, Koeberl C, Coccioni R (2004) Earth Planet Sci Lett
11. Clube V, Napier W (1984) Mon Not R Astronom Soc 211:953–968. 223:283–302.
12. Alvarez L, Alvarez W, Asaro F, Michel H (1980) Science 208:1095–1108. 39. Huber H, Koeberl C, Egger H (2003) Geochem J 37:123–134.
13. Becker L, Poreda RJ, Basu AR, Pope KO, Harrison TM, Nicholson C, Lasky R (2004) 40. Dyer BD, Lyalikova NN, Murray DP, Doyle M, Kolesov GM, Krumbein WE (1989) Geology
Science 304:1469–1476. 17(11):1036–1039.
41. Gabrielli P, Barbante C, Plane JM, Varga A, Hong S, Cozzi G, Gaspari V, Planchon FA,
14. Firestone RB, Topping W (2001) Mammoth Trumpet 16(9):1–5.
Cairns W, Ferrari C, et al. (2004) Nature 432:1011–1014.
15. Firestone RB, West A, Warwick-Smith S (2006) Cycle of Cosmic Catastrophes (Bear,
42. Mayewski PA, Meeker LD, Whitlow SI, Twickler MS, Morrison MG, Alley RB, Bloomfield
Rochester, VT), pp 19–35.
P, Taylor K (1993) Science 261:195–197.
GEOPHYSICS

16. Lougheed MS (1966) Ohio J Sci 66:274–283.

43. Mayewski PA, Meeker LD, Twickler MS, Whitlow SI, Yang Q, Lyons WB, Prentice M (1997)
17. Bi D (1993) Meteoritics 29:88–93.
J Geophys Res 102:26345–26366.
18. Alley RB (2000) Quat Sci Revs 19:213–226. 44. Legrand M, De Angelis M (1995) J Geophys Res 100:1445–1462.
19. Goodyear AC (2006) Curr Res Pleistocene, 23:100–101. 45. Martinez-Ruiz F, Ortega-Huertas M, Palomo I (1999) Terra Nova 11:290–296.
20. Melton FA, Schriever W (1933) J Geol 41:52–56. 46. Toon OB, Tukco RP, Covey C, Zahnle K, Morrison D (1997) Rev Geophys 35:41–78.
21. Prouty WF (1952) Bull GSA 63:167–224. 47. Schultz PH, Gault DE (1985) J Geophys Res 90:3701–3732.
22. Frey DJ (1955) Ecology 36(4):762–763. 48. Svetsov VV (2002) Palaeogeogr Palaeoclim Palaeoecol 185:403–405.
23. Anders E, Grevesse N (1989) Geochim Cosmo Acta 53:197–214. 49. Wasson JT (2003) Astrobiology 3:163–179.
24. McDonough W, Sun S (1995) Chem Geol 120:223–253. 50. Schultz PH (1992) J Geophys Res 97:16183–16248.
25. Rudnick R, Gao R (2003) Treatise on Geochemistry, eds Holland H, Turekian K (Elsevier, 51. Barnouin-Jha O, Schultz PH (1996) J Geophys Res 101:21099–21115.
Oxford, UK), Vol 3, pp 1–64. 52. Wrobel K, Schultz PH, Crawford D (2006) Meteorit Planet Sci 41:1539 –1550.
26. Wolbach WS, Lewis RS, Anders E (1985) Science 230:167–170. 53. Schultz PH, D’Hondt S (1996) Geology 24:963–967.
27. Kroto H (1988) Science 242:1139–1145. 54. Broecker WS (2006) Science 312:1146–1148.
28. Venkatesan MI, Dahl J (1989) Nature 338:57–60. 55. Anderson DG, Faught MK (2000) Antiquity 74:507–513.
29. Gilmour I, Russell SS, Arden JW, Lee MR, Franchi IA, Pillinger CT (1992) Science 56. Bamforth DB (2002) J World Prehistory 16(1):55–98.
258:1624–1626. 57. Kooyman B, Newman ME, Cluney C, Lobb M, Tolman S, McNeil P, Hills LV (2001) Am
30. Becker L, Poreda RJ, Bunch TE (2000) Proc Natl Acad Sci USA 97:2979–2983. Antiquity 66:686–691.
31. Haynes CV, Jr (1995) Geoarchaeology 10(5):317–388. 58. Boyd M, Running G, Havholm K (2003) Geoarchaeology 18:583–607.

Firestone et al. PNAS 兩 October 9, 2007 兩 vol. 104 兩 no. 41 兩 16021