PALEOCEANOGRAPHY, VOL. 10, NO. 4, PAGES 763-773, AUGUST 1995 Assessment of sea-surface temperature at 42øN in the California Current over the last 30,000 years FredrickG. Prahl, NicklasPisias,MargaretA. Sparrow,and AnneSabin Collegeof Oceanicand Atmospheric Sciences, OregonStateUniversity,Corvallis Abstract. Assessment of changes in surface oceanconditions, in particular, sea-surface temperature (SST),is essential to understand long-term changes in climateespecially in regions wherecontinental climateis strongly influenced by oceanographic processes. To evaluatechanges in SST in the northeastPacific,we haveanalyzedlong-chain alkenones of prymnesiophyte originat 38 depthsin a pistonand associated triggercorecollected beneaththe contemporary coreof the California CurrentSystem at 42øN,-0270km offthecoastof Oregon/California. The samples span30,000yearsofdeposition at thislocation.Unsaturation patterns k• (U37)in thealkenone series display a statistically significant difference (p((0.001) between interglacial (0.44q-0.02,n- 11)andglacial(0.29q-0.04,n- 20)intervals of thecores.Detailedexamination of othercompositional features of the C37,C38, C39alkenone series and a related C36 alkenoate series measured downcore suggests k• thepublished U37- temperature calibration (u3k•? -- 0.034x T q-0.039), defined for culturesof a strainof Emilianiahuxleyiisolatedfromthesubarctic Pacific,provides bestestimates of winterSST at ourstudysite. Thisinference is purelystatistical and doesnot imply, however,that the phytoplanktonsourceof thesebiomarkers iskmost productivein winter or at the oceansurface.The temperaturerecordfor • U37 implies (1) an -04øCshift occurredin winter SST from -0 7.5 4- 1.1øCat the lastglacialmaximum to -0 11.74-0.7øCin thepresent interglacial period,and(2) thiswarmingtrendwasconfined to the time frame14-10Ka withinthe glacialto interglacial transitionperiod.Theseconclusions arecorroborated entirelyby results froman independent SSTtransformation of radiolarian species assemblage data obtained from the same core materials. Introduction coincidentwith the point where wind stressdiverges [Hickey,1979]. Wind stressdivergence beginsin the mid-ocean(•-160øW) at •-40øN in summerand •-30øN The modernCaliforniaCurrentis a well-studied hydrographic featurein the NorthPacificOcean[Hickey, in winter. Separation of the wind field continuesto the 1979].Temporalmeasurements of the physicalandbio- east but at progressivelyhigher latitudes as the North (50øNin summer; logicalcharacteristics of this currentsystemhavebeen Americancontinentis approached 40øN in winter). The northward and southward flowing the focusof the CaliforniaCooperative Fisheries (CalCOFI) Programstarted in the early 1950s. CalCOFI currentsseparatein a broad transitionalzone (Figure data reveal variations in the character of the California Current on seasonal as well as interannual timescales. while that flowing southward becomesthe core of the Detailed studiesof the physicsof this systemand consequences of physicson regionalpatterns of primary productivityhaveevolvedin morerecentyearsthrough otherprograms [Strubet al., 1990;Hoodet al., 1991]. The California Current derives from the West Wind Drift (Figurela). The WestWind Drift separates into northwardand southwardflowingcomponents roughly l a). The northwardcomponent feedsthe AlaskanGyre, CaliforniaCurrent. North of 40øN and nearshore,the CaliforniaCurrentislargelySubarctic in type(lowtemperature and salinity,high dissolvedO•. and phosphate content).The percentage of Subtropical water(warmer and saltier,O•. and phosphate-poorer) mixedinto this system increases to the south and toward the west. The core of flow is located --150-300 km off the coast from the Oregon/California border(Figurelb). Flowintensity is always maintained southward but varies season- Copyright 1995 by the American GeophysicalUnion. Paper number 95PA01394. 0883-8305/95/95PA-01394510.00 ally, being fully developedand situated most nearshore in late summerto early fall and diminishedand situated most offshore in winter. Changes in the structure of the California Current betweenthe last glacialand the presentinterglacialpe- 764 PRAHL ET AL.' PALEOTEMPERATURES IN THE CALIFORNIA CURRENT (42øN) 60øN riod are known only vaguely at best. Some insight has been gleaned from paleoceanographicstudy of microfossil records in sediment cores from the region. The Subarctic boundary appearsto have shifted northward (Figurela) sincethe last glacialperiod.Moore[1973a] showedthe radiolarian assemblagenow representative of the Alaskan Gyre withdrew from the region -o41ø to 43ø30'N and 126025' to 127øW off Oregon/northern 50øN California and was replaced successivelyby a central Subarcticassemblageand finally a TransitionalZoneas- semblage.The samepattern of biotic chan•ehasbeen 40øN documented subsequently in a wider range of northeast Pacific cores spanning as far west as 132øW off Cape Blanco(-043øN)and as far southas 39øNoff the RussianRiver [Sabin,1995]. Theseobservations argue 180øW 160øW 140øW 120øW that the contemporary positions of divergencefor the West 46øN .................... 44ON ................. Wind Drift into northward and southward flow- ing currents[Hickey,1989]and of coastalupwellingoff the coast of the northwesternUnited States [Huyer, 1983]are significantlydifferenttoday fromthosewhich prevailed during the last glacial period. In glacial times, either the intensity of coastal upwelling diminished markedly in this region by an increasedinfluence of the Alaskan Low-PressureSystem or the whole area of episodiccoastalupwellingshiftedsouthward[Moore, 1973a]. This oceanographic implicationfrom the mi- ................... crofossil record 42øN in sediments is consistent with model predictionsfor large-scalechangesin atmosphericcirculation in the northeast Pacific resulting from deglacia- tion in North America[Cooperative HoloceneMapping Project(COHMAP), 1988,and references therein]. In this paper, we examine sedimentaryevidencefor a glacial to interglacial changein sea-surfacetemperature _•4ooo •Mendocino FZ I i::i::!iiiiiiiii!i!::i:: CA::iii::iiiii:: 40øN (SST) within the CaliforniaCurrentat a siteoff north- ern California. The site is located -0270 km offshore at -042øNwithin the contemporarycoreof this current sys- tem (Figurelb). Mooreet al. [1980]analyzedmicrofosi i i 130oW i i i 120ow 38ON Figure 1. (a) Map identifyingthe major surfacecurrents in the North Pacific and showingthe relationship of the California Current systemto the West Wind Drift (adaptedfromHickey[1989]).(b) Expandedmapindicating the coring site for W8709A-8TC and W8709A- 8PC examinedin this study (seestar; 3111 m water depth,42ø32.5'N127ø40.7'W).The maprelatesthe coring site to the longitudinalrange of the contemporary CaliforniaCurrentSystemoff Oregon/northernCalifornia (seeareaenclosed by bolddashedlines).For refer- sil speciesassemblages in a wide range of sedimentcores and reconstructeda map of SST for the entire glacial Pacific Ocean. Their resultssuggested<2øC colderwaters existed in this portion of the California Current during the last glacial period than at present,although SST estimatesalong the eastern limb of the North Pacific Current were not well constrainedby sampling. We employ indep, endent analyses of alkenone unsaturation patterns(Uak7) andradiolarian species assemblages on more recently collectedcorematerials to reconfirmprior assessmentof colder temperatures in the glacial California Current. Our results define winter SST in the northeast Pacific over the past 30,000 years and show ence purposes,three sediment trap sites for the MUL- (1) an -•4øC differencebetweenglacialand interglacial TITRACERS project (NS, MW, and G [Lyle et al., estimatesand (2) confinementof the shift from colder 1992]) and the collectionsite for two additionalcores to warmer conditions to the time frame of 14-10 Ka are alsoindicated(seesolidcircles;Y6604-10:3002 m water depth, 43ø16'N 126ø24'W; Y6910-2:2615 m wa- during the glacial to interglacial transition. ter depth,41ø16'N127ø01'W).Mooreet al. [1980]pre- Methods viously examined the microfossilrecords of the latter coresto reconstruct sea-surfacetemperatures at 18 Ka in this region of the northeast Pacific Ocean. Sample collection. A pistoncore(W8709A-8PC; 875 cm long, 10 cm diameter) and its associated trig- PRAHL ET AL.' PALEOTEMPERATURESIN THE CALIFORNIA CURRENT (42øN) 765 gercore(W8709A-8TC,190cm long)werecollectedin et al., 1988].ME/Ka7 measures the relativeabundance 1987 aboard RV Wecomafrom 3111 m water depth at of alkenoatesto alkenonesin a given biomarker series and is calculated from the abundance ratio of total Ca6 42ø32.5'N127ø40.7'W(Figurelb). Bothcoresweresec- tionedimmediately onboard shipinto-01.5-mlengths. methyl alkenoatesto total C37 alkenones[Prahl et al., Each length was capped,sealed,and returned to Corval- 1988]. The reproducibilityof measurements for each lis underrefrigeration(4øC) for storagein the Oregon State University (OSU) corerepository.At OSU, the compositional ratio is better than +5% in duplicate core sectionswere each split lengthwise to produce a working and archivalhalf. One-centimeter-thick,quarter round subsampleswere then taken as needed from samples. Radiolarian analysis. Radiolarian speciescensus data (Table 2) were collectedon thesecoresas part of a larger study of the northeastPacific[Sabin,1995]. the working half for purposesof organicgeochemical Cores 8TC and 8PC were subsampleddowncoreevery and microfossilanalysis. 10 cm yielding a total of 34 samples. Radiolarian miComparisonof detailed magnetic susceptibilityand croscopeslides were prepared using the technique of calcium carbonate records for 8TC and 8PC indicated Moore[1973b]asmodifiedby RoelofsandPisias[1986]. the piston coreoverpenetratedby 140 cm. So, all sam- After chemicalcleaning,sampleswere sievedwith a 63pling depthsfor the 8PC are reported +140 cm to cor- /•m screenand radiolarian fossilswere randomly settled onto microscopeslides. QUantitative countsof radiorect for overpenetration. A timescalefor these coreswas establishedusing combineddata for acceleratormass larian specieswere completed based on a total count of spectrometry (AMS) 14Cdatesonbulkorganiccarbon 800 radiolarian tests. The proportion of total radiolarand/or calciumcarbonateand oxygenisotope(5180) ian counted for 41 specieswas used in the estimation of mean annual SST. stratigraphy on benthie foraminifera. Sedimentation rate variedfrom 10 to 12 cm/kyr overthe 3.2 m of core examined in this study with lower values estimated at greater core depth. All data and the method used to establishchronostratigraphyfor $TC and 8PC are dis- cussedelsewhere[Lyleet al., 1992]. Alkenone analysis. Total extractablelipids(TEL) were recovered ultrasonically from 3 to 4 g of wet sediment using a 1:3 solvent mixture of toluene and methanol. Lipid fractions enriched in C37, C38, and C39 alkenonesand structurally related C36 alkenoates of prymnesiophyte origin [Marloweet al., 1990]were isolatedfrom TEL by column chromatographyand analyzedby capillarygaschromatography (GC). Further details of the extraction and analysisprocedureare dis- Mean annual SST were estimated from radiolarian population studies by a paleoecologicaltransfer function developed using the technique of Imbrie and Kipp r•a• •;•'•;• population data from l•n surface sediment samplesfrom the Pacific basin were analyzed usingQ modefactoranalysis(N. PisiasandA. Mix, unpublisheddata, 1994). This approachidentifiedseven radiolarian assemblages with discretegeographicdistributions and thereby allowed statistical distinction between surface waters of specificoceanographicsettings in the Pacific: Subtropical, Antarctic, Arctic, Transitional, Eastern Boundary Current, Gyre, and the Warm Western Basin. The relative abundance of these seven assemblageswas used to develop a paleotemperature transfer function for mean annual SST. cussedelsewhere [Prahlet al., 1989,1993]. The transfer function was developedusing nonlinear Alkenone/alkenoate signatures were measured multiple regressionanalysiswhere the assemblagedata throughoutthe triggercore (8TC) and in all intervals were the independent variable and mean annual SST at examinedfrom the top 1.8 m of the pistoncore(8PC). the location for each surfacesediment sample was used Various indices (U3k'7, K37:4/K37, K37/K38, U• e, and ME/Ka7; Table 1) are utilized to describethe overall prymnesiophyte biomarkercomposition identifiedby k• GC in each sample(Figure 2). U37, Ka7:4/Ka7,and U• e measurethe degreeof unsaturationin a given k• alkenone/alkenoate series.U37 is calculatedfrom the abundance ratio of the diunsaturated C37 ketone to the combineddiunsaturatedand triunsaturated C37 ketone. The latter two unsaturation indices are new to as the dependent variable. Mean annual SST was estimated for each sample location using data in the atlas of Levitus[1982].The standarderrorof the regression equation is +1.5øC. Residuals of the regressionfit for surfacesediment samplesfrom the northeast Pacific are less than +IøC. Thus we feel that this paleotemperature transfer function provides estimates for past seasurface conditions q- 1.5 ø C. in the northeast Pacific accurate to within the literature. K37:4/K37is calculatedfrom the abundance ratio of tetraunsaturated C37 ketone to the combined diunsaturated, triunsaturated, and tetraunsaturated C37 ketone. And U• e is calculatedfrom the abundanceratio of diunsaturatedC36methyl alkenoate to the combineddiunsaturated and triunsaturated C36 methylalkenonate.K37/K38measures the overallchain length in a given alkenoneseriesand is calculatedfrom the abundance ratioof total C37to C38alkenones [Prahl Results and Discussion Alkenone/alkenoate compositions. Figures2a and 2b illustrate typical gas chromatogramsof the alkenone/alkenoate seriescontainedin sedimentintervals of interglacial and glacial age, respectively. The most conspicuouscompositional differenceevident from this comparison is the virtual absenceof a tetraunsat- 766 PRAHL ET AL.' PALEOTEMPERATURES IN THE CALIFORNIA CURRENT (42øN) Table la. CompositionalProperties of Alkenoneand Alkenoate SeriesAnalyzed With Depth in a Trigger Core (W8709A-8TC) and PistonCore (W8709A-8PC)Fromthe NortheastPacificOcean Core Depth, TC8 TC8 TC8 TC8 TC8 TC8 TC8 TC8 TC8 TC8 TC8 TC8 TC8 TC8 TC8 TC8 TC8 TC8 TC8 TC8 TC8 TC8 TC8 TC8 TC8 TC8 TC8 TC8 PC8 PC8 PC8 PC8 PC8 PC8 PC8 PC8 PC8 PC8 See text Ka 2.5 4.5 23.5 26.0 47.5 65.5 67.5 88.5 89.5 97.5 107.5 117.5 127.5 128.5 134.5 137.5 140.5 141.5 144.5 146.5 149.5 152.5 157.0 163.5 166.5 175.0 183.5 185.5 159.5 178.5 199.5 219.5 239.5 259.5 259.5 279.5 299.5 319.5 0.2 0.4 2.1 2.3 4.5 6.4 6.6 8.9 9.0 9.9 10.9 12.0 13.1 13.2 13.9 14.2 14.6 14.7 15.0 15.2 15.5 15.8 16.3 17.0 17.4 18.1 19.0 19.2 16.6 18.5 20.5 22.2 23.7 25.1 25.1 26.5 28.0 29.4 for further Table lb. Age, cm U•; SST, U• e Ka7/K3s ME/K37 Ka•:4/Ka• øC 0.438 0.435 0.484 0.414 0.421 0.427 0.427 0.428 0.4 77 0.448 0.417 0.364 0.391 0.369 0.308 0.309 0.317 0.331 0.339 0.328 0.326 0.337 0.308 0.323 0.333 0.315 0.319 0.301 0.271 0.323 0.249 0.265 0.226 0.227 0.235 0.294 0.249 0.311 11.7 11.6 13.1 11.0 11.2 11.4 11.4 11.4 12.9 12.0 11.1 9.6 10.4 9.7 7.9 7.9 8.2 8.6 8.8 8.5 8.4 8.8 7.9 8.4 8.6 8.1 8.2 7.7 6.8 8.4 6.2 6.6 5.5 5.5 5.8 7.5 6.2 8.0 0.77 0.80 0.87 0.72 0.75 0.74 0.76 0.80 0.83 0.77 0.74 0.64 0.69 0.67 0.63 0.68 0.64 0.62 0.61 0.61 0.62 0.61 0.59 0.62 0.62 0.61 0.61 0.63 0.61 0.70 0.66 0.60 0.54 0.56 0.57 0.74 0.56 0.53 1.06 1.09 1.02 1.11 1.18 1.05 1.11 1.14 1.07 1.21 1.18 1.19 1.15 1.17 1.13 1.16 1.04 0.97 0.95 0.90 0.86 0.88 1.00 0.90 0.96 0.87 1.02 0.99 0.79 0.92 0.89 0.82 0.82 0.89 0.89 0.88 0.76 0.97 0.24 0.24 0.22 0.27 0.24 0.27 0.23 0.21 0.23 0.22 0.24 0.24 0.30 0.26 0.23 0.22 0.24 0.26 0.28 0.28 0.25 0.28 0.28 0.26 0.26 0.24 0.23 0.24 0.28 0.20 0.21 0.30 0.26 0.28 0.29 0.20 0.29 0.21 0.02 0.00 0.02 0.01 0.00 0.00 0.02 0.00 0.01 0.02 0.03 0.06 0.03 0.06 0.08 0.08 0.08 0.09 0.09 0.10 0.10 0.10 0.11 0.10 0.10 0.10 0.11 0.11 0.09 0.11 0.10 0.13 0.13 0.14 0.13 0.04 0.15 0.08 details. k! AverageAlkenoneCompositional Propertiesand U37-Be•sed Sea-Surface TemperatureEstimates Calculated for Interglacial and Glacial SedimentIntervals of Cores W8709A-8TC and 8PC From the Northeast Pacific Ocean Va•; Interglacial SSTøC VTM , 36 Ka•/Kas ME/Ka• KaT:4/Ka• n Ave Std Ave Std Ave Std Ave Std Ave Std Ave Std 12 0.438 0.022 11.7 0.7 0.78 0.04 1.11 0.06 0.24 0.02 0.01 0.01 20 0.294 0.038 7.5 1.1 0.61 0.05 0.90 0.07 0.26 0.03 0.11 0.02 34 0.344 0.070 9.0 2.1 0.67 0.08 1.00 0.13 0.25 0.03 0.07 0.05 (0-10 Ka) Glacial (15-30 Ka) Overall (0-30 Ka) Ave is average,Std is standarddeviation,and n is the total numberof samples in eachagecategory. PRAHLET AL.-PALEOTEMPERATURES IN THE CALIFORNIA CURRENT(42øN) • •E (A) E e _w• 767 A differenceof high statistical significanceis also observedwhen the same time averagesare comparedfor the alkenonechainlengthindex (K37/K3s,p < 0.001) • but not for the index of alkenoate to alkenone abun- dance(ME/Ks•, p • 0.1). Notably,all compositional changes occurduringthe periodof glacialto interglacial transition(i.e., •14-10 kyr B.P.). Alkenone ature. estimates of surface water ternperk• The alkenone unsaturation index U37 varies with phytoplankton growthtemperature[Brassellet al., (B) I I 1986]. For a singlestrainof Emilianiahuxleyi(55a, Northeast Pacific Culture Collection), isolated from k• the subarcticPacificOcean,U37changes quitelinearly I 59 60 61 RetentionTime (minutes) withgrowthtemperature (u3k'7 -- 0.034X T + 0.039, r 2 = 0.994[Prahlet al., 1988]). Prior fieldtestsink• Figure 2. Partial capillary gas chromatograms of alkenone/alkenoatefractionsisolatedfrom (a) interglacial and (b) glacial intervalsof W8709A-8TC and W8709A-8PC. The alkenonesare depicted by the code K followed by 37, 38, or 39 indicating carbon chain length; :4, :3, or :2 indicating the number of double bonds; and m or e indicating methyl or ethyl ketone. The alkenoates are depicted by the code A followed by 36 indicating carbon chain length, :3 or :2 indicating number of double bonds, and m or e indicating methyl or ethyl ester. The observed chromatographic separation was accomplishedusing on-column injection, a 0.25-mm i.d. x 30-m-long DB-1 fused silica capillary column, hydrogen carrier gas maintained dicatethe applicabilityof the U37 -T calibrationfor strain 55a of E. huxleyi to Pacific waters ranging in Table 2a. Sea-Surface Temperatures Predicted Using a Transform Function for Radiolarian SpeciesAssemblage Data Obtained With Depth in a Trigger Core (W8709A-STC) and PistonCore (W8709A-SPC)From the Northeast Core Pacific Ocean Depth• Age, cm Ka øC TC8 1 0.1 10.8 TC8 7 0.6 11.7 at constantheadpressure (0.35 kg/cm2) throughout TC8 16 1.3 10.5 the analysis,and temperatureprogramming(5øC/min from 100-300øCthen isothermal). As illustrated, a TC8 26 2.3 11.1 TC8 36 3.2 12.6 coelution problem existed between the ethyl alkenone K38:3e and the ethyl alkenoate A36:2e present in all TC8 TC8 46 56 4.2 5.3 12.1 11.5 samples.Quantitativemeasurement of the ME/K37 in- TC8 TC8 TC8 TC8 TC8 TC8 TC8 TC8 TC8 TC8 TC8 TC8 PC8 PC8 PC8 PC8 PC8 PC8 PC8 PC8 PC8 PC8 PC8 PC8 PC8 PC8 PC8 65 6.3 76 86 96 105 125 135 145 155 165 175 185 190 200 210 220 230 240 251 261 270 280 290 300 310 320 330 7.4 8.5 9.6 10.7 12.9 14.0 15.0 16.1 17.1 18.1 19.1 19.6 20.5 21.5 22.2 23.0 23.7 24.5 25.2 25.9 26.6 27.3 28.0 28.7 29.4 30.1 11.6 10.8 11.0 11.5 11.9 11.1 10.4 9.3 8.5 9.0 7.2 6.5 6.7 7.5 8.3 7.2 7.4 7.3 7.2 7.6 7.1 6.9 6.7 6.6 7.3 6.5 8.6 dex in selected glacial and interglacial samples before and after transesterification with 14% boron trifluoride in methanol[Prahl et aL, 1989]showedthe concentration of A36:2e was consistently60-70% of the concen- tration for A36:2m. All reportedvaluesfor K37/K•s have been correctedfor this coelutionproblem by mathematically subtracting away the A36:2e contributionto K38:3e usingthe followingexpression:K37/K38 (cor- rected)= [(K37/K38) -1 (measured)-0.65 x U•6e x ME/K37]-1. urated C37 ketone (K37:4m) in interglacialsediments and its clear occurrence in glacial sediments. Besides this feature, all analyzed sedimentintervalscontain the sameset of alkenone/alkenoate components. Table I displays complete data for the various atk• tributes(U37, K3?:q/K37,K3?/K3s,U• e, and ME/K3?) of the alkenone/alkenoate compositions measuredat 38 depths in the cores. Comparison of data averaged for the interglacial(0-10 Ka, n = 11) and glacial(15-30 Ka, n - 20) samplingintervalsshowssignificantdifferences(one-tailedStudent-ttest) for all threeunsaturak', K37:4/K37, andu•e). tion indices(p • 0.001 for U37 See text for further details. 768 PRAHL ET AL.' PALEOTEMPERATURES IN THE CALIFORNIACURRENT (42øN) Table 2b. Average Sea-SurfaceTemperature Esti- mates on a Transform Based Function for Radiolar- JanSpeciesAssemblagesCalculated for Interglacial and Glacial Sediment Intervals of Cores W8709A-8TC 8PC From the Northeast Pacific Ocean and k! inferredfromthe U37data and followsconsistently with the temperatureresponseof K3•/K3s observedin cultures of strain 555 (Figure 35). Second,valuesof the alkenoneunsaturationparameter,K3•:q/K3•, increase frominterglacialto glacialsedimentintervals(Table 1). SST, øC 2.0 n Ave Std (A) 1.8 Interglacial 12 11.4 0.6 20 7.5 0.8 34 9.1 2.1 (0-10 Ka) 1.6 Glacial 1.4 (15-30 Ka) Overall 1.2 (0-30 Ka) Ave is average, Std is standard deviation, and n is the total number of samples in each age category. 1.0 Strain 55a 8TCand8PC 0.8 , 0.6 , temperaturefrom -.•4 to >_25øC[Prahl and Wakeham, 1987; $ikes and Volkman,1993]. With this findingas precedent,the laboratory calibration equationwas used k• I , , , 10 , I , , , 15 , I Pacific into estimates i I i Temperature (øC) 0.20 to convertdowncoremeasurements of U37at our study site in the northeast i 20 (B) of absolute surfacewater temperature(Table 1). Estimatesaver- 0.15 age 11.7 + 0.7øC and 7.5 + 1.1øC for sediment intervals ofinterglacial andglacial age,respectively. The steplikedowncore decrease in u3k7 (Table1) al- most certainly depicts lower glacial relative to interglacial surface water temperatures. But we must now question the accuracy of the absolute water temperatures estimated above for two reasons. First, more re- 0.10 0.05 k• cent culturestudiesshowthe specificU37 -T relationship definedfor strain 555 of E. huxleyiis not applicable to all alkenone-producingprymnesiophytespotentially contributingto the marine sedimentaryrecord [Marloweet al., 1990]. Most notably,significantdeviations are evident betweencultured strainsof E. huxleyiisolated from variousmarine watersincludingoff the coast ofBermuda (u3k'7 -- 0.053x T- 0.55,calibration range: 15-25øC), theGulfofMaine(u3k'7 -- 0.016x T - 0.028, calibration range:10-20øC), a Norwegian fjord(u3k'7 = 0.033x T- 0.009, calibrationrange:10-20øC),and the 0.00 10 15 20 25 Temperature (øC) 0.30 (c) AA A& ß 0.25 &ß ß ß ß AA ß ß ß 0.20 ß 0.15 0.10 Sargasso Sea(u3k'7 = 0.013x T-0.12, calibration range: 15-25øC)(K. Amthor,unpublished results,1994). Seck• ond, anomalousrelationshipsbetweenU37 and water 0.05 , temperature now seem apparent in the North Atlantic [Conteet al., 1992],the BlackSea[Freemanand Wakeham, 1992], and in very cold (<4øC) watersfrom the SouthernOcean[Sikesand Volkman,1993]. However, despite such concern,our alkenone-derived estimatesof absolutewater temperature are judged as reasonableby comparisonof various compositionalfea- turesof the alkenone/alkenoate signatures preserved in the cores with culture data for strain 555 of E. hux- leyi. First, valuesof K3•/K3s, an indexof carbonchain length in the alkenoneseries,decreasefrom interglacial to glacialintervalsof the core (Table 1). This trend accompaniesthe -.•4øCdecreasein growth temperature , 0.00 5 10 15 20 25 Temperature (øC) 3. (a) K37/K38, an index of carbonchain length, (b) K37:4/K37,an index of alkenoneunsaturation, and (c) ME/K37, an index of the relativeabunFigure dance of alkenoatesto alkenonesin a given biomarker series plotted versus temperature for strain 55a of E. huxleyigrown in culture [Prahl et al., 1988]. For reference purposes, downcore data for each parameter are also plotted at growth temperatures predicted k' from U37 measurements usingthe calibrationequation k' U37- 0.034x T + 0.039 [Prahlet al., 1988]. PRAHL ET AL.- PALEOTEMPERATURESIN THE CALIFORNIA CURRENT (42øN) In cultures of strain 55a, this precisetrend is observed for K37:4/K37asgrowthtemperaturedecreases overthe inferredrangeof 11.7øto 7.5øC(Figure3b). So,a strain of E. huxleyi behaving like 55a is a probable, major alkenonecontributorto sedimentsin this regionof the northeastPacificoverthe last 30,000yearsof deposition and providesan appropriatetemperature calibrationfor k• U37. Evidence for selective degradation of prymnesiophyte biomarkers. Comparisonof culture data for strain 55a of E. huxleyi with core data reveals one noteworthy compositional difference that warrants 769 et al., 1988, and references therein]. A varietyof field k' observations now bolsterour view that Us7 is insensitive to early diageneticalteration [Prahl et al., 1989; McCaffreyet al., 1990]. Growthtemperatureremains the key factor recognizedto affect alkenone unsaturation patterns recorded in sedimentary samples. But, as describedabove, determination of the absolutequantitative relationship between alkenoneunsaturation patterns and growth temperature still requires a means of recognizing the specific algal source contributing the biomarker signal to a particular sedimentary setting. Current evidencefor the resistanceto diagenetic alter- discussion. The alkenoate to alkenonecomposition ation of indices based on the abundance ratio of two (ME/Ka7) of this straindisplaysconsiderable tempera- biosynthetically related but chemically distinct comture dependence with valuesof 0.10-0.15 expectedwhen poundclasses(e.g.,AA36, ME/Ks7) is tenuousat best growthoccursat _<10øC(Figure3c). But ME/Ka7 val- by comparison (see,for example,Conteet al. [1992]). uesshowlittle significantdifference(p -0 0.1) between Both diunsaturated and triunsaturated C•6 methyl interglacial(0.24q-0.02) andglacial(0.26q-0.03) sed- alkenoatesoccur in all depth intervals of 8TC and 8PC iment intervals(Table 1) and are --2-fold higherthan examinedin this study(Figure2). U• e, an indexanalo- expectedif growthoccurredin the inferredtemperature range of 7.5-11.7øC (Figure 3c). Assuminga strain of E. huxleyilike 55a representsthe major biomarker contributorto thesesediments,selectivedegradationof alkenonesrelativeto alkenoateswouldexplainthe dis- goustOU'37,k' isnow to quantify downcore variaß defined tion in alkenoateunsaturationpatterns(Table 1). U• • k' showssignificant,positivecorrelationwith U37 in the k' complete dataset(U• e -- 1.00xUa7+0.33,r 2 -- 0.74). parity. Alkenone/alkenoate data for sedimenttrap time series and surfacesedimentsfrom this regionof the north- may, in fact, be higher if analytical conditions were optimized to allow baseline peak resolution of the mi- eastPacific[Prahlet al., 1993]supporta selective degra- nor triunsaturatedC• methylester(A36:3m)and the dation hypothesis. ME/Ks7 measuredin three sur- major diunsaturatedC•7 ketone(K37:2m) (Figure2). face sedimentsbetween120, 270, and 630 km offshore Nonetheless,the observeddegreeof correlation supports along42øN (NS, MW, and G; Figurelb) are uniform our view that C•6 alkenoates and C•7, C•8, and (0.25q- 0.01), while averageparticulatematerialsset- alkenones derive from the same source and unsaturation patterns in thesetwo biosyntheticallyrelated compound lower and vary widely (0.09, 0.06, and 0.02, respec- classesconveythe same information about growth temtively). Flux comparisons showonly a smallfractionof peratures for E. huxleyi in the northeast Pacific. More the alkenonestransportedthrough 1000 m water depth significantly, however, if alkenonesare degraded selectively over alkenoatesin the sedimentary process,this survivesburial in sediment. If slight differencesin the degreeof degradation of alkenonesrelative to alkenoates correlation suggeststhat alkenoate degradation operoccurred,the dissimilaritybetweenME/Ks7 valuespre- ates nonselectivelyon a compound specificbasis, an inservedin sedimentsand expectedfor the presumedalgal ference completely consistent with our current view of alkenone geochemistry. source would be accountable. Relationship of biomarker estimates to seaThe possibilityof selectivedegradationof alkenones Using sedimenttraps deployrelative to alkenoatesin the sedimentaryprocessis rel- surface temperature. tling through 1000 m water depth above each site are ed in time series,Prahl et al. [1993]examinedthe com- evant to recentwork of Conte et al. [1992]. These investigatorscoinedan index (AA36) to quantifythe relativeabundance of alkenoates to alkenones in a given sampleand proposedits useeither independently or in k' conjunctionwith U37 to assesssurfacewater temperatures. If thesetwoclasses of biochemicals (estersof fatty acidsand ketones)are derivedfrom a commonsource [Marloweet al., 1990]but degraded selectively to any significantextent in the early stagesof sedimentation gardlessof samplingperiod, U37 ws•sremarkablyuniform throughoutthe year (0.41 q- 0.04) and compared wellwith valuesrecordedin underlyingsediment(0.43). k' Again, averageU37 valuesfor sedimenttrap materials as our data might suggest,the utility of AA36 as an unequivocaltool for paleotemperaturereconstructionis translated into growth temperatures of 10.9 q- 1.2øC using the existing calibration for strain 55a of E. huxleyi position and flux of alkenonesthrough 1000 m water depth at the coringsite for the presentstudy. Alkenone flux was evident in all samplingperiodsimplying some level of prymnesiophyteproductivity year round. Rek' compromised. [Prahlet al., 1988].This watertemperatureis observed Similar concernfor selectivedegradationof the more unsaturatedalkenoneisomerswas raised in the early at the sea surface along 42øN in the northeast Pacific throughout the winter and during summer periods of upwelling along the coast. But such cold water tern- k' development of Us7asa paleotemperature index[Prahl 770 PRAHL ET AL.: PALEOTEMPERATURES IN THE CALIFORNIA CURRENT (42øN) Radiolarian Assemblages perature occurs only within the upper thermocline at more offshoresites in the northeast Pacific during sum- mer to fall periodsof stratification[Prahlet al., 1993]. Interestingly, the inferred subsurfacedepth of alkenone production invariably correspondsto a zone of chloro- phyll maximumin theseoffshoreregions[Prahl et al., 1993]. Consequently, we surmisethat the alkenoneunsaturation patterns preservedin sedimentsfrom study regionssuchas ours beneaththe core of the contemporary California Current provide a proxy of winter SST even though prymnesiophyte productivity is not constrained to or maximal in this season. And, on this basis, we concludethat winter SST in the glacial California Current situated at 42øN in the northeast Pacific was at least 4øC colder than at present. Radiolarian species compositions. Figure 4 illustrates downcorechangesin radiolarian assemblages _ •80 .• 0.70 • 0.60 • 0.• observedin 8TC/8PC. Here we plot the four dominant factors found for modern and glacial sedimentsfrom the northeast Pacific, i.e., the Subtropical,Transitional, Eastern Boundary Current, and Arctic assemblages. The other three assemblagesidentified in the study of minor importance in this region. A marked changein radiolarian assemblagespreservedin these sedimentsis evident. The radiolarian population in glacial age sediments is dominated by a mixture of Arctic and Transitional assemblages.The speciesassociatedwith the Arctic assemblageare not found in modern sediments of the northeast Pacific but are presently found in sediment from the far North Pacific and Bering Sea. At 0,2 (18 to 14 Ka), the Arctic assemblage is replacedfirst by the return of the Transitional assemblage.Assemblages associatedwith the Subtropical Pacific and the Eastern Boundary Current subsequentlyincreasein importance. The reduced importance of the Eastern Boundary Current assemblages,and associatedreduction in the importance of upwellingrelated species,are consistent with inferred reduction of upwellingin the north- east Pacificduringthe last glacialepisode[Lyle et al., 1992]. This pattern suggests that the subpolarfront of the northeast Pacific was located much farther south t0 t5 20 25 • 0 5 tO t5 20 25 50 5 10 15 20 25 50 0.4 0,5 prevails as the Transitional assemblagebecomesmuch reduced in importance. At the onset of the glacial to interglacialtransition 5 0,5 N. Pisiasand A. Mix (unpublisheddata, 1994) are of the glacial maximum (18 Ka), the Arctic assemblage 0 0,4 0.5 0.2 0.1 [ 0.0 0 Age (kyr) Figure 4. Downcoreplotsof radiolarianassemblages derived from a Q Mode factor analysisof radiolarian populationsfrom 170 surfacesedimentsamplesspanningthe PacificBasin. Factorsshownare the Subtropical, Transitional,Arctic and EasternBoundaryCurrent factorsdefinedby N. Pisiasand A. Mix (unpublished results,1994). tions(Table2). The concordance of thesetwo,indepen- during the last glacial event. The observedchangesin the radiolarian assemblagesalso suggestthat deglacia- dent estimatesof past sea-surfaceconditionsis striking. Thus the conclusionfrom alkenoneanalysisseemsjusti- tion in the northeast fied that SST was at least 4øC colder at this site in the Pacific is related to a northward migration of the subpolar front acrossthe region. The pattern observedin 8TC/8PC is consistentwith the overall regional responseto deglaciation based on a northeastPacificduringthe last glacialmaximumthan at present. But why should a paleotemperatureequation, calibrated to mean annual SST, yield the same north-southtransectof coresfrom55øNto 30øN[Sabin, estimate as that provided by alkenoneunsaturation patterns reflecting winter SST? 1995]. Radiolarian species assemblages and SST estimates. In Figure 5, we plot versussedimentage both k' the SST estimatesbasedon U37 measurements (Table 1) and radiolarianpaleotemperature transferfunc- The radiolarian-based paleotemperature equation used in the presentwork was derivedthrough study of basinwide sediments from the Pacific and has been appliedto sedimentrecordsfrom the central and east- PRAHLET AL.. PALEOTEMPERATURES IN THE CALIFORNIACURRENT(42øN) 16 771 2.5 0 Radiolarian • uk'37 14 3.0 5180 X 14Cdates 12 3.5 10 4.0 o 4.5 6 5.0 I i 0 5 10 15 20 25 30 35 SedimentAge (kyr) k! Figure 5. Comparison of downcore SST predictions basedon Us7datawith thoseestimated usinga transformfunctionfor radiolarianspeciesassemblages. For purposes of stratigraphic reference,a downcoreprofile is shownfor 5180 measuredin tests of the benthic foraminifera Uvigerinaspp.,andcalendaragescalculated fromAMS 14Cdateson bulkorganiccarbonand calciumcarbonate samples are indicated(datafromLyle et al. [1992]).Seetext for further details. ern equatorial Pacific as well as the northeast Pacific (N. PisiasandA. Mix, unpublished results, 1994).The strategy of developingpaleoecological equationsfor bothmeansea-surface temperature andseasonality was chosenfor two reasons.First, prior approaches used paleotemperatureequationsto estimate a summerand winter SST. However,the correlationbetweenSST dur- returning to the question of why mean SST estimates derived from radiolarian assemblagedata shouldlook similar to winter SST estimates derived from alkenone data, we comparethe mean SST with water temperatures at the pycnoclinein the summerseason(Figure 6) where both parametershave been calculatedfrom the hydrographic data setof Levitus[1982].We assume that the temperature at the summerpycnoc!inepro- ingwarmandcoldseasons in thePacific(datafromLevitus[1982])isveryhigh(r = 0.93;N. PisiasandA. Mix, vides an estimate of oceantemperaturesat the baseof unpublished results,1994)andthusindependent regres- the euphoticzone which seemsto be the real physiosion equationscannot be developed,whereasthe cor- logical growth temperature reflectedby the alkenones relationbetweenmeanSSTandseasonality (datafrom Levitus [1982])ismuchreduced (r = -0.44) andthusindependent equationscanbe developed.Second, the definitionof winter and summerseasons nearthe equator becomes complicated because the thermalequatordoes not coincidenecessarily with the geographic equator. [Prahlet al., 1993].The correlationbetweenthesetwo temperatureparametersis very high (r - 0.90; Figure 6), and the two parametersare equivalentoverthe If the positionof the thermal equatoralsovarieswith pastclimate,then reconstructing pastseasonal temperatures is much more complicated at low latitudes. So temperature range of 0 to 15øC. The alkenonetemperature estimates seem to reflect winter SST or subsurface temperatures in summer[Prahlet al., 1993],andthe radiolarian transfer function, while calibrated to annual mean sea-surfacetemperatures, also provides a reasonable estimate of subsurfacetemperatures in summer. 772 PRAHL ET AL' PALEOTEMPERATURES IN THE CALIFORNIA CURRENT (42øN) 30- / Conclusions 1. Examination of various compositional properties of r=O.9 25 ß -** ß ..o/. ß %**ø ß %*'* ß,/?Y* .. 0 20 ß o o t- ß ß e/O 15 the alkenone/alkenoate seriespreserved in glacial/interglacial intervals of sedimentsfrom the northeast Pacific k/ showsthat the U37 - temperature calibration established for E. huxleyistrain 55a isolatedfrom the subarc- ticPacific (u3k'7 -- 0.034x T + 0.039[Prahl etal.,19881) providesrealistic estimates of absolutewater tempera- ß ture. •o lO ß 2. Early diageneticprocessespotentially alter the relative abundanceof alkenonesto alkenoatespreservedin oo k' ß ß o/ o I I I I I lO 16 20 26 30 MeanAnnualSST (TøC) sedimentseventhoughvaluesfor U37 and other compositionalindices(e.g., U•6, K37:4/Ks7,Ks7/Kss) based on ratios of compoundswithin a given chemical class appear faithfully translated from algal sourceto sediments. Given this possiblediagenetic effect, paleotemperature assessmentsbased on indices such as AA36 [Conteet al., 1992]shouldnowbe interpretedwith caution. k' 3. UsT-basedpaleotemperatureestimates for the Figure 6. Observed mean sea-surfacetemperatures (estimatedfrom averageof winter,spring,summer,and fall data) comparedto the mean temperatureat the northeast Pacific correspond to winter SST. This correspondenceis purely statistical and does not imply alkenone producers are most productive in the winter periodsor at the oceansurface[Prahlet al., 1993]. 4. Ocean temperature estimatesbasedon U3• and las of Levitus[1982].Data are basedon the geographic radiolaria transfer functions follow essentiallyidentical locationsof the 170 surfacesedimentsthroughoutthe summer pycnocline as determined from data in the at- Pacific Basin used to developthe sea- surfacetemperature transfer function for radiolarian speciesassem- blages(N. Pisiasand A. Mix, unpublished data, 1994). Thus it is not too surprisingthat our two independent estimatesof glacial/interglacialdifferencein oceantemperature are of the same magnitude in this core from the northeast Pacific. Our estimateof an •-4øC oceantemperaturechange since the last glacial maximum is higher than previ- ousestimates[Moore,1973a]. As in this study,Moore [1973a]usedradiolarian-based transferfunctions to estimate past sea-surfaceconditions and estimated a _<2øC temperature changeduring the last deglaciation.There are two factors which should be considered in evaluat- patterns of changeduring the last deglaciationand show that the northeast Pacific warmed by •-4øC since the glacial maximum. 5. Convergenceof the alkenoneand transfer functionderived temperature estimates supportsthe hypothesis that the oceanographicparameter being reconstructed by these techniquesin the northeast Pacific is a subsurface water temperature within the summer pycnocline. Acknowledgments. This manuscript benefited from thoughtful comments by Liz Sikes, Robert Thunell, and an anonymous reviewer. This research was supported by fundsfrom NSF grantsOCE-9000517(FGP), OCE-9203292 (FGP), and USGS-92-9460-1026(NP). We are also grateful to NSF for support of the MULTITRACERS project (OCE-8609366and OCE-8919956)and the corerepository at OregonState University (OCE-9402298). Graduatestudent assistantshipfor Anne Sabin was provided with support from the U.S. Geological Survey. ingthis difference:(1) the standarderrorof the transfer functionanalysisis of the orderof 1.5-2øCand (2) the References moderncalibrationdata set usedby Moore[1973a]and Brassell, S.C., G. Eglinton, I. T. Marlowe, U. Pfiaumann, the Climate:Long-RangeInvestigation,Mapping, and and M. Sarnthein, Molecular stratigraphy: A new tool for Prediction(CLIMAP) project (1976) is biasedand of climatic assessment,Nature, 320, 129-133, 1986. the order of 1øC too cold compared to other more re- centlycompiledmoderncalibrationdata sets(W. Prell, personalcommunication, 1995). Thus onepossibilityis that the •-2øC differenceis only reflectedin the standard error of the paleotemperatureregression equations or that the rangeof Moore [1973a]is underestimated Conte, M. H., G. Eglinton, and L. A. S. Madureira, Longchain alkenones and alkyl alkenoates as palaeotemperature indicators: Their production, flux and early sedimentary diagenesisin the eastern north Atlantic, Org. Geochem., 19, 287-298, 1992. CooperativeHoloceneMapping Project (COHMAP) mem- due to the bias in the modern temperature calibration bers, Climatic changes of the last 18,000 years: Observations and model simulations, Science, œ•1, 1043-1052, data set. 1988. PRAHLET AL.: PALEOTEMPERATURES IN THE CALIFORNIACURRENT(42øN) Freeman, K. H., and S.G. Wakeham, Variations in the distribution and isotopic compositions of alkenones in Black Sea particles and sediments, Org. Geochem., 19, 277-285, 1992. Hickey, B. M., The California Current system- Hypotheses and facts, Prog. Oceanogr., 8, 191-279, 1979. Hickey, B. M., Patterns and processesof circulation over the Washington continental shelf and slope, in Coastal Oceanographyof Washington and Oregon, edited by M. R. Landry and B. M. Hickey, pp. 41-115, Elsevier, New York, 1989. Hood, R. R., M. R. Abbott, and A. Huyer, Phytoplankton and photosynthetic light responsein the coastal transition zone off northern California in June 1987, J. Geophys. Res., 96(C8), 14,769-14,780,1991. Huyer, A., Coastal upwelling in the California Current system, Prog. Oceanogr., 12, 259-284, 1983. Imbrie, J., and N. G. Kipp, A new micropaleontologicmethod for quantitative paleoclimatology: Application to a late Pleistocene Caribbean core, in Late Cenozoic Glacial Ages, edited by K. K. Turekian, pp. 71-191, Yale Univ. Press, New Haven, Conn., 1971. Levitus, S., Climatological atlas of the World Ocean, NOAA Prof. Pap., 13, 1-173, 1982. Lyle, M., R. Zahn, F. G. Prahl, J. Dymond, R. Collier, N. Pisias, and E. Suess, Paleoproductivity and carbon burial across the California Current: The MULTITRAC- ERS transect, 42øN, Paleoceanography,7, 251-272, 1992. Marlowe, I. T., S.C. Brassell, G. Eglinton, and J. C. Green, Long-chain alkenones and alkyl alkenoates and the fossil coccolith record of marine sediments, Chem. Geol., 88, 349-375, 1990. McCaffrey, M. A., J. W. Farrington, and D. J. Repeta, The organic geochemistry of Peru margin surface sediments, I, A comparison of the Ca7 alkenone and historical E1 Nifio records, Geochim. Cosmochim. Acta, 5,i, 16711682, 1990. Moore, T. C., Jr., Late Pleistocene-Holoceneoceanographic changesin the northeastern Pacific, Quat. Res., 3, 99109, 1973a. Moore, T. C., Jr., Method of randomly distributing grains for microscopic examination, J. Sediment. Petrol., 904-906, 1973b. 77• Moore, T. C., Jr., et al., The reconstructionof sea surface temperatures in the Pacific Ocean of 18,000 B.P., Mar. Micropaleontol., 5, 215-247, 1980. Prahl, F. G., and S.G. Wakeham, Calibration of unsat- uration patterns in long-chainketonecompositions for palaeotemperatureassessment,Nature, 330, 367-369, 1987. Prahl, F. G., L. A. Muehlhausen, and D. L. Zahnle,Further evaluationof long-chainalkenonesas indicatorsof paleoceanographic conditions,Geochim.Cosmochim.Acta, 52, 2303-2310, 1988. Prahl, F. G., L. A. Muehlhausen,and M. Lyle, An organic geochemicalassessment of oceanographic conditionsat MANOP site C over the past 26,000 years, Paleoceanography, ,i, 495-510, 1989. Prahl, F. G., R. W. Collier, J. Dymond,M. Lyle, and M. A. Sparrow,A biomarkerperspectiveon prymnesiophyte productivityin the northeastPacificOcean,DeepSeaRes. Part I, ,iO, 2061-2076, 1993. Roelofs, A. K., and N. G. Pisias, Revised techniquefor preparingquantitative radiolarainslidesfrom deep-sea sediments,Micropaleontology, 32, 182-185, 1986. Sabin A. L., Holocene and latest Pleistocene paleoceanog- raphy of the northeastPacificand its relationshipto climate changein the PacificNorthwest,M.Sc. thesis,93 pp., Oreg. State Univ., Corvallis,1995. Sikes,E. L., and J. K. Volkman, Calibration of alkenone unsa•tura•tion ratios (U•',.) forpa!eotemperamre estimation in cold polar waters, Geochim. Cosmochim.Acta, 57, 1883-1889, 1993. Strub, P. T., C. James,A. C. Thomas, and M. Abbott, Seasonal and nonseasonalvariability of satellite-derived surfacepigmentconcentrationin the CaliforniaCurrent, J. Geophys.Res., 95(C7), 11,501-11,530,1990. N. Pisias, F. G. Prahl, A. Sabin, and M. A. Sparrow, College of Oceanic and AtmosphericSciences,Ore- gon State University, Corvallis, OR 97331. fprahl@oce.orst.edu) (e-mail: (ReceivedSeptember28, 1994;revisedMay 1, 1995; acceptedMay 1, 1995.)