LETTERS PUBLISHED ONLINE: 20 DECEMBER 2009 | DOI: 10.1038/NGEO724 High Earth-system climate sensitivity determined from Pliocene carbon dioxide concentrations Mark Pagani1*, Zhonghui Liu1,2, Jonathan LaRiviere3 and Ana Christina Ravelo3 Climate sensitivity—the mean global temperature response 90° N to a doubling of atmospheric CO2 concentrations through 60° N radiative forcing and associated feedbacks—is estimated at 882 982 ◦ 30° N 1012 1.5–4:5 C (ref. 1). However, this value incorporates only 1208 925 relatively rapid feedbacks such as changes in atmospheric 0° 806 water vapour concentrations, and the distributions of sea Latitude 30° S ice, clouds and aerosols2. Earth-system climate sensitivity, by contrast, additionally includes the effects of long-term 60° S feedbacks such as changes in continental ice-sheet extent, 90° S terrestrial ecosystems and the production of greenhouse gases 0° 60° E 120° E 180° 120° W60° W 0° other than CO2. Here we reconstruct atmospheric carbon Longitude dioxide concentrations for the early and middle Pliocene, when ◦ temperatures were about 3–4 C warmer than preindustrial 0 0.2 0.4 0.6 0.8 1.0 1.2 1.4 1.6 1.8 2.0 3–5 values , to estimate Earth-system climate sensitivity from a Phosphate (µM) fully equilibrated state of the planet. We demonstrate that only a relatively small rise in atmospheric CO2 levels was associated Figure 1 j Modern site locations and surface-water phosphate with substantial global warming about 4.5 million years ago, distributions30. and that CO2 levels at peak temperatures were between about 365 and 415 ppm. We conclude that the Earth-system climate represent a range of oceanographic and algal growth environments, sensitivity has been significantly higher over the past five including Ocean Drilling Program (ODP) Sites 806, 882, 1012 and million years than estimated from fast feedbacks alone. 1208 from the Pacific Ocean, and ODP Sites 925 and 982 from the The magnitude of Earth-system climate sensitivity can be Atlantic Ocean (Fig. 1) (Supplementary Information). assessed by evaluating warm time intervals in Earth history, such Alkenones are long-chained (C37–C39) unsaturated ethyl and as the peak warming of the early Pliocene ∼4–5 million years methyl ketones produced by a few species of photoautotrophic 19 ago (Myr). Mean annual temperatures during the middle Pliocene haptophyte algae in the modern ocean . Alkenone-CO2 estimates (∼3:0–3:3 Myr) and early Pliocene (4.0–4.2 Myr) were ∼2:5 ◦C are based on the stable carbon isotope compositions of the di- ◦ 13 (refs 3, 4), and 4 C (ref. 5) warmer than preindustrial conditions, unsaturated C37 methyl ketone (δ C37:2) and the total carbon- respectively. During the early Pliocene, the equatorial Pacific Ocean isotopic fractionation that occurred during algal growth ("p37:2). maintained an east–west sea surface temperature (SST) gradient Chemostat incubations with nitrate-limited conditions show that ◦ of only ∼1:5 C, which arguably resembles permanent El Niño- "p37:2 varies with the concentration of aqueous CO2, [CO2aq], like conditions6. Meridional5,7 and vertical ocean temperature specific growth rate, µ, and cell geometry20. In contrast, dilute gradients8 were reduced, and deep-ocean ventilation enhanced, batch cultures with nutrient-replete conditions yield substantially 9,10 relative to today . Deterioration in Earth's climate state from lower "p37:2 values, a different relationship for "p versus µ/CO2aq, a 3.5 to 2.5 Myr led to an increase in Northern Hemisphere minor response to ([CO2aq]) (ref. 21), and an irradiance effect on 11 glaciation . By ∼2 Myr, subtropical Pacific meridional SST the magnitude of "p37:2 (ref. 22). Given the available experimental gradients resembled modern conditions5, and the Pacific zonal SST results, the potential exists that different growth and environmental gradient (∼5 ◦C) was similar to the gradient observed today, with a conditions trigger different carbon isotopic responses23. However, 6 strong Walker circulation . validation of the alkenone–CO2 approach using sedimentary Tectonics and changes in ocean12–14 and atmospheric alkenones in the natural environment indicates that this technique circulation15,16 were potentially important factors in climate can be used to resolve relatively small differences in water column 3− evolution during this time. However, an assessment of the timing [CO2aq] when SST and phosphate concentrations ([PO4 ]) and of oceanographic and climate changes17, and the stability of the are reasonably constrained24. We use alkenone unsaturation indices 18 K0 Greenland ice sheet to a range of possible forcings , implicate (U37) to determine SST and assume haptophyte production depths 3− atmospheric CO2 as the primary factor driving the warmth of the between 0 and 75 m at each site, bounding a range of [PO4 ] early Pliocene and the onset of Northern Hemisphere glaciation. determined from modern mean-annual-phosphate depth profiles For this study, we evaluate the magnitude of CO2 change and (Supplementary Information) that is then applied to calculate a 3− Earth-system climate sensitivity during the Pliocene by using the range of pCO2 estimates (Fig. 2). We assume that Pliocene [PO4 ] alkenone–CO2 method to reconstruct Pleistocene–Pliocene pCO2 at each site was similar to modern values or within the range of 3− histories from six ocean localities. Ocean sites used in this study PO4 concentrations that encompass the modern photic zone. 1Department of Geology and Geophysics, Yale University, New Haven, Connecticut 06520, USA, 2Department of Earth Sciences, The University of Hong Kong, Hong Kong, China, 3Ocean Sciences Department, University of California, Santa Cruz, California 95064, USA. *e-mail: [email protected]. NATURE GEOSCIENCE j VOL 3 j JANUARY 2010 j www.nature.com/naturegeoscience 27 © 2010 Macmillan Publishers Limited. All rights reserved. LETTERS NATURE GEOSCIENCE DOI: 10.1038/NGEO724 ab450 475 c Site 925 Site 806 525 Site 1012 (1) 425 450 (1) (1) 475 400 425 425 375 400 (2) (2) 375 (ppm) (ppm) (ppm) (2) 2 2 2 350 375 325 CO CO CO p p p 325 350 275 300 (1) f(x) = 15x + 330 325 (1) f(x) = 9x + 385 225 (1) f(x) = 48x + 312 (2) f(x) = 14x + 295 (2) f(x) = 10x + 343 (2) f(x) = 32x + 207 275 300 175 0 1 2 3 4 5 6 0 1 2 3 4 5 6 0 1 2 3 4 5 6 Age (Myr) Age (Myr) Age (Myr) de425 475 f375 Site 1208 Site 882 Site 982 450 350 375 (1) 425 325 (1) 400 325 375 300 (ppm) (ppm) 350 (ppm) 275 2 (2) 2 2 (2) 275 325 CO CO (1) CO 250 p p 300 p 225 225 275 (1) f(x) = 38x + 251 (1) f(x) = 12x + 264 (1) f(x) = 19x + 280 250 (2) 200 (2) f(x) = 14x + 207 (2) f(x) = 29x + 188 (2) f(x) = 10x + 216 175 225 175 0 1 2 3 4 5 6 0 1 2 3 4 5 6 0 1 2 3 4 5 6 Age (Myr) Age (Myr) Age (Myr) 3− Figure 2 j Alkenone-based atmospheric CO2 concentrations and CO2slope. Carbon dioxide estimates are calculated from "p37:2 values and PO4 estimates (Supplementary Information). Upper and lower CO2 estimates represent a range calculated assuming production depths between 0 and 75 m depth. Dashed lines bounding CO2 estimates are linear regressions of maximum and minimum CO2 values versus time. Arrows in b and c represent CO2slope and early Pliocene CO2 estimates as the result of probable changes in physical oceanography and growth rates. Note the different scales for each panel. Our "p37:2 records all show a general trend of decreasing 5 to 0.5 Myr. Differences in the magnitude of early Pliocene values from ∼5 to 0.5 Myr, and alkenone-based temperature CO2 from site to site can be accounted for by differences in records (Supplementary Information) are consistent with other haptophyte growth rate, air–sea CO2 equilibrium and the level of alkenone- and Mg/Ca-based estimates5 that support regional irradiance22, as well as by differences in the regional expression 3− 23 differences in Pliocene temperature change. Warm-pool regions of the "p37:2–[PO4 ] calibration (Supplementary Information). in the tropical western Pacific Ocean (ODP Site 806) and western Instead of relying on absolute CO2 values, we focus on the slope Atlantic Ocean (ODP Site 925) show no to very little warming of CO2 change with time (CO2slope) for each site (represented by ◦ (perhaps ∼0:5 C) during the early Pliocene. In contrast, very linear regressions in Fig. 2) to determine the magnitude of CO2 large SST differences (6–10 ◦C) between the early Pliocene and change before preindustrial times (Fig. 2). The y intercepts of these 7 preindustrial times characterize mid- and high-latitude sites . regressions predict CO2 concentrations in the range of preindustrial Large site-to-site differences in SST trends imply that some to glacial CO2 levels (except for ODP Site 806, which predicts portion of observed SST change was influenced by regional twentieth-century values) (Fig. 2), supporting the robustness of changes in oceanographic conditions. For example, Site 1012 is the alkenone–CO2 methodology, as well as the observation of an subjected to coastal upwelling—a process that depresses mixed- approximate linear decrease in pCO2 since ∼4:5 Myr. Changes in layer temperatures by vertically transporting cold, nutrient-rich CO2 over the past 4.5 million years based on minimum estimates of 3− thermocline waters. In general, cold SSTs are notably absent from [PO4 ] (Supplementary Information) range from 45 to 144 ppm, 3− coastal upwelling regions, indicating that upwelling rates abated whereas changes in CO2 based on maximum [PO4 ] estimates and/or the subsurface source of upwelling water was warmer range between 41 and 216 ppm.