Mercury evidence for pulsed volcanism during theend-Triassic mass extinctionLawrence M. E. Percivala,1,2, Micha Ruhla, Stephen P. Hesselbob, Hugh C. Jenkynsa, Tamsin A. Mathera,and Jessica H. Whitesidec

aDepartment of Earth Sciences, University of Oxford, Oxford OX1 3AN, United Kingdom; bCamborne School of Mines and Environment and SustainabilityInstitute, University of Exeter, Penryn TR10 9FE, United Kingdom; and cOcean and Earth Science, National Oceanography Centre Southampton, Universityof Southampton, Southampton SO14 3ZH, United Kingdom

Edited by Edward A. Boyle, Massachusetts Institute of Technology, Cambridge, MA, and approved May 19, 2017 (received for review April 3, 2017)

The Central Atlantic Magmatic Province (CAMP) has long beenproposed as having a causal relationship with the end-Triassicextinction event (∼201.5 Ma). In North America and northernAfrica, CAMP is preserved as multiple basaltic units interbeddedwith uppermost Triassic to lowermost Jurassic sediments. How-ever, it has been unclear whether this apparent pulsing was a localfeature, or if pulses in the intensity of CAMP volcanism charac-terized the emplacement of the province as a whole. Here, sixgeographically widespread Triassic–Jurassic records, representingvaried paleoenvironments, are analyzed for mercury (Hg) concen-trations and Hg/total organic carbon (Hg/TOC) ratios. Volcanism isa major source of mercury to the modern environment. Clear in-creases in Hg and Hg/TOC are observed at the end-Triassic extinc-tion horizon, confirming that a volcanically induced global Hgcycle perturbation occurred at that time. The established correla-tion between the extinction horizon and lowest CAMP basaltsallows this sedimentary Hg excursion to be stratigraphically tiedto a specific flood basalt unit, strengthening the case for volcanicHg as the driver of sedimentary Hg/TOC spikes. Additional Hg/TOCpeaks are also documented between the extinction horizon andthe Triassic–Jurassic boundary (separated by ∼200 ky), supportingpulsatory intensity of CAMP volcanism across the entire provinceand providing direct evidence for episodic volatile release duringthe initial stages of CAMP emplacement. Pulsatory volcanism, andassociated perturbations in the ocean–atmosphere system, likelyhad profound implications for the rate and magnitude of the end-Triassic mass extinction and subsequent biotic recovery.

mercury | end-Triassic extinction | Central Atlantic Magmatic Province

The end of the Triassic Period was marked by a major massextinction event (∼201.5 Ma; e.g., refs. 1 and 2), one of the

five largest environmental perturbations of the Phanerozoic Eon.Significantly increased extinction rates of marine fauna, and majorturnovers in terrestrial vegetation and vertebrate groups, have beenwell documented (e.g., refs. 3–8). The end-Triassic mass extinctionpredated the onset of the Jurassic by ∼100 ky to 200 ky, as definedby the first occurrence of the Jurassic ammonite species Psilocerasspelae (9). The sedimentary record of the extinction correlates witha large (up to 6‰) negative excursion in organic carbon isotopes(δ13Corg: Fig. 1A), indicative of a severe carbon cycle perturbationcoincident with the biotic crisis (e.g., refs. 10–13). Moroccan stratathat record this global carbon cycle perturbation are transected bythe lowest documented flows of the Central Atlantic MagmaticProvince (CAMP). Consequently, the end-Triassic extinction hasbeen postulated as precisely coincident with the onset of knownCAMP volcanism (e.g., refs. 2, 12, 14, and 15).CAMP represents the most aerially expansive continental Large

Igneous Province (LIP) known on Earth, consisting of volumetri-cally large-scale flood basalt sequences covering at least 7 × 106 km2

across four continents and both hemispheres (Fig. 2A and ref. 16).In North America and Morocco, CAMP basalts are interbeddedwith continental sediments, which have precise temporal constraintsand are stratigraphically well correlated with marine sedimentary

records (Fig. 1A and refs. 12 and 14). The apparent episodic em-placement of CAMP basalts over several hundred kiloyears, at leastin Morocco and North America, is a key feature of this LIP.The oldest known CAMP basalts are the lower Moroccan unit

[termed the Lower Formation in the High Atlas and the TasguintFormation in the Argana basin (17, 18)]. This unit is overlain bytwo further Moroccan basalt units: the Middle and Upper for-mations in the High Atlas [the former named the Alemzi For-mation in Argana (17, 18)]. For clarity, the High Atlas names areused henceforth in this study. These three basalt groups areinterbedded with sedimentary deposits. A fourth extrusive unit,the Recurrent Formation, is locally preserved higher in theMoroccan sequence, with much thicker sediments between it andthe Upper Formation (Fig. 2B and refs. 14, 17, and 19). Thesefour basalt units are defined and correlated on the basis of distinctigneous geochemistry. Based on geochemical correlation withNorth American CAMP units, which are temporally constrainedby astrochronological and radioisotopic geochronology, theMoroccan Lower–Upper formations are thought to have eruptedin quick succession, with several hundred kiloyears then passingbefore the eruption of the Recurrent Formation (2, 14).At least three major CAMP units are documented in North

America: the Orange Mountain, Preakness, and Hook Mountainbasalts in the Newark Basin (20), with time-equivalent basaltsknown from other North American basins. The Orange Moun-tain Basalt overlies thin, usually lacustrine, sediments depositedabove the extinction horizon, and is thought to have been

Significance

The end of the Triassic Period (∼201.5 million years ago) wit-nessed one of the largest mass extinctions of animal life knownfrom Earth history. This extinction is suggested to have coin-cided with and been caused by one of the largest known epi-sodes of volcanic activity in Earth’s history. This study examinesmercury concentrations of sediments from around the worldthat record this extinction. Mercury is emitted in gaseous formduring volcanism, and subsequently deposited in sediments.We find numerous pulsed elevations of mercury concentrationsin end-Triassic sediments. These peaks show that the mass ex-tinction coincided with large-scale, episodic, volcanism. Suchepisodic volcanism likely perturbed the global environment overa long period of time and strongly delayed ecological recovery.

Author contributions: L.M.E.P. performed research; L.M.E.P., M.R., and J.H.W. analyzeddata; L.M.E.P., M.R., S.P.H., H.C.J., T.A.M., and J.H.W. wrote the paper; and M.R., S.P.H.,and J.H.W. provided rock samples.

The authors declare no conflict of interest.

This article is a PNAS Direct Submission.1To whom correspondence should be addressed. Email: lawrence.percival11@gmail.com.2Present address: Institute of Earth Sciences, Géopolis, University of Lausanne, CH-1015Lausanne, Switzerland.

This article contains supporting information online at www.pnas.org/lookup/suppl/doi:10.1073/pnas.1705378114/-/DCSupplemental.

www.pnas.org/cgi/doi/10.1073/pnas.1705378114 PNAS | July 25, 2017 | vol. 114 | no. 30 | 7929–7934

EART

H,A

TMOSP

HER

IC,

ANDPL

ANET

ARY

SCIENCE

S

Dow

nloa

ded

by g

uest

on

June

15,

202

0

extruded 14 ky to 20 ky after that event (2, 14, 21). Radioisotopicdating has demonstrated that the Preakness Basalt and HookMountain Basalt were emplaced 270 ky and 620 ky later, re-spectively (2). The Newark Basin Orange Mountain and HookMountain basalts have been geochemically established as equiv-alent to the Moroccan Middle and Recurrent formations, re-spectively (14); the Preakness Basalt has no known Moroccancounterpart (2, 14). Consequently, North American and Moroccanrecords suggest that CAMP was emplaced in at least three majorpulses of basalt extrusion over ∼700 ky, with the products of thefirst major pulse further divisible into at least three or fourgeochemically and stratigraphically distinct units. Thus, a total of atleast six CAMP units are documented in Morocco/North America(Fig. 2B), which were relatively close to one another at the end ofthe Triassic Period. However, more geographically dispersedCAMP basalts are also known from southern Europe, Brazil, andelsewhere in western Africa, with potentially different temporalrelationships with the end-Triassic extinction and the dated CAMPflows (16). Therefore, it is not clear whether the intensity of CAMPvolcanism was pulsatory across the entire province, or whether theapparent pulsing recorded in North America and Morocco was alocal feature of the much larger-scale LIP.Proxy records of volcanic volatiles can aid in reconstructing the

history of CAMP volcanism and its effects. Analyses of pedogeniccarbonates suggest increases in atmospheric pCO2 following em-placement of each of the Newark CAMP basalts (CAMP pulses 3,5, and 6 in Fig. 2B), supporting a pattern of globally incremental

emplacement (22). However, the effect of local processes, such asdiagenesis, on this record cannot be ruled out. Reconstructions ofpCO2 based on stomatal indices (albeit at low temporal resolu-tion) show no such pulsing during the Triassic–Jurassic transition(6, 23). Nor is there yet evidence of episodic CO2 increases as-sociated with the early Moroccan CAMP pulses (CAMP pulses 1,3, and 4 in Fig. 2B) that were extruded coincident with, and in theimmediate aftermath of, the extinction event.Here, the volatile emissions and ocean−atmosphere impact of

CAMP volcanism is investigated by analysis of sedimentarymercury (Hg) concentrations across multiple Triassic–Jurassicsedimentary archives. Volcanism is known to be a major naturalsource of Hg, emitting it as a trace volcanic gas (24). Gaseouselemental Hg has a typical atmospheric residence time of 0.5 y to2 y (25), allowing the element to be globally distributed beforebeing drawn down and eventually deposited in sediments. SeveralPhanerozoic events have previously been linked to approximatelycoeval LIPs through documented increases in sedimentary Hgconcentrations, including the end-Permian and end-Cretaceousextinctions and Toarcian Oceanic Anoxic Event (e.g., refs. 26–29).Importantly, sedimentary drawdown of Hg is typically achievedvia organic matter (30, 31), although sulphides and clays mayalso play a role (32–34). Consequently, sedimentary Hg con-centrations are typically normalized against total organic carbon(TOC) to account for the effect on Hg drawdown by changes inorganic matter deposition rates when looking for evidence of anelevated supply of Hg to the environment (26).

A B

Fig. 1. (A) Stratigraphic correlation of the end-Triassic extinction with the Moroccan Lower Formation CAMP basalt. Argana lithology, carbon isotope, andpaleomagnetic data are from ref. 14. Newark lithology and pCO2 data are from ref. 22; paleomagnetic and astrochronological data are from ref. 48; carbonisotope data are from ref. 12; and trilete spore data are from ref. 49. St Audries Bay lithology and astrochronology are from refs. 42 and 50; biostratigraphyand carbon isotope data are from refs. 10 and 50; trilete spore data are from ref. 51; and paleomagnetic data are from refs. 52 and 53. Stratigraphic cor-relation of CAMP units between Argana, Newark, and St Audries Bay is based on refs. 12 and 42. Kuhjoch biostratigraphy, lithology, and carbon isotope dataare from ref. 11. The end-Triassic extinction horizon (marked as ETE) and Triassic–Jurassic boundary (marked as TJB) are also shown. (B) Example of Hg/TOCdataset from this study (Kuhjoch, Fig. 3) is shown to stratigraphically correlate with the lowest CAMP basalt unit that intersects the end-Triassic extinctionhorizon at Argana. See SI Appendix, Fig. S2 for a full stratigraphic correlation of end-Triassic records.

7930 | www.pnas.org/cgi/doi/10.1073/pnas.1705378114 Percival et al.

Dow

nloa

ded

by g

uest

on

June

15,

202

0

In addition to interrogating the pulsatory history of CAMPvolatile emissions, Hg analysis of uppermost Triassic sedimentsalso provides a unique opportunity to test the assumption thatHg-enriched sediments were deposited precisely coincident withthe eruption of LIP basalts. There are excellent age constraintson numerous end-Triassic records (including those containingCAMP basalts), and the precise correlation between the end-Triassic extinction horizon and lowest Moroccan CAMP basalt(the Lower Formation) is well established. Such temporal con-straints may allow some Hg/TOC peaks in uppermost Triassicsediments to be directly correlated with specific CAMP basaltunits. This direct association between Hg/TOC excursions andspecific basalt flows has not been possible for other events due tothe poor preservation, or limited stratigraphic control relative tothe sedimentary record, of many LIPs.A recent study on the Triassic–Jurassic boundary section at

New York Canyon (Nevada) showed an abrupt increase in Hgconcentrations and Hg/TOC ratios correlated with the negativeexcursion in δ13Corg that marks the end-Triassic extinction ho-rizon (35). These Hg excursions were attributed to volcanicprocesses operating during the emplacement of CAMP. Here,the New York Canyon results are greatly expanded by analyzingsix further sedimentary records from around the world, to testwhether the end-Triassic Hg perturbation was a global phe-nomenon. The possibility of multiple episodic peaks in sedi-mentary Hg is also investigated, to examine whether thedocumented pulsatory nature of CAMP emplacement occurredprovince-wide or was limited to specific areas of the LIP. Thesynchrony of any Hg excursions with respect to the earliestCAMP flows is also assessed using the established stratigraphiccorrelation between the end-Triassic extinction horizon and thestratigraphically lowest known CAMP basalts.

Study AreasEnd-Triassic records of both marine and terrestrial environ-ments are known from a number of locations around the world(figure 1 in ref. 36). In this study, the Hg records from six geo-graphically widespread sections are presented, representing a

variety of marine and nonmarine paleoenvironments (Fig. 2A):St Audries Bay (United Kingdom: restricted shallow marine),Kuhjoch (Austria: open shallow marine), Arroyo Malo (Argentina:back-arc shallow marine), Astartekløft (Greenland: fluviodeltaic),Partridge Island (Canada: lacustrine), and Igounane (Morocco:evaporitic–lacustrine). See SI Appendix, Study Areas and Methodol-ogies for details on all of the studied sections, and see SI Appendix,Fig. S2 for a full correlation among all of the above sections andother end-Triassic records.

Results and DiscussionHg as a Recorder of CAMP Volcanism. Clear excursions in Hg/TOCratios and/or Hg concentrations are observed at five of the sixstudied locations (St Audries Bay, Kuhjoch, Arroyo Malo, Astar-tekløft, and Partridge Island). The onsets of these excursions arestratigraphically coincident with the globally observed δ13C negativeexcursion that marks the extinction horizon (Fig. 3). Four sectionsalso record additional peaks in Hg/TOC above that level. Crucially,at all sites where Hg has been normalized to TOC, the Hg/TOCpeaks result from elevated Hg concentrations rather than decreasedTOC content (see SI Appendix, Fig. S4). For Partridge Island andIgounane sediments, Hg/TOC ratios were deemed unreliable due tothe very low TOC content (typically below analytical uncertainty) insedimentary samples. Consequently, Hg signals at these two loca-tions are presented without normalization to TOC. The Hg trendsgenerated in this study are also compared, in Fig. 3, with the existingNew York Canyon record (35), which appears to have a subtlydifferent trend in sedimentary Hg/TOC increase, potentiallyresulting from atmospheric or local sedimentological processes.The observed peaks in sedimentary Hg and Hg/TOC strongly

suggest that a perturbation to the global Hg cycle took placeduring the Triassic–Jurassic transition, beginning coincidentallywith the end-Triassic extinction. The absence of a recorded Hgperturbation in sediments at Igounane is interpreted to resultfrom their deposition being below the oldest known CAMP flows(thus at a time preceding the onset of CAMP basalt extrusion). Alack of change in terrestrial spores from Argana sediments belowthe CAMP basalts further suggests that these sediments weredeposited before the end-Triassic extinction (14), and thus be-fore the onset of CAMP volcanism.The correlation between Hg excursions and the extinction

horizon in the other five studied records is strongly suggestive ofa perturbation to the global Hg cycle at that time. Variations inmarine redox during the extinction may have influenced themarine Hg cycle, but the records at both Kuhjoch and ArroyoMalo do not show a consistent stratigraphic correlation betweenthe observed Hg/TOC peaks and lithological or geochemicalevidence for redox changes (ref. 37 and this study). Additionally,the Hg excursions preserved in the terrestrial records fromAstartekløft and Partridge Island could not have been caused bychanges in the oceanic Hg inventory. Consequently, an atmo-spheric Hg perturbation is the most plausible cause.The Hg perturbation also coincided with the established onset

of an increase in atmospheric pCO2, based on Hg/TOC andstomatal density records from Astartekløft (23) (SI Appendix,Figs. S2 and S5). This correlation suggests a geologically simul-taneous increase in atmospheric Hg and CO2, plausibly origi-nating from magmatic degassing during CAMP emplacement.Emissions of both gases could also result from thermogenic gasrelease from kerogen in subsurface organic-rich sediments in-truded by (CAMP-associated) sills. Thermogenic emissions havebeen previously suggested as a key contributor to LIP atmo-spheric perturbations (e.g., refs. 38 and 39). Thermogenic vola-tiles also explain the observed negative excursion in δ13Corg atthe extinction horizon more satisfactorily than magmatic carbonemissions (40). However, peaks in Hg/TOC stratigraphicallyabove the extinction horizon are not marked by distinct negativeexcursions in δ13Corg. Consequently, there is less evidence for

CAMP

GE

F

A

D

B

C

(Ma)

200.9

201.0

201.1

201.2

201.3

201.4

201.5

201.6

Mo

rocc

o

N. A

me

rica

co

mp

osi

te

6

5

43

2ETE 1

5A

Recurrent Formation/

Hook Mountain Basalt

Upper Formation/

York Haven Intrusive

Middle Formation (Alemzi Fm.)/

Orange Mountain Basalt

North Mountain Basalt

Lower Formation (Tasguint Fm.)1:

2:

3:

4:

5:

6:

Preakness Basalt/

Hickory Grove Basalt (5A)

CAMP BASALT STRATIGRAPHYEND-TRIASSIC WORLD

Sediments deposited between basalt units

A B

Fig. 2. (A) Paleogeographic reconstruction of the end-Triassic world, withthe modern continents overlain. The locations of the six studied sections areindicated (A, St Audries Bay, United Kingdom; B, Kuhjoch, Austria; C, ArroyoMalo, Argentina; D, Astartekløft, Greenland; E, Partridge Island, Canada; andF, Igounane, Morocco). The New York Canyon section in Nevada (G; notedifferent color) studied by ref. 35, and CAMP are also shown (based onfigure 1 from refs. 2 and 12). (B) Summarized composite stratigraphy ofMoroccan and North American CAMP basalts, following the stratigraphicrelationships and ages (in million years) from refs. 2 and 14. The HickoryGrove Basalt is included with the Preakness, due to their geochemical simi-larity. The age of the end-Triassic extinction (ETE) is also indicated.

Percival et al. PNAS | July 25, 2017 | vol. 114 | no. 30 | 7931

EART

H,A

TMOSP

HER

IC,

ANDPL

ANET

ARY

SCIENCE

S

Dow

nloa

ded

by g

uest

on

June

15,

202

0

these later Hg perturbations resulting from thermogenic emis-sions, and magmatic Hg emissions are a more probable cause.Additional evidence for a volcanic origin of the perturbation

to the global Hg cycle during the end-Triassic extinction comesfrom the established correlation between the lowest known CAMPflow (the Lower Formation in the High Atlas) and the extinctionhorizon. This correlation allows the Hg/TOC increase at the ex-tinction horizon to be precisely stratigraphically matched with thatparticular unit of CAMP (Fig. 1B; see also SI Appendix, Fig. S2and refs. 14 and 15). Consequently, it is highly likely that volcanicHg associated with this lowest CAMP flow contributed to theglobal Hg perturbation during the end-Triassic extinction. Theability to stratigraphically correlate an Hg excursion directly withan individual CAMP basalt flow greatly strengthens the use of thiselement as a proxy for volcanism.

The Pulsatory Release of Magmatic Volatiles. In addition to the Hgexcursion at the extinction horizon, four sections record distinctpeaks in sedimentary Hg/TOC higher up in the stratigraphy.These higher peaks are most clearly distinct at Kuhjoch andArroyo Malo, but may also be recorded at New York Canyonand St Audries Bay (Fig. 3). Pulsatory CAMP volcanism hasbeen inferred from the stratigraphic record of CAMP basalts in

North America and Morocco (17–20). However, these litholog-ical records only prove an apparent pulsing of CAMP basalts inspecific locations of what is a much larger-scale province. Epi-sodic volcanism across the entire extent of CAMP has beeninferred from pedogenic carbonate pCO2 reconstructions (22), butthese concretionary carbonates can be impacted by local (diage-netic) processes. The observed pulsatory signal of Hg perturba-tions in multiple, globally distributed, sedimentary records acrossthe Triassic–Jurassic transition provides independent evidencethat the intensity of CAMP volcanism, and likely CAMP em-placement, occurred in a noncontinuous manner.Although the Hg records shown in Fig. 3 do show excursions in

Hg/TOC in Jurassic strata and support the continuation of CAMPvolcanism into the Jurassic, most of the Hg/TOC peaks are do-cumented between the end-Triassic extinction horizon and theTriassic–Jurassic boundary. The data (Fig. 3) suggest that at leasttwo, and possibly three, major volcanic episodes occurred betweenthe extinction and the beginning of the Jurassic (as defined byammonite biostratigraphy). This interval is estimated as lasting only100 ky to 200 ky based on U–Pb geochronology and astrochronol-ogy (e.g., refs. 1, 2, 41, and 42). Ar–Ar and U–Pb geochronologysuggests that the three oldest Moroccan basalt units (the Lower,

A

A E F G

B C D

Fig. 3. Comparison of Hg/TOC data from St Audries Bay, Kuhjoch, Arroyo Malo, Astartekløft, and New York Canyon (35), and Hg data from Partridge Islandand Iguonane (where TOC contents were below error). Letters next to locations refer to letters on the map in Fig. 2A. Carbon isotope data, biostratigraphy(ammonite first appearance), lithology, and magnetostratigraphy are also shown to allow stratigraphic correlation of the end-Triassic extinction (ETE) horizonand Triassic–Jurassic boundary (TJB). Lithological data are from St Audries Bay (10), Kuhjoch (11), Arroyo Malo (54), Astartekløft (10), Partridge Island (thisstudy; shown in figure), Argana (14), Igouane (this study; shown in figure), and New York Canyon (35). Carbon isotope and biostratigraphic data are from StAudries Bay (10, 50), Kuhjoch (11), Arroyo Malo (ref. 54 and this study), Astartekløft (10), Partridge Island (this study), Argana (14), and New York Canyon (35).Line M indicates the magnetostratigraphic correlation, below the extinction horizon, between St Audries Bay, Partridge Island, Argana, and Igounane. Thegray shading illustrates the stratigraphic gap between line M and line ETE. Magnetostratigraphic data are from St Audries Bay (52, 53), Partridge Island (55),and Argana (14). Note, for Astartekløft, the expanded horizontal scale, and the gaps in data due to sand beds with negligible TOC. Full δ13Corg, Hg, TOC, andHg/TOC data for each individual section are reported in SI Appendix, Table S3 and Fig. S4.

7932 | www.pnas.org/cgi/doi/10.1073/pnas.1705378114 Percival et al.

Dow

nloa

ded

by g

uest

on

June

15,

202

0

Middle, and Upper formations: units 1, 3, and 4 in Fig. 2B) and thefirst of the major North American CAMP pulses (units 2–4 in Fig.2B) were likely all emplaced during this interval (2, 14, 17). It ispossible that the observed three pulses in Hg/TOC observed instrata between the extinction horizon and the Triassic–Jurassicboundary may directly relate to atmospheric volatile release fromthe first three CAMP basalt units in the Moroccan Argana Basin,and their North American equivalents. However, further correlativework is needed to confirm such a hypothesis. The later NorthAmerican CAMP basalts, and the Moroccan Recurrent Formation(units 5 and 6, Fig. 2B), were emplaced 300 ky to 600 ky after theextinction (refs. 2, 14, and 21 and references therein). Thus, theHg/TOC peaks between the extinction horizon and the Triassic–Jurassic boundary cannot be related to these later flows.The observed record of multiple pulses in Hg/TOC is direct

evidence that volatile release associated with CAMP volcanismwas also pulsatory, likely including episodic emissions of carbon,sulfur, and Hg. Moreover, pulsatory emissions are shown to haveoccurred throughout the first 100 ky to 200 ky immediately fol-lowing the end-Triassic extinction, in addition to the docu-mented later pulses of volcanic emissions (22). Consequently, thelarge increase in atmospheric pCO2 during the extinction eventmay have arisen as a series of episodic carbon cycle perturbationsto the atmosphere (and subsequently the ocean). Pulsatoryperturbations of the ocean–atmosphere system caused by epi-sodic volcanic events (and release of volatiles) associated withCAMP may explain the documented prolonged period of eco-system deterioration and delayed recovery of benthic faunaduring the emplacement of CAMP (35, 43–45).

ConclusionsInvestigation of the global history of CAMP volatile emissions isimportant for understanding the development of this igneousprovince and its potential environmental impact. This studydemonstrates that the Hg cycle was perturbed on a global scaleduring the Triassic–Jurassic transition. Hg excursions are recordedin five of the six sections studied; the one section with no recordof Hg enrichment was likely deposited before the onset ofCAMP volcanism. The onset of Hg enrichment occurred syn-chronously across the globe, coincident with the end-Triassicextinction and associated global carbon cycle perturbation. Thepresence of Hg/TOC excursions in sedimentary records of ter-restrial and marine paleoenvironments, across both hemispheres,indicates that atmospheric mercury concentrations likely increasedsubstantially. This atmospheric perturbation probably resulted

from the emplacement of CAMP and the associated large-scaleemission of magmatic volatiles, and potentially thermogenicvolatiles from intruded country rock (including Hg). The ap-pearance of a global Hg excursion at the end-Triassic extinc-tion horizon and multiple Hg/TOC peaks between it and theTriassic–Jurassic boundary is further evidence that pulses in theintensity of CAMP volcanism (and associated volatile release)were not limited to North America and Morocco but were rep-resentative of the entire province. The direct correlation betweenthe oldest preserved flows of CAMP and the end-Triassic extinc-tion horizon allows the earliest pulse of elevated Hg in the sedi-mentary record of this time to be linked directly with these basalts.This correlation supports a volcanic origin for the increased Hgabundances and represents a unique tie between a globally ob-served mercury excursion and a specific basalt unit from a LIP.The recording of multiple Hg/TOC excursions between the ex-tinction horizon and the Triassic–Jurassic boundary highlights thatthe initial stages of CAMP emplacement were marked by multipleepisodes of volcanic volatile release. Repeated volcanically drivenperturbations of the ocean−atmosphere system in the 100 ky to200 ky during and immediately following the end-Triassic extinc-tion may have had important implications for the biospheric im-pact of CAMP.

MethodsHganalysiswasundertakenon theRA-915PortableMercuryAnalyzerwith PYRO-915 Pyrolyzer (Lumex) at the University of Oxford (26). Where previous TOCdeterminations were not available, new data were measured using either aStrohlein Coulomat 702 (46) or Rock-Eval VI (47) at the University of Oxford. Theδ13Corg analyses were performed on decarbonated Arroyo Malo samples (pre-pared at the University of Oxford) with a Thermo Scientific Flash 2000 HT Ele-mental Analyzer coupled to a Thermo Scientific MAT253 isotope ratio massspectrometer via a Conflo IV open split interface at the Stable Isotope Laboratoryat the Open University. For full method details, see SI Appendix, Study Areasand Methodologies.

ACKNOWLEDGMENTS. We greatly appreciate the two anonymous reviewersfor their reviews, which much improved this manuscript. We gratefullyacknowledge Alberto Riccardi, Susana Damborenea, and Miguel Manceñidofor their assistance in collecting material from Arroyo Malo, Argentina; PaulOlsen for assistance in collecting samples from Igounane, Morocco; and JohnFarmer and the University of Edinburgh for provision of geochemical stan-dards. We acknowledge Natural Environment Research Council Grant NE/G01700X/1 (to T.A.M.) and PhD studentship NE/L501530/1 (to L.M.E.P.), ShellInternational Exploration and Production Inc., a Niels Stensen FoundationResearch grant (to M.R.), the US National Science Foundation Grants EAR0801138 and EAR 1349650 (to J.H.W.), and the Leverhulme Trust for funding.

1. Schoene B, et al. (2010) Correlating the end-Triassic mass extinction and flood basalt

volcanism at the 100ka level. Geology 38:387–390.2. Blackburn TJ, et al. (2013) Zircon U-Pb geochronology links the end-Triassic extinction

with the Central Atlantic Magmatic Province. Science 340:941–945.3. Raup DM, Sepkoski JJ, Jr (1982) Mass extinctions in the marine fossil record. Science

215:1501–1503.4. Olsen PE, Shubin NH, Anders MH (1987) New early Jurassic tetrapod assemblages

constrain Triassic-Jurassic tetrapod extinction event. Science 237:1025–1029.5. Olsen PE, et al. (1990) The Triassic/Jurassic boundary in continental rocks of eastern

North America; A progress report. Spec Pap Geol Soc Am 247:585–594.6. McElwain JC, Beerling DJ, Woodward FI (1999) Fossil plants and global warming at the

Triassic-Jurassic boundary. Science 285:1386–1390.7. McElwain JC, Wagner PJ, Hesselbo SP (2009) Fossil plant relative abundances indicate

sudden loss of Late Triassic biodiversity in East Greenland. Science 324:1554–1556.8. Bonis NR, et al. (2009) A detailed palynological study of the Triassic–Jurassic transition

in key sections of the Eiberg Basin (Northern Calcareous Alps, Austria). Rev Palaeobot

Palynol 156:376–400.9. Hillebrandt Av, et al. (2013) The Global Stratotype Sections and Point (GSSP) for the

base of the Jurassic System at Kuhjoch (Karwendel Mountains, Northern Calcareous

Alps, Tyrol, Austria). Episodes 36:162–198.10. Hesselbo SP, et al. (2002) Terrestrial and marine extinction at the Triassic–Jurassic

boundary synchronized with major carbon-cycle perturbation: A link to initiation of

massive volcanism? Geology 30:251–254.11. Ruhl M, et al. (2009) Triassic–Jurassic organic carbon isotope stratigraphy of key

sections in the western Tethys realm (Austria). Earth Planet Sci Lett 281:169–187.

12. Whiteside JH, Olsen PE, Eglinton T, Brookfield ME, Sambrotto RN (2010) Compound-

specific carbon isotopes from Earth’s largest flood basalt eruptions directly linked to

the end-Triassic mass extinction. Proc Natl Acad Sci USA 107:6721–6725.13. Bartolini A, et al. (2012) Disentangling the Hettangian carbon isotope record: Implications

for the aftermath of the end-Triassicmass extinction.GeochemGeophys Geosyst 13:Q01007.14. Deenen MHL, et al. (2010) A new chronology for the end-Triassic mass extinction.

Earth Planet Sci Lett 291:113–125.15. Dal Corso J, et al. (2014) The dawn of CAMP volcanism and its bearing on the end-

Triassic carbon cycle disruption. J Geol Soc London 171:153–164.16. Marzoli A, Renne PR, Piccirillo EM, Ernesto M, Bellieni G; De Min A (1999) Extensive

200-million-year-old continental flood basalts of the Central Atlantic Magmatic

Province. Science 284:616–618.17. El Ghilani S, et al. (2017) Environmental implication of subaqueous lava flows from a

continental Large Igneous Province: Examples from the Moroccan Central Atlantic

Magmatic Province (CAMP). J Afr Earth Sci 127:211–221.18. El Hachimi H, et al. (2011) Morphology, internal architecture and emplacement

mechanisms of lava flows from the Central Atlantic Magmatic Province (CAMP) of

Argana Basin (Morocco). Geol Soc Lond Spec Publ 377:167–193.19. Marzoli A, et al. (2004) Synchrony of the Central Atlantic magmatic province and the

Triassic–Jurassic boundary climatic and biotic crisis. Geology 32:973–976.20. Olsen PE, et al. (1996) High-resolution stratigraphy of the Newark rift basin (early

Mesozoic, eastern North America). Geol Soc Am Bull 108:40–77.21. Kent DV, et al. (2017) Astrochronostratigraphic polarity time scale (APTS) for the Late

Triassic and Early Jurassic from continental sediments and correlation with standard

marine stages. Earth Sci Rev 166:153–180.

Percival et al. PNAS | July 25, 2017 | vol. 114 | no. 30 | 7933

EART

H,A

TMOSP

HER

IC,

ANDPL

ANET

ARY

SCIENCE

S

Dow

nloa

ded

by g

uest

on

June

15,

202

0

22. Schaller MF, Wright JD, Kent DV (2011) Atmospheric PCO2 perturbations associatedwith the Central Atlantic Magmatic Province. Science 331:1404–1409.

23. Steinthorsdottir M, et al. (2011) Extremely elevated CO2 concentrations at the Triassic/Jurassic boundary. Palaeogeogr Palaeoclimatol Palaeoecol 308:418–432.

24. Pyle DM, Mather TA (2003) The importance of volcanic emissions in the global at-mospheric mercury cycle. Atmos Environ 37:5115–5124.

25. Blum JD, et al. (2014) Mercury isotopes in earth and environmental sciences. Annu RevEarth Planet Sci 42:249–269.

26. Sanei H, et al. (2012) Latest Permian mercury anomalies. Geology 40:63–66.27. Percival LME, et al. (2015) Globally enhanced mercury deposition during the end-

Pliensbachian and Toarcian OAE: A link to the Karoo–Ferrar Large Igneous Prov-ince. Earth Planet Sci Lett 428:267–280.

28. Percival LME, et al. (2016) Osmium-isotope evidence for two pulses of increasedcontinental weathering linked to volcanism and climate change during the EarlyJurassic. Geology 44:759–762.

29. Font E, et al. (2016) Mercury anomaly, Deccan volcanism, and the end-Cretaceousmass extinction. Geology 44:171–174.

30. Benoit JM, et al. (2001) Constants for mercury binding by dissolved organic matterisolates from the Florida Everglades. Geochim Cosmochim Acta 65:4445–4451.

31. Outridge PM, Sanei LH, Stern GA, Hamilton PB, Goodarzi F (2007) Evidence for controlof mercury accumulation rates in Canadian High Arctic lake sediments by variations ofaquatic primary productivity. Environ Sci Technol 41:5259–5265.

32. Benoit JM, et al. (1999) Sulfide controls on mercury speciation and bioavailability tomethylation bacteria in sediment pore waters. Environ Sci Technol 33:951–957.

33. Niessen S, et al. (2003) Influence of sulphur cycle on mercury methylation in estuarinesediment (Seine estuary, France). J Phys IV 107:953–956.

34. Kongchum M, et al. (2011) Relationship between sediment clay minerals and totalmercury. J Environ Sci Health Part A 46:534–539.

35. Thibodeau AM, et al. (2016) Mercury anomalies and the timing of biotic recoveryfollowing the end-Triassic mass extinction. Nat Commun 7:11147.

36. Tanner LH, et al. (2004) Assessing the record and causes of Late Triassic extinctions.Earth Sci Rev 65:103–139.

37. Ruhl M, et al. (2010) Sedimentary organic matter characterization of the Triassic–Jurassic boundary GSSP at Kuhjoch (Austria). Earth Planet Sci Lett 292:17–26.

38. Svensen H, et al. (2009) Siberian gas venting and the end-Permian environmentalcrisis. Earth Planet Sci Lett 277:490–500.

39. Ganino C, Arndt NT (2009) Climate changes caused by degassing of sediments duringthe emplacement of large igneous provinces. Geology 37:323–326.

40. Beerling DJ, Berner RA (2002) Biogeochemical constraints on the Triassic–Jurassicboundary carbon cycle event. Global Biogeochem Cycles 16:1036.

41. Whiteside JH, et al. (2007) Synchrony between the Central Atlantic magmatic prov-ince and the Triassic–Jurassic mass-extinction event? Palaeogeogr PalaeoclimatolPalaeoecol 244:345–367.

42. Xu W, et al. (2017) Orbital pacing of the Early Jurassic carbon cycle, black-shale for-mation and seabed methane seepage. Sedimentology 64:127–149.

43. Self S, et al. (2014) Emplacement characteristics, time scales, and volatile release ratesof continental flood basalt eruptions on Earth. Spec Pap Geol Soc Am 505:319–338.

44. Schmidt A, et al. (2015) Selective environmental stress from sulphur emitted bycontinental flood basalt eruptions. Nat Geosci 9:77–82.

45. Ritterbush KA, et al. (2015) Andean sponges reveal long-term benthic ecosystem shiftsfollowing the end-Triassic mass extinction. Palaeogeogr Palaeoclimatol Palaeoecol 420:193–209.

46. Jenkyns HC (1988) The early Toarcian (Jurassic) anoxic event: Stratigraphic, sedi-mentary, and geochemical evidence. Am J Sci 288:101–151.

47. Espitalié J, et al. 1977, Source rock characterization methods for petroleum explo-ration. Proceedings of the 1977 Offshore Technology Conference, (Soc Petrol Eng,Richardson, TX), Vol 3, pp 439–443.

48. Kent DV, Olsen PE (1999) Astronomically tuned geomagnetic polarity timescale forthe Late Triassic. J Geophys Res 104:12831–12841.

49. Olsen PE, et al. (2002) Ascent of dinosaurs linked to an iridium anomaly at the Triassic-Jurassic boundary. Science 296:1305–1307.

50. Ruhl M, et al. (2010) Astronomical constraints on the duration of the early JurassicHettangian stage and recovery rates following the end-Triassic mass extinction (StAudrie’s Bay/East Quantoxhead, UK). Earth Planet Sci Lett 295:262–276.

51. Bonis NR, et al. (2010) Milankovitch-scale palynological turnover across the Triassic–Jurassic transition at St. Audrie’s Bay, SW UK. J Geol Soc London 167:877–888.

52. Hounslow MW, et al. (2004) Magnetostratigraphy and biostratigraphy of the upperTriassic and lowermost Jurassic succession, St Audrie’s Bay, UK. Palaeogeogr PalaeoclimatolPalaeoecol 213:331–358.

53. Hüsing SK, et al. (2014) Astronomically-calibrated magnetostratigraphy of the LowerJurassic marine successions at St. Audrie’s Bay and East Quantoxhead (Hettangian–Sinemurian; Somerset, UK). Palaeogeogr Palaeoclimatol Palaeoecol 403:43–56.

54. Riccardi AC, et al. (2004) The Triassic/Jurassic boundary in the Andes of Argentina. RivItal Paleontol Stratigr 110:69–76.

55. Deenen MHL, et al. (2011) The quest for E23r at Partridge Island, Bay of Fundy,Canada: CAMP emplacement postdates the end-Triassic extinction event at the NorthAmerican craton. Can J Earth Sci 48:1282–1291.

7934 | www.pnas.org/cgi/doi/10.1073/pnas.1705378114 Percival et al.

Dow

nloa

ded

by g

uest

on

June

15,

202

0