Extensive 200-Million-Year-Old Continental Flood Basalts of the Central Atlantic Magmatic Province

Abstract

The Central Atlantic Magmatic Province (CAMP) is defined by tholeiitic basalts that crop out in once-contiguous parts of North America, Europe, Africa, and South America and is associated with the breakup of Pangea. 40Ar/39Ar and paleomagnetic data indicate that CAMP magmatism extended over an area of 2.5 million square kilometers in north and central Brazil, and the total aerial extent of the magmatism exceeded 7 million square kilometers in a few million years, with peak activity at 200 million years ago. The magmatism coincided closely in time with a major mass extinction at the Triassic-Jurassic boundary.

Full text

Extensive 200-Million-Year-Old
Continental Flood Basalts
of the Central Atlantic
Magmatic Province
Andrea Marzoli,
1,2
* Paul R. Renne,
1,3
†Enzo M. Piccirillo,
2
Marcia Ernesto,
4
Giuliano Bellieni,
5
Angelo De Min
2
The Central Atlantic Magmatic Province (CAMP) is defined by tholeiitic basalts
that crop out in once-contiguous parts of North America, Europe, Africa, and
South America and is associated with the breakup of Pangea.
40
Ar/
39
Ar and
paleomagnetic data indicate that CAMP magmatism extended over an area of
2.5 million square kilometers in north and central Brazil, and the total aerial
extent of the magmatism exceeded 7 million square kilometers in a few million
years, with peak activity at 200 million years ago. The magmatism coincided
closely in time with a major mass extinction at the Triassic-Jurassic boundary.
Many aspects of the genesis and consequenc-
es of continental flood basalt provinces
(CFBPs) remain poorly known and contro-
versial. The definition of a CFBP, as well as
their genetic relations with other phenomena
such as rifting of continents, hot spot tracks,
and mass extinctions, commonly hinges on
geochronology. Such is the case for the Cen-
tral Atlantic Magmatic Province (CAMP),
which is associated with the disruption of
Pangea and the opening of the central Atlan-
tic Ocean (1–3). The CAMP (Fig. 1) is rep-
resented by tholeiitic dikes, sills, and lava
flows in North and South America, Africa,
and Europe. Components of the CAMP have
been studied for decades, but only through
recent high-precision geochronologic analy-
sis can magmatism represented by this prov-
ince now be related to a single brief magmat-
ic episode.
40
Ar/
39
Ar (4) and U/Pb (5) data
permit recognition of extensive basaltic mag-
matism in West Africa, eastern North Amer-
ica, and northernmost South America at
200 64 million years ago (Ma), with an
estimated original extent of the volcanism
over an area of at least 4.5 310
6
km
2
(4–7).
Here we present
40
Ar/
39
Ar data that indicate
that CAMP basalts are widespread also in
northern and central Brazil, over a previously
unrecognized area of about 2.5 310
6
km
2
.
Most of the South American CAMP mag-
matism occurred far inland, up to 2000 km
from the Atlantic margin, in northern and
central Brazil. These tholeiitic flows, sills,
and dikes occur in Archean–Early Proterozo-
ic cratonic areas and Late Proterozoic to Pa-
leozoic basins. Lava flows (for example, in
the Maranha˜o, Anari, and Tapirapua˜ subprov-
inces) are mostly preserved in the Paleozoic
sedimentary basins and presently cover a rel-
atively restricted area of 3 310
5
km
2
, reach-
ing a maximum thickness of 300 m and a
total estimated volume of 6 310
4
km
3
(8,9).
The sills of the Amazonian basin cover an
area of ;1310
6
km
2
and have an average
thickness of ;500 m and an estimated vol-
ume of ;4310
5
km
3
(10). The only known
extrusive remnant of the Amazonian magma-
tism is in the Maranha˜o basin, where the
so-called Mosquito basalts are geochemically
similar to the Amazonian sills (9,10). About
0.4 310
6
km
2
of strongly weathered and
deeply eroded areas in the Guyana and Am-
azonian cratons are intruded by dominantly
north-south–striking dike swarms. Some of
these swarms (for example, in the Roraima
and Cassipore´ subprovinces) are character-
ized by a high density of dikes 200 to 300 m
thick and up to 300 km long and therefore are
comparable to feeder dike swarms of other
CFBPs (11–13).
The basalts in Brazil have compositions
similar to those of CAMP magmatic rocks
from North America and West Africa. They
range in composition from tholeiitic basalts
to andesitic basalts, and alkaline and silicic
rocks are scarce. CAMP tholeiites are char-
acterized by low TiO
2
concentrations (typi-
cally ,2 weight %), negative mantle-normal-
ized Nb anomalies (relative to K and La), and
moderately to strongly enriched rare Earth
element patterns (6,8,9, our data). No sys-
tematic compositional differences are noted
between the different regions of CAMP. De-
spite this common signature, relatively few
evolved tholeiites from single localities in
Brazil as well as West Africa and eastern
North America have relatively variable trace
element and isotopic compositions (Fig. 2).
Such compositional variations cannot be at-
tributed uniquely to low-pressure differentia-
tion processes and suggest [as inferred for
other Gondwana CFBPs (12,13)] important
contributions of heterogeneous, possibly
lithospheric mantle in the petrogenesis of
these rocks.
We dated fresh tholeiitic dikes from Ro-
raima and Amapa´ (Cassipore´ dikes), tholei-
itic lava flows from the Maranha˜o basin and
the Anari-Tapirapua˜ region, and one alkaline
lava flow from Lavras da Mangabeira basin
(Ceara´) of northern and central Brazil by
40
Ar/
39
Ar incremental heating (11). Ages
were calculated relative to an age of 28.02
Ma for the Fish Canyon sanidine (FCs) neu-
tron fluence monitor (14 ).
Plateau ages, defined by 10 to 45 contig-
uous steps and by 50 to 84% of total
39
Ar
released, were obtained for 10 samples (Fig.
3). Apparent age spectra for lava flows are
little affected by argon loss or excess argon
1
Berkeley Geochronology Center, 2455 Ridge Road,
Berkeley, CA 94709, USA.
2
Dipartimento di Scienze
della Terra, Universita´ di Trieste, via Weiss 8, 34127
Trieste, Italy.
3
Department of Geology and Geophys-
ics, University of California, Berkeley, CA 94720, USA.
4
Instituto Astrono´mico e Geofı´sico, Universidade de
Sa˜o Paulo, R. do Mata˜o 1226, 015508-900 Sa˜o Paulo,
Brazil.
5
Dipartimento di Mineralogia e Petrologia, Uni-
versita´ di Padova, Corso Garibaldi 37, 35100 Padova,
Italy.
*Present address: Department de Mine´ralogie, Univer-
site´ de Gene`ve, 13 rue des Maraichers, 1211 Geneva
4, Switzerland.
†To whom correspondence should be addresssed. E-
mail: prenne@bgc.org
Fig. 1. Location map of the CAMP (4–7, present
data) in a Pangea reconstruction at 200 Ma,
also showing the Siberian and Karoo-Ferrar
CFBPs in the inset. The area presently recog-
nized as being part of CAMP is shown by a
dashed contour, with sample sites indicated: R,
Roraima; M, Maranha˜o; C, Cassipore´; Ce, Ceara´;
A, Anari; and T, Tapirapua˜.
REPORTS
23 APRIL 1999 VOL 284 SCIENCE www.sciencemag.org616
on July 16, 2020 http://science.sciencemag.org/Downloaded from

and yield plateau ages ranging from 190.5 6
1.6 to 198.5 60.8 Ma (Maranha˜o, Mosquito
basalts) and from 196.6 61.8 to 198.0 60.8
(Tapirapua˜ and Anari samples, respectively).
The alkaline basalt lava flow from Ceara´
yields a plateau age of 198.4 61.4 Ma. Dikes
from Roraima and Cassipore´, which intrude
Proterozoic basement, yield saddle-shaped
age spectra typical of excess argon contami-
nation, as is common for plagioclases of
dikes intruding wall rocks of much greater
age. Nonetheless, detailed step heating (32 to
84 steps for each sample) allowed definition
of plateaus on two Roraima (8818 and 8820)
and three Cassipore´ (8026, 8034, and 8041)
dike samples. Moreover, one dike from Ro-
raima (8804) yields a near plateau, defined by
43% of total
39
Ar released and 16 contiguous
steps. In some of the dikes,
39
Ar/
40
Ar versus
36
Ar/
40
Ar isotope correlation plots confirm
the presence of substantial excess
40
Ar and
permit determination of isochron ages, which
tend to be slightly younger than plateau ages
(Fig. 3). In these cases, to minimize the pos-
sible effect of excess argon, we adopted iso-
chron ages for the dikes (samples 8026, 8804,
8818, and 8820). In summary, ages for Ro-
raima and Cassipore´ dikes range from
197.4 61.9 to 201.1 60.7 and from 191.5 6
0.9 to 202.0 62.0 Ma, respectively.
All available radioisotopic dates for the
CAMP (4,5, present data) are between 191
and 205 Ma, with a mean age of 199.0 62.4
Ma and the peak at 200 Ma (Fig. 4). Our data
show that tholeiites from Brazil are similar in
age to those of the CAMP in Africa and
North America. The two distinguishably
younger samples (5013 and 8034, from the
Maranha˜o and Cassipore´, respectively) are
from the northernmost portion of the Brazil-
ian CAMP. A similar relation is shown on the
African continental margin in Guinea (4).
Thus, after the main pulse, magmatism may
have continued further toward the future rift-
ed margin of each continent [for example,
(15)].
Paleomagnetic data from widely distribut-
ed sites in the South American CAMP occur-
rences provide further evidence of a brief
coeval magmatic event in the circum-Atlantic
region (Fig. 5). Paleomagnetic poles for
South America are available from the Mar-
anha˜o, Guacamaya, Anari, and Tapirapua˜
volcanic rocks and the Guyana, Bolivar, Pe-
natecaua, and Cassipore´ dikes (9,16,17 ).
Data from seven independent sites in low-Ti
tholeiitic dikes and flows from Roraima yield
a paleomagnetic pole located at 235.0°E,
80.1°S (N57; 95% confidence angle a
95
5
6.6; concentration parameter k584). Con-
sidering uncertainties in the Euler poles used
to enact the continental reconstructions, the
average of these poles is in good agreement
with contemporaneous poles from western
Fig. 2. Initial (200 Ma) «Sr-«Nd isotopic com-
positions of CAMP basalts (6,8, and 32 data
samples from the present study). Compositions
of low-TiO
2
(LTi) and high-TiO
2
(HTi) basalts
from the Parana`(12) and the Karoo (13) Me-
sozoic Gondwana CFBPs, as well as mantle
components [EM, enriched mantle; HIMU,
high-m(
238
U/
204
Pb normalized for radioactive
decay to the present) mantle] and normal mid-
ocean ridge basalt (N-MORB) fields (29), are
shown for comparison. HTiP, Parana` high-TiO
2
basalts.
Fig. 3. Apparent age spectra of plagioclase
separated from dikes (8804, 8818, and
8820: Roraima; 8026, 8034, and 8041: Cas-
sipore´) and lava flows (5013 and 5042:
Maranha˜o; 8232: Ceara´; ANG-7: Anari; TRG-
10: Tapirapua˜) from north and central Bra-
zil. Isochron ages are reported in cases with
the statistic mean square of weighted de-
viates ,1.0. Plateau and isochron ages are
given in million years ago (Ma) with 2serrors, which include analytical uncertainty in neutron
fluence parameter Jvalue.
REPORTS
www.sciencemag.org SCIENCE VOL 284 23 APRIL 1999 617
on July 16, 2020 http://science.sciencemag.org/Downloaded from

Africa, southern Europe, and eastern North
America (18 –20). Magnetizations of most
CAMP tholeiites from South America (9,16 ,
17, present data), eastern North America (20),
southern Europe (18), and West Africa (19)
are of normal polarity. A ;4- to 5-million-
year normal polarity interval characterizes
the latest Triassic [uppermost Rhaetian (20)]
and earliest Jurassic [lower Hettangian (21)]
and is preceded and followed by long inter-
vals of reversed polarity or very frequent
polarity reversals (20,21). Thus, the CAMP
paleomagnetic data provide independent sup-
port for brevity of peak magmatism.
The CAMP thus includes the ;2.5 310
6
km
2
widespread magmatism in central and
northern Brazil and had a total extent of at
least 7 310
6
km
2
. Assuming conservatively
that preserved volcanic sections averaging
200 to 300 m thick in distal portions of the
CAMP are representative (6,8 –10), an orig-
inal volume of 2 310
6
km
3
is implied. The
geochronological and paleomagnetic data
suggest that most of this widespread magma-
tism occurred in a few million years, with a
peak at ;200 Ma. Similarly widespread and
short-lived tholeiitic magmatism characteriz-
es other well-studied CFBPs, for example,
the Siberian Traps (22) or the Karoo-Ferrar
(23,24 ) (Fig. 4).
Such brief and extremely widespread tho-
leiitic magmatism, occurring up to 2000 km
from the continental margin, implies that
anomalously hot mantle extended over a very
wide area and melted extensively. Consider-
ing the geochemical and isotopic composi-
tions of CAMP basalts, an important contri-
bution of lithospheric mantle is suggested.
The debate concerning the origin of CAMP is
open, and various models have been pro-
posed invoking the presence of either a man-
tle plume (1,2,25) or a shallow thermal
anomaly (7). In general, our data and previ-
ous (4–6,8 –10) geochemical and geochro-
nological data on CAMP are consistent with
models that suggest that an upwelling plume
head was trapped beneath the lithosphere and
separated from the plume tail (25), with
plume material spread over a very large area
by ambient mantle flux (2). However, these
models require modification to account for
CAMP magmatism not only in North Amer-
ica and Africa but also in South America.
The Triassic-Jurassic (T-J) boundary is
one of the five major Phanerozoic mass ex-
tinction events and involved marine and ter-
restrial genera and families (26 ). The timing
of the huge CAMP magmatic event overlaps
within errors with modern estimates for the
age of the T-J boundary. CAMP basaltic
dikes and flows (for example, the 201 Ma
Palisades and Gettysburg sills and the Orange
Mountain flow) of eastern North America
essentially define this boundary in the
Culpeper, Fundy, and Newark basins (5,20,
27 ) of North America. Documentation of the
enormous spatial extent of the CAMP, and its
temporal brevity, support the possibility of a
genetic relation with the T-J extinctions (28).
References and Notes
1. P. R. May, Geol. Soc. Am. Bull.82, 1285 (1971).
2. M. Wilson, J. Geol. Soc. London 154, 491 (1997).
3. M. P. Withjack, R. W. Schlische, P. E. Olsen, AAPG Bull.
82, 817 (1998).
4. J. F. Sutter, U.S. Geol. Surv. Bull. 1776, 194 (1988); A.
Sebai, G. Fe`raud, H. Bertrand, Earth Planet. Sci. Lett.
104, 455 (1991); L. Fiechtner, H. Friedrichsen, K.
Hammerschmidt, Geol. Rundsch. 81, 45 (1992); K.
Deckart, G. Fe`raud, H. Bertrand, Earth Planet. Sci.
Lett. 150, 205 (1997); A. K. Baksi and D. A. Archibald,
ibid. 151, 139 (1997).
5. G. R. Dunning and J. P. Hodych, Geology 18, 795
(1990); J. P. Hodych and G. R. Dunning, ibid. 20,51
(1992).
6. C. Alibert, Earth Planet Sci. Lett. 73, 81 (1985); C.
Dupuy et al.,ibid. 87, 100 (1988); W. J. Pegram, ibid.
97, 316 (1990); J. G. McHone, Geology 24, 319
(1996).
7. W. S. Holbrook and P. B. Kelemen, Nature 364, 433
(1993).
8. G. Bellieni et al.,Neues Jahrb. Mineral. Abh. 162,1
(1990).
9. C. R. Montes-Lauar et al.,Earth Planet. Sci. Lett. 128,
357 (1995).
10. F. M. F. De Almeida, Rev. Bras. Geociencias 16, 325
(1986).
11. P. R. Renne et al.,Earth Planet. Sci. Lett. 144, 199
(1996).
12. E. M. Piccirillo and A. J. Melfi, Eds., The Mesozoic Flood
Volcanism of the Parana` Basin: Petrogenetic and Geo-
physical Aspects (University of Sa˜o Paulo, Sa˜o Paulo,
Brazil, 1988).
13. K. G. Cox, in Continental Flood Basalts, J. D. Macdou-
gall, Ed. (Kluwer Academic, Dordrecht, Netherlands,
1988), pp. 239–272.
14. P. R. Renne et al.,Chem. Geol. (Isot. Geosci. Sect.)
145, 117 (1998).
15. M. A. Richards, R. A. Duncan, V. E. Courtillot, Science
246, 103 (1989); I. H. Campbell and R. W. Griffiths,
Earth Planet. Sci. Lett. 99, 79 (1990).
16. S. D. C. Guerreiro and A. Schult, Mu¨nchner Geophys.
Mitt. 1, 37 (1986).
17. M. Ernesto, I. G. Pacca, R. Siqueira, paper presented at
the International Union of Geodesy and Geophysics
Meeting, Boulder, CO, 1995.
18. J. J. Schott, R. Montigny, R. Thuizat, Earth Planet. Sci.
Lett. 53, 457 (1981).
19. A. E. Rapalini, A. L. Abdeldayen, D. H. Tarling, Tec-
tonophysics 220, 127 (1993).
20. W. K. Witte, D. V. Kent, P. E. Olsen, Geol. Soc. Am.
Bull. 103, 1648 (1991); D. V. Kent, P. E. Olsen, W. K.
Witte, J. Geophys. Res. 100, 14965 (1995).
21. Z. Yang et al.,J. Geophys. Res. 101, 8025 (1996).
22. P. R. Renne and A. R. Basu, Science 253, 176 (1991).
23. R. A. Duncan et al.,J. Geophys. Res. 102, 18127
(1997).
24. A. Heimann et al.,Earth Planet. Sci. Lett. 121,19
(1994).
25. A. M. Leitch, G. F. Davies, M. Wells, ibid. 161, 161
(1998).
26. D. M. Raup and J. J. Sepkoski, Science 231, 833
(1986); P. E. Olsen, N. H. Schubin, M. H. Anders, ibid.
237, 1025 (1987).
27. S. Fowell and A. Traverse, Rev. Palaeobot. Palynol. 86,
211 (1995).
28. V. E. Courtillot et al.,Geol. Soc. Am. Spec. Pap. 307,
513 (1996).
29. S. Hart and A. Zindler, in Mantle Convection. Plate
Tectonics and Global Dynamics, W. R. Peltier, Ed.
(Gordon and Breach Science, New York, 1989), pp.
261–388.
30. E. Irving and G. A. Irving, Geophys. Surv. 5, 141
(1982); J. Besse and V. E. Courtillot, J. Geophys. Res.
96, 4029 (1991).
31. P. D. Rabinowitz and J. LaBrecque, J. Geophys. Res.
84, 5973 (1979).
32. This research was funded by Italian (Ministero
dell’Universita` e della Ricerca Scientifica e Tecno-
logica and Consiglio Nazionale delle Ricerche), Bra-
zilian (Programa da Apoio ao Desenvolvimento Cien-
tı´fico e Tecno´logico/Financiadora de Estudos e Pro-
jectos and Conselho Nacional de Desenvolvimento
Cientı´fico e Tecnolo´gico), and U.S. (NSF grant EAR-
9496326) agencies and the Ann and Gordon Getty
Foundation. We thank M. Iacumin, P. Antonini, M.
Hirschmann, and W. Sharp for discussions; Compan-
hia de Pesquisa de Recursos Minerais for logistical
support in sampling the Roraima dikes; C. Montes-
Lauar (Anari and Tapirapua˜) and A. Novello (Ceara´)
for providing samples; and T. Becker for argon labo-
ratory assistance. Special thanks are due to geologist
Sandoval Pinheiro for his help during fieldwork.
14 December 1998; accepted 10 March 1999
Fig. 4. Age probability spectra for CAMP (4,5,
present study) from North America (five data
samples), West Africa (20 data samples), and
South America (16 data samples). The curve
labeled “CAMP All” (41 data samples) allows
comparison of the
40
Ar/
39
Ar with the U/Pb data
by including uncertainties in decay constants,
errors in K-Ar data for
40
Ar/
39
Ar standards, and
2sanalytical errors (14). The curve labeled
“CAMP Ar/Ar” (36
40
Ar/
39
Ar data samples) in-
cludes only 1sanalytical errors and provides
the best estimate of duration of a brief mag-
matism, similar to those of the Karoo (23) and
Siberian Traps (22) CFBPs. All
40
Ar/
39
Ar data
are normalized to the same standard basis
[FCs 528.02 Ma (14)].
Fig. 5. Circles show the mean paleomagnetic
poles (95% confidence) of CAMP tholeiites of
South America (SA) (9,16,17, present data),
southern Europe (SE) (18), western Africa
( WA) [compiled by (19)], and North America
(NA) (20). The Apparent Polar Wander Path
from 300 to 125 Ma (open diamonds) for
North America is shown for comparison (30).
Rotation poles of South America to Africa
and North America to Africa are after (17)
and (31), respectively.
REPORTS
23 APRIL 1999 VOL 284 SCIENCE www.sciencemag.org618
on July 16, 2020 http://science.sciencemag.org/Downloaded from

Province
Extensive 200-Million-Year-Old Continental Flood Basalts of the Central Atlantic Magmatic
Andrea Marzoli, Paul R. Renne, Enzo M. Piccirillo, Marcia Ernesto, Giuliano Bellieni and Angelo De Min
DOI: 10.1126/science.284.5414.616
(5414), 616-618.284Science
ARTICLE TOOLS http://science.sciencemag.org/content/284/5414/616
CONTENT
RELATED file:/content
REFERENCES http://science.sciencemag.org/content/284/5414/616#BIBL
This article cites 38 articles, 9 of which you can access for free
PERMISSIONS http://www.sciencemag.org/help/reprints-and-permissions
Terms of ServiceUse of this article is subject to the
is a registered trademark of AAAS.ScienceScience, 1200 New York Avenue NW, Washington, DC 20005. The title
(print ISSN 0036-8075; online ISSN 1095-9203) is published by the American Association for the Advancement ofScience
No claim to original U.S. Government Works.
Copyright © 1999 The Authors, some rights reserved; exclusive licensee American Association for the Advancement of Science.
on July 16, 2020 http://science.sciencemag.org/Downloaded from