You are on page 1of 15

Quaternary Science Reviews, Vol. 17, pp.

395409, 1998
( 1998 Elsevier Science Ltd.
Printed in Great Britain. All rights reserved.
PII: S02773791(97)00048-6 02773791/98, $19.00

SAHARAN WIND REGIMES TRACED BY THE SrNd ISOTOPIC


COMPOSITION OF SUBTROPICAL ATLANTIC SEDIMENTS:
LAST GLACIAL MAXIMUM vs TODAY

F.E. GROUSSET*, M. PARRA*, A. BORY-, P. MARTINEZ*, P. BERTRAND*,


G. SHIMMIELD- and R.M. ELLAM
* Departement Geologie et Oceanographie, UMR CNRS 5805, Universite Bordeaux I, Avenue des Facultes,
33405 Talence Cedex, France
(e-mail: grousset@geocean.u-bordeaux.fr.)
- Marine Geosciences Unit, Department of Geology and Geophysics, University of Edinburgh,
West Mains Road, Edinburgh EH9 3JW, U.K.
Isotope Geosciences Unit, Scottish Universities Research and Reactor Centre, East Kilbride G75 0QF, U.K.
AbstractNew NdSr isotopic data on the (30 lm lithic particles of surface and Last Glacial
Maximum sediments recovered along the African margin between the Equator and the
Gibraltar Strait are presented in combination with grain-size measurements. This (30 lm size
fraction allows us to eliminate any hemipelagic contribution that could occur in the coarser
fractions. In the eolian fraction, both Sr and Nd isotopic tracers reveal the same major
northwestern origin (Mauritania, Mali, southern Algeria and Morocco). The Archaean forma-
tions of the western Saharan shield could be the source of the very unradiogenic ratios observed
here. The more southern regions (Senegal, Guinea) act only as secondary sources. A similar
pattern is observed for the LGM. Lithic particles are mostly transported by both Trade and
Saharan Air Layer (SAL) winds, along an approximate NESW axis; this main feature matches
the southern plume, characterizing the dust transport observed during winter. No significant
latitudinal shift of the belt winds is observed between the LGM and today. At the LGM,
however, dust fluxes were 24 times higher than today, leading to a more Archaean-type
imprint in the deposits. We do not observe any clear relationship between the latitudinal
variability of the upwelling systems identified in this region at the LGM and the location of the
major wind systems. Both enhanced aridity on the continent and increased wind speed
probably occurred together over western tropical Africa during the Last Glacial period. ( 1998
Elsevier Science Ltd. All rights reserved

INTRODUCTION derived from the more southerly SaharanSahelian


region (Rognon et al., 1996), indicating transport by
It is well known that every year, millions of tons of
EW winds. Depending on their altitude, these EW
Saharan dust are transported over, and deposited in,
winds are called: continental trade winds, Saharan Air
the tropical Atlantic (Buat-Menard and Chesselet,
Layer (SAL) or African Easterly Jets. According to
1979; Kolla et al., 1979; Sarnthein et al., 1981; Ganor
dAlmeida (1986), they mostly originated from source
and Mamane, 1982; Duce et al., 1991), with the finest
regions located roughly at the same latitude in the
particles even reaching the Caribbean (Prospero, 1981).
Sahelian belt (Mali, Niger, southern Algeria and even
The study of dust accumulation in marine sediments
Chad). These two major wind systems are still active
leads to a better understanding of palaeo-atmospheric
to-day over this region (Schutz, 1980; Coude-Gaussen
circulation. Previous studies have demonstrated that
et al., 1987), their relative intensity being seasonally
Saharan dust is transported by two major wind sys-
modulated. They are at the origin of two major sea-
tems. Thus, the crust-derived fraction of the Holocene
sonal dust plumes over the ocean: one centered around
loess that accumulated over the Canary Islands was
20N during the summer, the other shifting southward
derived from Morocco (Grousset et al., 1992) and
around 10N during the winter (Kalu, 1979; Moulin
transported from the NE by the marine trade winds.
et al., 1997). The seasonal latitudinal shifts of about 10
By contrast, the crust-derived fraction of the Holocene
of the ITCZ (Inter Tropical Convergence Zone) from
loess accumulated over the Cape Verde Islands was
a summer northern location to a winter southern loca-
tion would control the location of these plumes. In
oceanic surface sediments, the distribution of lacustrine
Present address: Centre des Faibles Radioactivites, Laboratoire diatoms would reflect the transport of aerosols by
Mixte CNRS-CEA, Avenue de la Terrasse, 91198 Gif-sur-Yvette,
France.
these two major wind regimes (Pokras and Mix, 1985).
Present address: Dunstaffnage Marine Laboratory, P.O. Box 3., According to Kolla et al. (1979), who studied the
Oban, Argyll PA34 3AD, U.K. quartz distribution in sediment cores, the main dust

395
396 Quaternary Science Reviews: Volume 17

transport axis was centered around about 8N at the the aerosol samples were collected aboard research
LGM and has shifted northward since that period, as vessels, during oceanographic cruises conducted over
far north as about 18N in the late Holocene. Sarnthein the last two decades, off NW Africa between the Ca-
et al. (1981) studied the distribution of the terrigenous nary Islands and the Equator.
silt fraction (%'6 lm of the carbonate-and-opal-free) Loess, fine powdered dust (feshfesh) and silt sam-
in both surface and LGM levels. This fraction should ples were collected in the northwestern regions of the
reflect the input of African aerosols transported west- African continent from Morocco to as far south as
ward, mostly by the SAL winds, across the Atlantic Guinea, mostly in dune fields (ergs) and Sahelian pow-
ocean. However, these authors do not observe any dered dust accumulations. Although these deposits are
clear latitudinal shift of the patterns between these two drifted by winds over long distances, we assume that
periods, but do infer an increase, by a factor of 2, in the they are mainly derived from their underlying outcrops
silt particle flux at LGM, with the latitudinal location and thus they provide an average composition of the
remaining roughly the same. Thus, the ITCZ may not regional geological units. We also documented the
have shifted substantially between Glacial and Inter- areas that are considered as the main sources of the
glacial periods, and the observed difference would be present-day aerosols (dAlmeida, 1986; Chiapello, 1996;
the result of variable prevailing winds, NE trades car- Chiapello et al., 1997).
rying white quartz, chlorite and illite from the northern Marine sediment samples: we sampled the first cen-
region and EW SAL winds transporting iron-oxide- timeter of both the surface sediments (using multi- and
coated quartz and kaolinite from the more southern box-cores) and the LGM levels (using piston- and
Sahelian zone (Sarnthein and Koopmann, 1979). On karsten-corers) in cores recovered along the African
the other hand, Ruddiman (1997) recently suggested margin, between the Gibraltar Strait and the Equator.
that, as changes in monsoonally driven source-area Ages were derived from the organic carbon records
aridity had not left a strong dust-flux imprint in tropi- measured on all cores and calibrated against the d18O
cal Atlantic sediments in the late Pleistocene, source- records obtained on planktic foraminifera from three
area aridity could not be the main factor controlling of these cores (Bertrand et al., 1996; Martinez et al.,
dust fluxes. In this scenario, enhanced glacial influxes 1996). In the four BOFS cores, the LGM levels were
of dust would be more likely to the due to changes in derived from the d18O records obtained on planktic
transport by winter winds, rather than to glacial hypera- foraminifera (Matthewson et al., 1995). Due to biotur-
ridity. Furthermore, Rea (1994) suggested that aridity bation processes, surface sediments generally represent
was the dominating factor that controls the increase in a mixture of particles deposited over the last few centu-
the dust flux emitted by the desert areas at the LGM. ries and frequently the last two millennia. Thus, their
Clearly, the origin of LGM lithic particles, their isotopic composition will most likely reflect an integra-
glacial atmospheric context and their marine sedimen- tion of the historical period, rather than the present-
tary regimes are highly controversial issues. In this day situation. For both the surface and the LGM
context, we studied the (30 lm lithic particles (the levels, we sampled hemipelagic deposits after checking
carbonate-free fraction) of surface sediments (box- optically that there was no visible evidence of turbidi-
cores) and the LGM levels (piston-cores) recovered ties (hiatus, vertical grain-size gradient, laminae)
along the African margin between the Equator and the or of bioturbation (worm burrows). Nonetheless, we
Gibraltar Strait. This size fraction ((30 lm) allows us cannot rule out the possibility of a hardly visible diffuse
to eliminate any hemipelagic contribution that could bioturbation, mostly due to benthic foraminifera
occur in the coarser fractions. The purpose of studying movements (Bornmalm et al., 1997).
the NdSr isotopic composition and the grain-size
distribution of these lithic particles was to fingerprint
Methods
their origins and identify their distribution patterns for
both periods (LGM and today). These two approaches Grain-size distributions were measured on the lithic
might help to constrain better the atmospheric trans- fraction, after leaching the carbonate fraction, using
port patterns and strength, to identify possible latitudi- a Malvern 2600 E laser sizer, based on the near-for-
nal shifts that could characterize the contrast between ward scattering of a laser beam by particles suspended
glacial and interglacial regimes and thus to assess in water (McCave and Syvitski, 1991).
better the hypothesized temporal latitudinal shifts of Sufficient sample was weighed out to yield about
the Mauritanian upwelling system. 100 mg of alumino-silicate material after dissolution of
the calcium carbonates and crushed in a grinder. Car-
bonate was dissolved using mechanical agitation of
a pH sodium acetate solution buffered at a pH of 4.5.
SAMPLES AND METHODS This carbonate free fraction was then sieved at 30 lm
in order to analyze the (30 lm fraction. The carbon-
Samples ate-free material may include some opaline silica, but
Sample locations are reported in Fig. 1 and Table 1. since the concentrations of Sr and Nd in marine bio-
Aerosol samples: except for two samples collected genic opal are low (+5 ppm and +3 ppm, respective-
over the Canary Islands (dA18) and Niger (dNya), all ly, unpublished data), we do not consider this to be
TABLE 1. Sample location, Sr and Nd concentrations and isotopic ratios

Sample Name Location Nd 143Nd/144Nd E(Nd)o Sr 87Sr/86Sr Source*


(ppm) ($2sig]10~6) (ppm) ($2sig]10~6)

Possible source areas


Ouj Oujda Morocco 29.9 0.512031 (12) !11.8 111.6 0.716593 (22) 1
Elra El Rachidia Morocco 32.7 0.511938 (21) !13.6 87.2 0.726932 (08) 1
Foum Foum el Hassan Morocco 38.7 0.511761 (14) !17.1 90.3 0.728197 (09) 1
Smara Essmarra Morocco 40.2 0.511798 (16) !16.3 104.6 0.734041 (18) 1
Zouerat Zouerat Mauritania 38.0 0.511725 (12) !17.8 91.6 0.735679 (37) 1
Atar Atar Mauritania 3.5 0.511717 (12) !17.9 23.1 0.727284 (19) 1
Erg Erg sud Atar Mauritania 3.2 0.511943 (07) !13.5 78.1 0.737645 (17) 1
Elm El Mroiti Mauritania 35.5 0.511784 (12) !16.6 94.1 0.737535 (16) 1
Tich Tichit Mauritania 31.4 0.511856 (08) !15.2 78.2 0.731493 (11) 1
Kif Kiffa Mauritania 31.5 0.511923 (07) !13.9 77.8 0.728389 (16) 1
Nou Nouakchott Mauritania 55.9 0.511760 (09) !17.1 126.5 0.720021 (10) 1
Tim Timimoun Algeria 21.1 0.726253 (13) 1
MEK-21 Sebkra Mekkerane Algeria 164.7 0.720515 (16) 1
MEK-58 Sebkra Mekkerane Algeria 168.6 0.724395 (35) 1
ATK-35 Atakor Algeria 39.7 0.512015 (08) !12.1 1
Tas Tassendjanet Algeria 0.511247 !27.1 0.713 2
Kay Kayes Mali 30.9 0.511855 (11) !15.2 66.9 0.736840 (10) 1
Tamba Tambacounda Senegal 31.4 0.511894 (08) !14.5 74.0 0.718296 (14) 1
Est Est Dakar Senegal 3.2 0.511807 (13) !16.2 10.6 0.726326 (11) 1
Foun Foundiougne Senegal 28.4 0.511901 (10) !14.3 127.5 0.714707 (18) 1
Ro continental shelf Senegal 41.9 0.511894 (11) !14.5 97.5 0.715296 (20) 1
Sen Riv Senegal river Senegal 19.2 0.511838 (31) !13.1 47.7 0.728581 (42) 1
Til Tillaberi Niger 29.0 0.511999 (12) !12.4 87.1 0.723832 (28) 1
Lab Labbe Guinea 22.0 0.512013 (08) !12.1 37.5 0.723603 (43) 1
F.E. Grousset et al.: Saharan wind regimes

Ivo Ivory Coast 0.511493 !23.2 0.7146 2


Aerosols Latitude Longitude
A17 Canaries Islands +28N +04W 33.9 0.511949 (10) !13.4 95.7 0.725225 (16) 1
SL off St. Louis (Senegal) +17N +1830W 13.9 0.511890 (14) !14.6 154.3 0.713499 (30) 1
A9- Cape Verde Basin 1114N +26W 35.6 0.512062 (17) !11.2 140.9 0.721871 (09) 1
Nya Niamey (Niger) 14N +0230E 32.0 0.512123 (11) !10.0 90.5 0.714660 (09) 1
A18 Sierra Leone Basin 813N +1330W 37.9 0.511897 (12) !14.4 147.3 0.721116 (14) 1
A4- Cape Verde Basin 03N +2930W 38.6 0.511980 (40) !12.8 127.7 0.720972 (46) 1
n4 Corsica 42N +09E 23.2 0.512059 !11.3 128.6 0.72071 3
n10 Corsica 42N +09E 25.9 0.511959 !13.2 92.5 0.72451 3
n17 Corsica 42N +09E 33.4 0.511913 !14.1 128.1 0.71925 3
n27 Corsica 42N +09E 31.3 0.511890 !14.6 116.0 0.72426 3
n59 Corsica 42N +09E 26.3 0.512025 !12.0 110.2 0.72227 3
397
398
TABLE 1. (continued)

Sample Cruise Latitude Longitude Depth Nd 143Nd/144Nd E(Nd)o Sr 87Sr/86Sr Source*


(m) (ppm) ($2sig]10~6) (ppm) ($2sig]10~6)

Surface sediments
K23 Sedorqua 3420N 0701W 730 35.3 0.512037 (08) !11.7 102.2 0.719567 (19) 4
K25 Sedorqua 3224N 1025W 3165 33.0 0.511969 (10) !13.0 112.5 0.724544 (19) 4
K22 Sedorqua 2659N 1413W 1245 142.2 0.719791 (10) 1
K17d Sedorqua 2653N 1441W 2597 165.3 0.718151 (09) 1
K20b Sedorqua 2502N 1639W 1445 32.2 0.511911 (10) !15.3 200.0 0.720657 (11) 1
28 BOFS 2434N 2249W 4855 30.7 0.511947 (07) !13.4 131.0 0.720137 (19) 4
K15 Sedorqua 2344N 1716W 1000 27.4 0.511899 (09) !14.4 150.6 0.720385 (12) 1
32 BOFS 2231N 2200W 4560 29.4 0.511940 (10) !13.6 142.4 0.719635 (66) 4
K09 Sedorqua 2120N 1815W 2002 24.5 0.511876 (10) !14.8 137.7 0.720191 (08) 1
29 BOFS 2034N 2107W 4005 27.1 0.511892 (06) !14.6 133.1 0.720367 (19) 4
CV2 Dialante 2000N 3000W 5000 32.1 0.512024 (15) !12.0 121.0 0.720419 (27) 5
K02 Sedorqua 1929N 1717W 1407 21.9 0.511901 (58) !14.3 125.0 0.721866 (27) 1
31 BOFS 1900N 2010W 3300 27.4 0.511940 (10) !13.6 138.6 0.719497 (53) 4
VM22-196 L-DEO (Vema cores) 1350N 1857W 3728 32.4 0.511832 (19) !15.7 138.7 0.718021 (18) 1
VM27-175 L-DEO (Vema cores) 0848N 2212W 4548 34.6 0.512733 (11) #1.9 476.8 0.704635 (16) 1
VM22-189 L-DEO (Vema cores) 0455N 2107W 2525 40.0 0.511684 (12) !18.6 124.5 0.719184 (13) 1
VM30-41K L-DEO (Vema cores) 0013N 2303W 3874 26.6 0.512058 (14) !11.3 111.7 0.719789 (11) 1
Glacial sediments (LGM)
K17d Sedorqua 2653N 1441W 2597 205.3 0.716317 (25) 1
K20b Sedorqua 2502N 1639W 1445 24.8 0.511853 (28) !14.1 122.9 0.719334 (28) 1
28 BOFS 2434N 2249W 4855 28.5 0.512001 (08) !12.4 126.7 0.720436 (17) 4
K15 Sedorqua 2344N 1716W 1000 36.9 0.511634 (09) !19.5 187.3 0.719040 (50) 1
Quaternary Science Reviews: Volume 17

32 BOFS 2231N 2200W 4560 27.0 0.511964 (08) !13.1 104.0 0.720932 (33) 4
K11 Sedorqua 2129N 1757W 1200 30.7 0.511720 (11) !17.9 199.6 0.716994 (39) 1
29 BOFS 2034N 2107W 4005 28.2 0.511933 (11) !13.7 168.5 0.717435 (34) 4
K02 Sedorqua 1929N 1717W 1407 32.1 0.511723 (11) !17.8 113.1 0.723080 (44) 1
31 BOFS 1900N 2010W 3300 26.3 0.511957 (07) !13.3 158.7 0.717439 (38) 4
VM22-196 L-DEO (Vema cores) 1350N 1857W 3728 27.8 0.511697 (09) !18.3 79.1 0.719912 (13) 1
VM27-175 L-DEO (Vema cores) 0848N 2212W 4548 39.2 0.511819 (08) !15.9 145.1 0.718480 (73) 1
VM22-189 L-DEO (Vema cores) 0455N 2107W 2525 35.1 0.511928 (09) !13.8 112.1 0.718046 (11) 1
VM30-41K L-DEO (Vema cores) 0013N 2303W 3874 25.4 0.512053 (09) !11.4 126.2 0.717470 (07) 1

* 1"This work; 2"Allegre and Ben Othman (1982); 3"Colin (1993); 4"Bory (1997); 5"Grousset and Biscaye (1989).
-
In a previous paper we had already analyzed the bulk fraction of these samples (Grousset et al., 1988).

Cores from DGO (Sedorqua and Dialante cruises).

R/V Charles Darwin cruise 53 BOFS Leg C (U.K.).


F.E. Grousset et al.: Saharan wind regimes 399

for individual analyses on each mass spectrometer. The


measured 87Sr/86Sr and 143Nd/144Nd ratios were cor-
rected for mass fractionation by normalizing to 86Sr/
88Sr"0.1194 and 146Nd/144Nd"0.7219, respective-
ly. Blanks were measured (Sr(10~9 g and Nd(
10~10 g) and are considered to be negligible in all cases.
For convenience, 143Nd/144Nd ratios are normalized
and reported as follows: e (o)"((143Nd/144Nd/
N$
0.512636)!1)]104, in which 0.512636 is the Chon-
dritic Uniform Reservoir value (Jacobsen and Wasser-
burg, 1980). The spike contribution to the isotope
ratios for samples was corrected using an iterative
calculation. Strontium standard NIST 987 was meas-
ured 6 times at Toulouse with an average 87Sr/86Sr"
0.710205 ($0.00002), and 14 times at SURRC with an
average 87Sr/86Sr"0.710214 ($0.00002) vs the certi-
fied value of 0.710245. Neodymium standard LaJolla
was analyzed at Toulouse 5 times with an average
143Nd/144Nd"0.511846 ($0.000015) vs the certified
value of 0.511865. This standard was analyzed at
SURRC 20 times during instrument installation, when
it gave 0.511851 ($0.000012). Neodymium standard
Johnson-Matthey was analyzed at SURRC 25 times
with an average 143Nd/144Nd"0.511504 ($0.00002)
vs the long-term mean value of 0.511503 ($0.000016).
Differences between the analyzed and certified values
FIG. 1. Sample location. Symbols: Possible source areas (solid are within the range of the uncertainty of the sample
circles), marine cores (squares), aerosols (crossed open squares). measurements.
Along with our sediment cores (solid squares), are some cores from
Dasch (1969) and Grousset et al. (1988) (open squares).
Possible Effect of Grain-size on Isotopic Compositions

a serious problem. We will refer to this carbonate-free Dasch (1969) reported increasing 87Sr/86Sr ratios
fraction as the lithic fraction. This fraction may some- with decreasing grain-size in carbonate-free sediment
times also contain a minor authigenic component, e.g. samples. A grain-size effect on Sr-isotope composition
FeMnoxides, that has a negligible effect on the iso- must therefore be eliminated or at least considered.
topic composition of the lithic fraction (Grousset and This possibility has already been discussed for the
Biscaye, 1989). &20 mg aliquots of carbonate-free Saharan-derived aerosols (Grousset et al., 1992;
samples (and &150 mg for the BOFS cores) were Rognon et al., 1996). We minimized this potential
dissolved with a mixture of teflon-distilled [HF# grain-size effect by studying the same grain-size frac-
HClO #HNO ]. Sr and Nd were chemically sepa- tion ((30 lm) in all samples. The choice of this size
4 3 range is justified below (Results section). However, for
rated through ionic chromatographic columns, in
a class-1000 clean laboratory. We followed chemical the BOFS samples (see Table 1), we analyzed the car-
and mass spectrometer techniques previously des- bonate-free bulk samples: nonetheless we do not sus-
cribed (Grousset et al., 1988; Grousset and Biscaye, pect any significant bias in these four samples as they
1989; Fitton et al., 1997). TIMS analysis of the samples were collected far away from the continent ('500 km
was done on two different mass spectrometers: analysis off the coasts), thus precluding any significant input of
of Sr and Nd in the BOFS cores (Table 1) was done on coarse ('30 lm) particles, making the sediments
a VG Sector 54-30 multi-detector mass spectrometer at themselves effectively (30 lm.
Scottish Universities Research and Reactor Centre By contrast, possible grain-size effects on Nd iso-
(SURRC) [U.K.]; analysis of Sr and Nd on the other topic compositions are of little concern because the Nd
samples was done at the Universite de Toulouse isotope ratios are almost not affected by grain size
[France] on a Finnigan MAT 261 multi-detector mass variations, either in riverine sediments (Goldstein et al.,
spectrometer. At Toulouse, Sr was mounted as nitrate 1984) or in aerosols and loess (Grousset et al., 1992).
on a W filament and Nd as nitrate on a Re-filament; at
SURRC, Sr was mounted as nitrate onto Ta filament
RESULTS
and Nd as nitrate onto TaReTa triple filaments. All
the measurements, on both mass spectrometers, were
Grain-size Distributions
made using multiple Faraday cups. We have intercali-
brated the results of both mass spectrometers with The grain-size distributions of source samples,
standards and the results are within the range of errors aerosol and marine sediment samples reveal three
400 Quaternary Science Reviews: Volume 17

40% of '63 lm particles at the LGM, and less than


20% today; such a high value at the LGM could be
explained by an advected hemipelagic contribution.
Indeed, at the LGM the low sea-level stand was
located at about !120 m (Fairbridge, 1961) and, thus,
dune fields invaded the narrow continental shelf and
might have directly fed the slope with coarse particles.
This is confirmed by the fact that the size of '80% of
the particles from both the Saharan dune fields and the
sediments from the Senegal shelf, range from &60 to
&100 lm.
For this reason, in order to fingerprint only the
wind-borne particles directly derived from the atmo-
sphere, isotopic analyses have been conducted only on
the (30 lm grain-size fraction. Moreover, most of the
dust particles collected over this oceanic region are
smaller than &30 lm (Beltagy et al., 1972; Schutz,
1980; Coude-Gaussen, 1989; Wefer and Fischer, 1993;
Ratmeyer et al., 1997), which is in agreement with the
fact that in the surface sediments more than 80% of the
particles are smaller than 30 lm (mode centered
around 20 lm).

Isotopic Results
Sr/ 86Sr ratios and eNd (o) distributions
87

In considering the surface sediments, we have aug-


mented the new 87Sr/86Sr ratios measured in this
study, with previously published Sr isotope data
FIG. 2. Grain-size distribution of the carbonate-free fraction of
some characteristic samples: uni-modal aerosols and sand dune fields (Dasch, 1969; Grousset and Biscaye, 1989). In the detri-
(a), present-day (c) and LGM (b) marine sediments; some rare LGM tal fraction of the surface sediments, 87Sr/86Sr ratios
samples display a bi-modal distribution, explained by a mixture of range from &0.715 to &0.725 (with an average of
aerosols and hemipelagic deposits, advected from the continental 0.7194 for 25 values) and are thus rather homogeneous
shelf (d). if compared to the values observed in the entire Atlan-
tic Ocean: &0.7033 to &0.7450 (Biscaye et al., sub-
mitted). Moreover, there is no obvious geographic
major features (Fig. 2). A first family (Fig. 2a) displays 87Sr/86Sr zonation. Over the same region, aerosol Sr
a uni-modal distribution, with more than 80% of the isotope ratios define a similar range (&0.7135 to
particles ranging between &70 and &90 lm. This &0.7253). Data from the literature (Biscaye et al.,
sandy population is very well sorted, is characteristic of 1974; Grousset and Biscaye, 1989), even though they
the Saharan sand dune fields and of the continental also fall within this range, are not directly comparable
shelf deposits, and is very different from the aerosol because they were obtained on bulk aerosols and are
distribution in which the mode is centered around thus modified by the carbonate fraction.
&20 lm. A second family also displays a uni-modal The geographic distribution of e (o) in the detrital
N$
distribution, with a mode centered around &20 fraction of surface sediments is displayed in Fig. 3a.
25 lm (Fig. 2bc). In some samples, this mode shifts Previous data from the literature obtained on both
slightly toward the coarser diameters, up to &30 sediments (Grousset et al., 1988) and aerosols (Gold-
40 lm. The fine particles ((10 lm) account for less stein et al., 1984; Grousset et al., 1988) are also included.
than 20%. These silty populations characterize most of The general pattern displayed in Fig. 3a corresponds
the marine sediments. A third bi-modal family (Fig. 2d) roughly to that obtained previously, albeit at much
characterizes a few LGM marine sediments from the lower resolution (Grousset et al., 1988). Sixteen e (o)
N$
continental slope of Mauritania, the two previous values range from !11.3 to !18.6 (with an average
modes being superimposed (e.g. LGM levels in cores of !13.8). Over the same region, present-day aerosols
K11 and K20). In this last group of samples, we suspect display a narrower range (!12.8 to !14.6). These
that such coarse particles (coarse mode) could not have values are much less radiogenic than the nearby Medi-
been introduced by atmospheric transport, but more terranean Sea sediments, !2 to !10 (Frost et al.,
likely reflect the addition of particles advected by 1986), that are significantly influenced by the local
gravitational processes from the upper slope or the active volcanism. A similar local volcanic influence is
external shelf. In core K20 (Table 1), we observe about observed in core V27-175 (e (o)"#1.9; 87Sr/86Sr
N$
F.E. Grousset et al.: Saharan wind regimes 401

FIG. 3. (a) Present-day distribution of e (o). (b) Distribution of e (o) at the LGM. Symbols: Possible source areas (solid circles), marine cores
N$ N$
(data from the literature: open squares; our data: solid squares), aerosols (crossed squares), Senegal river (solid diamond). Isoline !20 is drawn
according to data from Chardon (1996) and Potrel et al. (1996).

&0.7046), in which some volcanogenic particles were


observed, probably inherited from the nearby Cape
Verde volcanic islands through turbiditic currents.
This sample cannot therefore be representative of
eolian material.
Although their contribution is probably negligible in
this region, we have also to consider local rivers as
possible sources of lithic particles. Potential riverine
sources include the minor local Senegal River, which
enters the Atlantic at the latitude of the Cape Verde
Islands (e (o)"!13.1), and the remote Niger River,
N$
which has e (o) of about !10.5 (Goldstein et al.,
N$
1984). Fluvial particles could be advected by oceanic
bottom currents flowing westward towards the region
studied here (Biscaye, 1965) and thereby influence the
sediment samples.
Finally, in the detrital fraction of the LGM sedi-
ments, 87Sr/86Sr ratios (Table 1) range from &0.717
to &0.723 (with an average of 0.72 for 14 values). The
e (o) values of the LGM sediments (Fig. 3b) range
N$ FIG. 4. 87Sr/86Sr ratios plotted against 1000/Sr of surface and last-
from !11.4 down to !19.5 (with an average of !15),
thereby revealing a significantly less radiogenic Nd glacial maximum (LGM) sediments and both literature values and
measured values of (30 lm fractions of aerosol, loess and desertic
than for the surface sediments. samples from possible source areas. Symbols: possible source areas
(solid circles; and open diamond for sample named Tim in Table 1),
87 marine surface sediments (solid squares), marine LGM sediments
Sr/ 86Sr ratios vs Sr concentrations (open squares), aerosols (crossed squares), Archaean Rocks (data
In considering the Sr isotope data, it is useful to plot from Potrel et al., 1996; Chardon, 1996) (solid triangles), Precam-
brian granitoids (data from Allegre and Ben Othman, 1980) (reversed
87Sr/86Sr vs 1000/Sr (Fig. 4). Inverting the concentra- solid triangles), sea-water (black cross), Cape Verde basalts (open
tion is avantageous because mixtures between two circle), Senegal river (solid diamond). For end-member A: see
end-members are constrained to lie upon straight lines Results section.
402 Quaternary Science Reviews: Volume 17

(instead of hyperbolae), making the identification of II, along with dust and loess samples from Senegal and
binary mixing straight forward. On this plot, the data from coastal Mauritania (Nouakchott, named Nou
define two distinct fields (called Field I and Field II in on Table 1) and one aerosol collected offshore Saint-
Fig. 4). Both fields have a common end-member (low Louis-Senegal (SL).
87Sr/86Sr ratios and high Sr concentration) labelled A As most of the LGM sediment compositions match
in Fig. 4. Field I links end-member A to the most the Senegal-and-coastal-Mauritania (Nouakchott) trend
radiogenic Sr-rich source samples (Morocco, Maurita- (Field II), we could consider that they derived from this
nia and Mali). Field II links A to more southerly southern area. However, they display more radiogenic
source samples (Senegal, Guinea). Oceanic surface 87Sr/86Sr ratios and also unradiogenic e (o) values
N$
sediments, as well as most of the aerosols, fall into Field when compared to these sources.
I. LGM sediments located north of the Cape Verde
islands also fall into Field I, but a few LGM sediments
located between Cape Verde islands and the Equator,
together with one aerosol collected off Saint-Louis DISCUSSION
(Senegal), tend to shift toward Field II.
Are bottom sediments reliable imprints of the eolian
input?
Sr/ 86Sr ratios plotted against eNd (o)
87
From the data obtained with sediment traps moored
In this kind of diagram (Fig. 5), mixtures obtained by at 20N and 20W, Wefer and Fischer (1993) concluded
mixing particles from two end-members, define hyper- that high eolian inputs coincide with enhanced biolo-
bolae, the shape of which are determined by the Sr and gical productivity and thus may favour high and rapid
Nd concentrations and isotopic compositions of the particle sedimentation in the boreal summer off Cap
two end-members (Langmuir et al., 1978; Faure, 1986). Blanc. Thus, the particle transfer rate from the atmo-
The sediment samples plot within (calculated) hyper- sphere down to the surface sediments should be short
bolic envelopes encompassing the northwestern enough to avoid any strong advection by currents or
source-areas (Field I: Morocco, Mali, Mauritania) and any significant chemical exchange during their short
the southern source-areas (Field II: Senegal), respec- residence time in the water column. Taking into ac-
tively. These envelopes are hyperbolic mixing curves count both the present-day horizontal speed of the
calculated by mixing a volcanic end member and a de- watermasses at about 20N and the residence time of
sert-derived dust component. Almost all the surface the particles in the water column, a possible advection
sediment and aerosol samples plot in Field I, clearly of these particles could not exceed 3 degrees in latitude
reflecting their northwestern origin. On the other (Jeandel, pers. com.; Siegel et al., 1990). This assump-
hand, most of the LGM sediment samples plot in Field tion is confirmed by other studies which demonstrated

FIG. 5. 87Sr/86Sr vs e (o) of surface and LGM sediments and both literature values and measured values of (30 lm fractions of aerosol, loess
N$
and desertic samples from possible source areas. Symbols as in Fig. 4.
F.E. Grousset et al.: Saharan wind regimes 403

that the mineralogical composition of surface sedi- the nature of which has to be discussed. This end-
ments reflected the aerosol composition (Blank et al., member A is very close in isotopic composition to
1985). Two other studies of the vertical evolution sea-water. Sea-water might be a suitable end-member if
of e (o) in the tropical Atlantic water column, also we had kept the carbonate fraction, but since we
N$
revealed that the surface sediment composition reflec- leached our samples specifically to remove the sea-
ted the aerosol composition (Jeandel et al., 1995; water signature in the carbonate fraction, this is not an
Tachikawa et al., 1997; Bory, 1997), as aerosols were appealling possibility. Another possibility could be
transferred mainly as large particles, which sink rapidly a volcanogenic end-member, but A also displays
through the water column. Thus, we consider that the a radiogenic value (&0.709). The only possibility is
isotopic composition of the surface sediments closely provided by plotting the field of values obtained for the
reflects the isotopic composition of the present-day Archaean rocks (Chardon, 1996; Potrel et al., 1996),
aerosols. This assumption is verified by the fact that which goes exactly across point A.
e (o) values of aerosols correspond with the geo- Now, if we agree with the fact that the main source of
N$
graphic distribution of the surface sediment e (o) dust is located in the northen part of the western
N$
values (Fig. 3a). As far as the 87Sr/86Sr ratios are Sahara (western Algeria, ex-Spanish Sahara, Southern
concerned, it has also been demonstrated that the Morocco), we have to explain why we observe an
aerosol imprint was retained in the underlying sedi- apparent NESW distribution of the dust inherited
ments (Biscaye et al., 1974). For the LGM samples, it from these regions. North of &20N, such a NESW
could be argued that their 87Sr/86Sr ratios may be oriented gradient could be partly explained by advec-
affected by post-depositional exchange with interstitial tion in the NESW Canaria surface current (Melia,
water. However, the possibility of Sr mobility in the 1984); south of &20N, it requires, however, a simi-
lithic fraction is challenged by the observations of larly oriented dominant wind regime.
Perry and Turekian (1974), who showed that Sr iso- Thus, if we look at the present-day airmass trajecto-
topes do not begin to move and equilibrate between ries responsible for the transport of present-day dust,
phases in P conditions corresponding to less than we observe that the dominant pathway is a NESW
thousands of metres of burial. direction over both the continent (Chiapello et al.,
1997) and the tropical Atlantic (Kalu, 1979; Moulin
et al., 1997), matching the feature described above for
Present-day dust: origin, transport trajectories, fluxes
e (o) (Fig. 3a). Moreover, for many authors, one of the
N$
The average e (o) value in the surface sediment of most important sources of dust in northwest Africa is
N$
the North Atlantic Ocean is about &!11 (Goldstein precisely located in the northwestern Archaean pro-
et al., 1984; Grousset et al., 1988). This is typically the vince (region dA1 in dAlmedia, 1986; sector d3 in
range of values that we observe at the boundaries of the Chiapello et al., 1997). Such a northwestern Saharan
studied area, off the Gibraltar Strait and in the Equa- origin matches precisely the contours of the nothwes-
tor region (Fig. 3a). Between these two regions, charac- tern Archaean province described above. According to
terized by the Atlantic background values, e (o) Chiapello et al. (1997), dust events of northwestern
N$
values become less radiogenic, with the highest values and western Saharan origins (southwestern Algeria,
ranging between !14 and !18, between the Cape ex-Spanish Sahara, southern Morocco) are the most
Verde Islands and the African coasts. Such high values frequent nowadays. This dominant regime, typically
require a relatively unradiogenic source on the conti- a trade pattern, would explain why such a NESW
nent. Such values are found in the region stretching elongated feature is recorded in the oceanic surface
from Nouakchott to the western tip of Algeria, which sediments.
corresponds to Archaean outcrops of northeastern
Mauritania (the Reguibat Rise (Potrel et al., 1996)) and
LGM dust: origin, transport trajectories, fluxes
to surrounding Palaeozoic formations. In this pro-
vince, Archaean rocks can display e (o) as low as The e (o) of the LGM sediments are also less
N$ N$
!54, with an average of about !35 (Chardon, 1996; radiogenic than the so-called Atlantic background
Potrel et al., 1996). In the superimposed dune fields (!11), ranging from &!13 to &!20. Paradoxi-
accumulated over these formations (samples Nou, cally, they display the same geographic pattern as the
Atar, Zouerat on Fig. 1), e (o) values range from one observed for the surface sediments, with the same
N$
&!17 to &!18, which reflect the unradiogenic major NESW negative gradient, roughly located at
composition of their source rocks. This region is clearly the same latitudes.
the source of unradiogenic Nd which controls the NE- Sarnthein et al. (1981) did not observe any clear
SW 143Nd/144Nd zonation observed in the oceanic northward shift of the patterns between these two
surface sediments (Fig. 3). Moreover, present-day aero- periods, but more likely an increase by a factor 2 in the
sols derived from this region are enriched in illite- silt particle flux at LGM, the latitudinal location re-
micas, a clay-mineral which is abundant in these maining roughly the same. We have already observed
Archaean rocks (Chamley, 1988). a similar increase in core K20 (Table 1; Bertrand et al.,
In Fig. 4, we identified a second end-member called 1996) and even in the remote Cape Verde abyssal plain
A (low 87Sr/86Sr ratios and high Sr concentration), (30W; depth: 5000 m, CV2 in Fig. 1) (Grousset et al.,
404 Quaternary Science Reviews: Volume 17

1989). Kolla et al. (1979) studied the temporal evolu- (e (o)&!11), which largely originates from the peri-
N$
tion of quartz (%), in a core located &300 km off the Atlantic rivers (Goldstein et al., 1984) and is mostly
Senegal coast (core V22-196, Fig. 1). They observed composed of (10 lm particles (Biscaye, 1965). Off the
about 40% of quartz in the two last glacial maxima Mauritanian coasts there are more than 40% of clay
(stages 2 and 6), compared to 1020% during the minerals in the surface sediments and more than 60%
interglacial periods. In terms of the flux of quartz, the south of 15N (Johnson, 1979). Indeed, a simple mixing
LGM fluxes were &3 to &6 times higher than today. calculation demonstrates that a doubling of the dust
In the more southerly cores (as far south as the Equa- flux can induce the observed e (o) shift (from !14 to
N$
tor), mean LGM dust fluxes are &2 to &3 times !18). An argument confirms this explanation: the
higher than today (Franiois et al., 1990; Ruddiman, most negative value (e (o)"!18.6) is observed at
N$
1997). Such a factor, 23, does not characterize solely about 6N (core VM22-189; Fig. 1) on the top of
the LGM, but is characteristic of at least the last a seamount, less affected by possible advections, thus
8 climatic cycles in this area (Parkin and Shackleton, recording a purer eolian signal. It must be pointed out
1973). It is interesting to point out that a similar pat- that, when we look at the ratio of the LGM to present
tern was observed in another core also located in the interglacial fluxes (Ruddiman, 1997), we observe the
tropical latitudes, but in the eastern Pacific Ocean same NESW elongation axis, roughly linking
(Boyle, 1983). Was it possible that advection processes Mauritania to the northeastern tip of Brazil. Thus, this
could have artificially increased apparent fluxes? Al- LGM NESW direction is reflected by both qualitative
though most of the material must have initially been (e (o)) and quantitative (terrigenous fluxes) parameters.
N$
delivered by winds, there was then possibly local-scale Finally, overlapping this NESW trade dust signal,
redistribution on the sea-floor. Thus, flux increase a SAL dust signal can be recognized at more than
evaluations should account for sea-floor redistribution about 250 km offshore, west of +21W, south of 23N
(Ruddiman, pers. com.). (Fig. 3b). Elongated along an EW axis, the e (o)
N$
In terms of Sr isotopic composition, most of the values range from !12 to !15, reflecting a more
LGM and present-day samples are very similar, both Sahelian origin (i.e. Field II, in Fig. 5). This SAL signal
lying within Field I (Fig. 4). Only the three southern is deposited precisely where previous studies recog-
samples (south of 10N) display slightly different values nized it in the grain-size (Sarnthein et al., 1981) and
(shifting through Field II): they are thus more in- clay mineral (Lange, 1982) features.
fluenced by the SenegalGuinea province. When we
compare present-day surface and LGM samples on the
LGM vs today
e (o) vs 87Sr/86Sr plot (Fig. 5), most of the LGM
N$
values tend to shift through Field II, toward the bot- According to Ruddiman (1997), the winter dust
tom-left part of the plot (Fig. 5), where a possible third plume would have been increased at the LGM com-
end-member might exist. The SenegalGuinea pro- pared to nowadays and this change, involving wind
vince thus cannot be the unique source for these depo- changes, could have been induced: 1) by a southward
sits. A possible candidate for such an end-member, shift; 2) or by an increase in aridity and/or by increas-
characterized by both unradiogenic Sr and Nd isopic ing the deflation efficiency or 3) by increasing their
compositions, could be the old Archaean (Potrel et al., velocity. Are there arguments favoring one of these
1996; Chardon, 1996) and Proterozoic (Allegre and possible explanations?
Ben Othman, 1980; Ben Othman et al., 1984) forma-
tions of the Saharan shield.
A southward shift in transport trajectories
Samples with even lower negative e (o) values are
N$
observed along the margin over &200 km from the According to Kolla et al. (1979) who studied the
coasts. For example, along the longitude 18W, e (o) quartz distribution in sediment cores, the main dust
N$
ranges between !15 and !20 at LGM, and between transport axis was centered around &8N at the LGM
!14 and !15 at the present day. This less radiogenic and shifted northward since that period, as far north as
LGM pattern could reflect an increased contribution &18N in the late Holocene. Such a shift, however, is
of the Archaean-derived source region at the LGM. neither reflected by our isotopic data nor by Ruddi-
Such an increase would require increased weathering/ mans accumulation rate maps (1997), nor by the
erosion processes on the continent, but it is difficult to '6 lm coarse particle distribution (Sarnthein et al.,
imagine that it would preferentially affect the Archaean 1981).
outcrops and not the surrounding giant Paleozoic What do we know about wind trajectories? The
formations. Moreover, it is difficult to admit that it present-day wind regime has largely been described in
would preferentially affect the most unradiogenic previous works (Kalu, 1979; Sarnthein et al., 1981;
(87Sr/86Sr(0.71) outcrops (end-member A suggested Chamley, 1988; Coude-Gaussen, 1989; Rognon et al.,
on Fig. 4). 1996). Nowadays, the most intense events come from
A more plausible explanation can be proposed. In the more southern Sahelian source and occur only
the LGM sediments, lower (e (o) down to !19.5) during the coldest period (December and January)
N$
could be explained by an increased dust input, which (Chiapello et al., 1997; Ratmeyer et al., 1997). They are
would then prevail upon the Atlantic background induced by the so-called continental trade winds and
F.E. Grousset et al.: Saharan wind regimes 405

are considered to be responsible for most of the dust On the other hand, a few arguments against aridity
transport to the ocean (Lezine et al., 1995). This last have to be considered. If aridity had been the only
hypothesis is in agreement with higher influx of wind- reason for the dust flux increase, both wind regimes
blown pollens, interpreted as indicating stronger trade (northeast trades and African easterly winds) would
winds (Melia, 1984; Hooghiemstra, 1989). On the other have transported more dust particles and we would
hand, during the summer period, the African easterly observe the imprints of both the northern summer
jets reaching their maximum strength in the summer, plume and the southern winter plume (as described
would be responsible for the transport of the main in Kalu, 1979). Pokras and Mix (1985) observed
pollen release from southern Sahara and Sahel zones that LGM sediments from both plumes were highly
(Hooghiemstra, 1989), which is in agreement with the enriched in the freshwater diatom Melosira, when
increase in '6 lm coarse particles reported by Sarn- compared to the lower Holocene (up to a factor 10).
thein et al. (1981). At the LGM, due to the extension of Such an enrichment could reflect an enhanced aridity
the ice-sheets in the Northern Atlantic, the subsequent on the continent, as already suggested by Diester-Haas
cooling (Rind, 1987) and the southward shift of the (1976). However, their abundance is very similar
westerlies (Rognon and Coude-Gaussen, 1996), the nowadays and at the LGM. Finally, if aridity was
wind regimes characterizing northwest Africa could predominant, time-series might display a precession
have been intensified significantly (Sarnthein et al., variability (23 ky) and if northern glaciations were
1981; Ruddiman, 1997). The LGM increase in eolian the dominant controlling factor, time-series might
sedimentation occurred in both winterspring and be more likely to display an obliquity variability
summer dust plumes (Balsam et al., 1995); this would (41 ky). Matthewson et al., (1995) who studied a core
explain why we do not identify any clear latitudinal from the Cape Verde Terrace (at roughly 20N20W)
shift. Finally, atmospheric global circulation models observed only the precession frequency by carrying
(AGCMs) do not simulate any shift between the LGM out a cross spectral analysis on the Zr/Rb ratio,
and today, but more likely a wind speed strengthening a grain-size proxy, and concluded that aridity was the
(Joussaume, 1993). governing factor. Thus, considering these drawbacks,
we have to envisage other simultaneous forcing para-
meters.
An increase in source-area aridity?
According to Rea (1994), more dust would reflect
An alternative (but not mutually exclusive) explanation
more aridity, but coarser grains would reflect stronger
could be an increase in the wind velocity
winds. Thus, at the LGM the dust flux increase could
be the sole result of an aridity increase. There are some Increased winds should transport coarser particles
arguments favoring this hypothesis. At the LGM, dune (Rea, 1994). However, the grain-size distributions of
fields were more extended in latitude over the Saha- our samples (Fig. 2) do not show any strong grain-size
ranSahel area (Sarnthein, 1978), which could be the increase at the LGM: most of the present-day and
result of increased aridity. The most efficient sources of LGM samples display an eolian mode between 10 and
fine particles are, however, more likely to be located in 40 lm. This is typically what has been found in a series
the semi-desertic regions, surrounding the main deserts of sediment traps moored at &500 km off the Mauri-
(Pye, 1987; Rea, 1994). Over the last 30 years, the tanian coasts, in which the present-day lithogenic frac-
increase in dustiness over West Africa was caused by tion reaches more than 50% of the total flux and is
a decrease in rainfall (NTchayi et al., 1994). The sub- largely composed of quartz grains of &1050 lm size
sequent decline in vegetation facilitates soil erosion (Wefer and Fischer, 1993). At the LGM the coarsest
and thus air dustiness. An increase in the deflation sediments are certainly observed between 10 and 20N.
efficiency might have contributed too (Marticorena, For example, in core VM22-196 (Fig. 1), the LGM
1995). This possibility has been suggested by Balsam mode ranges from &20 to &60 lm (see Fig. 2b).
et al. (1995), who observed a higher iron oxide content Although Saharan sands and aerosols from central
in sediments at the LGM, which could reflect an in- Sahara would display a uni-modal distribution cen-
creased deflation of the Sahara and Sahel dust during tered around &4050 lm (Schutz, 1980), we have
glacial time. Similarly, it has been proposed that varia- observed much coarser grains (around &60100 lm)
tions in Chinese dust flux to the North Pacific could in the southern Algeria sand dunes and the Senegal
reflect to some degree variations in the intensity of continental shelf (Fig. 2a). Increased winds could have
vertical dust lifting events (Pye and Zhou, 1989). Un- brought these coarse particles to the ocean. We cannot
fortunately, we do not have relevant data to address be sure, however, that this coarse mode reflects only an
this problem seriously. If we follow Rea (1994), more increased wind-speed, as these cores are the closest to
dust would reflect more aridity. Indeed, increased dust the margin and thus could be suspected of contamina-
fluxes were observed in the LGM marine sediments at tion by hemipelagic inputs (Fabre, 1995)
different latitudes: a factor 4 for quartz at &25N Increased winds should also transport heavier par-
(Thiede et al., 1982), a factor 3 for Al at &20N ticles. In cores K02 and CV2 (Fig. 1), we have observed
(Matthewson et al., 1995), thus supporting the aridity a significant increase in the Ti/Al ratios in the LGM
enhancement hypothesis. levels (&0.06) compared to the present-day ratios
406 Quaternary Science Reviews: Volume 17

(&0.04), this increase being correlated with an increase Finally, arguments advanced for increased glacial
in the coarse fraction (Bertrand et al., 1996). Volcanic aridity are not very strong, but the lack of much change
rocks from western Sahara and the lateritic formation in grain-sizes is an argument against increased wind
of the Sahelian zone are particularly enriched in ilme- strength. Thus, a last possible explanation could be
nite (Dumon, 1977). Titanium is known to be preferen- that the season of generally strong winds lengthened,
tially concentrated in the coarse fraction, as heavy carrying more dust of about the same size. This would
minerals (ilmenite density &4.72). Thus, at the LGM, mean that there was an increase, not only in wind
these coarse particles were much heavier than today. strength, but also in glacial winter regime duration
Therefore, a high Ti/Al ratio could reflect intensified (Ruddiman, pers. com.).
wind transport, as previously suggested for dust fluxes
(Boyle, 1983). Good correlations between Ti/Al and the
aluminosilicate fluxes, and even with the quartz con- Implications for upwellings
tent (Kolla et al., 1979) have already been attributed to It has been suggested that in the southern Ocean,
both more arid source regions and increased intensity aerosols could have a fertilizing influence on the sur-
(Boyle, 1983). It is interesting to point out that Boyle face waters (Martin, 1990) and could have enhanced
(1983) studied a core located in the same latitudinal the biological productivity by a factor +5 at the
band, but off the Peru coast: this coherence could LGM in the Southern Atlantic (Kumar et al., 1995).
therefore reflect synchronous climate variations in The possible influence of the dust input has not been
different oceans. investigated so far in this area. From the data they
Another wind-speed proxy can be considered: the obtained with sediment traps moored at 20N and
Zr/Rb ratio indicates an increase in grain-size (Mat- 20W, Wefer and Fischer (1993) concluded that in the
thewson et al., 1995) and grain density at the LGM, boreal summer, high eolian inputs coincided with en-
and thus could be considered as a wind velocity proxy. hanced biological productivity. The idea of increased
Thus, the wind speed had to be significantly intensified fertilization by windblown nutrients has also been en-
at LGM. Finally, the southwesterly imprint of the dust visaged by Ruddiman (1997) in the tropical eastern
fallout in both glacial and interglacial periods identi- Atlantic. On the other hand, the high productivity
fied in this study could, however, be presumably caused characterizing this region can be also due to the strong
by stronger (Ruddiman, 1997) or longer (Boyle, 1983) upwelling activity. In the hypothesis of Lezine et al.
winter transport. Indeed, geographical patterns dis- (1995), the NESW marine trade winds would pre-
played on Fig. 1 roughly fit that previously observed dominantly generate coastal upwellings. At the LGM,
for aerosols collected exclusively in March and April African aerosols transported westward (mostly by the
(Grousset et al., 1988), and with the winter and spring SAL winds) across the Atlantic ocean, could have
patterns simulations produced in model (Moulin et al., strongly influenced the intensity of biological produc-
1997). Due to the fact that one-centimetre thick sedi- tion in the upwelling zone (Sarnthein et al., 1981),
ment core samples integrate a few centuries, it may not triggering a strong enhancement of the carbon fluxes
be possible, however, to identify any short term varia- (Muller and Suess, 1979). However, such a systematic
tion, either the seasonal waning and waxing of the enhancement of carbon fluxes at the LGM has to be
ITCZ or any decadal trends of dust exportation such reconsidered, as in some cores from the Mauritanian
as the one driven nowadays by the North Atlantic margin (e.g. K11), we observed a decrease in carbon
oscillation (Moulin et al., 1997). fluxes at the LGM (Bertrand et al., 1996; Martinez
This last explanation, strong winter winds, is the et al., 1996). Thus, we cannot easily link wind and
hypothesis which best explains most of the data. How- upwelling patterns. The dust distribution that we ob-
ever, as airborne particle fluxes were also much higher served (Fig. 3) could, however, be explained by both
than today, an increase in the aridification of the processes. Further work needs to be done if we are
source area cannot be ruled out and we must conclude to address fully the short scale variability of this
that both increases in source-area aridity and in wind upwelling system.
velocity occurred together during the LGM. On a glo-
bal scale, this conclusion is consistent with the fact that
LGM dust fluxes are enriched by factors 10-to-20 in
CONCLUSIONS
Greenland and Antarctic ice-cores (Thompson and
Mosley-Thompson, 1981; Petit et al., 1981; Biscaye Along the NW African margin, the eolian input of
et al., 1997), which implies not only a general aridifica- the oceanic sediments is mostly concentrated in the
tion of the desert areas, but also a more intense atmo- 1040 lm grain-size range. In the eolian fraction of the
spheric circulation (Joussaume, 1993). Our data allow sediments deposited between the Canary Islands and
us to propose that both processes were enhanced the equator, both Sr and Nd isotopic tracers reveal the
simultaneously at the LGM, which is an adequate way same major northwestern origin (Mauritania, Mali,
of reconciling the increased aridity advocates (Kolla southern Algeria and Morocco). The Archaean forma-
et al., 1979; Rea, 1994; Matthewson et al., 1996) with tions of the western Saharan shield could supply the
the increased wind strength advocates (Sarnthein et al., very unradiogenic Nd ratios characterizing this frac-
1981; Boyle, 1983; Ruddiman, 1997). tion off Africa. The more southern regions (Senegal,
F.E. Grousset et al.: Saharan wind regimes 407

Guinea) act only as secondary sources. A similar pat- Ruddiman and M. Sarnthein for their very helpful and
tern is observed for the Last Glacial Maximum. pertinent reviews, C. Jeandel for fruitful discussions
Nowadays, the isotopic composition of oceanic sedi- and A. Gonialves who polished up the English lan-
ments reveals that lithic particles are transported by guage. Aloys Borys research was funded by the Euro-
both Trade and SAL winds, along an average NESW pean Commission as part of a Community Training
axis, linking roughly Dakar to the northeastern tip of Project. This work was mostly funded by the French
Brazil. The same main transport direction has also Centre National de la Recherche Scientifique.
been identified by looking at airmass trajectories
(Chiapello, 1996). This main feature matches the south-
ern plume, characterizing the dust transport observed
REFERENCES
during winter.
No significant latitudinal shift of the wind belts is
Allegre, C.J. and Ben Othmann, D. (1980) NdSr isotopic relation-
observed between the LGM and today; the sole major ship in granitoid rocks and continental crust development: a
difference between these two periods is that, at the chemical approach to orogenesis. Nature, 286, 335342.
LGM, the dust fluxes were 24 times higher than dAlmeida, G.A. (1986) On the variability of desert aerosol radiative
today, leading to a more Archaean-type imprint in the characteristics. Journal of Climatic Applied Meteorology, 24, 903916.
deposits. Thus, if the winds are the major upwelling Balsam, W.L., Otto-Bliesner, B.L. and Deaton, B.C. (1995) Modern
and last glacial maximum eolian sedimentation patterns in the
driving force, the latitudinal variability of the upwell- Atlantic Ocean interpreted from sediment iron oxide content.
ing systems identified in this region at the LGM Paleoceanography, 10, 493507.
(Bertrand et al., 1996), cannot be explained by simple Behairy, A.K., Chester, R., Griffiths, A.J., Johnson, L.R. and Stoner,
latitudinal shifts of the major wind systems. J.H. (1975) The clay mineralogy of particulate material from
Many arguments (isotopic tracers, geographic pat- some surface sea water of the eastern Atlantic ocean. Marine
Geology, 18, 4556.
terns, dust fluxes, grain-size distribution, particle den- Beltagy, A.I., Chester, R. and Padgham, R.C. (1972) The particle-size
sity proxies) tend to demonstrate that, at the LGM, distribution of quartz in North Atlantic deep-sea sediments.
both enhanced aridity on the continent and increased Marine Geology, 13, 297310.
wind speed occurred together over the western tropical Ben Othmann, D., Polve, M. and Allegre, C.J. (1984) NdSr isotopic
Africa and confirm that transports were enhanced composition of granulites and constraints on the evolution of the
lower continental crust. Nature, 307, 510515.
mostly during winter, as recently proposed by Ruddi- Bergametti, G., Dutot, A.L., Buat-Menard, P., Losno, R. and
man (1997). Remoudaki E. (1989) Seasonal variability of the elemental com-
position of atmospheric aerosol particles over the northwestern
Mediterranean. ellus, 41(B), 353361.
Bertrand, P., Shimmield, G., Martinez, P., Grousset, F.E., Jorissen, F.,
ACKNOWLEDGEMENTS Paterne, M., Pujol, C., Bouloubassi, I., Buat-Menard, P., Pe-
ypouquet, J.P., Beaufort, L., Sicre, M.A., Lallier-Vergez, E., Foster,
We thank the officers and crew of the the R/V Le J.M., Ternois, Y. and the other participants of the Sedorqua Pro-
Suroit who helped us to recover sediment cores during gram (1996) The glacial ocean productivity hypothesis: the import-
the Sedorqua cruise (March 1994). We thank all the ance of regional and temporal studies. Marine Geology, 130, 19.
colleagues who provided us with aerosols (R. Chester, Biscaye, P.E. (1965) Mineralogy and sedimentation of recent deep-
sea clay in the Atlantic Ocean and adjacent seas and oceans.
J. Prospero), loess (G. Coude-Gaussen, P. Rognon), Geological Society American Bulletin, 76, 803832.
powdered soils (G. Bergametti, G. Berliet, T. Biscaye, Biscaye, P.E., Chesselet, R. and Prospero, J. (1974) Rb-Sr, 87Sr/86Sr
G. Coude-Gaussen, P. Rognon) or marine sediment isotope system as an index of the provenance of continental dust
samples (G. Shimmield, I.N. McCave) from various in the open Atlantic Ocean. Journal de Recherches Atmosphe& ri-
potential source areas of northwestern Africa and the ques, 8, 819829.
Biscaye, P.E., Grousset, F.E., Revel, M., Van der Gaast, S., Zielinsky,
L-DEO Deep-Sea Sample Repository, and its curator, G.A., Vaars, A. and Kukla, G. (1997) Asian provenance of glacial
Rusty Lotti, who kindly provided us with sediment dust (Stage 2) in the GISP2 ice core, Summit (Greenland). Journal
samples from the 0-to-15N latitudinal band (support of Geophysical Research, 102, C12, 26,76526,781.
for the curating facilities of the L-DEO Deep-Sea Biscaye, P.E., Grousset, F.E., Dasch, J. and Huon, S. (1998) Rb-Sr
Sample Repository is provided by the NSF through isotope system as a tracer of provenance in marine sediments and
aerosols: Atlantic ocean. Marine Geology (in press).
grant OCE94-02150 and the Office of Naval Research Blank, M., Leinen, M. and Prospero, J.M. (1985) Major Asian
through grant N00014-96-I-0186). Special thanks to aeolian inputs indicated by the mineralogy of aerosols and sedi-
G. Berliet, who collected soil samples at each stage ments in the western North Pacific. Nature, 314, 8486.
bivouac, during the 1996 Paris-Dakar car rally! Bornmalm, L., Corliss, B.H. and Tedesko, K. (1997) Laboratory obser-
Thanks are due to A. Matthewson (Univ. of Edin- vations of rates and patterns of movement of continental margin
benthic foraminifera. Marine Micropaleontology, 29, 175184.
burgh, U.K.) who provided the d18O stratigraphy on Bory, A. (1997) Etude de limpact de laerosol desertique sur la
BOFS cores. We are grateful to P. Pedemay and to A. colonne deau dans lAtlantique subtropical Nord-Est. Doctorate
Kelly and V. Gallagher, for their help in the Sr-Nd Thesis Dissertation, Universite Paris VII, 261 pp.
chemistry carried out at University Bordeaux Boyle, E.A. (1983) Chemical accumulation variations under the Peru
I (France) and at the SURRC, respectively. We are current during the past 130,000 years. Journal of Geophysical
Research, 88, 76677680.
also grateful to M. Roy-Barman for access to the Buat-Menard, P. and Chesselet, R. (1979) Variable influence of the
mass spectrometer at the University Paul Sabatier atmospheric flux on the trace metal chemistry of oceanic sus-
(Toulouse, France). We thank P. Buat-Menard, W. pended matter. Earth and Planetary Science etters, 42, 399411.
408 Quaternary Science Reviews: Volume 17

Chamley, H. (1988) Contribution eolienne a la sedimentation marine Grousset, F.E., Biscaye, P.E., Zindler, A., Prospero, J. and Chester, R.
au large du Sahara. Bulletin de la Socie& te& Ge& ologique de France, (1988) Neodymium isotopes as tracers in marine sediments and
8 (IV-6), 10911100. aerosols: North Atlantic. Earth and Planetary Science etters, 87,
Chamley, H., Coude-Gaussen, G., Debrabant, P. and Rognon, P. 367378.
(1987) Contribution des aerosols a la sedimentation Quaternaire Grousset, F.E. and Biscaye, P.E. (1989) Nd and Sr isotopes as tracers
de lle de Fuerteventura (Canaries). Bulletin de la Socie& te& of wind transport: Atlantic aerosols and surface sediments. In:
Ge& ologique de France, 8 (III-5), 939952. Leinen, M. and Sarnthein, M. (eds.), Paleoclimatology and
Chardon, D. (1996) Les deformations continentales archeennes. Paleometeorology, NATO ASI Series, G, 282, 385400.
Exemples naturels et modelisation thermomecanique. Doctorate Grousset, F., Buat-Menard, P., Boust, D., Tian, R., Baudel, S., Pujol,
Thesis Dissertation, Universite Rennes 1, 257p. C. and Vergnaud-Grazzini, C. (1989) Temporal changes of eolian
Chiapello, I. (1996) Les aerosols atmospheriques au dessus de Saharan input in the Cape Verde abyssal plain since the last
lAtlantique nord: approche physico-chimique et meteoro- glacial period. Oceanologica Acta, 12, 177185.
logique. Doctorate Thesis Dissertation, Universite Paris VII, Grousset, F.E., Rognon, P., Coude-Gaussen, G. and Pedemay, P.
93p. (1992) Origins of peri-Saharan dust deposits traced by their Nd
Chiapello, I., Bergametti, G., Chatenet, B., Bousquet, P., Dulac, F. and Sr isotopic composition. Palaeogeography, Palaeoclimatology,
and Santos Suarez, E. (1997) Origins of African dust transported Palaeoecology, 93, 203212.
over the north-eastern tropical Atlantic. Journal of Geophysical Hooghiemstra, H. (1989) Variations of the NW African trade wind
Research, 102, D12, 13,70113,709. regime during the last 140,000 years: changes in pollen influx
Colin, C. (1993) Etude isotopique des retombees atmospheriques evidenced by marine sediment records. In: Leinen, M. and Sar-
totales en Mediterranee occidentale. DEA Diploma, Universite nthein, M. (eds.), Paleoclimatology and Paleometeorology, NATO
Paris XI, 98p. ASI Series, G, 282, 733770.
Coude-Gaussen, G. (1989) Local proximal and distal Saharan dust: Jacobsen, S.B. and Wasserburg, G.J. (1980) Sm-Nd isotopic evolu-
characterization and contribution to the sedimentation. In: tion of chondrites. Earth and Planetary Science etters, 50,
M. Leinen and M. Sarnthein (eds), Paleoclimatology and 139155.
Paleometeorology, NATO ASI Series, G, 282, 339358. Jeandel, C., Bishop, J.K.B. and Zindler, A. (1995) Exchange of
Coude-Gaussen, G., Rognon, P., Bergametti, G., Gomes, L., Strauss, Nd and its isotopes between seawater and small and large par-
B. and Gros, J.M. (1987) Saharan dust on the Fuerteventura ticles in the Sargasso Sea. Geochimica Cosmochimica Acta, 59,
island (Canaries). Chemical and mineralogical characteristics, 535547.
airmass trajectories and probable sources. Journal of Geophysical Joussaume, S. (1993) Paleoclimatic tracers: An investigation using an
Research, 92, 97539771. atmospheric general circulation model under ice age condi-
Dasch, E.J. (1969) Strontium isotopes in weathering profiles, deep- tions. Journal of Geophysical Research, 98(D2), 27672805, 1993.
sea sediments and sedimentary rocks. Geochimica Cosmochimica Kalu, A.E. (1979) The African Dust Plume: its characteristics and
Acta, 33, 15211552. propagation across West Africa in winter. In: Morales (ed.),
Diester-Hass, L. (1976) Late Quaternary climate variations in north Saharan Dust: Mobilisation, ransport, Deposition, SCOPE 14-C,
west Africa deduced from East Atlantic sediments cores. Quater- pp. 95118, Wiley, Chichester.
nary Research, 6, 299314. Kolla, V., Biscaye, P.E. and Hanley, A.F. (1979) Distribution of
Duce, R.A., Liss P.S., Merill, J.T., Atlas, E.L., Buat-Menard, P., Hicks quartz in late Quaternary Atlantic sediments in relation to
B.B., Miller, J.M., Prospero, J.M., Arimoto, R., Church, T.M., climate. Quaternary Research, 11, 261277.
Ellis W., Galloway, J.N., Hansen, L., Jickells T.D., Knapp, A.H., Kumar, N., Anderson, R.F., Mortlock, R.A., Froelich, P.N., Kubik,
Reinhardt K.H., Schneider B., Soudine, A., Tokos, J.J., Tsunogai, P., Dittrich-Hannen, B. and Suter, M. (1995) Increased biological
S., Wollast, R. and Zhou, M. (1991) The atmospheric input of productivity and export production in the glacial Southern
trace species to the world ocean? Global Biogeochemical Cycles, 5, Ocean. Nature, 378, 675680.
139259. Lange, H. (1982) Distribution of chlorite and kaolinite in eastern
Dumon, J.C. (1977) Recherche de lorigine des mineraux titanes des Atlantic sediments off North Africa. Sedimentology, 29, 427431.
plages senegalaises. Bulletin de lInstitut de Ge& ologie du Bassin Langmuir, C., Vocke, R., Hanson, G.N. and Hart, S.R. (1978) A
dAquitaine, 21, 207231. general mixing equation with applications to Icelandic basalts,
Fabre S. (1995) Caracterisation geochimique et granulometrique des Earth and Planetary Science etters, 37, 380392.
paleo-apports eoliens dans la zone de lupwelling de Mauritanie. Lezine, A.M., Leroux, M., Turon, J.L., Buchet, G. and Tastet, J.P.
DEA Diploma, Univ. Bordeaux I, pp. 33. (1995) Transport pollinique et circulation atmospherique au
Fairbridge, R.W. (1961) Eustatic changes in sea level, Physics and large de lAfrique tropicale occidentale au cours de la derniere
Chemistry of the Earth, 4, 99186. deglaciation. Bulletin de la Socie& te& Ge& ologique de France, 166,
Faure, G. (1986) Principles of Isotope Geology. Wiley, New York. 247257.
Fitton, J.G., Hardarson, B.S., Ellam, R.M. and Rogers, G. (1997) Sr-, McCave, I.N. and Syvitski, J.P.N. (1991) Principles and methods of
Nd-, and Pb-isotopic composition of volcanic rocks from the geological particle size analysis. In: Syvitski, J.P.M. (ed.), Prin-
southeast-Greenland margin, at 63N: temporal variation in ciples, Methods and Applications of Particle Size Analysis, pp.
crustal contamination during continental break-up. In: College 321. Cambridge University Press, Cambridge.
Station TX (ed.), Proceedings of the ODP Scientific Results, Leg McCulloch, M.T. and Wasserburg, G.J. (1978) Sm-Nd and Rb-Sr
152 (in press). chronology of continental crust formation, Science, 200, 1003
Franiois, R., Bacon, M. and Suman, D.O. (1990) 230Th profiling in 1011.
deep-sea sediments: high resolution records of flux and dissolu- Marticorena, B. (1995) Modelisation de la production daerosols
tion of carbonate in the equatorial Atlantic during the last 24,000 desertiques en regions arides et semi-arides: developpement et
years. Paleoceanography, 5, 761787. validation dun code de calcul adapte au transport a grande
Frost, C.D., ONions, R.K. and Goldstein, S.L. (1986) Mass bal- echelle. Doctorate Thesis Dissertation, Universite Paris VII,
ance for Nd in the Mediterranean Sea Chemical Geology, 55, 268p.
4550. Martin, J.H. (1990) Glacialinterglacial CO change: the iron hy-
2
Ganor, E. and Mamane, Y. (1982) Transport of Saharan dust across pothesis. Paleoceanography, 5, 113.
the Eastern Mediterranean. Atmospheric Environment, 16(3), Martinez, P., Bertrand, P., Bouloubassi, I., Bareille, G., Vautravers,
581587. B., Grousset, F.E., Shimmield, G., Guichard, S., Ternois, Y. and
Goldstein, S.L., ONions, R.K. and Hamilton, P.J. (1984) A Sm-Nd Sicre, M.-A. (1996) An integrated view of inorganic and organic
isotopic study of dusts and particulates from major river systems, biogeochemical indicators of paleoproductivity changes in a
Earth and Planetary Science etters, 70, 221236. coastal upwelling area. Organic Geochemistry, 24, 411420.
F.E. Grousset et al.: Saharan wind regimes 409

Matthewson, A.P., Shimmield, G.B., Kroon, D. and Fallick, A.E. Rind, D. (1987) Components of the Ice Age circulation. Journal of
(1995) A 300 kyr high-resolution aridity record of the North Geophysical Research, 92, 42414281.
African continent. Paleoceanography, 10, 677692. Rognon, P., Coude-Gaussen, G., Revel, M., Grousset, F.E. and
Melia, M.B. (1984) The distribution and relationship between Pedemay, P. (1996) Holocene Saharan dust deposition on the
palynomorphs in aerosols and deep-sea sediments off the coasts Cape Verde islands: sedimentological and Nd-Sr isotopic argu-
of northwest Africa. Marine Geology, 58, 345371. ments. Sedimentology, 43, 359366.
Moulin, C., Lambert, C.E., Dulac, F. and Dayan, U. (1997) Control Rognon, P. and Coude-Gaussen, G. (1996) Paleoclimates off North-
of atmospheric export of dust from North Africa by the North west Africa (2835N) about 18,000 yr B.P. based on continental
Atlantic Oscillation. Nature, 387, 691694. eolian deposits. Quaternary Research, 46, 118126.
Muller, P.J. and Suess, E. (1979) Productivity, sedimentation rate, Ruddiman, W.F. (1997) Tropical Atlantic terrigenous fluxes since
and sedimentary organic matter in the oceans. I. Organic carbon 25,000 yrs B.P. Marine Geology, 136, 189207.
preservation. Deep-Sea Research, 26, 13471362. Sarnthein, M. (1978) Sand deserts during glacial maximum and
NTchayi, G.M., Bertrand, J., Legrand, M. and Baudet, J. (1994) climatic optimum. Nature, 271, 4346.
Temporal and spatial varaitions of the atmospheric dust loading Sarnthein, M. and Koopmann, B. (1979) Late Quaternary deep-
throughout west Africa over the last thirty years. Annalae Geo- sea record on northwest African dust supply and wind cir-
physicae, 12, 265273. culation. In: Balkema, A. (ed.), Paleoecology of Africa, 12,
Parkin, D.W. and Shackleton, N.J. (1973) Trade wind and tempera- 239253.
ture correlation down a deep sea core off the Saharan coast. Sarnthein, M., Tetzlaff, G., Koopmann, B., Wolter, K. and Pflau-
Nature, 245, 455457. mann, U. (1981) Glacial and interglacial wind regimes over the
Perry, E.A. and Turekian, K.K. (1974) The effects of diagenesis on the eastern subtropical Atlantic and north-west Africa. Nature, 293,
distribution of Sr isotopes in shales. Geochimica Cosmochimica 193196.
Acta, 38, 929935. Sarnthein, M., Thiede, J., Pflaumann, U., Erlenkeuser, H., Futterer,
Petit, J.-R., Briat, M. and Royer, A. (1981) Ice age aerosol content D., Koopmann, B., Lange, H. and Seibold, E. (1982) Atmospheric
from East Antarctic ice core samples and past wind strength, and oceanic patterns of northwest Africa during the past 25
Nature, 293, 391394. million years. In: Von Rad, U. Linz, K. Sarnthein, M. and
Pokras, E.M. and Mix, A.C. (1985) Evidence for spatial variability of Seibold, E. (eds.), Geology of the Northwest African Continental
late Quaternary climates in tropical Africa. Quaternary Research, Margin, 545604.
24, 137149. Schutz, L. (1980) Long-range transport of desert dust with special
Porter, S.C. and An, Z. (1995) Correlation between climate events in emphasis on the Sahara. Annals of the New ork Academy of
the North Atlantic and China during the last glaciation, Nature, Science, 338, 515532.
375, 305308. Siegel, D.A., Granata, T.C., Michaels, A.F. and Dickey, T.M. (1990)
Potrel, A., Peucat J.J., Mark Fanning, C., Auvray, B., Burg, J.P. and Mesoscale eddy diffusion, particle sinking, and the interpretation
Caruba, C. (1996) 3.5 Ga old terranes in the west African Craton, of sediment trap data. Journal of Geophysical Research, 95, C4,
Mauritania. Journal of the Geological Society of ondon, 153, 53055311.
507510. Tachikawa, K., Jeandel, C. and Dupre, B. (1997) Distribution of
Prospero, J. (1981) Eolian transport to the world ocean. In: Emiliani, rare-earth elements and neodymium isotopes in settling partic-
C. (ed.), he Oceanic ithosphere: the Sea, Vol. 7, pp. 801874. ulate material of the tropical Atlantic Ocean (EUMELI site).
Wiley, Chichester. Deep-Sea Research I, 44, 17691792.
Pye, K. (1987) Aeolian Dust and Dust Deposits. Academic Press, San Thiede, B., Suess, E. and Muller, P.J. (1982) Late Quaternary
Diego USA, 334p. fluxes of major sediment components to the sea floor at
Pye, K. and Zhou, L.P. (1989) Late Pleistocene and Holocene aeolian the northwest African continental slope. In: von Rad, U.
dust deposition in North China and the northwest Pacific Ocean. Hinz, K. Sarnthein, M. and Seibold, E. (eds.), Geology of
Palaeogeography, Palaeoclimatology, Palaeoecology, 73, 1123. the Northwest African Margin, Vol. 25, 605631. Springer,
Ratmeyer, V., Fischer, G. and Wefer, G. (1997) Aeolian dust input Berlin.
and vertical particle transport recorded with sediment traps in Thompson, L.G. and Mosley-Thompson, E. (1981) Microparticle
the deep ocean off NW Africa. European nion Geology Meeting, concentration variations linked with climatic change: Evidence
Strasbourg, March 97, 604. from Polar ice. Science, 212, 812815.
Rea, D.K. (1994) The paleoclimatic record provided by eolian depo- Wefer, G. and Fischer, G. (1993) Seasonal patterns of vertical particle
sition in the deep sea: the geologic history of wind. Review of flux in equatorial and coastal upwelling areas of eastern Atlantic.
Geophysics, 32, 159195. Deep-Sea Research, 40, 16131645.

You might also like