Evidence for a Volcanic Cooling Signal in a 335-Year Coral Record from New Caledonia

Home , Mount Logan

University of South Florida Scholar Commons

Marine Science Faculty Publications College of Marine Science

10-1997

Thomas J. Crowley Texas A & M University

Terrence M. Quinn University of South Florida, [email protected]

Frederick W. Taylor University of Texas

Christian Henin lnstitut Francais de Recherche Scientifique pour de Developpement and Cooperation

Pascale Joannot Aquarium de Noumea

Follow this and additional works at: https://scholarcommons.usf.edu/msc_facpub

Part of the Marine Biology Commons

Scholar Commons Citation Crowley, Thomas J.; Quinn, Terrence M.; Taylor, Frederick W.; Henin, Christian; and Joannot, Pascale, "Evidence for a Volcanic Cooling Signal in a 335-Year Coral Record from New Caledonia" (1997). Marine Science Faculty Publications. 108. https://scholarcommons.usf.edu/msc_facpub/108

This Article is brought to you for free and open access by the College of Marine Science at Scholar Commons. It has been accepted for inclusion in Marine Science Faculty Publications by an authorized administrator of Scholar Commons. For more information, please contact [email protected]. PALEOCEANOGRAPHY, VOL. 12, NO. 5, PAGES 633-639, OCTOBER 1997

Evidence for a volcanic cooling signal in a 335-year coral record from New Caledonia

ThomasJ. Crowley,1 Terrence M. Quinn,2 Frederick W. Taylor,3 ChristianHenin, 4 and PascaleJoannot, 5

Abstract. Althoughvolcanic cooling events have been detected in tree ring records,their occurrencein marinerecords hasreceived much less attention. Herein we reportresults from a 335-yearoxygen isotope record (1657-1992) from a New Caledoniacoral indicating that asmany as 16 interannual-scalecooling events occur within 1 year of a volcaniceruption as determinedby ice corerecords. There are alsopentadal/decadal-scale cooling events beginning in 1675, 1813,and 1903 that immediatelypostdate volcanic eruptions. However, the interannualcorrespondences are complicatedby the fact that someof the coolingevents also coincide with E1Nifios, which cause cooling in this part of the westernSouth Pacific. If ourconclusions are substantiated by furtherwork, occurrence of distinctvolcanic cooling signals may enablerefinement of coral chronologiesby use of the "eventstratigraphic" approach, with the mostpromising correlation horizons being associatedwith the following eruptions:1808 (Unknown), 1813-1821 (severaleruptions), 1835 (Coseguina),1883 (Krakatau),and possibly 1963 (Agung).

Introduction Although Genin et al. [ 1995] found evidencefor significant The role of volcanism as an agent of climate changehas overturn in their coral record, they speculatedthat isotopic been a topic of interest for some time becausestratospheric variations may be more difficult to detect in coral records. injections of sulphateaerosols should increasebackscatter of However, Gagan and Chives [1995] found evidence for solar radiationand causecooling. After many yearsof debate, Pinatubocooling in a northwestAustralia record. In this paper studiesof large volcaniceruptions indicate a significanteffect we provide further evidence for imprint of large volcanic on interannualclimate change [e.g., Bradley, 1988; Angell, eruptions on a 335-year coral record from New Caledonia. 1990]. Volcanic signals have also been detectedin tree ring There is also some evidence linking smaller isotope records [e.g., LaMarche and Hirschboeck, 1984; Joneset al., excursionswith smaller volcanic eruptions. 1995]. The role of volcanism with respect to decadal-scale variability is more uncertain. Although earlier studies Methods suggestedsuch an influence [Robock, 1979; Hammer et al., 1980; Porter, 1986], more recentwork has called into question We cored a living coral near Amedee Island, New Caledonia the strengthof this correlation[Wigley, 1991; Crowley et al., (22øS, 166øE), in September1992. This region was chosen 1993]. becauseof the very large data gap in the South Pacific; for Coral recordshave recentlyemerged as an importantsource example, many sea surface temperature(SST) records are of information on interannual-decadalclimate change [e.g., relatively incomplete before about 1960. The selection of Cole et al., 1993; Quinn et al., 1993; Dunbar et al., 1994; T. New Caledonia was based on availability of an excellent M. Quinn et al., A multicentury coral stable isotope record calibration data set (daily SST and salinity measurements) from New Caledonia, submittedto Paleoceanography, 1997 developed by the French oceanographicgroup at Institut (hereinafter referred to as T. M. Quinn et al., submitted Francaisde RechercheScientifique pour le D6veloppementon manuscript, 1997)]. The primary featuresof interest in these Coop6ration (ORSTOM) (cf. T. M. Quinn et al., submitted studieshave been variationsin the tropical Pacific circulation manuscript,1997]. Details of the site and a full descriptionof on E1 Nifio and decadaltimescales. Recently, however, Genin the long record are given by Quinn et al. [1996a; submitted et al. [1995] reportedon an anomalousalgal bloom in a coral manuscript, 1997]. Correlation of the monthly oxygen record from the Gulf of Eilat following the eruptionof Mount isotope record with a nearby high-quality sea surface Pinatubo in June 1991. The increase in water column overturn temperature (SST) record from Amedee Island is 0.86, is consistentwith observationsof a strong winter cooling in indicatingthat the coral is primarily recordinginformation on this region after volcanic eruptions [cf. Graf et al., 1993; SSTs on the seasonaltime scale. Although the mean annual Robock and Mao, 1995]. /5•80/SST calibrationis different, the coral still can be usedto estimateSSTs on time scaleslonger than seasonal(c.f. Quinn 1Department.of Oceanography,Texas A&M University,College et al., submittedmanuscript, 1997). Station. The chronology in the coral was determined by band 2Departmentof Geology, University of SouthFlorida, Tampa. counting (T. M. Quinn et al., submittedmanuscript, 1997) 31nstitutefor Geophysics, University of Texas,Austin. 41nstitut Francais de Recherche Scientifique pour de densitybands in this coral are quite distinct. The dark bands Developpementen Cooperation,Noumea, New Caledonia. are laid down in australwinter when growth rates slow; this 5Aquariumde Noumea, Noumea, New Caledonia. marks the beginningof the "coral year" in our record. Higher Copyright 1997 by the AmericanGeophysical Union. resolution sampling (12/yr) over a calibration interval

Paper number 97PA01348. indicatesthat 4/yr samplingyields estimatesof mean annual 0883-8305/97/97PA-01348512.00. conditions similar to 6/yr and 12/yr [Quinn et al., 1996a].

633 634 CROWLEY ET AL.: VOLCANIC SIGNAL IN A CORAL RECORD

The data set will be availablein the National GeophysicalData volcanic compilation of Robock and Free [!995] but not in Center following publication of the full descriptionof the our record(again, seediscussion in appendix). coral record (T. M. Quinn et al., submittedmanuscript, 1997). With respectto the pre-1800 record, where the origin and In order to compare our coral results with the volcanic chronologyof the volcanic events are less well known, six of record for the southernhemisphere, we compileda composite the nine volcanic events (1693, 1720, 1744, 1752, 1754, and estimateof middle- to low-latitudeeruptions based on records 1788) coincide within a year of an eruption. There are two of conductivityand sulphurin ice core recordsfrom Antarctica other possible volcanic events (1668 (Pacaye?), 1799 (Table 1). The details involved in the constructionof this (Fuego)) occurringin the Moore et al. [1991] recordthat are index are discussedin the appendix. The volcanicpeaks have below the thresholdlevel usedin this study(see appendix) but been compared to catalogs of volcanism, primarily from which also coincidewith coolingevents in the New Caledonia Sirekin and Siebert [1994] but augmentedby Lamb [1970]. record. Although many of the volcanicpeaks are basedon occurrence Even if furtheranalysis modifies the suggestedlow-latitude in more than one ice core, a numberof peaksin the eighteenth origin of someof the volcanicpeaks (Table 1), there is still a centuryare from only one core [Moore et al., 1991]. Thus the large amount of evidenceindicating that many of the coral identificationof peaks and attributionsin this sectionrequire cooling events occur at the same time as low-latitude more verification. As the most compelling evidence for a eruptions. If there were no other complicating factors volcano/coral cooling occurs in the post-1800 record (see associated with these cooling events (see below), the below), we do not consider the uncertainty with respectto likelihoodof 16 volcanic"hits" occurring in sucha recordis volcano identificationsin the eighteenthcentury as being exceedingly small (p < 0.00001). This probability was critical to the main conclusionof our paper. determinedin the followingmanner. First, we determinedthe numberof interannualcooling events (47) and multipliedthis Results numberby 2 to accountfor the fact that we are determining whetherthe volcanopeak is coincidentor predatesby a year a The principal results of the coral-volcanocomparison are cooling event in the coral record. Then we divided the entire shown in Figure 1. In addition to a long-term trend and length of the record (335 years) by this 2-year criteria to multidecadalvariability (which are discussedby T. M. Quinnet determine the number of independentopportunities for a al. (submitted manuscript, 1997)), there are striking volcanoto recorda hit. Coolingevents of greaterthan 3 years excursions in the mean annual isotope record coincident duration (a total of 70 years) were first removed from the within the accuracyof the coral recordwith the eruptionsof originaltime seriesbefore determining that there are 132 two- Krakatau (1883), Coseguina(1835), a large unknowneruption year "samples"that are not associatedwith multiyearcooling in 1808 [cf. Dai et al., 1991], and possiblya paired set of events. Of this number,approximately one third (47) actually eruptionsin 1752 (Little Sunda)and 1754 (Taal). In order to have cooling events (Figure 2). Then we determined the investigate this phenomenon more closely, we identified chancethat 16 of the 20 volcanic eruptionswould coincide cooling events of less than 3 years duration and compared with the smaller categoryof 47 cooling events. A way to them to the subset of the volcanism compositethat is not envisionthis exerciseis to imagine a table with 132 pockets associatedwith cooling eventsof greaterthan 3 years duration in it, 47 of which are blue. Then 20 red balls (volcanic (the rationale being that it is less easy to distinguishbetween eruptions)are randomly cast on the table. By chancealone causeand effect for the longercooling events). The magnitude one would expect about one third of the red balls to fall within of the cooling event was determined as the decline from a the blue pockets. However the actual numberis about3 times preexistingvalue representingeither a peak or a "ledge"in the that amount;thus the very low probability of the coincidence 15180 record. occurringby chance. Comparison(Figure 2) of the interannualoxygen isotope Althoughthe previousexercise is instructive,a significant excursionswith the compositevolcanic record indicatesthat complication results from the fact that El Nifio/Southern 16 of the 20 volcaniceruptions coincide with or are followed Oscillation(ENSO) eventsare associatedwith coolingin the within 1 year by an interannual isotope excursion. The westernSouth Pacific. For example,the strong1940-1941 El correlations are most striking after 1800, where there is a Nifio is clearly recorded as a cooling event in the New sequenceof nine "directhits" in a row between 1803 and 1886. Caledonia record (Figures 1 and 2). Likewise, the weaker The noncorrespondenceof the 1917 and 1920 ice core peaks 1976-1977El Nifio is associatedwith a significantisotopic with a cooling in the coral record may reflect the unidentified excursion(the stronger1982-1983 ENSO is more muted and sourcesof the 1917 and 1920 peaks (see footnotesd and e in blendsinto a 6-year "cool"interval). In fact, a recordof ENSO Table 1). Although cooling occurs after the powerful 1963 eventsfor the pastfew centuries[Quinn, 1992] indicatesthat Agung eruption, maximum cooling postdatesthe eruption by 12 of the 16 volcano/coralhits also occur within a year of 2 years and blends with the strong El Nifio event documented cooling events linked with volcanism. The only four by •uinn [1992] and others. This complicationwill be further volcano/coral excursions that do not seem to be associated discussedbelow. However, the results indicate that if the 1917 with ENSOs are the 1746 cooling following the November and 1920 volcanic events are ignored, then all 10 post-1800 1744 eruption of Cotopaxi,an unknown 1788 eruptionthat peaks identified in ice core records from the southern hasalso been found in a reanalysisof northernhemisphere ice hemisphere are associatedwith cooling events in the coral core [Crowley et al., 1993; cf. appendixnote c], the strong record. An eleventhcooling event (1857) could be associated 1808 eruption of unknownorigin [Dai et al., 1991], and a with the ! 855 eruptionof Cotopaxi and has been listed in the weak coolingfollowing the 1822 eruptionof Galungung. CROWLEYET AL.: VOLCANIC SIGNAL IN A CORALRECORD 635

Table1. Conductivity/SulphatePeaks in AntarcticIce Core Records Used to Constructa Composite Index of Volcanism That CouldHave a Hemispheric-ScaleImpact on SouthernHemisphere Climate

Volcano Scaled Ice Core Year Volcano Latitude Longitude Reference Amplitude Reference

1600 Huaynaputina,Peru 16.6øS 70.9øW S-5 1.21 CH,Z, DM,M 1605 Momotombo,Nicaragua 12.4øN 86.5øW S-4 0.45 Z, M 1622 Colima?,Mexico 19.5øN 103.6øW S-4 1.02 CH, DM, M 1641 Kelut,Indonesia 7.9øS 112.3øE S-4 0.98 CH,Z, DM, M Awu, Indonesia 3.7øN 125.5øE S-5 1675a LongIsland, N.G.? 5.4øS 147.1øE S-6 0.91 CH,M 1680 Tongkoko,Indonesia 1.5øN 125.2øE S-5 0.89 CH,M Krakatau,Indonesia 6.0øS 105.4øE LMB 1693 Serua,Indonesia 6.3øS 130.0øE S-4 0.67 Z, M GunungApi (1694) 4.5øS 130.0øE LMB 1711b ?? -- 0.91 M 1720 BravoCerro, Columbia 5.1øN 75.3øW S-4 0.45 M 1744 Cotopaxi,Ecuador 0.7øS 78.4øW S-4 0.45 M 1752 LittleSunda, Indonesia 8.0øS 118.0øE LMB 0.50 M 1754 Taal,Philippines 14.0øN 121.0øE S-4 0.45 M 1768 Cotopaxi,Ecuador 0.7øS 78.4øW S-4 0.45 M 1772 Gunung 7.5øS 108.0øE LMB 0.56 M Papandajan,Indonesia 1788c ?? -- 0.64 CH, M 1803 Cotopaxi,Ecuador 0.7øS 78.4øW LMB 0.50 M 1808 ?? 1.93 LW, Z, DM, D, LG, M 1812 Soufriere,Caribbean 13.3øN 61.2øW S-4 0.50 M, D? Awu, Indonesia 3.7øN 125.5øE S-4 1815 Tambora,Indonesia 8.0øS 118.0øE S-7 2.83 CH,Z, LW,DM, D, LG,M 1817 Roung,Indonesia 8.1øS 114.0øE S-4 0.98 D,M 1822 Galungung,Indonesia 7.2øS 108.0øE S-5 1.28 D, M 1826 Kelut,Indonesia 7.9øS 112.3øE S-4 0.67 M 1831 Babuyan,Philippines 19.5øN 121.9øE S-4 0.68 Z, DM,LG 1835 Coseguina,Nicaragua 13.0øN 87.6øW S-5 0.90 Z, LW,DM, LG, M 1877 Cotopaxi,Ecuador 0.7øS 78.4øW S-4 0.49 DM,M 1883 Krakatau,Indonesia 6.1øS 105.4øE S-6 1.00 CH,Z, LW,DM, LG,M 1886 Tarawera,N.Z. 38.2øS ! 76.5øE S-5 0.66 DM, M 1902 SantaMaria, Guatemala 14.8øN 91.6øW S-6 0.57 Z, LG,M? Soufriere,Caribbean 13.3øN 61.2øW S-4 Pelee,Caribbean 14.8øN 61.2øW S-4 1917d Agrigan,Marianas? 18.8øN 145.7øE S-4 0.54 Z?,M 1920e ?? 40.6øS 72.1øW S-4 0.57 M i 963 Agung,Indonesia 8.3øS 150.5øE S-4 1.23 DM,M 1991f Pinatubo,Philippines 15.1øN 120.4øE S-6 0.78 Mc

Theeruption chronology wasderived from published analyses ofAntarctic records (see below), with peaks defined asincreases above local background [cf.Crowley etai.. 1993] and amplitudes scaled tothe 1883 Krakatau peak in all records except for the Dai et ai. [1991] core, which was scaled toTambora andthen rescaled to Krakatau using the Krakatau/Tambora relationship inother cores. "Year" refers to estimated year of theeruption. "Volcano reference" refersto the suggested volcano(s) responsible forthe ice core peak, as determined bya search inthe volcano chronologies ofLamb [I970] and Sirekin and Siebert[1994]. The number after "S" refers tothe relative magnitude oferuption asestimated bySirnkin and $iebert [1994]. These latter references wereused toeliminate some ice core peaks that appear tocorrelate with high- latitude eruptions. "Ice core reference" refers to the ice core reference used to identify or corroboratepeak, with northern hemisphere sitesused as an aid in assigning a low-latitude origin (CH, Crowley etal. øs [1993] analysis ofHammer etal. øs [1980]results; D,Dai et al. [1991]; DM, Delrnas etal. [1992]; LG, Legrand and Delrnas [1987]; LW, Langway etai. [t995]; M, Moore et ai., [1991]; Mc, McCormicket al. [1995];and Z, Zielinski[1995]). Note that low-latitude northern hemisphere eruptions can influence the southern hemisphere. See appendixand footnotesfor furtherdiscussion. a Theattribution ofthis peak to the Long Island eruption isvery uncertain. The 14C age determination of 1660 + 20years [Sirnkin and Siebert, I994] has alsoresulted in thiseruption being identified with other events in thistime interval [Jones et al., 1995;Zielinski, 1995]. b Thispeak could be a localeruption; it is includedin the list because there is no reason to exclude it. c Eventhough no volcano compilations listan eruption atthis time, occurrence ofa prominent peak in the Crete core at the same time [Crowley eta!., 1993]suggests it could be a low-latitudeeruption. This peak also occurs inan ice core from Mount Logan, Yukon Territories, Canada [Mayewski etal., 1993]. However,further verification of thispeak is neededin othercores. d Theattribution ofthis peak to Agrigan isvery uncertain; it could be an unknown eruption inthe high latitudes ofthe southern hemisphere. Theonly reasonit is listedas possibly having a low-latitude influence isbecause a well-dated I917 peak occurs in theGreenland IceSheet Project 2 record I year earlierthan the localIcelandic (Katla) eruption it hasbeen attributed to [Zielinski,1995]. e Theorigin of thispeak is unknown;it could be noise or some undocumented eruption in the high latitudes of the southern hemisphere. f As discussedin the appendix, the Pinatubo estimate is basedon satellitemeasurements. Although the global Pinatubo aerosol loading was approximately20%greater than Agung [McCormick etai., 1995], the Pinatubo aerosol veil was approximately evenly distributed between the northern and southernhemispheres, while the Agung forcing was primarily in thesouthern hemisphere. Thus the estimated magnitude of the Pinatubo versus Agung eruptionsin this table reflects only the estimated relative magnitude oftheir southern hemisphere aerosol loading. 636 CROWLEYET AL.:VOLCANIC SIGNAL IN A CORALRECORD

Volcanism vs New Caledonia O18 Record In addition to the interannualcooling association,there are

ß also several cases where volcanic events (1675, 1812, and 1902) occurjust prior to pentadal-scalecooling eventsin the coral record. The most prominent cooling in the coral record •- 4.0 from approximately 1813 to 1821 correlates with the most _ intense cluster of eruptionsfound in the compositeice core 3.0 index. In fact, the interval 1803-1835 contains 9 of the 27

_ volcanic peaks identified in the composite ice core index. m 2.0 Lamb [1970] documentsa further 1814 eruption of Mayon

• -3.6 _ (Philippines) that cannot be distinguishedclearly in the ice z core records. This time interval was probably the coolest -3.2 period in the last 200 years[Groveman and Landsberg,1979; I 1.0 Bradley and Jones, 1993]. Because there is also some evidence for a solar-inducedcooling in the early nineteenth 650 1700 1750 1800 ! 850 1900 ! 950 2000 Year century [Lean et al., 1995; Crowley and Kim, 1996], the total cooling in the 1813-1822 interval may reflect a combined Figure 1. Comparisonof meanannual b i80 isotope(per mil) record effect of volcanism and solar irradiance decreases. However, (top) from a Porites coral head near Amedee Island (22ø29'S, 166ø27'E),New Caledonia,with a compositesouthern hemisphere ice the sharp onset of cooling seems most likely due to corerecord (botto m ) of volcanicsulphur; note that low-latitude northern volcanism. hemisphereeruptions can also affect the southernhemisphere (Table The prominent cooling from 1813 to 1821 is part of a 1). longer pattern that occurs in the above hemisphericindices. There is a warm-cold-warm-coldset of pentadal/decadal-scale oscillationsbetween about 1803 and 1837 occurringin these Because of the coincidenceof some of the coral cooling records, with the boundariesbetween the oscillationsbeing eventswith ENSOs, it is thereforedifficult to unambiguously 1813, 1821, and (approximately) 1835. Because of the ascribe all the volcano-cooling coincidencesto cause and similarityof the •5•80 changesin New Caledoniaand northern effect,especially given the uncertaintyin someof the volcano hemisphereclimate recordsover this longer interval, and the identifications. It is also beyond the scopeof this paper to large geographicscale of the coolingto be expectedfrom such address the more complicated relationship from the an extensiveamount of volcanic aerosol(and possiblysolar) probabilisticviewpoint. However,given the well-established forcing, we tentatively suggestthat the warm-cold-warm-cold cooling following significant volcanic eruptions[Bradley, oscillationsmay be usedto refine chronologiesin other coral 1988; Angell, 1990], a volcanic causefor someof the New records. To test this idea we illustrate (Figure 3) two other Caledoniacooling events seems quite plausible. Furthermore, Pacific coral records [Dunbar et al., 1994; Quinn et al., some of the cooling events(1746, 1754, and 1836) are quite 1996a], one from "nearby" Vanuatu (15øS, 167øE) and the large comparedwith the magnitudeof the E1 Nifios estimated other from the eastern equatorial Pacific (Galapagos,0øS, by Quinn [1992]. With respectto the largest interannual 91øW). We alsoidentify tentative time lines that could be-used isotope fluctuationsdiscussed above (1752/1754, 1836, and from our volcanic forcing index, assumingthat the response 1884), the correspondenceis quite compelling for a strong from large-scaleforcing will occurwithin a year. volcano imprint. Evaluation of the smaller peaks would Comparisonof the different coral recordssuggests that all requiremore verification from futurestudies. illustrated corals appear to record the unidentified volcanic One interesting feature of the interannual isotope excursionsinvolves comparison of the b]3C andb180 records from these events. Results (not shown) indicate a general Volcanism vs Coral del O18 Excursions -0.8__._L_L I • J• f•.._.t_[_••, • J,•,• J• Jt• J• • 2.4 tendencyfor •5•3C to becomemore positivewhen •5•80 - _

_ VOLC. - increases(indicating cooler temperature). This responseis -0.6 • oppositeto what would be expectedfrom a cooling large - enoughto causewater column overturn, as a negative•513C -0.4• J _ - -- 0.8 signalwould be expectedfrom sucha response.As •513C -0.2 variations in the symbioticalgae in corals [McConnaughey, - II - 1989; Winter et al., 1991] have also been linked to changing - II IS - photosyntheticactivity due to light availability (cloudiness), 0.2-- -

_ one possibleexplanation for the observed•513C response at oo.o_ Jl'' ' -o.o_ New Caledoniainvolves changes in cloudinessdue to air mass OI8 - changesassociated with cooler temperatures. For example, 0.4i ' • _---o.s 4, - increased outbreaks of high-pressuresystems from higher 0.8 4T-r-r-r-T-r-T-r-r-j-T-r• i • •, 1 i •j • • i • i f i i • i • • - -2.4 latitudes could causeboth cooling and decreasedcloudiness. 1650 1700 1750 1800 1850 1900 1950 2000 Becausethe entire subjectof interpretationof •5•3Cvariations Year in corals is open to uncertaintyand often contentious,this Figure 2. Comparisonof interannualcoral isotopeexcursions (bottom) with the subsetof Table I conductivity/sulphatepeaks (top) that do not problem would have to be investigatedfurther before any coincidewith pentadaldecadal-scale cooling events in the coral record. firmer conclusionscould be made as to the explanation. We Arrows denote E! Nifio-SouthernOscillations (ENSOs) that correlate discussit here for the record only. with volcanicspikes. Seetext for furtherdiscussion. CROWLEY ET AL.: VOLCANIC SIGNAL IN A CORAL RECORD 637

eruption in i808-i809 [Dai et al., 199i; Zielinski, 1995]. Potential Correlation Horizons in Pacific Corals Then there is the longer cooling subsequentto the start of the .....i...... •...... •...... •...... l ...... • ...... •...... •__•._• ...... •..... : ,. •. •_1 •..... ;...... •._•... main pulse of eruptionsin 1812. A prominentcooling in 1836 seems to mark the very large eruption on Coseguina (1835) and the end of the main burst of volcanic activity. Although earlier studies [Palais/and Sigurdsson, 1989; = -50 Crowley et al., 1993] suggestedthat the magnitude of the Coseguinaeruption had been overestimatedby Lamb [1970], • 40" there are clear sulphate/conductivitypeaks in Antarctic ice • -35 core records [Legrand and Delrnas, 1987; Moore et al., 1991; • 40

Delmas et al., 1992] indicatingthat this eruptionresulted in a -30 . significant stratosphericperturbation. Zielinski [1995] and • • • [ ..... OI8'Snl...... 018 UB - -3.5 Langway et al. [1995] have also found evidence for the

Coseguina eruption in Greenland ice cores (Greenland Ice 820 i 840 1860 1880 1900 SheetProject 2 (GISP2) and Dye 3). Year Results of this trial exercise (Figure 3) suggestthat the Figure 3. Comparisonof volcanic "events" in New Caledonia 6180 chronologyof the Vanuaturecord may be 2-3 yearstoo young record(middle) with meanannual 6180 variationsat EspirituSanto, and the Galapagosrecord 2-3 years too old in the sections Vanuatu[Quinn et al., 1996b](top) andUrvina Bay (UB, Galapagos) analyzed. We emphasize again that the results from this (bottom) [Dunbar et al., 1994]. Arrows indicateproposed time stratigraphichorizons related to volcaniceruptions. These involve exercise need to be treated with caution until they can be coolingassociated with the 1808eruption, the beginning (1813) andend further tested with other records. Nevertheless, if this "event (1821) of a decadaloscalecold event associatedwith several volcanic stratigraphic"approach is substantiatedby other studies, it eruptions,and the coolings associated with the 1835Coseguina and 1883 could be of considerable value in further refinement of coral Krakataueruptions. Note that the proposedcorrelation with the 1883 Krakataueruption in the Galapagosrecord is in our opinionmore chronologiesfor testing, for example, model predictionsof uncertain than some of the other events. phase relationshipsof decadal-scaleoscillations in different parts of the Pacific Basin [cf. Latif and Barnett, 1994; Gu and Philander, 1997]. For example, a 1.7% lengtheningin the responseto Southern Oscillation changes. The decreased chronologyof the Vanuatu recordcould changethe computed imprint of ENSO events in this region could therefore decadal oscillation from 14.5 to 14.8 years [cf. Quinn et al., facilitate clearer detection of other sources of interannual 1996b], bringing it closer in line to the 15.4 yr. periodicity variability. An additional factor to consider is that the determinedby T. M. Quinn et al. (submittedmanuscript, 1997) responsefrom volcaniceruptions may differ by latitude. For for the New Caledonia record. example, althoughradiative forcing from the 1991 Pinatubo eruption was greatest in low latitudes [Stenchikov et al., Discussion and Conclusions 1997], the cooling resulting from the eruption persisted longer in higher southern latitudes [Dutton and Christy, To summarize,evidence is quite good for the effect of large 1992]. This cooling extendedapproximately to the latitude of volcanic eruptions on the New Caledonia oxygen isotope New Caledonia. As the latter analysis utilized zonally record. Smallereruptions may also have an impact,although it averageddata, furtherwork would have to be doneto determine is not always easy to separatethe volcanic signal from E1 if thereare preferredsectors of the southernsubtropics that are Nifio-induced cooling. There is also good evidence that the influenced by cold air outbreaks from higher latitudes. abrupt cooling in 1813 may have been triggeredby volcanic Whatever the explanation(s), future studies of volcanic eruptions,but solar irradiancechanges may contributeto the cooling eventsin coral recordsmay be of interestnot only to full magnitudeof the decadal-scalecooling. Finally, someof the paleoceanographicand climate communities but also to the large volcaniceruptions, and the decadal-scalewarm-cold- thoseinterested in the effect of environmentalchange on coral warm-cold oscillationsfrom about 1800-1840, may be useful populations. for correlativepurposes in coral studies. The reason volcanic cooling events in corals have rarely been identified or discussed before is difficult to determine. Appendix For oxygen isotope time series dominated by precipitation The composite record of volcanism from 1600 to the variations [Cole et al., 1993], the apparentlack of a volcanic present[Table 1] was determinedin the followingmanner. cooling contribution to the records may reflect the fact that Prior work [Robockand Free, 1995, 1996] suggeststhat the precipitationresponse to volcanismis difficult to predict. compositingof volcanicsulphur/conductivity spikes in ice Some investigators may simply have overlooked the coresrepresents a promisingapproach to developmentof an relationship; it is difficult to detect unless mean annual objectiveindex of hemisphericvolcanism. This approachwas averaging is used, and not all authorschoose this option. adoptedin the presentstudy. However,the methodologywas Another considerationinvolves the observationthat although somewhatdifferent than that usedby Robockand Free [1995, some of Quinn's [1992] ENSO events occur in the New 1996],who usedthe standarddeviation of short-intervalpeaks Caledonia record, almost two thirds do not. This lack of above local background[cf. Crowleyet al., 1993] and then responsemay reflectthe fact that New Caledoniais just westof averagedthe valuesfor differentsites. In this studywe took the "hinge line" in the SouthernOscillation Index [Trenberth the published estimatesof some peaks from two studies and $hea, 1987] separatingregions of positive and negative [Delrnaset al., 1992;Langway et al., 1995] and then identified 638 CROWLEY ET AL.: VOLCANIC SIGNAL IN A CO• RECORD

additional peaks from three other cores by determining the from 1850 to the present. Resultsindicated general agreement magnitude of increase above local background. However in relative magnitude for the largest peaks but some becauseof chronology uncertainties[Legrand and Delmas, differencesin the smallerpeaks, perhapsbecause we only 1987] and different levels of noisinessin different cores, we averaged sites for which peaks occurred, and not all cores definedpeaks as only thosepeaks larger than a smallpeak that record all the volcanic events. Another difference between our could be relatively unambiguouslytied to a volcaniceruption composite and that of Robock and Free [I995] is that their from publishedvolcano referencesources. For example, the record has more volcanic events than ours. This subset of 1803 Cotopaxi eruption was set as the lower cutoff in the volcanicpeaks in their compositerecord, which do not occur Moore et al. [1991] time series. By contrast,only the largest in our index, is consistentlysmaller in amplitude than our peaks in the Legrand and Delmas [1987] time seriescould be small peaks. This relationshipsuggests that the Robock and identified with any confidence,especially since the timescale Free [1995] criterion for inclusion of volcanic "events" was more difficult to interpretin this record. included a lower cutoff level than ours. The relative merits of Relative magnitudesof the peaks were then scaledto the the two approachesare a subjectfor further discussion. The 1883 Krakataueruption, except for the peaksin the Dai et al. approachof Robock and Free [1995] may include more [1991] core, which only includedvariations from the first half eruptionsbut it also includesa numberof peakswe could not of the nineteenthcentury. Here the peakswere first scaledto relate to known eruptions. Our compositecan be confidently Tamboraand then rescaledto Krakataubased on the average tied to most eruptionssince I800, but it has a higher cutoff Tambora/Krakatau ratios in other southern hemisphereice level that might excludesome legitimate volcanic peaks. We cores. Compositing involved averaging values from only neverthelessbelieve our approachis a viable alternative.One those cores that recorded a volcanic eruption. The rationale could also arguethat the smallerpeaks included in the Robock here is that wind scouringcan remove the sulphaterecord from and Free [1995] record, even if they all reflected volcanic snow but can less easily magnify them, so only cores that eruptions,are likely to have a proportionatelysmaller climate containedpeaks shouldbe considered. Low latitudenorthern impact. Since the latter is the objectiveof the presentstudy, hemisphere eruptions are included because they can affect we believe our composite is of some merit, all the while southern hemisphere climate. Finally, data from the 1991 recognizingit is subjectto further modification. Pinatubo eruption were added for completeness,based on Acknowledgments.This research was supported in partby the relating estimated global stratospheric loading from National Oceanic and Atmospheric Administration (NOAA McCormick et al. [ 1995] to the Agung peak, the latter which NA36GP0546). We collectedsamples with supportfrom the Aquarium appearsprominently in southernhemisphere ice core records de Noumea and ORSTOM New Caledonia and from NSF grant (see also footnotef of Table 1). ATM8922114 to F. Taylor. We alsothank the followingfor discussion, To test the validity of our method,we first comparedthe comments,and assistance:R. Bradley,R. Free,H. Graf, W. Hyde, A. resultsof our compositewith thoseof Robockand Free [ 1995] Robock,G. Zielinski,and an anonymousreviewer.

References

Angel!, J. K., Variations in global tropospheric climate variability in Galfipagos corals, volcanismand its climatic impact, Nature, 288, temperature after adjustment for the El Nifio Paleoceanography, 9, 29 I-315, 1994. 230-255, ! 980. influence, 1958-1989, Geophys.Res. Lett., 17, Dutton, E.G., and J. R. Christy, Solar radiative Jones,P. D., K. R. Briffa, and F. H. Schweingruber, ! 093-1096, 1990. forcing at selected locations and evidence for Tree-ring evidenceof the widespreadeffects of Bradley, R. S., The explosive volcanic eruption global lower troposphericcooling following the explosive volcanic eruptions, Geophys. Res. signal in northern hemisphere continental eruption of E! Chichon and Pinatubo, Geophys. Lett., 22, 1333-1336, 1995. temperaturerecords, Clint. Change, I2, 221-243, Res. Lett., I9, 2313-2316, 1992. LaMarche,V. C., and K. K. Hirschboeck,Frost rings 1988. Gagan, M. K., and A. R. Chivas, Oxygen isotopes in trees as recordsof major volcaniceruptions, Bradley, R. S., and P. D. Jones, 'Little Ice Age' in western Australian coral reveal Pinatubo Nature, 307, 121-126, 1984. summertemperature variations: Their natureand aerosol-inducedcooling in the Western Pacific Lamb, H. H., Volcanicdust in the atmosphere:With relevance to recent global warming trends, Warm Pool, Geophys.Res. Lett., 22, 1069-1072, a chronology and assessment of its Holocene, 3, 367-376, 1993. 1995. meteorological significance, Philos. Trans. R. Cole, J. E., R. G. Fairbanks, and G. T. Shen, Recent Genin,A., B. Lazar,and S. Brenner,Vertical mixing Soc. London, A266, 425-533, 1970. variability in the SouthernOscillation: Isotopic and coral death in the Red Sea following the Langway, C. C., Jr., K. Osada, H. B. C!ausen,C. U. results from a Tarawa Atoll coral, Science, 260, eruption of Mount Pinatubo, Nature, 377, 507- ! 790- ! 793, ! 993. Hammer,and H. Shoji, A 10-centurycomparison 510, 1995. of prominentbipolar volcanicevents in ice cores, CrowIcy, T. J., and K.-Y. Kim, Comparison of Graf, H.-F., I. Kirchner, A. Robock, and I. Schult, d. Geophys.Res., I00, 16,241- ! 6,247, ! 995. proxy records of climate change and solar Pinatuboeruption winter climate effects: Model forcing, Geophys.Res. Lett., 23, 359-362, 1996. Latif, M., and T. P. Barnett, Causes of decadal versus observations, Clint. Dyn., 9, 81-93, climate variability over the North Pacific and Crowicy, T. J., T. A. Criste, and N. R. Smith, 1993. Reassessment of Crete (Greenland) ice core North America, Science, 266, 634-637, 1994. acidity/volcanism !ink to climate change, Graham, N. E., Simulation of recent global Lean, J., J. Beer, and R. Bradley,Reconstruction of Geophys.Res. Lett., 20, 209-212, 1993. temperature trends, Science, 267, 666-671, solar irradiance since 1610: Implications for 1995. Dai, J., E. Mosley-Thompson,and L. G. Thompson, climatechange, Geophys. Res. Lett., 22, 3195- Ice core evidence for an explosive tropical Groveman, B. S., and H. E. Landsberg,Simulated 3198, 1995. volcaniceruption 6 yearspreceding Tambora, d. northern hemisphere temperature departures Legrand, M., and R. J. Delmas, A 220-year Geophys.Res., 96, ! 736 i- ! 7366, 1991. 1579-1880, Geophys. Res. Lett., 6, 767-769, continuousrecord of volcanic H2SO4 in the Delmas, R. J., S. Kirchner, J. M. Palais, and J.-R. 1979. Antarcticice sheet,Nature, 327, 671-676, 1987. Petit, 1000 years of explosive volcanism Gu, D., and S. G. H. Philander, Interdecadal climate Mayewski, P. A., eta!., Ice core sulphatefrom three recorded at the South Pole, Tellus, 44B, 335- fluctuationsthat depend on exchangesbetween northern hemisphere sites: Source and 350, ! 992. the tropics and extratropics,Science, 275, 805- temperature forcing implications, ,4trnos. Dunbar, R. B., G. W. Wellington, M. W. Colgan, 807, ! 997. Environ., 27A, 2915-2919, i 993. and P. W. G!ynn, Eastern Pacific sea surface Hammer, C. U., H. B. C!ausen,and W. Dansgaard, McConnaughey,T. A., 13C and 180 isotopic temperaturesince 1600 A.D.: The•5180 record of Greenland ice sheet evidence of post-glacial disequilibrium in biological carbonates, l, CROWLEY ET AL.: VOLCANIC SIGNAL IN A CORAL RECORD 639

Patterns, Geochim. Cosmochim. Acta, 53, 151- Historical and Paleoclimatic Aspects o.! the Winter, A., C. Goenaga, and G. A. Maul, Carbon 162, 1989. Southern Oscillation, edited by H. F. Diaz, and and oxygen isotopetime seriesfrom an 18-year McCormick, M.P., L. W. Thomason, and C. R. V. Markgraf, pp. 119-149, Cambridge Univ. Caribbean reef coral, d. Geophys. Res., Trepte, Atmosphericeffects of the Mt. Pinatubo Press, New York, 1992. 96,16,673-16,678, 1991. eruption, Nature, 373, 399-404, 1995. Robock, A., The "Little Ice Age": Northern Woodruff, S. D., R. J. Slutz, R. L. Jenne, and P.M. Moore, J. C., H. Narita, and N. Maeno, A hemisphere average observations and model Steurer,A comprehensiveocean-atmosphere data continuous 770-year record of volcanic activity calculations, Science, 206, 1402-1404, 1979. set, Bull. Am. Meteorol. Soc., 68, 1239-1250, from East Antarctica, d. Geophys. Res., 96, Robock, A., and M.P. Free, Ice cores as an index of 1987. 17,353-17,359, 1991. global volcanism from 1850 to the present, d. Zielinski, G. A., Stratosphericloading and optical Palais, J. M., and H. Sigurdsson, Petrologic Geophys. Res., 100, I 1,549-11,567, 1995. depth estimatesof explosive volcanismover the evidenceof volatile emissionsfrom major historic Robock, A., and M. P. Free, The volcanic record in last 2100 years derived from the Greenland Ice and prehistoric volcanic eruptions, in ice cores for the past 2000 years, in Climatic Sheet Project2 ice core, d. Geopl[vs.Res., 100, Understanding Climate Change, Geophys. Variations and Forcing Mechanisms oj' the Last 20,937-20,955, 1995. Monogr. Ser., vol. 52, edited by A. Berger, R. 2000 Years, edited by P. D. Jones, R. S. E. Dickinson, and J. W. Kidson, pp. 31-53, Bradley, and J. Jouzel, pp. 533-546, Springer- AGU, Washington, D.C., 1989. Verlag, New York, 1996. T. J. Crowley, Department of Oceanography, Porter, S.C., Pattern and forcing of northern Robock, A., and J. Mao, The volcanic signal in Texas A&M University, College Station, TX 77843- hemisphere glacier variations during the last surface temperature observations,d. C!im., 8. 3146. (e-mail: tcrowley(•ocean.tamu.edu) millennium, Quat. Res., 26, 27-48, 1986. 1086-1103, 1995. C. Henin, lnstitut Francais de Recherche Quinn, T. M., F. W. Taylor, and T. J. CrowIcy, A Simkin, T., and L. Siebert, Volcanoesof the World, Scientifique pour le Develloppement en 173 year stable isotope record from a tropical 2nd ed., 349 pp., Geoscience, Tucson, Ariz., Cooperation, M. de New Caledonia, B.P. A-5, South Pacific coral, Quat. Sci. Rev., 12, 407- 1994. Nourmea, New Caledonia. (e-mail: 418, 1993. Stenchikov, G. L., I. Kirchner, A. Robock, H.-F. henin•noumea.orstom.nc) Quinn, T. M., F. W Taylor, T. J. CrowIcy, and S. M. Graf, J. C. Antuqa, R. Grainger,A. Lambert,and P. Joannot, Aquarium de Noumea, B.P. 395, Link, Evaluationof samplingresolution in coral L. Thomason,Radiative Forcing .[i'om the 1991 Noumea, New Caledonia. stable isotope records: A case study using Mount Pinatubo Volcanic Eruption, Rep. 231, T. M. Quinn, Departmentof Geology, University monthly stable isotope records from New 40 pp., Max-Planck Inst. far Meteorol., of South Florida, Tampa, FL 33620. (e-mail: Caledonia and Tarawa, Paleoceanography, 11, Hamburg, Germany, 1997. quinn(•chuma.cas.usf.edu) 529-542, 1996a. Trenberth, K. E., and D. J. Shea, On the evolution F. W. Taylor, Institute for Geophysics, Quinn, T. M., T. J. CrowIcy, and F. W. Taylor, New of the southern oscillation, Mon. Weather Rev., University of Texas, 4412 Spicewood Springs stableisotope results from a 173-yearcoral record 115, 3078-3096, 1987. Road, Bldg. 600, Austin, TX 78759. (e-mail: from Espiritu Santo, Vanuatu, Geophys. Res. Wigley, T. M. L., Climate variability on the 10-100 fred@ ut i g.ig. u texa s.edu) Lett., 23, 2413-2416, 1996b. year time scale: Observations and possible Quinn, w. H., A study of Southern Oscillation- causes, in Global Changes of the Past, edited (Received May 15, 1996; related climatic activity for A.D. 622-1990 by R. S. Bradley, pp. 83-101, Univ. Corp. for revised May 1, 1997; incorporatingNile River flood data, in El Nifo: Atmos. Res., Boulder, Colo., 1991. acceptedMay 8, 1997.)