6. K. L. Smith Jr., R. S. Kaufmann, R. J. Baldwin, Limnol.Oceanogr. 39, 1101 (1994).7. R. J. Baldwin, R. C. Glatts, K. L. Smith Jr., Deep-SeaRes. Part II Top. Stud. Oceanogr. 45, 643 (1998).8. K. L. Smith Jr., Limnol. Oceanogr. 32, 201 (1987).9. K.L.SmithJr.,R.S.Kaufmann,Science 284, 1174(1999).10. K. L. Smith Jr., R. J. Baldwin, D. M. Karl, A. Boetius,Deep-Sea Res. Part I Oceanogr. Res. Pap. 49, 971(2002).11.D.Roemmich,J.A.McGowan,Science 267, 1324(1995).12. D. Roemmich, J. A. McGowan, Science 268, 352 (1995).13. J. A. McGowan, D. R. Cayan, L. M. Dorman, Science281, 210 (1998).14. C. E. Reimers, R. A. Jahnke, D. C. McCorkle, GlobalBiogeochem. Cycles 6, 199 (1992).15. E. M. Druffel, B. H. Robison, Science 284, 1139 (1999).16. M. W. Silver, S. L. Coale, C. H. Pilskaln, D. R. Steinberg,Limnol. Oceanogr. 43, 498 (1998).17. W. M. Hamner, B. H. Robison, Deep-Sea Res. 39,1299 (1992).18. E. G. Barham, Science 205, 1129 (1979).19. P. J. Davoll, M. J. Youngbluth, Deep-Sea Res. 37, 285(1990).20.J.C.Hunt,D.J.Lindsay,Plankt. Biol. Ecol. 46,75(1999).21. C. P. Galt, Fish. Bull. 77, 514 (1979).22. S. E. Beaulieu, K. L. Smith Jr., Deep-Sea Res. Part IITop. Stud. Oceanogr. 45, 781 (1998).23. Small larvacean species are often very abundant innear-surface waters. Most have bodies less than 10 mmlong, with house diameters commonly twice as large.Their houses may be produced at a rate of six or moreeach day, depending on the density of food particles.Discarded small houses are important components oforganic aggregate flux in the ocean’s upper layers, butthey rarely reach the deep sea floor (34–36).24. Materials and methods are available as supportingmaterial on Science Online.25. Sediment traps catch what they were designed to catch,namely, small, slowly sinking particles. Although sedi-ment traps may occasionally collect sinker fragments,physical contact, particularly with traps that haveinterior baffles, is certain to exclude, disrupt, or dispersethis material (16). The easily recognized rectangularmesh structures of larvacean filters have not beenreported in analyses of sediment trap contents.26. R. Fenaux, Q. Bone, D. Deibel, in The Biology of PelagicTunicates, Q. Bone, Ed. (Oxford Univ. Press, New York,1998), chap. 15.27. Bathochordaeus sp. is found chiefly at depths from100to300m;Mesochordaeus erythrocephalusoccurs principally between 300 and 500 m (17, 28).28. R. R. Hopcroft, B. H. Robison, J. Plankton Res. 21,1923 (1999).29. J. H. Martin, G. A. Knauer, D. M. Karl, W. W. Broenkow,Deep-Sea Res. 34, 267 (1987).30. C. H. Pilskaln, J. B. Paduan, F. P. Chavez, R. Y. Anderson,W. M. Berelson, J. Mar. Res. 54, 1149 (1996).31. This value considerably exceeds the amount of flux es-timated by Silver, Coale, Pilskaln, and Steinberg (16),for Bathochordaeus inthesameregion.Althoughourmeasurements of the abundance and turnover ofhouses and sinkers agree, our measurements of thecarbon content of sinkers are substantially greater, prin-cipally because of incomplete sampling in the earlierstudy.32. F. P. Chavez et al., Prog. Oceanogr. 54, 205 (2002).33. A. B. Burd, G. A. Jackson, R. S. Lampitt, M. Follows,Eos 83, 573 (2002).34. A. L. Alldredge, Science 177, 885 (1972).35. A. L. Alldredge, Limnol. Oceanogr. 21, 14 (1976).36. D. Deibel, Mar. Biol. 93, 429 (1986).37. We thank the pilots of the ROVs Ventana andTiburon, for their skills and patience in the difficulttask of collecting these specimens, and the officersand crews of the research vessels Point Lobos andWestern Flyer. Supported by the David and LucilePackard Foundation.Supporting Online Materialwww.sciencemag.org/cgi/content/full/308/5728/1609/DC1Materials and MethodsSOM TextFigs. S1 to S4References and Notes23 December 2004; accepted 15 April 200510.1126/science.1109104Rapid Acidification of the OceanDuring the Paleocene-EoceneThermal MaximumJames C. Zachos,1*Ursula Ro¨hl,2Stephen A. Schellenberg,3Appy Sluijs,4David A. Hodell,6Daniel C. Kelly,7Ellen Thomas,8,9Micah Nicolo,10Isabella Raffi,11Lucas J. Lourens,5Heather McCarren,1Dick Kroon12The Paleocene-Eocene thermal maximum (PETM) has been attributed to therapid release of È2000  109metric tons of carbon in the form of methane. Intheory, oxidation and ocean absorption of this carbon should have lowereddeep-sea pH, thereby triggering a rapid (G10,000-year) shoaling of the calcitecompensation depth (CCD), followed by gradual recovery. Here we presentgeochemical data from five new South Atlantic deep-sea sections thatconstrain the timing and extent of massive sea-floor carbonate dissolutioncoincident with the PETM. The sections, from between 2.7 and 4.8 kilometerswater depth, are marked by a prominent clay layer, the character of whichindicates that the CCD shoaled rapidly (G10,000 years) by more than 2kilometers and recovered gradually (9100,000 years). These findings indicatethat a large mass of carbon (d2000  109metric tons of carbon) dissolved inthe ocean at the Paleocene-Eocene boundary and that permanent seques-tration of this carbon occurred through silicate weathering feedback.During the Paleocene-Eocene thermal maxi-mum (PETM), sea surface temperature (SST)rose by 5-C in the tropics and as much as 9-Cat high latitudes (1–3), whereas bottom-watertemperatures increased by 4- to 5-C(4). Theinitial SST rise was rapid, on the order of È103years, although the full extent of warming wasnot reached until some È30,000 years (30 ky)Fig. 3. Comparative plot of activehouses of giant larvaceans (blueline) and discarded sinkers (redline) versus depth, in square metersof area swept. The data are derivedfrom a 10-year time series ofquantitative video transects atdepth intervals between 100 and1000 m (n 0 679 transects). Withan average sinking rate of 800 mdayj1, the difference betweenthe integrated areas beneath thecurves indicates that these ani-mals produce a new house eachday (24).www.sciencemag.org SCIENCE VOL 308 10 JUNE 20051611REPORTSlater (5). The most compelling evidence forgreenhouse forcing is a coeval global carbonisotope excursion (CIE) of roughly –3.0 permil (°)indeep-seacores(4). The pattern ofthe CIE—an initial rapid decrease (È20 ky)followed by a more gradual recovery (130 to190 ky) (1, 6–8)—indicates the input of a largemass of isotopically depleted carbon into theocean and atmosphere. Quantitatively, meth-ane hydrates, with a mean d13CofG–60°,appear to be the most plausible source of thiscarbon (9). For example, only È1200  109metric tons of carbon (GtC) of