RESEARCH | REPORTS 30. M. X. Kirby, B. MacFadden, Palaeogeogr. Palaeoclimatol. OISE, EAR, DRL 0966884, Colciencias, and the National Geographic Supplementary Text Palaeoecol. 228, 193–202 (2005). Society. We thank N. Hoyos, D. Villagomez, A. O’Dea, C. Bustamante, Figs. S1 to S3 O. Montenegro, and C. Ojeda. All the data reported in this manuscript Tables S1 to S5 ACKNOWLEDGMENTS are presented in the main paper and in the supplementary materials. References (31–42) Supported by Ecopetrol-ICP “Cronología de la Deformación en las Cuencas Subandinas,” Smithsonian Institution, Uniandes P12. SUPPLEMENTARY MATERIALS 160422.002/001, Autoridad del Canal de Panama (ACP), the Mark www.sciencemag.org/content/348/6231/226/suppl/DC1 12 November 2014; accepted 2 March 2015 Tupper Fellowship, Ricardo Perez S.A.; NSF grant EAR 0824299 and Materials and Methods 10.1126/science.aaa2815 EARTH HISTORY The PTB in the Tethys is characterized by two negative d13C excursions interrupted by a short- term positive event (10). There is no consensus as Ocean acidification and the to the cause of this “rebound” event and so we instead focus on the broader d13Ctrend.Ourd13C Permo-Triassic mass extinction transect (Fig. 1B) starts in the Changhsingian (Late Permian) with a gradual decreasing trend, interrupted by the first negative shift in d13Cat M. O. Clarkson,1*† S. A. Kasemann,2 R. A. Wood,1 T. M. Lenton,3 S. J. Daines,3 EP1 (at 53 m, ~251.96 Ma) (Figs. 1B and 2). This is S. Richoz,4 F. Ohnemueller,2 A. Meixner,2 S. W. Poulton,5 E. T. Tipper6 followed by the minor positive rebound event (at 54 m, ~251.95 Ma) (Figs. 1B and 2) before the Ocean acidification triggered by Siberian Trap volcanism was a possible kill mechanism minima of the second phase of the negative CIE for the Permo-Triassic Boundary mass extinction, but direct evidence for an acidification (58to60m,~251.92Ma)(Figs.1Band2)that event is lacking. We present a high-resolution seawater pH record across this interval, marks the PTB itself. After the CIE minimum, using boron isotope data combined with a quantitative modeling approach. In the latest d13C gradually increases to ~1.8 per mil (‰) and Permian, increased ocean alkalinity primed the Earth system with a low level of remains relatively stable during the earliest atmospheric CO and a high ocean buffering capacity. The first phase of extinction on April 10, 2015 2 Triassic and across EP2. was coincident with a slow injection of carbon into the atmosphere, and ocean Our boron isotope record shows a different pH remained stable. During the second extinction pulse, however, a rapid and large pattern to the carbon isotope excursion. The injection of carbon caused an abrupt acidification event that drove the preferential loss boron isotope ratio (d11B) is persistently low of heavily calcified marine biota. (Fig. 1C) at the start of our record during the late-Changhsingian, with an average of 10.9 T he Permo-Triassic Boundary (PTB) mass (7). A rapid input of carbon is also potentially 0.9‰ (1s). This is in agreement with d11Bvalues extinction, at ~252 million years ago (Ma), recorded in the negative carbon isotope excur- (average of 10.6 T 0.6‰,1s) reported for early- represents the most catastrophic loss of sion (CIE) that characterizes the PTB interval Permian brachiopods (19). Further up the section 10 11 (at ~40 m, ~252.04 Ma) (Fig. 1C), there is a stepped biodiversity in geological history and played ( , ). The interpretation of these records is, www.sciencemag.org T 12–16 d11 T ‰ a major role in dictating the subsequent however, debated ( ) and is of great impor- increase in Bto15.3 0.8 (propagated evolution of modern ecosystems (1). The PTB ex- tance to understanding the current threat of uncertainty, 2sf) and by implication an increase tinction event spanned ~60,000 years (2)and anthropogenically driven ocean acidification (11). inoceanpHof~0.4to0.5(Fig.2).d11Bvalues can be resolved into two distinct marine extinc- To test the ocean acidification hypothesis, we then remain relatively stable, scattering around tion pulses (3). The first occurred in the latest have constructed a proxy record of ocean pH 14.7 T 1.0‰ (1s) and implying variations within Permian [Extinction Pulse 1 (EP1)] and was fol- across the PTB using the boron isotope compo- 0.1 to 0.2 pH, into the Early Griesbachian (Early lowed by an interval of temporary recovery be- sition of marine carbonates (d11B) (17). We then Triassic) and hence across EP1 and the period of fore the second pulse (EP2), which occurred in used a carbon cycle model (supplementary text) carbon cycle disturbance (Figs. 1 and 2). Downloaded from the earliest Triassic. The direct cause of the mass to explore ocean carbonate chemistry and pH After the d13C increase and stabilization (at extinction is widely debated, with a diverse range scenarios that are consistent with our d11Bdata ~85 m, ~251.88 Ma) (Fig. 1), d11B begins to de- of overlapping mechanisms proposed, including and published records of carbon cycle distur- crease rapidly to 8.2 T 1.2‰ (2sf), implying a widespread water column anoxia (4), euxinia (5), bance and environmental conditions. Through sharpdropinpHof~0.6to0.7.Thed11Bmin- global warming (6), and ocean acidification (7). this combined geochemical, geological, and model- imum is coincident with the interval identified Models of PTB ocean acidification suggest that ing approach, we are able to produce an envelope as EP2. This ocean acidification event is short- 11 a massive and rapid release of CO2 from Siberian that encompasses the most realistic range in pH, lived (~10,000 years), and d B values quickly re- Trap volcanism acidified the ocean (7). Indirect which then allows us to resolve three distinct cover toward the more alkaline values evident evidence for acidification comes from the inter- chronological phases of carbon cycle perturba- during EP1 (average of ~14‰). pretation of faunal turnover records (3, 8), poten- tion,eachwithverydifferent environmental con- The initial rise in ocean pH of ~0.4 to 0.5 units tial dissolution surfaces (9), and Ca isotope data sequences for the Late Permian–Early Triassic duringtheLatePermian(Fig.2)suggestsalarge Earth system. increase in carbonate alkalinity (20). We are able 1 School of Geosciences, University of Edinburgh, West Mains We analyzed boron and carbon isotope data to simulate the observed rise in d11B and pH Road, Edinburgh EH9 3FE, UK. 2Faculty of Geosciences and MARUM–Center for Marine Environmental Sciences, from two complementary transects in a shallow through different model combinations of in- University of Bremen, 28334 Bremen, Germany. 3College marine, open-water carbonate succession from the creasing silicate weathering, increased pyrite dep- of Life and Environmental Sciences, University of Exeter, United Arab Emirates (U.A.E.), where deposi- osition (21), an increase in carbonate weathering, Laver Building, North Parks Road, Exeter EX4 4QE, UK. 4 tional facies and stable carbon isotope ratio and a decrease in shallow marine carbonate dep- Institute of Earth Sciences, NAWI Graz, University of Graz, 13 Heinrichstraße 26, 8010 Graz, Austria. 5School of Earth and (d C) are well constrained (18). During the ositional area (supplementary text). Both sili- Environment, University of Leeds, Leeds LS2 9JT, UK. PTB interval, the U.A.E. formed an expansive cate weathering and pyrite deposition result in a 6Department of Earth Sciences, University of Cambridge, carbonate platform that remained connected large drop in partial pressure of CO2 (PCO2)(and Downing Street, Cambridge CB2 3EQ, UK. to the central Neo-Tethyan Ocean (Fig. 1A) (18). temperature) for a given increase in pH and *Corresponding author. E-mail: [email protected] d13 W †Present address: Department of Chemistry, University of Otago, Conodont stratigraphy and the distinct Ccurve saturation state ( ).Thereisnoevidencefora Union Street, Dunedin, 9016, Post Office Box 56, New Zealand. are used to constrain the age model (17). large drop in PCO2, and independent proxy data SCIENCE sciencemag.org 10 APRIL 2015 • VOL 348 ISSUE 6231 229 RESEARCH | REPORTS Fig. 1. Site locality and high-resolution Bioclastic pack-grainstone Transect carbon and boron Calcisphere grainstone WSA SHA1 isotope data. (A) Laminated to massive dolomudstone PALEOTETHYS Paleogeographic Bioturbated mud- to packstone N EO reconstruction for the -T E T Unidentified (estimated texture) A H Late Permian showing E Y S G the studied section Foraminifera N Gastropods A Wadi Bih, in the P Calcareous algae Musandam Mountains Bivalves of U.A.E., that formed Thinsection Outcrop an extensive carbon- ate platform in the 100 Neo-Tethyan Ocean. mid-Bih breccia [Modified from (35).] 13 90 (B) Shallow water d C I.isarcica EP2 record (18). (C) Boron 85 isotope (d11B) record (propagated uncer- tainty given as 2sƒ ) 80 and average Early Per- mian brachiopod value 75 ? n 19 Griesbachian ( = 5 samples) ( ). Early Triassic Lithology, biota, and 70 H . parvus H . transect key are Thrombolite provided in (A). Only 65 Hindeodus parvus has been found so far in 60 this section (18), and PTB the conodont zones praeparvus 55 with dashed lines are H. EP1 identified from the 13 d C record and 50 C.yini regional stratigraphy (36–38). 45 40 35 30 25 Late Permian Changhsingian 20 changxingensis C . 15 10 Average Early-Permian 5 Brachiopod 0 (m) WP G MFRB Fossils 012345 6 8 1012141618 11 Zone δ13 δ B (‰) Conodont C (‰) indicate only a minor temperature decrease of A decrease in carbonate sedimentation is con- Sea-level fall also exposed carbonates to weath- a few degrees celsius during the Changhsingian sistent with the decrease in depositional shelf ering (23), which would have further augmented (22), suggesting that these mechanisms alone can- area that occurred because of the second-order thealkalinityinflux.ThepHincreaseeventsup- not explain the pH increase (fig.