Origin of Bermuda's Clay-Rich Quaternary Paleosols and Their Paleoclimatic Significance

University of Nebraska - Lincoln DigitalCommons@University of Nebraska - Lincoln

USGS Staff -- Published Research US Geological Survey

1996

Origin of Bermuda's clay-rich Quaternary paleosols and their paleoclimatic significance

Stanley R. Herwitz Clark University

Daniel R. Muhs U.S. Geological Survey, [email protected]

Joseph M. Prospero University of Miami

Shannon Mahan U.S. Geological Survey

Bruce Vaughn U.S. Geological Survey

Follow this and additional works at: https://digitalcommons.unl.edu/usgsstaffpub

Part of the Earth Sciences Commons

Herwitz, Stanley R.; Muhs, Daniel R.; Prospero, Joseph M.; Mahan, Shannon; and Vaughn, Bruce, "Origin of Bermuda's clay-rich Quaternary paleosols and their paleoclimatic significance" (1996). USGS Staff -- Published Research. 165. https://digitalcommons.unl.edu/usgsstaffpub/165

This Article is brought to you for free and open access by the US Geological Survey at DigitalCommons@University of Nebraska - Lincoln. It has been accepted for inclusion in USGS Staff -- Published Research by an authorized administrator of DigitalCommons@University of Nebraska - Lincoln. JOURNAL OF GEOPHYSICAL RESEARCH, VOL. 101, NO. D18, PAGES 23,389-23,400, OCTOBER 27, 1996

Origin of Bermuda's clay-rich Quaternary paleosols and their paleoclimatic significance StanleyR. Herwitz,1 Daniel R. Muhs,2 JosephM. Prospero,3 ShannonMahan, 2 and Bruce Vaughn2

Abstract. Red clayeypaleosols that are chieflythe productof aerosolicdust deposition are interbeddedin the Quaternarycarbonate formations of the Bermuda oceanicisland system.These paleosolsprovide a basisfor reconstructingQuaternary atmospheric circulationpatterns in the northwesternAtlantic. Geochemicalanalyses were performed on representativepaleosol samples to identify their parent dust source.Fine-grained fractionswere analyzedby energy-dispersiveX ray fluorescenceto determine trace element (Zr, Y, La, Ti, and Nb) concentrationsand to derive geochemicalsignatures basedon immobile element ratios. These ratios were comparedwith geochemical signaturesdetermined for three possiblesources of airborne dust: (1) Great Plains loess, (2) MississippiRiver Valley loess,and (3) Saharandust. The Zr/Y and Zr/La ratios providedthe clearestdistinction between the hypothesizeddust sources.The low ratios in the paleosolB horizonsmost closely resemble Saharan dust in the <2-/•m size class fraction. Contributions from the two North American loessial source areas could not be clearlydetected. Thus Bermudapaleosols have a predominantlySaharan aerosolic dust signature.Saharan dust depositionon Bermuda during successiveQuaternary glacial periodsis consistentwith patternsof generalcirculation models, which indicatethat during glacial maxima the northeastsummer trade winds were strongerthan at present and reachedlatitudes higher than 30øN despitelower-than-present sea surface temperaturesin the North Atlantic.

Introduction derived from biogenicbeach sedimentsthat constitute>90% of the total rock volume above sea level [Land et al., 1967; The usefulnessof reconstructingatmospheric circulation Vacher, 1973; Harmon et al., 1981; Vacherand Hearty, 1989]. during the glacial-interglacialcycles of the last 2 m.y. has be- These eolianites overlie a submergedvolcanic seamount. In- come increasinglyimportant due to growing concernsabout terbedded between the eolianites and some minor marine car- global climate change. General circulation modeling of the bonatesare clay-rich"terra rossa"paleosols [Ruhe et al., 1961; northernhemisphere suggests that the northeastsummer trade Rowe, 1990;Hearty and Vacher,1994; Herwitz and Muhs, 1995]. winds between 24ø and 36øN latitude were stronger during Paleosolsinterbedded between datable rocksprovide valuable Quaternary glacial periods than at present [Kutzbach,1987]. opportunitiesfor reconstructingpast conditions over definable Palynologicalanalyses of northwestAfrican marine sediments time periods. are consistentwith this interpretation [Hooghiemstraet al., The Bermuda paleosolsare reddish to reddish brown in 1987; Hooghiemstra,1988]. Kefferet al. [1988], on the other color and consistof high concentrationsof aluminosilicate hand, concluded that the reduction in North Atlantic sea sur- clays [Ruhe et al., 1961; Vacherand Hearty, 1989]. Reddish face temperaturesduring glacial maxima causeda significant brown, clayey soils that formed by different processeshave southward summer shift in the Gulf Stream-North Atlantic been studiedin many parts of the world [Pye, 1983]. In situ Current systemand the subtropical-subpolarfrontal system. weatheringof calcareousrock formationshas been suggested The geologicalrecord of Bermuda, located at 32øN latitude in as a possiblepedogenic mechanism in other areas [Prasad, the northwestAtlantic, providesa unique opportunityfor ad- 1983;Moresi and Mongelli, 1988;Pye, 1992]. Bermuda'shigh- dressingthese contrastingviews of Quaternary paleocircula- purity eolianitescontain insutficient Si- and Al-bearingminer- tion. als to generate reddishbrown, clay-richsoil over periods of The oceanic island systemof Bermuda currently exposed <10s years[Herwitz and Muhs, 1995].The mineralogyand abovesea level is a lithified dune field composedof calcareous geochemistryof Bermuda's paleosolsindicate that the paleosols are not the insoluble residue of weathered eolianite. eolianites.The eolianitesare Quaternaryin age and are locally The generallyaccepted view is that most of the red clay frac- •Departmentsof Geographyand Biology, Clark University, Worces- tion of Bermudanpaleosols represents accumulations of fine- ter, Massachusetts. grained airborne dust originating from distant continental 2U.S.Geological Survey, Denver, Colorado. sourceareas [Bricker and Prospero,1969; Bricker and Macken- 3RosenstielSchool of Marineand Atmospheric Sciences, University zie, 1970; Herwitz and Muhs, 1995]. of Miami, Miami, Florida. Determination of the sourceof Bermuda'spaleosols would Copyright 1996 by the American GeophysicalUnion. providea basisfor testingmodels of Quaternaryatmospheric Paper number 96JD02333. circulationpatterns in the North Atlantic. Given the long res- 0148-0227/96/96JD-02333509.00 idencetime of atmosphericdust and the great distancesit can

23,389 23,390 HERWITZ ET AL.: BERMUDA PALEOSOLS AND PALEOCLIMATE travel [Rahn et al., 1979], any stable subaerialsurface is a Materials and Methods potential site of deposition. Long-term accumulationsat a Paleosol Collection given site may therefore representa mix of dust from widely Red claysoils are distributedthroughout Bermuda. Plate la separated source areas. The objective of our study was to showsa subaeriallyexposed red clay soil surfacerepresenting identify the predominantsource area(s) that contributedto these accumulations on Bermuda. Immobile trace elements their presumedcondition before burial by calcareouseolian deposits.We studiedred claypaleosols overlain by carbonates that do not experiencedifferential chemical weathering, re- (Plate lb) at sitesdelineated on Vacheret al.'s [1989]geological lease,and removal have been used as a basisfor geochemical map of Bermuda.Soil samples(100-300 g) were collectedat fingerprintingbecause they maintainthe samerelative propor- siteswhere the differenteolianite formations were represented tionsover time [Muhset al., 1990].We identifiedthe geochemi- (Figure 1). Most of the samplesites were exposedat road or cal signaturesof Bermudanpaleosols and related them to the railway cuts or in the walls of abandonedquarries (Plate 2). signaturesof aerosolic dust from likely continental sources The numberof samplescollected at eachsite depended on soil usingimmobile element ratios. These data provideda basisfor thicknessand the numberof horizons;commonly, two to seven evaluatingmodels of Quaternary atmosphericcirculation pat- sampleswere collectedat each site. terns in the northwestern Atlantic. All sampleswere from paleosolprofiles with A and B horizons (Plate 3a), and, in somecases, distinguishable subhori- Quaternary History of Bermuda zons(e.g., Btl; Bt2). Paleosolsthat did not appearto have B The calcareous dunes that lithified to eolianite were first horizonswere not selected(Plate 3b) becauseof their high thoughtto have formed duringglacial low seastands when the carbonatecontent (> 50%) and their relativelylow proportions of red clay. Trace element concentrationsin soil A horizons Bermuda platform was more exposed,whereas the palcosols formed during intcrglacials [Sayles,1931; Livingston,1944; were so low that they could not be used for a meaningful definitionof a geochemicalsignature. Signatures described in O•cer et al., 1952].This interpretationwas challenged by Bretz this paper are mainly limited to samplescollected from B [1960],who arguedthat the dunesdeveloped during intergla- horizons.In preliminary studiesof Bermuda paleosols,our cial high seastands and the clay-richsoils formed duringglacial analyseswere limited to samplescollected from the interiorsof maxima. Submerged eolianite units, however, have been re- soil-filled solution pipes [Herwitz, 1993; Herwitz and Muhs, ported on the Bermudaplatform at depthsranging from 18 to 1995].Paleosol samples collected from the interiorsof solution 75 m below present-daysea level [Pirsson,1914; Stanleyand pipesat WhaleboneBay (Figure 1) were includedin this study Swift, 1967]. These submergedeolianites and other strati- becauseof their closeproximity to Whalebone Bay's Upper graphic evidenceindicate that the large lithified dune ridges Town Hill marine facies(Qtm), which contains the most con- that constitute the bulk of the island systemdid not form centratedoccurrence of volcanicdetritus on Bermuda.Of par- duringhigh or low seastands but rather formedrapidly during ticular interestwas how these volcanicconstituents may have the first phasesof sea level regressionsat the end of intergla- affectedthe geochemicalsignatures of Bermuda'spaleosols. cials [Vacherand Hearty, 1989]. The successionof onlapping and locally overlapping eolianites reflects progradation of GeochemicalAnalyses of Paleosols and Eolian Sources younger dune units over older units [Vacher,1973; Vacheret The fine-grainedfractions of 38 paleosolsamples from the al., 1995]. 15 localitiesshown in Figure 1 (includingWhalebone Bay) Six eolianiteformations ranging in age from >880 to 85 ka were analyzedby energy-dispersiveX ray fluorescence[John- have been distinguishedon the basisof stratigraphicposition, sonand King, 1987].Each samplewas passed through a 2-mm relative degree of lithification,uranium seriesdating of spe- sieve,crushed to <200 mesh,and homogenizedprior to anal- leothemsand fossilcoral samples,and amino acid racemiza- ysis.The powderedsamples were analyzedfor concentrations tion dating of eolianite, fossil land snail, and fossil marine of the immobile trace elements Zr, Y, La, Ti, and Nb. Mean mollusksamples [Harmon et al., 1983;Vacher and Hearty, 1989; immobile element ratios were computedfor the five paleosol Vacheret al., 1989;Hearty et al., 1992].From oldestto youngest, units positionedbetween the eolianiteformations Qtu/Qtl and theseeolianites are (1) the WalsinghamFormation (Qw, >880 Qw, Qb and Qtu, Qr and Qtu, Qr and Qb, and Qs and Qr, where ka), (2) the Lower Town Hill Formation(Qt•, 430-475 ka), (3) Qw is Walsingham,Qtl is Lower Town Hill, Qm is Upper Town the Upper Town Hill Formation (Qtu, 325-350 ka), (4) the Hill, Qb is Belmont, Qr is Rocky Bay, and Qs is Southampton. BelmontFormation (Qb, 200-225 ka), (5) the RockyBay For- The geochemicalsignatures defined by the elemental ratios mation (Qr, -125 ka), and (6) the SouthamptonFormation were comparedwith the signaturesof airbornedust from likely (Qs, -85 ka) [Vacheret al., 1995].The agesof the Southamp- sourceareas. This approachis similar to that usedby Muhs et ton, Rocky Bay, and Belmont Formations are based on ura- al. [1990] in their studyof soilson Quaternary carbonatesof nium seriesdating of speleothemsand corals from the beachor the Caribbean islands, which revealed that Saharan dust was marine facies[Harmon et al., 1983].Age estimatesof the older the dominant parent material. Bricker and Prospero[1969] units are based on amino acid ratios of whole-rock eolianite were the first to suggestthat Bermuda'spaleosols represent samplesand amino acid ratios of land snailsfrom the eolian- accumulationsof aerosolic dust originating from northwest ites, calibratedby the uranium-series-datedunits [Heartyet al., Africa. Other possibleaerosolic dust sources are North Amer- 1992].Given theseages of Bermuda'seolianite formations, the ican loessdeposits in the Great Plains[Holliday, 1991] and the interbedded paleosolswere subaeriallyexposed and formed lower MississippiRiver Valley [Fisk, 1951;Pye and Johnson, over time periods ranging from -40,000 to >400,000 years. 1988]. These North American sourcesshould have geochemi- Thosesoils that formedover the longertime periodsrepresent cal signaturesdistinct from those of northwestAfrica because compositeunits. of differencesin parent material mineralogyand in trace ele- HERWITZ ET AL.: BERMUDA PALEOSOLS AND PALEOCLIMATE 23,391

Qtu

0 lm 2

Plate 1. Clay-rich red soils of Bermuda: (a) subaeriallyexposed (S) and underlain by the Walsingham Formation(Qw) adjacentto a sinkholein the Idwal HughesNature Reserve,and (b) underlainby the Upper Town Hill Formation (Qtu) and overlain by the Rocky Bay Formation (Qr) exposedin fresh cut at a constructionsite on Gibbon Road near North Shore Road southwestof Bailey'sBay.

ment compositionsin the same minerals [Herwitzand Muhs, collectedfrom 8 samplesites in Mississippi,11 in Louisiana,7 19951. in Arkansas, and 1 in Tennessee.Great Plains loess samples Becausethe fine-grainedfraction of North American loesses were from 9 sites in eastern Colorado and from 13 sites in the has the highestprobability of reachingthe westernAtlantic area of eastern New Mexico and western Texas. Ocean during dust storms [Windom and Chamberlain,1978; The geochemicalsignatures of Saharan dust were based on McCauleyet al., 1981],we isolatedthe fine-grained(<10 bulk filter samplesof Saharandust collectedfrom Sal Island fractionsfrom whole-sedimentsamples collected from repre- (the northeasternmostisland in the Cape Verde archipelago), sentativelocalities. Mississippi River Valley loesssamples were Barbados,and Miami, Florida (Figure 1) by J. Prosperoand 23,392 HERWITZ ET AL.: BERMUDA PALEOSOLS AND PALEOCLIMATE

60 ø

W B "".-..,..

•:;:.',.. • •

CASTLEHARBOUR 90 ø 60 ø 30 ø 0 o O' ,q -32 ø 20' N Bermuda

0 2 4 km I I I I I

64 ø 45'W 64 ø 40'W I I ¾igure 1. Locationof Bermuda(BE), Barbados(BR), and Miami (M) in thewestern Atlantic and Sal Island (SI) in the easternAtlantic (inset) and of Bermuda'sWhalebone Bay (WB) and paleosolsample sites 1-14. Sampleswere taken between the following eolianite formations: 1, Qr (RockyBay) and Qtu (Upper Town Hill) alongRailway Trail, St. George'sIsland; 2, Qr and Qtu on PendleHill, St. George'sIsland; 3, Qtu and Qw (Walsingham)at ShoreHills siteon Essofuel storageproperty, St. George'sIsland; 4, Q, (Southampton)and Qr at intersectionof SwingBridge and Ferry ReachRoads, St. George'sIsland; 5, Qtl (LowerTown Hill) and Qw at entranceto CastleHarbour Marriott Hotel; 6, Qr and Qb (Belmont)on RockyBay Hill; 7, Qb and Qtu alongRailway Trail at entranceto Grape Bay Road adjacentto SouthRoad; 8, Qb and Qtu on Ord Road adjacentto CobbHill Road intersection;9, Q• and Qb on SpiceHill Road; 10, Q• and Qtu adjacentto Mount Zion Church;11, Q• and Qtu on Whale Bay Road in Southampton;12, Qs and Q• on SomersetIsland; 13, Qtl and Qw on Ireland Island North; and 14, Qr and Qb on Coney Island. analyzedusing emissionspectroscopy. To evaluatepossible paleosols(Table 1) are lower than the ratios of all three hy- differencesin the geochemicalsignatures as a function of par- pothesizedsource areas. Bermuda paleosols interbedded be- ticle size, nylon mesh samplesof Saharan dust collectedby tween the relativelyyoung Southampton and Rocky Bay For- Prosperoon Barbadoswere fractionatedinto four particlesize mationshave Zr/Y ratiosthat are suggestiveof a Saharandust classes(10-20/zm, 5-10/zm, 2-5 /zm, and <2/zm) and ana- source(Figure 2). One outlier is the paleosolbetween the lyzed using instrumentalneutron activation.The procedures RockyBay andBelmont Formations at site 14 on ConeyIsland used for protectingthe samplesfrom local contaminationare (Figure I and Plate 2b), whichhas Zr/Y ratiosconsistent with describedby Delany et al. [1967] and Prospero[1968]. a Mississippiloess source; however, closer examination of the Immobile element ratios from the different sample sites site 14 samplesusing light microscopyrevealed that the soil were determined, and the range and weighted mean values profile at this site may be more accuratelycharacterized as a were computed for each of the three possiblesources: (1) protosol. MississippiRiver Valley loess(hereinafter referred to as Mis- Of the three hypothesizedsources, the Zr/Y ratios in Sa- sissippiloess), (2) Great Plainsloess, and (3) Saharandust. haran dust (Table 2) are closestto thoseof Bermuda'spaleosols.Nevertheless, the mean Zr/Y values for all five pa-

Results leosol groupsshown in Figure 2 are lower than those for Saharandust, suggesting either (1) a differentparent material Eolian Source and Paleosol Geochemical Signatures or (2) lowerZr/Y valuesin Saharandust reaching Bermuda as Of all the ratios examined,the Zr/Y ratio providesone of comparedwith Saharandust reachingSal Island, Barbados, the clearest distinctionsbetween the three hypothesizeddust and Miami. We favor the latter hypothesisbecause the atmo- sources,with Great Plains loess having the highest values sphericcirculation path from northwestAfrica to Bermuda (mean _+ I s.d. = 9.44 +_ 1.42) and Saharandust havingthe may be severalhundred kilometers greater and the aerosolic lowest(4.26 _+0.80). Mississippiloess has intermediate values dust may not be maintaininga stablesize distribution.Zirco- (7.04 + 1.26) that overlapslightly with thoseof Great Plains nium, which is found almostexclusively in the heavymineral loess (Figure 2). The ZffY ratios of most of the Bermuda zircon,is likely to be depletedmore thanY. Zircon is knownto HERWITZ ET AL.: BERMUDA PALEOSOLS AND PALEOCLIMATE 23,393

o ß . 23,394 HERWITZ ET AL.: BERMUDA PALEOSOLS AND PALEOCLIMATE

Plate 3. Close-upviews of Bermudapaleosol profiles illustrating (a) a well-developedpaleosol with A and 'B horizonsat site 10 (Plate 2d) and (b) a protosolalong Paynter'sRoad approximately0.5 km southwestof site 5.

concentratepreferentially in the coarsersilt fraction of parti- 10 clesderived from rock weathering[Chapman and Horn, 1968]. Size fractionation thus may be greater for Zr than for Y. On the basis of the data set forth in Table 2, the mean Zr/Y ratio of 4.5 at Sal Island decreases to a value of 4.1 at Barbados. Although this decreaseis not pronounced,it raisesthe possi- ]MississippiVall'e•)l bility of further size fractionation as the dust continueson a •. longer travel route to Bermuda. The Zr/La ratios differ from the Zr/Y ratios in that the Mississippiand Great Plains loessesare not as readily distin- [SaharaDesert guishedfrom each other. However, the Saharandust signature remainsdistinct, with a lower mean value (_+ 1 s.d.) of 2.63 _+ 0.50. The Zr/La ratios of Bermudapaleosols are most concen- t2

trated in this range (Figure 3). Most of the paleosol Zr/La 2 - ratios are statisticallyindistinguishable from Saharandust. Ex- ceptionsare the youngerpaleosols between the Rocky Bay and SouthamptonFormations, which have valuesmore consistent o I I I I I with those of the Mississippi1oess signature, and the anoma- lOO 2OO 4OO 6OO 8OO lous site 14 outlier that has a mean Zr/La value exceeding Age (ka) thoseof all three sourceareas. Although there is someoverlap with Mississippiloess, most of the paleosolZr/La ratios are Figure 2. Mean Zr/Y ratios (_+ 1 s.d.) from paleosolsample stronglysuggestive of a Saharandust source. sites as a function of age of underlying eolianite shown in relation to the range of values (mean + 1 s.d.) of the three The complicatingissue is why the Zr/La ratios are more possiblesources areas: Great Plains, MississippiValley, and consistentwith Saharandust than the Zr/Y ratios. One possi- Sahara Desert. The numbers next to each mean value show the ble explanationis variability in the immobile element ratios as number of paleosol sample sites represented.The boxed-in a function of particle size. How representativeare analyses outlier representsthe Zr/Y ratio from site 14. HERWITZ ET AL.' BERMUDA PALEOSOLS AND PALEOCLIMATE 23,395 involvingthe entire <10-/xm particle size fraction for charac- Table 2. Trace Element Concentrations and Ratios for terizingthe aerosolsthat reachedBermuda? This questionwas Bulk Filter Samplesof Saharan Dust Determined by addressedby the neutron activationdeterminations of Zr/La EmissionSpectroscopy valuesfor Saharandust samplessubdivided into the four par- Concentration,ppm Ratio ticle sizeclass groupings of 10-20/xm, 5-10/xm, 2-5/xm, and Sample <2/xm (Table 3). The result is a declinein the Zr/La ratio as Number Zr Y La Zr/Y Zr/La a function of decreasingparticle size (Figure 4). The mean Sal Island Zr/La ratio is 6.35 for the 5- to 10-/xm size classand only 2.87 1 185 34 51 5.4 3.6 for the <2-/xm size class.This decreaseis attributable to the 2 126 32 56 3.9 2.3 marked decrease in the Zr concentrations in Saharan dust with 3 151 32 51 4.7 3.0 decreasingparticle size. In contrast, the La concentrations 4 100 37 51 2.7 2.0 increasedonly slightlywith decreasingparticle size. The low 5 141 32 58 4.4 2.4 6 117 28 48 4.2 2.4 Zr/Y ratiosof the Bermudapaleosols thus may be attributedto 7 140 35 51 4.0 2.7 their clori,ratlnn œrnrn • finer fraction •,e q•h•an A..o, ,h ...... 8 226 35 54 6.5 4.2 previouslythought. 9 139 31 58 4.5 2.4 The best fit equation shown in Figure 4 can be used to 10 153 34 54 4.5 2.8 estimatethe mean particle sizefraction of Saharandust reach- Barbados ing Bermuda. The mean Zr/La ratio of 3.03 for all of the 11 158 41 64 3.9 2.5 sampledB horizonBermuda paleosols (excluding the anoma- 12 150 41 63 3.7 2.4 13 192 43 70 4.5 2.7 loussite 14) suggestsa meandust particle size of 1.9/xm,which 14 139 40 67 3.5 2.1 is slightlyfiner grained than the mean dust particle size reach- 15 191 43 69 4.4 2.8 16 203 49 70 4.1 2.9 17 125 32 57 3.9 2.2 18 137 34 58 4.0 2.4 Table 1. Trace Element Concentrations and Ratios of 19 128 41 41 3.1 3.1 Bermuda PaleosolsDetermined by Energy-DispersiveX Ray 20 234 41 67 5.7 3.5 21 111 32 64 3.5 1.7 Fluorescence 22 232 43 84 5.4 2.8

Sample Concentration,ppm Ratio Miami Stratigraphic 23 190 35 48 5.4 4.0 Unit* Site No. Zr Y La Zr/Y Zr/La 24 125 32 48 3.9 2.6 25 135 32 58 4.2 2.3 12 sm-2 96.3 26.0 28.8 3.7 3.3 26 95 23 51 4.1 1.9 4 ff-2 188.8 64.1 48.4 2.9 3.9 27 114 32 51 3.6 2.2 6 rb-2 179.0 79.0 72.9 2.3 2.5 28 147 34 51 4.3 2.9 9 sh-1 61.6 94.6 20.6 0.7 3.0 29 133 34 69 3.9 1.9 sh-2 112.4 45.1 45.6 2.5 2.4 30 119 32 56 3.7 2.1 14 cn-1 116.3 12.3 10.4 9.5 11.2 cn-2 123.1 27.2 25.8 4.5 4.8 1 rr-2 131.1 26.9 25.0 4.9 5.2 rr-3 122.4 44.2 29.4 2.8 4.2 2 49.6 2.0 3.2 pr-2 157.4 79.1 ins Barbados[Prospero et al., 1970]. Longer travel distances pr-3 213.5 173.6 117.1 1.2 1.8 pr-4 289.2 377.1 214.7 0.8 1.3 and residencetimes for African dust reachingBermuda thus 10 zi-3 282.6 400.3 161.1 0.7 1.8 may explain some of the geochemicalcharacteristics of Ber- zi-4 289.6 460.4 175.1 0.6 1.7 muda's paleosols. zi-5 168.4 347.8 96.6 0.5 1.7 In situ differences in the Zr/Y and Zr/La ratios also are xi-7 296.0 311.0 164.0 1.0 1.8 11 zz-10 111.9 41.1 17.7 2.7 6.3 evidentwithin the profilesof Bermuda'spaleosols (Figure 5). zz-ll 87.8 48.3 28.8 1.8 3.0 At the two samplesites with the best developedprofiles (sites zz-12 122.9 43.4 29.4 2.8 4.2 5 and 10), the decreasein the ratios betweenthe A and B zz- 16 97.6 27.8 21.0 3.5 4.6 horizons exceeds 50%; the B horizon contains a greater WB wb-40 265.4 152.2 554.7 1.7 0.5 wb-42 391.0 181.7 595.0 2.2 0.7 amountof illuviatedfine clays.Vertical movementof claysand wb-44 294.3 113.3 477.1 2.6 0.6 their accumulationin the B horizon is common in tropical Qb/Qm gr-1 195.1 326.7 137.0 0.6 1.4 soils.Illuviated material in the Bermuda paleosolspermits a gr-3 211.6 163.8 99.1 1.3 2.1 clearer resolution of the geochemicalsignature of the atmo- or-1 130.4 68.8 35.7 1.9 3.7 sphericdust from which the soilsare derivedand further jus- Qt./Q,. es-20 875.0 386.1 248.8 2.3 3.5 es-21 875.5 376.5 256.8 2.3 3.4 tifieslimiting the geochemicalfingerprinting to B horizon sam- es-22 889.4 352.9 331.8 2.5 2.7 ples. The low Zr/Y and Zr/La ratios determinedfor the B Qt,/Q,. mh-1 334.1 257.3 121.0 1.3 2.8 horizon of Bermuda'spaleosols are more consistentwith Sa- mh-3 122.4 103.9 46.2 1.2 2.6 haran dust than with the two North American loess source mh-5 222.2 200.3 103.4 1.1 2.1 areas. mh-6 360.5 631.8 225.3 0.6 1.6 mh-7 272.5 684.4 214.9 0.4 1.3 Given the fact that the calcareniticsediments comprising the mh-8 360.6 310.4 150.3 1.2 2.4 eolianites of Bermuda are cemented as a result of freshwater mh-9 158.2 367.0 133.9 0.4 1.2 circulationthrough the meteoric-vadoseand meteoric-phreatic mh-10 359.0 416.8 197.0 0.9 1.8 13 Jr-1 177.4 104.5 55.2 1.7 3.2 zones[Plummet et al., 1976; Vacher,1978; Vacherand Wallis, 1992], one might considerthe possibilityof subsurfacesedi- *Overlyingand underlyingeolianite formations. ment transportbetween stratigraphic units as an explanation 23,396 HERWITZ ET AL.: BERMUDA PALEOSOLS AND PALEOCLIMATE

been a sourceof contamination. Second,the soil-eolianite con- i i i i tactson Bermuda are zonesof enhancedcarbonate diagenesis characterizedby well-cementedlayers of fine-grainedcalcite rangingin thicknessfrom 1 to 3 cm [Plummeret al., 1976];the extremelylow permeabilityof these calcite layers limits the downwarddispersion of particulatesby percolatingmeteoric water. The more likely avenueof particle transportto lower levelswould be along irregular fracture planes and solution pipes,which are alsolined by calcitewalls [Herwitz, 1993]. With Sahara Deser' the exceptionof the solutionpipe siteat WhaleboneBay, none of the paleosolsamples were collectedfrom sectionsof pa- leosolunits where there was evidence of theseflow pathsin the overlyingeolianite. 01 One of the remainingissues that has not been completely lOO 200 400 600 800 resolvedis the possibleinfluence of volcanicdetritus derived Age (ka) from the Bermudaplatform on the geochemistryof Bermuda's paleosols[Blackburn and Taylor,1969, 1970;Bricker and Mack- Figure 3. Mean Zr/La ratios(+_ 1 s.d.)from paleosolsample enzie, 1970]. sites as a function of age of underlyingeolianite shownin relation to the range of values(mean +_ 1 s.d.) of the three Possible Contributions From the Volcanic Platform possiblesource areas: Great Plains, MississippiValley, and Sahara Desert. The numbers next to each mean value show the The immobile elementsTi and Nb did not yield ratios that number of paleosolsample sites represented.The boxed-in could be used to distinguishbetween the three hypothesized outlier representsthe Zr/La ratio from site 14. source areas. The ratios involvingTi and Nb also were too variable amongthe Bermudapaleosols to providea consistent geochemicalsignature. However, the concentrationsof Ti, Nb, for some of the geochemicalvariation amongstthe paleosol and La did revealthat the paleosolat site 3 (Plate 2f) is more samples.This explanation,however, is of limited significance similarto the solutionpipe soilsat WhaleboneBay (WB) than for the followingreasons. First, the sixeolianite formations are to the other paleosols(Figure 6). The site 3 paleosolappears not stackedin a vertical successionbut rather represent a to have been influencedby volcanicdetritus from the older successionof mainly onlappingdune unitsproduced by lateral carbonateformations (>300 ka) on the west side of Castle accretion;as a result, none of the paleosolsexamined in this Harbour in the northeasternsection of the Bermudaplatform studywere overlainby youngerpaleosol units that couldhave (Figure 1). Seismicsurveys have reported volcanic rock at depths of only 30.5 m below sea level on the rim of Castle Harbour [Geesand Medioli, 1970]. Table 3. Trace Element Concentrations and Ratios for The Upper Town Hill marine facies(Qtm) at Whalebone Size-FractionatedSamples From Barbadosof SaharanDust Bay is noted for dark bands [Woolardand Ewing, 1939] con- Determined by InstrumentalNeutron ActivationAnalysis

Concentration, Size ppm I I I I Fraction, Sample Zr/La /am Number Zr La Ratio 10

0-2 27.5.67 152 53 2.9 / ./ 6.6.67 204 60 3.4 ./ ./ ./ 23.8.67 141 52 2.7 ./ ./ 26.9.67 98 44 2.2 ./ 1.10.68 184 54 3.4 ./ 6.7.69 161 61 2.6 2-5 27.5.67 183 46 4.0 6.6.67 188 53 3.5

/. ,/././t 3.7.67 202 50 4.0 23.8.67 166 47 3.5 26.9.67 196 45 4.4 1.10.68 204 49 4.2 /. ./ 6.7.69 249 51 4.9 5-10 27.5.67 225 42 5.4 Saharan dust 6.6.67 252 40 6.3 3.7.67 290 41 7.1 y z (0.36x) + 2,33 23.8.67 232 42 5.5 26.9.67 269 43 6.3 <2 <5 <10 <15 <20 1.10.68 334 44 7.6 6.7.69 257 41 6.3 Particle size fraction (p.) 10-20 27.5.67 311 33 9.4 Figure 4. Zr/La ratios in Saharandust as a functionof par- 6.6.67 459 41 11.2 3.7.67 319 41 7.8 ticlesize. The best fit relationshiphas an r 2 valueof 0.99 (P < 23.8.67 327 39 8.4 0.001). Analyseswere performedon samplescollected by J. 6.7.69 412 41 10.0 Prospero at the aerosolic dust collection site on Barbados [Delanyet al., 1967]. HERWITZ ET AL.' BERMUDA PALEOSOLS AND PALEOCLIMATE 23,397

Zr/Y Zr/La 0.2 0.6 1 1.4 1.4 1.8 2.2

A- o i i i i I c)

AB- o/

BA-

o N '•- Btl - o

o Bt2- _

Bt3 /

Bt4I o--o SITE 5 ß --ß SITE 10 o Bst I o I I I I I

Figure 5. Zr/Y and Zr/La ratios as function of soil horizon for two selected sites: site 5 and site 10. The arrowsrefer to the much larger valuesin the A horizon at site 10 (2.7 for Zr/Y and 4.4 for Zr/La). sistingof rounded, vesicularvolcanic sand and some obsidian currence of red clayey soils throughout Bermuda is inconsis- grains cemented in the carbonatematrix (Plate 4a). Band tent with the soils originating from locally weathered volcanic thicknessranges from 0.2 to 1.5 cm at elevationsbetween 1 and materials. This interpretation is further supported by the ex- 1.3 m above mean sea level. Loose black sand derived from ceptionallyhigh Ti, La, and Nb concentrationsin the Whale- these dark bands by chemical weathering and wave abrasion bone Bay solution pipe soils located near the black sand de- can be found in the swashzone of the present-dayshoreline posits.Only one palcosolunit has comparableconcentrations immediately beneath the erosionalface of the unit Qt,• out- and that is the palcosolbetween the Upper Town Hill (Qtu) crop. Shallow cores of the modern beach reveal bands of and Walsingham(Q,•) Formationsat Site 3 (Plates2f and 4b). sorted volcanicparticulates similar to those in the marine fa- Other palcosolsin the sameimmediate area (sites1, 4, and 14) cies.The black sand in unit Qt,, thus marks the time when the do not have the same geochemicalsignal. There is no signifi- Castle Harbour volcanicrim (a sectionof the Bermuda plat- cant relationship between immobile element concentrations form) was subaeriallyexposed and the Town Hill eolianites and distancefrom the WhaleboneBay exposure(Figure 6). were deposited[Hearty et al., 1992];the volcanicplatform was Volcanic detritus thus appears to have had a significant never again exposed[Land et al., 196'7]. effect only on the site 3 geochemicalsignature rather than on Using X ray diffractometry,Blackburn and Taylor [1969] all the soilsthroughout Bermuda. The Ti, La, and Nb concen- reported that both the Whalebone Bay volcanic sand and trations at site 2 are slightly higher than the overall average, nearby surfacesoils contain varying amountsof apatite, chro- but this signalis entirely lacking at site 1, which also is located mite, garnet, gold, goethite, perovskite,anatase, sphene, cal- on St. George's Island near sites 2 and 3. The higher specific cite, magnetite, and volcanic glass.These results led them to gravityof the volcanicsediment compared with the calcareous suggestthat Bermuda's palcosolswere derived from volcanic biogenic sedimentslimits their colJantransport and explains detritus from the Bermuda platform. Bricker and Mackenzie why they were not found in any eolianite formation >5 m [19'70]countered with the argument that the widespreadoc- above sea level or in any eolianite formation west of Har- rington Sound. During Upper Town Hill time, a major coastal storm overwashevent would be a likely causeof volcanic sed-

30K iment transport from Whalebone Bay to nearby inland loca- Ti I ø S,T•3 I A tions such as site 3. Subsequentburial of the Whalebone Bay 15K Qtm marine faciesby eolianiteswould explain the limited spa- E 0 0 0 0 0 0 tial distribution of volcanic detritus in the younger carbonate • 0 c La formations and palcosols.It is clear that volcanic sediment o '.• from the Bermuda platform is not the primary or even the • 300 ß o o o o secondaryparent material of Bermuda'spalcosols. • 0 I o

c)o Nb 3OO Discussion

0 • • 0 I I (D IN I I'• I I I n The geochemical signatures of Bermuda's palcosols are 0 2 4 6 8 10 12 4 16 18 20 most consistentwith an influx of Saharan dust mainly in the Distance from Whalebone Bay site (km) <2-/•m particle size range. This finding is in agreement with Figure 6. Ti, La, and Nb concentrations as a function of the view that aerosolicdust having a mean particle size of •2 distancefrom WhaleboneBay (WB), the sourceof volcanic /•m is kept aloft by the trade wind inversionlayer [Carlsonand sediment. No significant statistical correlation with distance Pro•79ero,1972] and transportedlong distancesacross the At- was found. Only site 3 is consistentlysimilar to WB. lantic Ocean [Prosperoet al., 1970, 1979;Prospero and Carlson, 23,398 HERWITZ ET AL.: BERMUDA PALEOSOLS AND PALEOCLIMATE

ß t

Plate 4. (a) Volcanicblack sand (v 0 in the intertidalswash zone at WhaleboneBay (WB) derivedfrom band of cemented volcanic sand (v) in the Upper Town Hill marine facies ((•tm); geochemicalanalyses were performed on samplesfrom the interiorsof soil-filled(s) solutionpipes (p). (b) Close-upview of the Shore t tills palcosol at site 3, located 1.5 km from Whalebone Bay and the only palcosol unit clearly affected geochemicallyby volcanicsediment.

1972].According to Ralm et al. [I 979], smaller averageparticle seaboard,however, would not be a likely sourcearea for aero- sizcsare associatedwith longer travel distances.Larger parti- solic dust becauseof its moist climatic regime and forest cover cles unaffectedby the inversion layer are deposited mainly throughout the Quaternary. t towever, if Saharan dust can within 1000 km of the west coast of Africa [Schiitz, 1979; travel this far north of the Caribbean, then a path from north- Jaenicke, 1979; Grottssetet al., 1989]. Northwest Africa thus west Africa to the Caribbean that then shifts northeast to appearsto havebeen the dominantQuaternary source area for Bermuda at 32øN latitude is certainly a possibility. Atmo- Bermuda'spalcosols, as was originally suggested by Brickerand spheric transport of Asian dust to Alaska over distancesex- Prospero[ 1969]. ceeding 6000 km has been documented [Rahn et al., 1979]. Evidencefor long-distancetransport of Saharandust beyond Bermuda is only half this distancefrom northwestAfrica. the Caribbeanwas first providedby Syerset al. [1969] working The most extensivelong-term field records of present-day in North Carolina. The oxygenisotope composition of quartz African dust transport are measurementsmade on Barbados grains in soilswas used as a natural tracer for tracking atmo- since 1965 [Delanyet al., 1967; Prosperoand Nees, 1986]. Sum- sphericdust contributionsto Quaternary soilsin nonglaciated mer is the seasonof maximum dust depositionon Barbados, regions.They found that metamorphicquartz grainscommon with the trade wind regime providing an extremely efficient in the lower part of red soil profiles are a product of in situ transport systemfor Saharan dust. In Miami, Florida, dust saproliteweathering of graywacke.But in the upper part of the concentrationsfollow the same seasonalcycle as on Barbados same soil profiles, the oxygen isotope composition of the [Prosperoet al., 1987], although mean concentrationsare typ- quartz fraction resemblesthat of the red soilsof Barbadosand ically only a third to a half of those on Barbados [Prospero, the Bahamas rather than the locally derived metamorphic 19951. quartz. Syerset al. [1969] suggestedthat aerosolicdust origi- Arimoto et al. [1995] measuredrates of AI, aluminum dep- natingfrom the Saharadesert was carried into the Caribbean- osition at a coastal site on the west end of Bermuda over a Gulf of Mexico region by the trade winds, then northward in 2-year period (1989-1991), but only during onshorewind con- the rising,warm, moist air massesalong the easternseaboard. ditionsthat precludedcontamination from local islandsources. They further suggestedthat precipitationwas the mechanism After convertingtheir A1 measurementsto equivalentamounts by which the aerosolicdust was washedfrom the atmosphere ofmineral dust by assuming the average concentration ofAI in and depositedon the southerncoastal plain. crustal materials, they report a pronouncedseasonal cycle of The red soils of the southeastern United States are similar to dust concentrations on Bermuda similar to that measured on Bermudanpalcosols in terms of color and texture. The eastern Barbados:a maximum during summer and a minimum during HERWITZ ET AL.: BERMUDA PALEOSOLS AND PALEOCLIMATE 23,399 winter. The summer dust events on Bermuda are attributed to aturesin the North Atlantic during glacial summers[Keller et air mass trajectories consistentwith the strong circulation al., 1988] may not have been low enoughto displacethe Gulf around the Bermuda-Azoreshigh-pressure center. Mean min- Stream and Bermuda's subtropical atmospheric circulation eral dust concentrations on Bermuda, however, are less than pattern south of 32øN. thoseon Barbados[Prospero, 1996]. Somedust transportevents associated with sourcesin North Acknowledgments. We thank David Lines, Rip Heminway, Ste- America have been detected on Bermuda [Chen and Duce, phen Kerns, Joel Frisch, Isaac Phillips, Denise Lajoie, Samantha 1983],but the dustconcentrations associated with theseevents Kaplan, Robert Redpath, Michael McGeehan, Patrick Burns, Sonya are much lower than those for African dust events[Prospero, Sundaram,and Adam Sadowskyfor their assistancewith the sample 1996].,4rimoto et al. [1995] groupedBermudan aerosolsam- collection.Thanks also to Marith Reheis and Ralph Shrobafor their ples accordingto air masstrajectory and found that the mean many constructivecomments on an earlier draft of this paper. This studywas funded by a grant from the National GeographicSociety A1 concentrationfor trajectoriesfrom the eastis approximately Committee for Researchand Exploration. 10 timesgreater than that for trajectoriesfrom North America. When consideringthese findingsin relation to the origin of Bermudanpaleosols, the challengeis ascertainingthe extent to References which the present-dayinterglacial is representativeof Quater- Arimoto, R., R. A. Duce, B. J. Ray, W. G. Ellis Jr., J. D. Cullen, and nary dust depositionover the period 880 to 85 ka. Given the J. T. Merrill, Trace elements in the atmosphereover the North effectsof agriculture and poor land use practiceson soil dust Atlantic, J. Geophys.Res., 100, 1199-1214, 1995. deflation [Prospero,1996], one must first recognizethe possi- Blackburn,G., and R. M. Taylor, Limestoneand red soilsof Bermuda, bility that present-daypatterns of dust transport may not be Geol. Soc. Am. Bull., 80, 1595-1598, 1969. Blackburn,G., and R. M. Taylor, Limestoneand red soilsof Bermuda: representativeof past glacialsand interglacials.However, the Reply, Geol. Soc.Am. Bull., 81, 2525-2526, 1970. paleosolgeochemistry reported in this paper suggeststhat the Bowles, F. A., Palcoclimaticsignificance of quartz/illite variations in current atmosphericcirculation pattern in the North Atlantic coresfrom easternequatorial North Atlantic, Quat. Res.,5, 225-235, has persistedand may have been even more pronouncedthan 1975. Bretz, H., Bermuda: A partially drowned, late mature, Pleistocene it is at present. karst, Geol. Soc. Am. Bull. 71, 1729-1754, 1960. Bricker, O., and F. T. Mackenzie, Limestones and red soils of Ber- muda: Discussion, Geol. Soc. Am. Bull., 81, 2523-2524, 1970. Conclusion Bricker, O., and J. M. Prospero,Airborne duston the Bermuda Islands Saharanaerosolic dust has contributedsignificant amounts and Barbados, Eos Trans. AGU, 50, 176, 1969. Carlson, T. N., and J. M. Prospero, The large-scalemovement of of silicatesand other mineralsnot only to the red soilsof the Saharanair outbreaksover the equatorial North Atlantic, J. Appl. Caribbean and the eastern seaboard of North America but also Meteorol., 11,283-297, 1972. to the red soilsof Bermuda. Our study highlightsthe impor- Chapman, S. L., and M. E. Horn, Parent material uniformity and tance of particle size in the identificationof geochemicalsig- origin of silty soils in northwest Arkansas based on zirconium- titanium contents, Soil Sci. Soc. ,4m. Proc., 32, 265-271, 1968. naturesusing immobile element ratios. Volcanic detritus from Chen, L., and R. A. Duce, The sources of sulfate, vanadium and the Bermuda platform has had no detectable influence on mineral matter in aerosol particles over Bermuda, ,4ttnos.Environ., Bcrmudan paleosols,except at one location. Aerosolic dust 17, 2055-2064, 1983. originating from the American-Canadian continental land- Delany, A. C., D. W. Parkin, J. J. Griffin, E. D. Goldberg, and B. E. F. massesand carried by the jet streamwesterlies may have con- Reimann, Airborne dust collected at Barbados, Geochitn. Costno- chitn. `Acta, 31,885-909, 1967. tributed, but this contribution cannot be clearly detected in Fisk, H. N., Loess and Quaternary geologyof the lower Mississippi Bermuda'spalcosols. This interpretation is a slight departure Valley, J. Geol., 59, 333-356, 1951. from our original study,based on Bermuda soil material col- Francois,R., and M.P. Bacon,Variations in terrigenousinput into the lected from subaeriallyexposed solution pipes, in which we deep equatorial Atlantic during the past 24,000 years, Science,251, 1473-1476, 1991. suggestedthe possibilityof mixing of aerosolicdust from both Gees, R. A., and F. Medioli, A continuous seismic survey of the the Saharaand an unknownNorth American sourcearea [tIer- Bermuda platform, I, Castle Harbour, Marit. Sediments',6, 21-25, witz and Muhs, 1995]. 1970. Changesin accumulationrates of eolian dust in deep-sea Grousset,F., ?. Buat-Menard, D. Boust, R. Tian, S. Baudel, C. Pujol, cores [Rea, 1994] from the easternAtlantic near Africa indi- and C. Vergnaud-Grazzini, Temporal changesof aeolian Saharan input in the Cape Verde abyssalplain sincethe last glacial period, cate that wind systemswere more vigorousand Saharan dust Oceanol..Acta, 11, 177-185, 1989. transportwas significantlygreater during glacial maxima than Harmon, R. S., L. S. Land, R. M. Mitterer, P. Garrett, H. P. Schwarcz, duringinterglacials [Bowles, 1975; Kolla et al., 1979;Pokras and and G. J. Larson, Bermuda sea level during the last interglacial, Mix, 1985; Groussetet al., 1989; Francoisand Bacon, 1991; Rea, Nature, 289, 481-483, 1981. Harmon, R. S., R. M. Mitterer, N. Kriausakul, L. S. Land, H. P. 1994]. More recently, Yung et al. [1996] concluded that the Schwarcz, P. Garrett, G. J. Larson, H. L. Vacher, and M. Rowe, increaseddust concentrations found in polar ice coresfrom the Uranium-seriesand amino-acidracemization geochronology of Ber- last glacial maximum are the result of stronger winds and a muda: Implications for eustatic sea-levelfluctuations over the past weakened hydrologiccycle as diagnosedby reduced rates of 250,000 years, Palaeogeogr.Palaeoclitnatol. Palaeoecol., 44, 41-70, 1983. wet depositionand longer washoutlifetimes of aerosolicdust Hearty, P. J., and H. L. Vacher, Quaternary stratigraphyof Bermuda: particles.If Bermuda'sred clay soilsformed most rapidly dur- A high-resolutionpre-Sangamonian rock record, Quat. Sci. Rev., 13, ing glacial maxima, then the Saharan signaturewould be in 685- 697, 1994. agreementwith Kutzbach's[1987] general circulationmodel. Hearty, P. J., H. L. Vacher, and R. M. Mitterer, Aminostratigraphy His model indicatesthat during glacial maxima the northeast and ages of Pleistocene limestones of Bermuda, Geol. Soc..Am. Bull., 104, 471-480, 1992. summer trade winds reached latitudes higher than 30øN and Herwitz, S. R., Stemflow influenceson the formation of solution pipes were strongerthan at present.The dominanceof Saharandust in Bermuda eolianite, Geotnorphology,6, 253-271, 1993. in Bermudanpaleosols thus indicatesthat sea surfacetemper- Herwitz, S. R., and D. R. Muhs, Bermuda solution pipe soils: A 23,400 HERWITZ ET AL.: BERMUDA PALEOSOLS AND PALEOCLIMATE

geochemicalevaluation of eolian parent materials,in Terrestrialand Prospero, J. M., R. T. Nees, and M. Uematsu, Deposition rate of ShallowMarine Geologyof the Bahamasand Bermuda,edited by H. particulate and dissolvedaluminum derived from Saharan dust in A. Curran and B. White, Spec.Pap. Geol. Soc.Am., 300, 311-323, precipitationat Miami, Florida, J. Geophys.Res., 92, 14,723-14,731, 1995. 1987. Holliday, V. T., The geologicrecord of wind erosion, eolian deposi- Pye, K., Red beds,in ChemicalSediments' and Geomorphology:Precip- tion, and aridity on the Southern High Plains, Great PlainsRes., 1, itatesand Residuain the Near-SurfaceEnvironment, edited by A. S. 6-25, 1991. Goudie and K. Pye, pp. 227-263, Academic,San Diego, Calif., 1983. Hooghiemstra,H., Palynologicalrecords from northwestAfrica ma- Pye, K., Aeolian dusttransport and depositionover Crete and adjacent rine sediments:A generaloutline of the interpretationof the pollen partsof the MediterraneanSea, Earth Surf.Processes Landforms, 17, signal,Philos. Trans.R. Soc. London, Sec.B, 318, 431-449, 1988. 271-288, 1992. Hooghiemstra,H., A. Bechler, and H.-J. Beug, Isopollen maps for Pye, K., and R. Johnson,Stratigraphy, geochemistry, and thermolumi- 18,000 years B.P. of the Atlantic offshoreof northwestAfrica: Evi- nescenceages of lower MississippiValley loess,Earth Surf.Processes dencefor paleowindcirculation, Paleoceanography, 2, 561-582, 1987. Landforms,13, 103-124, 1988. Jaenicke, R., Monitoring and critical review of the estimatedsource Rahn, K. A., R. D. Borys, G. E. Shaw, L. Schfitz, and R. Jaenicke, strength of mineral dust from the Sahara, in SaharanDust: Mobili- Long-rangeimpact of desertaerosol on atmosphericchemistry: Two zation, Transport,Deposition, edited by C. Morales, pp. 233-242, examples,in SaharanDust: Mobilization, Transport,Deposition, ed- John Wiley, New York, 1979. ited by C. Morales, pp. 243-266, John Wiley, New York, 1979. Johnson,R. G., and B. King, Energy-dispersivex-ray fluorescence Rea, D. K., The paleoclimaticrecord provided by eolian depositionin spectrometry,in Methods'for GeochemicalAnalysis, edited by P. A. the deep sea--The geologic history of wind, Rev. Geophys.,32, Baedecker, U.S. Geol. Surv. Bull., 1770, F1-F5, 1987. 159-195, 1994. Keffer, T., D. G. Martinson, and B. H. Corliss,The positionof the Gulf Rowe, M. P.,An Explanationof the Geologyof Bermuda,Min. of Works Stream during Quaternary glaciations,Science, 241,440-442, 1988. and Eng., Hamilton, Bermuda, 1990. Kolla, V., P. E. Biscaye,and A. F. Hanley, Distributionof quartz in late Ruhe, R. V., J. G. Cady, and R. S. Gomez, Paleosolsof Bermuda, Quaternary Atlantic sedimentsin relation to climate, Quat. Res., 11, Geol. Soc. Am. Bull., 72, 1121-1142, 1961. 261-277, 1979. Sayles,R. W., Bermuda during the ice age,Proc. Am. Acad. Arts'Sci., Kutzbach, J. E., The influence of changingorbital parameters and 66, 380-467, 1931. surface boundary conditions on climate simulationsfor the past Schfitz, L., Sahara dust transport over the North Atlantic Ocean-- 18000 years,J. Atmos. Sci., 43, 1726-1759, 1987. Model calculations and measurements, in Saharan Dust.' Mobiliza- Land, L. S., F. T. Mackenzie, and S. J. Gould, Pleistocenehistory of tion, Transport,Deposition, edited by C. Morales, pp. 267-278, John Bermuda, Geol. Soc. Am. Bull., 78, 993-1006, 1967. Wiley, New York, 1979. Livingston,W., Observationson the structure of Bermuda, Geogr.J., Stanley,D. J., and D. J.P. Swift, Bermuda'ssouthern aeolianite reef 104, 40-48, 1944. tract, Science, 157, 677-681, 1967. McCauley, J. F., C. S. Breed, M. J. Grolier, and D. J. Mackinnon, The Syers,J. K., M. L. Jackson,V. E. Berkheiser,R. N. Clayton,and R. W. U.S. dust storm of February 1977, Spec.Pap. Geol. Soc.Am., 186, Rex, Eolian sedimentinfluence on pedogenesisduring the Quater- 123-147, 1981. nary, Soil Sci., 107, 421-427, 1969. Moresi, M., and G. Mongelli, The relationshipbetween terra rossa and Vacher, H. L., Coastal dunes of younger Bermuda, in Coastal Geo- the carbonate-freeresidue of the underlyinglimestones and dolos- morphology,edited by D. R. Coates,pp. 355-391, Publ. in Geomor- tones in Apulia, Italy, ClaysClay Miner., 23, 439-446, 1988. phol., State Univ. of N.Y., Binghamton,1973. Muhs, D. R., C. A. Bush, K. C. Stewart, T. R. Rowland, and R. C. Vacher, H. L., Hydrogeologyof Bermuda--Significanceof an across- Crittenden, Geochemicalevidence for Saharandust parent materi- the-islandvariation in permeability,J. Hydrol.,39, 207-226, 1978. als for soilsdeveloped on Quaternarylimestones of Caribbeanand Vacher, H. L., and P. J. Hearty, History of stage5 sea level in Ber- western Atlantic islands, Quat. Res., 33, 157-177, 1990. muda: Review with new evidenceof a brief rise to presentsea level Officer, C. B., M. Ewing, and P. C. Wuenschel, Seismic refraction during substage5A, Quat. Sci. Rev., 8, 159-168, 1989. measurements in the Atlantic Ocean, IV, Bermuda, Bermuda Rise, Vacher, H. L., and T. N. Wallis, Comparativehydrogeology of fresh- and Nares Basin, Geol. Soc. Am. Bull., 63, 777-808, 1952. water lenses of Bermuda and Great Exuma Island, Bahamas, Pirsson, L. V., Geology of Bermuda Island: The igneousplatform, Ground Water, 30, 15-20, 1992. Am. J. Sci., 38, 189-206, 331-344, 1914. Vacher, H. L., M.P. Rowe, and P. Garrett, The GeologicalMap of Plummer, L. N., H. L. Vacher, F. T. Mackenzie, O. P. Bricker, and Bermuda,1:25,000 scale. Public Works Dep., Hamilton, Bermuda, L. S. Land, Hydrochemistryof Bermuda: A casehistory of ground- 1989. water diagenesisof biocalcarenites,Geol. Soc.Am. Bull., 87, 1301- Vacher, H. L., P. J. Hearty, and M.P. Rowe, Stratigraphyof Bermuda: 1316, 1976. Nomenclature,concepts, and statusof multiple systemsof classifi- Pokras,E. M., and A. C. Mix, Eolian evidencefor spatialvariability of cation,in Terrestrialand ShallowMarine Geologyof theBahamas and late Quaternaryclimates in tropicalAfrica, Quat. Res.,24, 137-149, Bermuda,edited by H. A. Curran and B. White, Spec.Pap. Geol.Soc. 1985. Am., 300, 271-294, 1995. Prasad,G., On the origin of red clayson Quaternarylimestone along Windom, H. L., and C. F. Chamberlain,Dust-storm transport of sed- the East African coast,Z. Geomorph.Suppl., 48, 51-64, 1983. iments to the North Atlantic Ocean, J. Sediment.Petrol., 48, 385-388, Prospero, J. M., Atmospheric dust studies on Barbados, Bull. Am. 1978. Meteorol. Soc., 49, 645-652, 1968. Woolard, G. P., and M. Ewing, Structuralgeology of the Bermuda Prospero,J. M., The atmospherictransport of particlesto the ocean,in Islands, Nature, 143, 898, 1939. Particle Flux in the Ocean, edited by V. Ittekkot, P. Schafer, S. Yung, Y. L., T. Lee, C.-H. Wang, and Y.-T. Shieh,Dust: A diagnostic Honjo, and P. J. Depetris, pp. 19-52, John Wiley, New York, 1996. of the hydrologiccycle during the lastglacial maximum, Science, 271, Prospero,J. M., and T. N. Carlson, Vertical and areal distributionof 962-963, 1996. Saharandust over the westernequatorial North Atlantic Ocean,J. Geophys.Res., 77, 5255-5265, 1972. S. R. Herwitz, Departmentsof Geographyand Biology,Clark Uni- Prospero,J. M., and R. T. Nees, Impact of North African droughtand versity,Worcester, MA 01610. (e-mail: [email protected]) E1 Nifio on mineral dust in the Barbados trade wind, Nature, 320, S. Mahan, D. R. Muhs, and B. Vaughn, U.S. GeologicalSurvey, MS 735-738, 1986. 974, Box 25046, Denver Federal Center, Denver, CO 80255. Prospero,J. M., E. Bonatti, C. Schubert,and T. N. Carlson,Dust in the J. M. Prospero,Rosenstiel School of Marine and AtmosphericSci- Caribbeanatmosphere traced to an African duststorm, Earth Planet. ences,University of Miami, 4600 RickenbackerCauseway, Miami, FL Sci. Lett., 9, 287-293, 1970. 33149. Prospero, J. M., D. L. Savoie, T. N. Carlson, and R. T. Nees, Moni- toring Saharanaerosol tranqport by meansof atmosphericturbiditv measurements,in SaharanDust: Mobilization, Transport, Deposition, (ReceivedMarch 20, 1996;revised July 12, 1996; edited by C. Morales, pp. 171-186, JohnWiley, New York, 1979. acceptedJuly 12, 1996.)