A 300 kyr high-resolution aridity record of the North African continent

A 300 kyr high-resolution aridity record of the North African continent
of 16
All materials on our website are shared by users. If you have any questions about copyright issues, please report us to resolve them. We are always happy to assist you.
Related Documents
  PALEOCEANOGRAPHY, OL. 10, NO. 3, PAGES 677-692, JUNE 1995 A 300 kyr high-resolution aridity record of the North African continent A. P. Matthewson, G. B. Shimmield, and D. Kroon Marine Geosciences nit, Department f Geology nd Geophysics, niversity f Edinburgh, dinburgh United Kingdom A. E. Fallick Isotope Geosciences nit, Scottish niversities esearch nd Reactor Centre, Glasgow, nited Kingdom Abstract. New oxygen sotope data and a high-resolution arbonate ecord of a core rom the northwest frican margin are presented, n combination ith geochemical Zr/Rb ratio, A1, Ba, organic arbon, nd biogenic pal luxes) and sedimentological median grain size) parameters, o show ariations n African continental ridity, wind strength, nd upwelling roductivity. he carbonate ecord s diluted by the supply of aeolian ust rom the arid Saharan egion of north Africa and s seen o vary with a characteristic awtooth attern ndicating apid ncreases n dust flux followed by gradual eclines. Spectral nalysis f the carbonate rofile and comparison ith SPECMAP stacked xygen sotope) nd ETP (solar adiation) rofiles hows hat he preces- sional 23 kyr) insolation-induced frican climate system s decoupled rom global ice volume) climate changes perating n a 100 kyr frequency. Detailed studies f glacial erminations , II, and II confirm he presence f rapid, short imescale limatic variations in aridity). Introduction The Saharan egion of the African continent s one of the most arid places n Earth, and he major source f dust particles o the deep-sea ediments f the eastern quatorial tlantic. Movement of the dust s controlled y two different wind regimes. Most of the ransport f terrestrial aterial ccurs n the Saharan ir ayer [Carlson and Prospero, 1972]. This is related o the African Easterly Jet, a mid-tropospheric onal wind between 1000 and 5000 m, which s coincident ith he ntertropical onvergence Zone (ITCZ) at 18øN during he northern ummer. At altitudes below 1000 m, the meridional ortheast rade winds provide he principal ransport Figure 1). The northwest frican margin s one of the major areas of upwelling n the world's oceans Smith, 1968]. Cold, nutrient- rich waters re brought o the surface y the offshore movement Of surface aters timulated y he northeast rade inds Figure 1). The upwelling nutrients promote a dramatic ncrease n primary productivity f phytoplankton Huntsman nd Barber, 1977] resulting n a large component f the sediment eing biogenic n srcin. Other minor sedimentary nputs re fluvial sedimentation ffshore rom ephemeral iver outlets uring umid periods, nd mass low deposits Tiedemann t al., 1989]. All inputs re controlled y variations n global nd ocal limate, which are modulated y astronomical orcing Milankovitch, 1941]. Copyright 1995 by the American Geophysical nion. Paper number 94PA03348. 0883-8305/95/94PA-03348510.00 Numerous tudies ave hown hat he nput f terrigenous material ff northwest frica, n the form of aeolian ust, increased uring lacials, ainly ue o ncreased vailability f source material nd strength f the ransporting inds Broecker et al., 1958; Hays and Peruzza, 972; Bacon, 984; Grousset t al., 1988; Francois t al., 1990]. Furthermore, iester-Haass [1976a, ] showed hat he nput f terrigenous ust ppeared o dilute carbonate ontents uring lacial eriods. Kolla et al. [1979] dentified zone of high-percent uartz ediment ff northwest frica (reflecting he trade wind transport f desert dust) which migrated outhward y about ø of latitude uring he last glacial maximum due to increased ridity and the southern shift of the trade wind belt. Evidence from lake and fluvial sediments, nd palaeosols onfirms hat a 50-8 southerly movement of the ITCZ could cause a less humid northwest African climate [Nicholson and Flohn, 1980; Nicholson, 1982; Newell and Kidson, 1979; Pokras and Mix, 1985]. Sarnthein et al. [1981], however, concluded hat aridity changes ould not be explained by the fluctuating positions of climatic zones and proposed hat varying wind strengths of tropical storms and supply of water vapor rom the equatorial Atlantic were the most important factors. The hemipelagic sediments off northwest Africa can herefore e used o develop n understanding f the processes f palaeoceanographic nd climate change associated with desertification of northern Africa. Tiedemann t al. [ 1989] proposed hat short-term ariations n dust lux over the last 500,000 years ollowed precessional 23 kyr) cycles, hus providing a link between Sahelian aridity fluctuations and evaporation in the tropical Atlantic. Consequently, olar radiation appears o control the dust flux from he north African continent. This precessional omponent s  678 MATTHEWSON ET AL.: A HIGH-RESOLUTION RIDITY RECORD OF NORTH AFRICA I - 30'N 5'W øN- Easterly •• I 5'N 25'W 20'W 16'W -o., / L ' o'.- •ape Verde Isla d 15'N 15'N , , , Figure . Wind dgimes ver orthwest frica, howing ower midtropospheric low patterns, onal winds f the Saharan ir ayer summer), nd meridional rade winds; umbers efer o wind elocities n meters er second; ITCZ, ntertropical onvergence one; otted egion ndicates he egion f perennial oastal pwelling modified from Stein and Sarnthein 1984] and Tetzlaaf nd Peters 1986]). (inset) detailed athymetry nd ocation f CD53-30 and other cores mentioned in text. also evealed n freshwater iatom abundance ecords Pokras and Mix, 1985] and n lithogenic rain ize and ithogenic ass accumulation ate MAR) records rom he Arabian ea Clemens and Prell, 1990]. Climatic hanges ith century-scale requencies ave been revealed n ice core records Dansgaard et al., 1993] and in further studies off northwest Africa. Carbon and oxygen sotope studies ndicate hat short-term Younger Dryas-style" limatic  MATTHEWSON ET AL.: A HIGH-RESOLUTION ARIDITY RECORD OF NORTH AFRICA 679 rebounds may be triggered by perturbations n North Atlantic Deep Water (NADW) circulation patterns [Sarnthein and Tiedemann, 1990], and short, intense periods of cold upwelling waters re apparent n the alkenone uk37) ecord f Eglinton t al. [ 1992]. This study will concentrate on the aeolian input to the sediment, and the extent to which it dilutes the biogenic component, as recorded n a core from the northwest African margin, at the edge of the upwelling zone. Geochemical and sedimentological arameters re used, in comparison with geo- chemical ndicators of palaeoproductivity nd climate proxy data from other sources, o present a high-resolution dust flux and climate record for the last 300,000 years. Using this, we investigate he climatic instability shown by the small-scale events which can be recognised, as well as the mechanisms behind he "sawtooth" attern of the profile. Analytical Methods Sampling The core for this study (CD53-30) was recovered rom the Cape Verde Terrace, at 19ø42.78'N, 0ø42.81'W n 3565 m water depth (Figure 1) and 300 km from the northwest African upwelling centre. Samples 20 cm ) were aken t 10-cm ntervals, ried t 50øC, and ground in a ball mill prior to geochemical analysis. Subsamples 10 cm ) were aken t ntermediary 0-cm ntervals for foraminiferal work, dried at 50øC and sieved at 63 gm. The <63-gm fraction was retained or particle size analysis. For the high resolution tudy, approximately cm slices of sediment were sampled t 1-cm ntervals down the core, dried at 50øC, and sieved at 63 gm. Both size fractions were dried and retained or analysis. Water contents were determined by weight loss on drying. All data are stored n the Biogeochemical cean Flux Study (BOFS) database at the British Oceanographic Data Centre (BODC). Stable Oxygen sotope Analysis Approximately even ndividuals f Globorotalia nfiata 250- 300 gm size fraction) and three individuals of Cibicidoides wuellerstorfi 300-425 gm) were equired o achieve n analytical weight of between .05 and 0.10 mg. Picked oraminifera ere cleaned n an ultrasonic ath prior o stable sotope nalysis n a VG Isotech PRISM isotope atio mass spectrometer, ith a VG Autocarb ommon-bath reparation ystem nd microinlet. O was produced y reaction with orthophosphoric cid at 90øC and passed hrough stainless teel coil water rap maintained t -80 ø C with a liquid N 2 probe. Calibration as made o the Pee Dee Belemnite PDB) standard, hrough analysis f NBS 19. The precision or oxygen sotope analysis was 0.08%0 (standard deviation or 100 analyses f a standard arbonate, M1, with weights f 0.05 to 0.10 mg, conducted ver several months). XRF Analysis Bulk sediment eochemistry as determined ith a Philips PW1480 sequential avelength ispersive ray spectrometer and PW1510 sample hanger. Glass iscs were prepared or major lement etermination Norfish nd Hutton, 969], using around 1 g of ground bulk sediment with a La-Li tetraborate lux. Discs were produced or trace element etermination y pressing around 3 g of ground bulk sediment nto a briquet with boric acid backing Shimmield, 985]. All analyses were corrected or the contribution and dilution effect of the sea-salt content of the dried sediment. Biogenic Silica Analysis The biogenic silica content of the sediment was determined using a wet chemical each technique developed by Dobbie [1988], after Eggimann t al. [1980]. Approximately 0 mg of dried, ground bulk sediment was digested in 2M sodium carbonate solution for 4 hours at 90ø-100øC. We used a colourimetric method nvolving he reduction o molybdenum blue of yellow silicomolybdate cid. The absorption f the resulting solution was measured at 812 nm in 2-cm cells in a Perkin-Elmer 550 spectrophotometer, o give an apparent biogenic silica value. This was then corrected for the effect of nonamorphous ilica derived rom clays by deducting wice the A1203 ontent f the solution, ssuming he SiO2/A1203 atio of clays in this region to be approximately . A colourimetric method nvolving he formation of a catechol iolet complex [Dougen and Wilson, 1974] was employed o determine he A1203 ontent f the solutions, he absorption eing measured t 585 nm in 1-cm cells. Organic Carbon Analysis Organic carbon analyses were carried out on a Carlo Erba NA1500 nitrogen/carbon/sulphur nalyzer. Carbonate was removed y slowly adding % sulphurous cid o 10-15 mg of dried, ground sediment n 2-mL plastic vials, and the samples were reeze-dried nd encapsulated n 30-mm in discs prior to analysis. Particle Size Analysis Two methods f particle size analysis were employed: 1. The biogenic component of the <63-gm size fraction was removed Clemens nd Prell, 1990]. Carbonate was emoved y employing n overnight digestion n 20% acetic acid. Following three rinses n deionized water the sediment ample was digested in 2M sodium carbonate at 90ø-100øC for 4 hours to remove biogenic opal. Removal of organic carbon was unneccessary because of the very low concentrations between 0.5% and 1.0%). Analysis was performed using a Coulter LS 100 laser particle size analyzer. This method was used or the low-resolution ecord and will hereafter e referred o as ithogenic rain size LGS). 2. Dried bulk sediment was wet-sieved at 63 gm and the >63 -gm content calculated as a percentage of total weight. This method was used or the high-resolution ecord and will hereafter be referred o as bulk grain size (BGS). Carbonate Analysis Carbonate content in the high-resolution samples was determined y titration Grimaldi, 1966]. An initial digestion n 0.5M hydrochloric acid solution was followed by back-titration against .35M sodium hydroxide solution. The precision or this method of carbonate analysis was 0.13% (standard deviation or six analyses of a homogenized ample). Twelve samples of a pure carbonate standard 99.5%) were analyzed, he resulting mean being 99.15%.  680 MATTHEWSON T AL.: A HIGH-RESOLUTION RIDITY RECORD F NORTH AFRICA Cross-Spectral nalysis In order to examine the frequency domain of the proxy indicators escribed bove, ross-spectral nalysis as employed using he Arand software ackage eveloped y P. Howell (Brown niversity) nd xtensively sed n the iterature Imbrie et al., 1993]. Each periodogram ndicates he proportion f the total observed variance accounted for by individual frequency components. he amount f variance er requency and log variance ensity) s plotted gainst requency cycles/unit ime). The 80% confidence nterval CI) is represented n the plot by a vertical bar. The shorter horizontal bar defines he bandwidth of the filter used n the smoothing f the spectra n the frequency domain and s inversely elated o the number of lags used o calculate he autocorrelation/autocovariance unction (40 lags in every ase). The highest requency hich an be resolved s the Nyquist requency fn)=l/(2 time nterval etween ata oints). In our analyses, n=0.1316 7600 ears). ur plots o not extend to frequencies bove .1 cycles/1000 ears. Coherency epresents he inear correlation etween he two frequency pectra ver a given requency and when he phase difference s converted o zero. The coherency plot indicates what requency omponents he X series nd he Y series hare. An 80% confidence level is set, above which statistical significance n the coherency elationship s considered o exist. The phase ifferences etween he wo requency pectra ver any given requency re displayed n the phase lot. The phase angle s the shift of the Y series elative o the X series. The amplitude f the phase ngle nd ts sign eveal he ime domain relationship etween he wo series nder tudy. Cross-spectral nalysis as arried ut on a number f climate proxies gainst n orbital ime series. Lithogenic rain size (LGS), Zr/Rb ratio, biogenic arium and ithogenic omponent (low resolution) rofiles were all analyzed, with the records truncated bove he thin turbidite t 198,000 years B.P. Hence all frequency omain nalyses over he ime eriod -195,000 ears B.P., and have been nterpolated o an equal ample nterval f 3000 years rom an average pacing f 3800 years. We are aware that n a frequency omain tudy f a record slightly) ess han 200 kyr long, he nherent rrors with respect o the 100 kyr frequency and are large, but the results re presented or completeness. Results Sediment Description Core CD53-30 consists of 7.62 m of foraminifer-bearing nannofossil oze with occasional rganic-rich aminations nd one thin (8 cm) sandy unit. Occasional races of burrowing occur. Distinct, yclic color changes re evident hroughout he core, rom light grey nterglacial ediment, o darker buff to brown lacial ediment. etween 17 and 525 cm depth here s a dark olive green, coarse quartz sand with shell fragments, grading p o dark olive green mud nd medium and, nd with a sharp asal oundary. his s nterpreted s a mass low deposit. Oxygen sotope tratigraphy nd Age Model The oxygen sotope stratigraphy or core CD53-30 was produced rom planktic Globorotalia nfiata) and benthic (Cibicidoides uellerstorfi) oraminifera nd correlated ith the orbitally uned, stacked ecord of Martinson et al. [1987]. Age control points at isotope events over the last 30,000 years were corrected o calendar years [Bard et al., 1990, 1993] in the following manner: 12.05 kyr was converted o 14.10 kyr, 17.85 kyr to 21.29 kyr, and 24.11 kyr to 28.30 kyr (Figure 2). Nineteen events were identified for correlation purposes Table 1). Where an event seemed o occur at differing depths n the planktic and benthic records, a feature often attributed to bioturbation, the depth aken was an average of the planktic and benthic values. Table 1 also shows sediment accumulation rates for CD53-30, which vary between 1.5 and 5.6 cm kyr •. All ages eported below are based on this sidereal timescale. As a result of the relatively coarse 10 cm) sampling nterval and the slow accumulation ates (Table 1), not every isotopic event dentified by Martinson et al. [1987] was recognized, ut the major erminations f the last 300,000 years are all recorded in the isotope ecord. The Younger Dryas event, identified at ODP Site 658 [Sarnthein and Tiedemann, 1989, 1990] from this general ocality (Figure 1), is possibly epresented n the benthic record, but not in the planktic record. Despite he lack of radiocarbon ates, we feel confident n referring o this event as he Younger Dryas as t also manifests tself in the lithogenic grain size, Zr/Rb, and high- resolution carbonate records. An anomalously ow 15•80 alue or event 3.3, at the beginning of Stage 3, is recorded in both records 0.4-0.5%0 less than expected), but is most convincing n the planktic record with a 15•aO, lmost as low as present-day alues 0.34%0). The very low value is constrained by only one datum point and the anomaly s not seen n the isotope ecords f other cores rom this region e.g., ODP Site 659 [Sarnthein nd Tiedernann, 989]), so this eature may result rom sediment eworking. Stage 6 displays a number of isotopic fluctuations n the planktic record which may be correlated with those by Martinson et al. [ 1987], but it is not possible o do so with a high degree of certainty. Most of Stage 7 is missing, at the location of a thin turbidite which probably roded he underlying ediment esulting n a hiatus of approximately 7 kyr. In the planktic ecord anomalously igh 15•aO alues (approximately .5%0 more than expected) occur near he end of Stage 8 (event 8.2) and have been confirmed y duplicate nalysis. The amplitude f this positive spike s not anomalous n the benthic ecord, but it lags he planktic ecord by approximately 000 years. This ag, and he very high values n the planktonic ecord also occur at Site 658 [Sarnthein nd Tiedemann, 989] and so could represent feature peculiar o this region, perhaps period of wind-induced urface water cooling (by approximately øC). Benthic vent 6.2 also ags he planktic record, his time by approximately 000 years, hough his could be an artefact of the relatively coarse sampling interval. Sarnthein and Tiedemann [1989] noted disagreements n the timing of glacial o interglacial hange etween heir planktic and benthic records for Terminations I and III, but in their results the benthic ecord eads he planktic by 4000 years. Sediment Geochemistry: Lithogenic We attempted o identify he parameters hich can indicate variations in the dust flux record or wind strength or a combination f both. Lithogenic grain size, the bulk sediment Zr/Rb ratio, aluminium lux, and carbonate ontent profiles or core CD53-30 are presented Figure 3).  MATTHEWSON ET AL.: A HIGH-RESOLUTION ARIDITY RECORD OF NORTH AFRICA 681 ½-.4 :;:':';;ili ':' "44'. • , ' 0 100 200 300 0 100 200 300 Age (kyr) Figure . The oxygen sotope ecords f Globorotalia nfiata nd Cibicidoides uellerstorfi n CD53-30 ompared to he stacked ecord f Martinson t al. [ 1987]; haded reas ndicate lacial sotope tages. The aeolian dust component of east Atlantic sediments increases uring glacials, and at <800 km from the continental source hows rain izes p o 125 gm and silt modes f 24 gm. Sediment ontaining oarser rains s deposited uring onditions of higher wind strength Sarnthein t al., 1981]. Work in the Indian Ocean Clemens nd Prell, 1990] and Pacific Ocean Rea and Janecek, 981 on he median article ize of the ithogenic fraction f the sediment upports his hypothesis. Lithogenic rain size LGS) n CD53-30 approximately 50 km due west f the African oast) aries etween pproximately gm in interglacials nd 10-11 gm at peak glacial tages. LGS also ncreases n the Stage /2 boundary the Younger ryas event) nd during nterglacial tage (substages b and 5d). The Zr/Rb ratio, can be used o indicate he proportion f heavy minerals present within the sediment because irconium occurs n the heavy mineral fraction, whereas ubidium is a
Similar documents
View more...
Related Search
We Need Your Support
Thank you for visiting our website and your interest in our free products and services. We are nonprofit website to share and download documents. To the running of this website, we need your help to support us.

Thanks to everyone for your continued support.

No, Thanks