PALEOCEANOGRAPHY, VOL. 19, PA3012, doi:10.

1029/2004PA001006, 2004

Iron control of past productivity in the coastal upwelling system

off the Atacama Desert, Chile
Laurent Dezileau,1,2 Osvaldo Ulloa,1 Dierk Hebbeln,3 Frank Lamy,3
Jean-Louis Reyss,4 and Michel Fontugne4
Received 12 January 2004; revised 25 May 2004; accepted 25 June 2004; published 16 September 2004.

[1] Biogenic opal and organic carbon vertical rain rates in sediment cores reveal a strong cyclicity in the
productivity of the upwelling system off presently arid northern Chile during the last 100,000 years.
Changes in productivity are found to be in phase with the precessional cycle (
20,000 years) and with
inputs of iron from the continent. During austral summer insolation maxima, increased precipitation and
river runoff in the region appear to have brought high inputs of iron, mainly from the Andes, to the coastal
ocean enhancing primary productivity there. We interpret our results as providing evidence for iron control
of past productivity in this upwelling system and for a tight link between productivity and orbital forcing at
midlatitudes. INDEX TERMS: 4267 Oceanography: General: Paleoceanography; 4825 Oceanography: Biological and Chemical:
Geochemistry; 4860 Oceanography: Biological and Chemical: Radioactivity and radioisotopes; KEYWORDS: paleoproductivity,
thorium-230, iron, coastal upwelling, biogenic opal, organic carbon
Citation: Dezileau, L., O. Ulloa, D. Hebbeln, F. Lamy, J.-L. Reyss, and M. Fontugne (2004), Iron control of past productivity in the
coastal upwelling system off the Atacama Desert, Chile, Paleoceanography, 19, PA3012, doi:10.1029/2004PA001006.

1. Introduction ability of macro nutrients [Arrhenius, 1952; Berger et al.,

1989; Sarnthein et al., 1988]. However, recent experi-
[2] Since the hypothesis was established that variations
ments [Hutchins and Bruland, 1998; Hutchins et al.,
of marine biological productivity might be one of the
2002] have demonstrated that in two major coastal
mechanisms responsible for glacial/interglacial pCO2
upwelling systems, the California and the Peru-Chile
changes [Broecker, 1982] much effort has been put into
ecosystems, phytoplankton growth can also be limited
the reconstruction of paleoproductivity. Most of the
by iron, a micronutrient which has additional sources to
studies have been performed on sediments underlying
upwelling deep waters. In these regions where riverine
high nutrient-low chlorophyll (HNLC) regions and the
and atmospheric inputs are negligible, and where the
high productivity areas of the world’s oceans: the South-
continental shelf is very narrow, iron limitation is a major
ern Ocean, the equatorial upwelling regions and the
constraint on phytoplankton growth. Despite accumulating
eastern boundary current systems.
evidence for iron limitation in these modern coastal
[3] Eastern boundary currents are systems of enhanced
upwelling systems, paleoceanographic studies have not
biological productivity that contribute a significant frac-
yet revealed evidence for a link between past changes of
tion of the global primary production and burial of
iron availability and biological productivity. Here we
organic carbon [Brink and Cowles, 1991; Longhurst et
report on biogenic opal, organic carbon and iron accu-
al., 1995; Summerhayes et al., 1995]. There wind-driven
mulation rates (corrected for sediment redistribution by
coastal upwelling brings subsurface waters rich in major 230
Th-normalization) obtained from sediment cores
nutrients into the photic zone where they are capable of
retrieved from the upwelling system off presently arid
fertilizing the otherwise nutrient-poor surface waters. Past
northern Chile and we provide evidence for iron control
changes in the productivity of these regions have com-
of past productivity in this upwelling system.
monly been explained in terms of changes in the intensity
of upwelling-favorable winds, and therefore in the avail-
2. Study Area, Material, and Method
[4] We have analyzed variations of ocean productivity
1
Centro de Investigación Oceanográfica, Programa Regional de through time in the upwelling system of the eastern South
Oceanografia Fisica y Clima, Departamento de Oceanografı́a, Universidad Pacific, as recorded in sediments from two gravity cores
de Concepción, Concepción, Chile.
2
Now at Laboratoire de Dynamique de la Lithosphère, Université de
(GeoB 7101-1 and GeoB 3375-1, Figure 1) retrieved at the
Montpellier 2, Montpellier, France. same location from the continental slope off northern Chile
3
Fachbereich Geowissenschaften, Universität Bremen, Bremen, (27
280S, 71
150W, 1956 and 1947 m water depth respec-
Germany. tively). This perennial upwelling system is characterized by
4
Laboratoire des Sciences du Climat et de l’Environnement, CNRS/
Commissariat à l’Energie Atomique, Gif-sur-Yvette, France.
an extremely narrow shelf, only 10 to 15 km wide on
average. Its surface waters present relatively high concen-
Copyright 2004 by the American Geophysical Union. trations of nitrate and phosphate, low concentrations of
0883-8305/04/2004PA001006$12.00 silicate and phytoplankton pigment (Figure 1), as well as

PA3012 1 of 15
PA3012 DEZILEAU ET AL.: PAST PRODUCTIVITY OFF CHILE PA3012

Figure 1. Composite of annual average of phytoplankton pigment concentration (Chl) of the eastern
South Pacific and annual maximum normalized difference vegetation index (NDVI) of the central and
southern Andean region (both for year 2000). These data were derived from the SeaWiFS sensor
(SeaWiFS Project) Distributed Active Archive Center. Top left inset: map of the study area with location
of the two cores GeoB 3375-1 and GeoB 7101-1 (27
280S, 71
150W). Bottom left inset: chlorophyll and
nutrient profiles of the upper 100 m depth at station GeoB 3377-1 (27
280S; 71
31.50W). Right inset:
average of annual continental rainfall data in Chile (http://www.meteochile.cl). See color version of this
figure at back of this issue.

a relatively low primary production [Daneri et al., 2000], all sediments was measured by weighting a fixed sample
characteristics of iron limited HNLC systems [Hutchins and volume before and after freeze-drying (Table 1). Uranium
Bruland, 1998; Hutchins et al., 2002]. The modern climate and thorium concentrations were measured by a spectrom-
of the continental region adjacent to the sampling position etry following separation on anion exchange columns and
(the southern limit of the Atacama Desert) is hyperarid deposition onto aluminum foil. The analytical uncertainty of
(Figure 1). the 230Th concentration data from counting statistics was
[5] Core GeoB7101-1 was sampled at 10 cm intervals 5%. Samples were only occasionally run in duplicate and
using 10-ml syringes. Two samples series were taken. The uncertainties of the reproductibilities were less than or equal
first was designated for uranium, thorium (Table 1) and rare to those from counting statistics. The analytical uncertainty
Earth elements analyses and the second for carbonate of 238U concentration measurements from counting statis-
concentration measurements. The dry bulk density of the tics was 6%. The carbonate content (% CaCO3) has been

2 of 15
 PA3012

Table 1. Results Obtained on Marine Sediment Samples for Cores GeoB 3375-1 and GeoB 7101-1a
GeoB 7101-1 Age, DBD, 238U, 232Th, 230Th, 230Thex, 230Thex
, 238U auth, Geob 3375 Age, Org C, Opal, Detrital Material, Iron,
Depth, cm kyr g/cm3 dpm/g dpm/g dpm/g dpm/g dpm/g dpm/g Depth, cm kyr % % % ppm
3 9.7 0.87 1.51 ± 0.11 1.39 ± 0.11 3.94 ± 0.24 2.86 ± 0.25 3.13 ± 0.28 0.47 ± 0.13 3 9.7 0.43 ± 0.02 0.70 ± 0.14 81.94 26047
8 10.3 1.07 3.08 ± 0.18 1.41 ± 0.10 4.06 ± 0.24 2.82 ± 0.25 3.10 ± 0.28 2.02 ± 0.19 8 10.3 0.41 ± 0.02 0.60 ± 0.12 77.65 25997
18 12.2 1.05 5.03 ± 0.19 1.42 ± 0.10 4.08 ± 0.21 2.59 ± 0.22 2.90 ± 0.25 3.97 ± 0.20 18 12.2 0.52 ± 0.03 0.40 ± 0.08 74.07 24589
23 13.2 0.96 5.14 ± 0.22 1.47 ± 0.19 3.94 ± 0.37 2.38 ± 0.38 2.68 ± 0.43 4.04 ± 0.25 23 13.2 0.65 ± 0.03 0.30 ± 0.06 75.19 26123
33 15.2 1.00 3.88 ± 0.19 1.57 ± 0.10 4.02 ± 0.18 2.49 ± 0.19 2.86 ± 0.22 2.70 ± 0.20 33 15.2 0.53 ± 0.03 0.60 ± 0.12 80.90 27356
48 16.7 1.09 2.60 ± 0.55 1.95 ± 0.17 3.25 ± 0.25 1.62 ± 0.28 1.89 ± 0.33 1.13 ± 0.56 48 16.7 0.52 ± 0.03 0.75 ± 0.15 85.91 28353
63 18.1 1.18 2.15 ± 0.15 1.79 ± 0.15 2.90 ± 0.20 1.44 ± 0.22 1.70 ± 0.26 0.81 ± 0.17 63 18.1 0.48 ± 0.02 0.80 ± 0.16 87.73 28177
78 19.5 1.12 2.23 ± 0.12 2.03 ± 0.08 3.20 ± 0.12 1.56 ± 0.13 1.87 ± 0.15 0.71 ± 0.13 78 19.5 0.46 ± 0.02 0.75 ± 0.15 86.36 32282
93 21.1 1.14 2.14 ± 0.16 1.85 ± 0.13 2.67 ± 0.17 1.16 ± 0.19 1.40 ± 0.23 0.76 ± 0.18 93 21.1 0.48 ± 0.02 0.70 ± 0.14 87.64 34005
103 22.2 1.17 1.95 ± 0.08 1.72 ± 0.04 2.67 ± 0.06 1.26 ± 0.07 1.55 ± 0.09 0.66 ± 0.08 103 22.2 0.45 ± 0.02 0.80 ± 0.16 87.20 32400
128 25.2 1.11 2.69 ± 0.33 1.71 ± 0.12 3.01 ± 0.19 1.44 ± 0.22 1.81 ± 0.27 1.40 ± 0.34 128 25.2 0.75 ± 0.04 0.95 ± 0.19 85.00 32245
153 28.7 1.12 2.48 ± 0.11 1.57 ± 0.07 3.94 ± 0.15 2.46 ± 0.16 3.20 ± 0.20 1.30 ± 0.12 153 28.7 0.69 ± 0.03 1.10 ± 0.22 85.07 30279
163 30.0 1.16 2.93 ± 0.30 1.43 ± 0.05 3.58 ± 0.10 2.06 ± 0.12 2.71 ± 0.16 1.86 ± 0.30 163 30.0 0.80 ± 0.04 0.70 ± 0.14 81.09 25247
173 34.6 1.23 3.30 ± 0.07 1.57 ± 0.08 3.71 ± 0.15 1.95 ± 0.16 2.69 ± 0.22 2.13 ± 0.08 173 34.6 0.59 ± 0.03 0.60 ± 0.12 81.37 25740
178 36.9 1.23 3.65 ± 0.11 1.42 ± 0.06 3.63 ± 0.13 1.82 ± 0.14 2.56 ± 0.19 2.58 ± 0.11 178 36.9 0.41 ± 0.02 0.65 ± 0.13 79.41 24572
193 40.3 1.18 3.36 ± 0.10 1.36 ± 0.08 3.18 ± 0.14 1.43 ± 0.16 2.07 ± 0.22 2.35 ± 0.11 193 40.3 0.58 ± 0.03 0.60 ± 0.12 80.35 29429
208 42.7 1.17 2.32 ± 0.14 1.51 ± 0.09 2.81 ± 0.15 1.29 ± 0.17 1.91 ± 0.25 1.19 ± 0.15 208 42.7 0.58 ± 0.03 0.80 ± 0.16 81.47 28684

3 of 15
223 47.2 1.18 2.73 ± 0.15 1.74 ± 0.14 2.85 ± 0.20 1.05 ± 0.22 1.62 ± 0.34 1.42 ± 0.17 223 47.2 0.69 ± 0.03 0.70 ± 0.14 83.40 31029
238 51.0 1.25 2.38 ± 0.09 1.94 ± 0.07 3.02 ± 0.10 1.21 ± 0.12 1.94 ± 0.19 0.93 ± 0.10 238 51.0 0.53 ± 0.03 0.80 ± 0.16 83.23 30481
253 54.8 1.18 2.63 ± 0.14 2.00 ± 0.13 3.69 ± 0.21 1.74 ± 0.23 2.88 ± 0.38 1.14 ± 0.16 253 54.8 0.65 ± 0.03 0.70 ± 0.14 79.53 26514
263 57.4 1.24 3.22 ± 0.16 1.62 ± 0.13 3.47 ± 0.23 1.43 ± 0.25 2.43 ± 0.43 2.00 ± 0.18 263 57.4 0.69 ± 0.03 0.50 ± 0.10 81.68 26040
273 59.9 1.29 2.38 ± 0.17 1.44 ± 0.09 3.37 ± 0.16 1.74 ± 0.19 3.02 ± 0.33 1.30 ± 0.18 273 59.9 0.60 ± 0.03 0.50 ± 0.10 83.15 26148
293 65.0 1.17 2.98 ± 0.07 1.39 ± 0.10 3.03 ± 0.17 1.11 ± 0.18 2.02 ± 0.33 1.94 ± 0.09 293 65.0 0.80 ± 0.04 0.70 ± 0.14 81.88 28981
313 69.8 1.24 2.88 ± 0.08 1.74 ± 0.06 3.03 ± 0.09 0.98 ± 0.10 1.87 ± 0.20 1.57 ± 0.09 313 70.5 0.58 ± 0.03 0.80 ± 0.16 83.86 32879
333 73.9 1.22 2.66 ± 0.20 1.88 ± 0.07 2.86 ± 0.10 0.83 ± 0.15 1.65 ± 0.30 1.24 ± 0.21 323 73.2 0.54 ± 0.03 0.70 ± 0.14 86.39 33223
343 75.9 1.24 2.38 ± 0.09 1.95 ± 0.07 2.97 ± 0.10 1.04 ± 0.12 2.10 ± 0.24 0.91 ± 0.10 333 75.9 0.63 ± 0.03 1.00 ± 0.20 87.84 33039
353 78.1 1.13 2.62 ± 0.12 1.76 ± 0.09 3.24 ± 0.14 1.26 ± 0.16 2.58 ± 0.33 1.31 ± 0.13 343 78.6 0.56 ± 0.03 1.40 ± 0.28 84.59 30914
DEZILEAU ET AL.: PAST PRODUCTIVITY OFF CHILE

363 80.3 1.19 2.95 ± 0.15 1.54 ± 0.06 3.17 ± 0.10 1.07 ± 0.13 2.25 ± 0.28 1.79 ± 0.15 353 81.4 0.67 ± 0.03 1.10 ± 0.22 80.10 28266
373 82.5 1.17 3.52 ± 0.13 1.33 ± 0.08 3.54 ± 0.17 1.20 ± 0.19 2.57 ± 0.40 2.53 ± 0.14 363 84.1 0.63 ± 0.03 1.00 ± 0.20 78.61 25225
383 89.5 1.11 3.85 ± 0.20 1.50 ± 0.10 3.57 ± 0.17 0.91 ± 0.22 2.08 ± 0.49 2.72 ± 0.20 383 89.5 0.50 ± 0.03 0.80 ± 0.16 78.03 28111
403 95.0 1.21 2.75 ± 0.13 1.77 ± 0.08 2.90 ± 0.11 0.74 ± 0.14 1.78 ± 0.34 1.42 ± 0.14 403 95.0 0.39 ± 0.02 1.00 ± 0.20 87.92 33803
409 96.4 1.25 2.45 ± 0.10 1.83 ± 0.11 2.89 ± 0.16 0.88 ± 0.18 2.13 ± 0.45 1.08 ± 0.12 408 96.4 0.50 ± 0.02 1.00 ± 0.20 87.21 31965
a
The dry bulk density (DBD) of the sediments was measured by weighting a fixed sample volume before and after freeze-drying. Uranium and thorium concentrations were measured by a spectrometry. 230Th
ex
refers to excess 230Th activity in sediments decay-corrected to the time of deposition. 238U auth refers to excess U above detrital background levels. The biogenic opal content was measured applying the method of
Müller and Schneider [1993]. Organic carbon analyses were conducted with an Hereaus CHN-O-rapid. The detrital fraction of the sediment was estimated from bulk sediment weight minus biogenic components.
The iron data have been previously published in the work of Lamy et al. [2000].
PA3012
PA3012 DEZILEAU ET AL.: PAST PRODUCTIVITY OFF CHILE PA3012

were conducted with an elemental analyzer (Hereaus CHN-

O-Rapid) on HCl-treated samples (Table 1). The detrital
fraction of the sediment was estimated from the bulk
sediment weight minus the biogenic components (opal,
carbonate, organic matter) with detrital fraction = 1  (opal
fraction + carbonate fraction + 3  organic carbon fraction)
[Dezileau et al., 2000]. The iron (Table 1) and aluminum
data used in this study were previously reported in the work
of Lamy et al. [2000].

3. Chronology
[7] The age model for core GeoB 3375-1 is based on both
AMS 14C ages and @ 18O records of N. pachyderma (sin.)
[Lamy et al., 1998] resulting in an age of
117.000 cal
years BP at the base of the core (Figure 2). The chronology
for core GeoB 7101-1 has been obtained by correlating
graphically its carbonate content variations with those
of core GeoB 3375-1, using the Analyseries Software
[Paillard et al., 1996] (Figure 2). On this basis, core GeoB
7101-1 covers the last 100,000 cal years BP, with an
average sedimentation rate of 5 cm/1000 years. These twin
cores come from almost exactly the same location and have
the same age at each depth up to 300 cm. This suggests that
if these cores have been affected by sediment redistribu-
tions, these effects were of same intensities in the past.

Figure 2. Age model for sediment core GeoB 3375-1 and 4. Vertical Fluxes Calculations
GeoB 7101-1. (top) @ 18O(%) curve of the planktic [8] Although past changes in ocean productivity can be
foraminifera N. pachyderma (sin.) and age control points assessed, in principle, from the record of biogenic detritus
(indicated by triangle) include seven AMS 14C dates and buried in the seabed, their interpretation is often hampered
two graphical correlation points of the @ 18O records to a by sediment redistribution processes (focusing or winnow-
standard @ 18O curve. (bottom) Carbonate content for the ing) and variable dissolution. We have calculated preserved
cores GeoB 3375-1 and GeoB 7101-1. We obtained the vertical rain rates to the seafloor by normalizing biogenic
chronology for the core GeoB 7101-1 by graphically component fluxes to the corresponding excess 230Th activ-
correlating its carbonate content variations with those of ity in the sediments, a method that corrects for sediment
core GeoB 3375-1. The used standard methods for @ 18O focusing and winnowing [Bacon, 1984; Suman and Bacon,
analysis, AMS dating, and carbonate content determinations 1989; François et al., 1990, 2004].
are described elsewhere in detail [Lamy et al., 1998]. [9] Thorium-230 normalized flux calculations are based on
the assumption that the flux of 230Th to the seafloor is constant
and equal to the rate of production [Bacon, 1984; Suman and
determined with a calcimeter by the CO2 volumetric method.
Bacon, 1989; François et al., 1990, 2004]. Excess 230Th
The precision is estimated to be about 2%. Analyses of La
activity in settling particulates is then inversely related to total
and Yb were carried out by Epithermal Neutron Activation
mass flux, and excess 230Th activities (to see how is obtained
Analysis (ENAA) at the Pierre Sue Laboratory (Saclay,
the age-corrected excess 230Th activity see Appendix A) in
France). Two USGS standards, Mag 1 and A1, were used.
the sediment can be used as reference against which the flux
Sediment samples and standards were enclosed in plastic
of other sedimentary components can be estimated:
bags. After irradiation, sediment samples and standards
were measured with an HP Ge detector (FWMH 1.7 keV
b:Z:fi
at 1332 keV; relative efficiency: 20%) at the Laboratoire des Fi ¼ 230  ; ð1Þ
Sciences du Climat et de l’Environnement. The precision is Thex
estimated to be about 5%.
[6] The biogenic opal content was measured on core where Fi is the normalized rain rate of the sedimentary
GeoB 3375-1 (D. Hebbeln, unpublished results, 2001) component i; b is the constant production rate of 230Th in
applying the method of Müller and Schneider [1993] and the water column (2.63  105 dpm cm3 kyr1); Z is the
the values obtained were between 0.3 and 1.4%. For water depth (cm); fi is the percentage of the sedimentary
samples with low opal contents (<1 wt %) the uncertainty component i; and 230Thex
is the activity of decay- and
of the method used is of 20% and for samples with very low ingrowth-corrected excess 230Th in sediment.
opal contents (<0.5 wt %) the uncertainty is about 60% [10] Arguments for the validity of this approach have
(Table 1). Organic carbon analyses of core GeoB 3375-1 been examined elsewhere [Henderson et al., 1999; Yu et al.,

4 of 15
PA3012 DEZILEAU ET AL.: PAST PRODUCTIVITY OFF CHILE PA3012

tion rates on the seafloor during the Holocene and marine

oxygen isotope stage (MIS) 2, were 5.0 and 3.7 times
higher, respectively, than the vertical rain rates. During
MIS 5 to 3, the accumulation rates were only 2.1 times
higher than the vertical rain rates.
[12] These differences are probably due to an increase
of focusing during these periods, however, it may also be
possible that the estimation of the vertical rain rates by
the 230Thex
method is biased because the assumption
underlying the constant flux model of 230Th is not valid.
Having shown that these problems are negligible for our
cores (see Appendix B), we conclude that the focusing
factor is a reliable quantitative proxy to correct for lateral
sediment redistribution in this region.
[13] The difference between the sediment accumulation
rate and the vertical sediment rain rate is pronounced and
has mainly been caused by an increase of focusing intensity,
in particular during the Holocene and LGM. The recon-
struction of paleoproductivity would be quite different if
only the accumulation rates had been used without the
230

-normalization. The nonnormalized accumulation
Thex
rates do not show any climate related variation. With
the 230Thex
-normalization, the vertical sediment rain
rates reveal a strong cyclicity which is dominated by
precessional cycles. This illustrates the importance of the
230

-normalization method for reconstructions of past
Thex
particle fluxes.

5.2. Records of Terrigenous Composition off the Norte

Chico (27.5

S)
[14] The sediments consist mainly of terrigenous material
with contents between 75 and 88 wt % (Figure 4a). The iron
content (Figure 4b) varies between 24000 and 34000 ppm
[Lamy et al., 2000]. Lamy et al. [2000] used the Fe/Al ratio
Figure 3. (a) Comparison between sediment accumulation in gravity core GeoB 3375-1 (located in the same area) to
rate and vertical particle rain rate for core GeoB 7101-1. estimate past changes in terrigenous sediment provenance
Sediment accumulation rates (dotted line) for the last (Figure 4c). In their study, they interpreted changes in the
100,000 years based on mass accumulated between oxygen Fe/Al ratio of the bulk sediment as a terrigenous signal,
isotope stage boundaries. Vertical particle rain rate (solid which is determined by relative contributions of two source
line) is based on point-by-point normalization to areas, i.e., the Andes and the Coastal Range. Observed
230

applying the constant flux model. (b) 230Thex
Thex
flux variations could also have been produced by changes in the
in dpm cm2 kyr1 versus age for each @ 18O isotope proportions of two components in the sediment, a detrital
stages. The stippled line marks the expected flux of component with a low and constant Fe/Al ratio and a
5.14 dpm cm2 kyr1 at this location. variable amount of an authigenic phase (Fe oxide, pyrite,
apatite, etc.) with a high Fe/Al ratio. However, we have also
2001]. This approach was successfully applied where lateral analyzed rare Earth elements (REE) by neutron activation in
re-sedimentation is important [Suman and Bacon, 1989; the same core. In particular we have used the La/Yb to
François et al., 1993, 1997; Kumar et al., 1995; Anderson determine the origin of the detrital particles (Figure 4c). The
et al., 1998, 2002; Frank et al., 1995, 2000; Dezileau et al., advantage of the La/Yb ratio over the Fe/Al is that the
2000, 2003; Chase et al., 2003]. lithogenic signal of the bulk sediment is not altered by
changes in redox conditions. The Coastal Range consists
primarily of Mesozoic calc-alkaline plutonic rocks (gran-
5. Results ites) and is characterized by La/Yb ratios between 8 and 9.2
5.1. Vertical Sediment Rain Rates Compared [Pichowiak, 1994]. The Andes consist primarily of Plio-
With Sediment Accumulations Rates Quaternary volcanism, including andesitic and rhyolitic
[11] When comparing the sediment accumulation rates to rocks characterized by La/Yb ratios between 10.6 and 18
the vertical particle rain rates calculated by the 230Th
ex [Kay et al., 1991]. The La/Yb ratios in our core are below
method [see Suman and Bacon, 1989, equation (A2); 9.5 during periods of low iron contents, indicating a
François et al., 1993, 1997, 2004], we note striking differ- dominating contribution of terrigenous particles from the
ences between the two records (Figure 3a). The accumula- Coastal Range source (Figure 4c). These ratios increase

5 of 15
PA3012 DEZILEAU ET AL.: PAST PRODUCTIVITY OFF CHILE PA3012

ered to be exclusively present in detrital minerals; r2 = 0,91,

Figure 4d).
5.3. Records of Biogenic Constituents off the Norte
Chico (27.5

S)
5.3.1. Organic Carbon
[15] Varying organic carbon fluxes in hemipelagic sedi-
ments below eastern boundary current systems are primarily
controlled by surface water productivity and thus can be
used as a direct proxy for paleoproductivity in such environ-
ments [Lyle et al., 1988; Sarnthein et al., 1988; Berger et
al., 1989; Pedersen and Calvert, 1990; Schneider et al.,
1997]. However, in a region marked by a strong terrigenous
input, the organic carbon content of the marine sediments
can be increased by a certain portion of terrigenous organic
matter, which might bias paleoproductivity estimates based
on organic carbon accumulation. This problem can be
addressed by using d13Corg data, which are often used to
detect contributions of terrestrial organic matter to marine
sediment. Terrestrial organic matter usually has d13Corg
values of 28 to 26% [Emerson and Hedges, 1988],
while marine plankton has typical d13Corg values of 22
to 19% for water temperatures between 13 and 18
C
[Fontugne and Duplessy, 1981]. The d13Corg data in the
surface sediments vary between 21.97 and 20.48. Thus,
although our gravity core has been collected very close to
the coast, the d13Corg data do not give any hint for a
substantial input of terrigenous organic matter. For all our
samples d13Corg lies well inside the typical ranges for
marine organic matter, which seems to clearly dominate
the total organic carbon content. Keeping in mind that on
average 80% of the sediment material is of terrigenous
origin, the terrestrial organic carbon input seems to have
been rather low. On the other hand, a significant contribu-
tion from C4 plants, characterized by high d13C values, can
be discarded due to the very low vegetation biomass of this
desert or steppic environment. Furthermore, Quay et al.
[1992] showed that the contribution of C4 plants to open
ocean sediments is low, because of their preferential oxida-
tion in rivers. Mariotti et al. [1991] showed that the organic
carbon isotopic composition of suspended material in the
Congo river presents typical C3 plant values, in spite of the
occurrence of large savanna areas covered by C4 plants in
Figure 4. Comparison of geochemical data for the last the hydrographic basin. Similar results were found in sedi-
100,000 years derived from deep-sea sediment cores GeoB ments off India by Bentaleb et al. [1997].
3375-1 and GeoB 7101-1. (a) The detrital fraction (%). 5.3.2. Opal and Organic Carbon Vertical Fluxes
(b) Iron content in ppm( 104). (c) The Fe/Al (line) and La/ [16] The calculated vertical sediment rain rates were
Yb (square) ratio. (d) Comparison between iron (solid line) combined with the sedimentological analyses in order to
and 232Th vertical fluxes (dotted line). assess component vertical rain rates (Table 2). Organic
carbon vertical rain rates varied by more than a factor of
3, and the opal vertical rain rates varied by a factor of 6
during periods of high iron contents, indicating a likely (Figures 5a and 5b). These variations were in phase during
relative increase in the volcanic contribution, i.e., from the the last 100,000 years. Variation in sedimentation rates can
Andes. The La/Yb ratios thus support the conclusions affect opal and organic matter preservation. Thus it could be
obtained from Fe/Al ratios. We then assume that the argued that the ups and downs in opal and organic carbon
presence of authigenic iron is negligible in our core and fluxes are mainly driven by preservation effects. However,
that the Fe/Al ratio can be used to reconstruct past changes we do not find any correlation between organic matter
of the lithogenic material in this core. The terrigenous origin content and sediment accumulation/vertical particle rain
of iron is also confirmed by an excellent correlation rate, nor between opal content and sediment accumula-
between iron and 232Th vertical rain rates (232Th is consid- tion/vertical particle rain rate. Thus we do not have any

6 of 15
PA3012 DEZILEAU ET AL.: PAST PRODUCTIVITY OFF CHILE PA3012

Table 2. Vertical Flux Refers to 230Th Normalized Total Mass Accumulation Rate, Organic Carbon (Opal and Iron) Vertical Flux Refers
to 230Th Normalized Organic Carbon (Opal and Iron) Accumulation Ratea
Org C Opal Iron
Vertical Flux Vertical Flux Vertical Flux Vertical Flux Authigenic Uranium
Age, 230Thex
, (230Th Method), (230Th Method), (230Th Method), (230Th Method), Accumulation Rate,
Kyr dpm/g g/cm2/kyr mg/cm2/kyr mg/cm2/kyr mg/cm2/kyr Opal/CaCO3 dpm/cm2/kyr
9.7 3.13 ± 0.28 1.64 ± 0.14 7.09 ± 0.62 11.51 ± 1.01 42.82 ± 3.77 0.044 ± 0.009 3.48
10.3 3.10 ± 0.28 1.66 ± 0.15 6.75 ± 0.60 9.97 ± 0.89 43.20 ± 3.85 0.029 ± 0.006 10.96
12.2 2.90 ± 0.25 1.77 ± 0.15 9.13 ± 0.78 7.09 ± 0.61 43.58 ± 3.74 0.017 ± 0.003 21.14
13.2 2.68 ± 0.43 1.92 ± 0.31 12.42 ± 2.01 5.75 ± 0.93 50.07 ± 8.09 0.013 ± 0.003 19.65
15.2 2.86 ± 0.22 1.80 ± 0.14 9.46 ± 0.73 10.79 ± 0.84 49.20 ± 3.82 0.035 ± 0.007 28.01
16.7 1.89 ± 0.33 2.72 ± 0.47 14.02 ± 2.42 20.41 ± 3.52 77.17 ± 13.31 0.064 ± 0.013 12.83
18.1 1.70 ± 0.26 3.02 ± 0.46 14.54 ± 2.21 24.19 ± 3.67 85.20 ± 12.94 0.080 ± 0.016 9.91
19.5 1.87 ± 0.15 2.76 ± 0.23 12.69 ± 1.04 20.69 ± 1.69 89.04 ± 7.26 0.065 ± 0.013 7.64
21.1 1.40 ± 0.23 3.66 ± 0.60 17.72 ± 2.90 25.63 ± 4.19 124.53 ± 20.36 0.069 ± 0.014 8.26
22.2 1.55 ± 0.09 3.33 ± 0.19 14.94 ± 0.84 26.63 ± 1.49 107.84 ± 6.04 0.075 ± 0.015 6.39
25.2 1.81 ± 0.27 2.84 ± 0.43 21.31 ± 3.21 26.99 ± 4.07 91.60 ± 13.81 0.081 ± 0.016 11.15
28.7 3.20 ± 0.20 1.61 ± 0.10 11.11 ± 0.71 17.69 ± 1.13 48.70 ± 3.10 0.094 ± 0.019 10.53
30.0 2.71 ± 0.16 1.90 ± 0.11 15.11 ± 0.91 13.28 ± 0.80 47.91 ± 2.89 0.044 ± 0.009 4.74
34.6 2.69 ± 0.22 1.92 ± 0.15 11.23 ± 0.91 11.50 ± 0.93 49.31 ± 3.98 0.037 ± 0.007 5.74
36.9 2.56 ± 0.19 2.01 ± 0.15 8.23 ± 0.62 13.08 ± 0.98 49.44 ± 3.70 0.040 ± 0.008 13.98
40.3 2.07 ± 0.22 2.48 ± 0.27 14.33 ± 1.55 14.88 ± 1.61 72.96 ± 7.91 0.035 ± 0.007 17.02
42.7 1.91 ± 0.25 2.69 ± 0.35 15.62 ± 2.01 21.50 ± 2.77 77.10 ± 9.93 0.050 ± 0.010 4.68
47.2 1.62 ± 0.34 3.18 ± 0.67 22.10 ± 4.64 22.29 ± 4.68 98.80 ± 20.73 0.051 ± 0.010 6.58
51.0 1.94 ± 0.19 2.65 ± 0.25 14.07 ± 1.34 21.20 ± 2.02 80.79 ± 7.70 0.056 ± 0.011 4.55
54.8 2.88 ± 0.38 1.79 ± 0.24 11.61 ± 1.54 12.50 ± 1.66 47.37 ± 6.30 0.039 ± 0.008 5.27
57.4 2.43 ± 0.43 2.12 ± 0.37 14.67 ± 2.59 10.58 ± 1.87 55.11 ± 9.72 0.032 ± 0.006 9.76
59.9 3.02 ± 0.33 1.70 ± 0.18 10.28 ± 1.11 8.51 ± 0.92 44.50 ± 4.80 0.034 ± 0.007 6.59
65.0 2.02 ± 0.33 2.55 ± 0.41 20.36 ± 3.30 17.84 ± 2.89 73.85 ± 11.98 0.047 ± 0.009 9.50
69.8 1.87 ± 0.20 2.75 ± 0.29 15.89 ± 1.66 21.99 ± 2.30 90.37 ± 9.45 0.059 ± 0.012 9.55
73.9 1.65 ± 0.30 3.12 ± 0.56 16.78 ± 3.03 21.84 ± 3.94 103.64 ± 18.72 0.062 ± 0.012 7.42
75.9 2.10 ± 0.24 2.45 ± 0.28 15.53 ± 1.80 24.50 ± 2.84 80.94 ± 9.39 0.108 ± 0.022 5.18
78.1 2.58 ± 0.33 1.99 ± 0.25 11.09 ± 1.40 27.93 ± 3.53 61.66 ± 7.79 0.113 ± 0.023 6.76
80.3 2.25 ± 0.28 2.28 ± 0.29 15.31 ± 1.92 25.13 ± 3.15 64.57 ± 8.10 0.066 ± 0.013 9.78
82.5 2.57 ± 0.40 2.00 ± 0.32 12.63 ± 1.99 20.04 ± 3.15 50.56 ± 7.95 0.054 ± 0.011 4.17
89.5 2.08 ± 0.49 2.47 ± 0.58 12.44 ± 2.93 19.74 ± 4.65 69.37 ± 16.34 0.041 ± 0.008 11.08
95.0 1.78 ± 0.34 2.90 ± 0.56 11.18 ± 2.17 28.97 ± 5.61 97.93 ± 18.98 0.101 ± 0.020 7.37
96.4 2.13 ± 0.45 2.41 ± 0.51 11.97 ± 2.52 24.13 ± 5.08 77.13 ± 16.24 0.097 ± 0.019 5.74
a
The authigenic uranium accumulation rate is estimated by multiplying the total mass accumulation rate (g cm2 kyr1) with the authigenic uranium
concentration (dpm g1).

evidence that the signal we observe is due to changes in 3375-1 in order to know if variable dissolution of CaCO3
preservation. occurred (X. Mohtadi et al., manuscript in preparation,
5.3.3. Opal//CaCO3 Ratio 2004). Mohtadi finds that the preservation of the plank-
[17] We have used the opal/CaCO3 ratio as another tonic forams is generally good to very good.
index of productivity (Figure 5c). This ratio gives an idea [18] To sum up, we do not want to argue that preservation
about the siliceous/carbonate productivity, taking into has had no effect on our data, but the similar pattern for all the
account that when the productivity increases in a region paleoproductivity indicators we studied, brings us to the
this ratio generally increases. However, there are some conclusion that we really see a paleoproductivity signal in
uncertainties, which can induce false conclusions when the data.
evaluating paleoproductivity from the opal/CaCO3 ratio.
Both opal and carbonate may be variably dissolved and 5.4. Authigenic Uranium: Proxy for Export
the opal/CaCO3 ratio can be controlled by dissolution. Productivity?
Opal and carbonate respond differently to dissolution, as [19] In oxic pore water, U is present as a soluble carbon-
carbonate is dissolved under more acidic conditions, ate complex. When conditions become sufficiently reducing
while opal gets dissolved under basic conditions. Thus to initiate sulphate reduction, U is reduced to an insoluble
dissolution problems can be considered negligible only if form and precipitates. The U added to the sediment by this
the Opal/CaCO3 record closely parallels the other paleo- process (referred to as authigenic U) accumulates at a rate
productivity proxies. Here, there is a good correlation that depends on the level of oxygen in bottom water and on
between the Opal/CaCO3 record and the organic carbon the flux of organic carbon. Kumar et al. [1995] showed a
and opal vertical flux. Therefore we do not have any positive relationship between organic carbon flux (export
evidence that the opal/CaCO3 ratio in our core has been productivity) and the accumulation rate of authigenic ura-
controlled by dissolution. Moreover, M. Mohtadi (Uni- nium and, assuming that a similar relationship existed in the
versity of Bremen, Germany) has counted and analyzed past, proposed that down-core records of authigenic U flux
planktonic foraminifera assemblages in the core GeoB could be used to constrain past changes in the flux of carbon

7 of 15
PA3012 DEZILEAU ET AL.: PAST PRODUCTIVITY OFF CHILE PA3012

Figure 5. Comparison of geochemical data for the last 100,000 years derived from deep-sea sediment
cores GeoB 3375-1 and GeoB 7101-1. (a) Organic carbon and (b) biogenic opal vertical rain rates based
on point-by-point normalization to 230Thex
. (c) Biogenic opal/carbonate ratio is also a proxy for
productivity. (d) Iron content (dotted line) and iron vertical rain rates (line) based on point-by-point
normalization to 230Thex
. Terrigenous particle origin can be determined by using the Fe/Al ratio. The
Fe/Al (line) (and La/Yb square) ratio of basaltic-andesitic rocks as found in the Andes is characterized by
a value above 0.45 (9.5) and the Fe/Al (La/Yb) ratio of acidic-plutonic rocks, as found in the coastal
range is characterized by a ratio under 0.45 (9.5). Measured Fe/Al ratios clearly show a higher
contribution of terrigenous particles from the Andes during (e) precessional (summer insolation) maxima.

to the deep sea. Thus we have decided to estimate authi- Holocene/MIS 2 periods, the authigenic U fluxes are high
genic U flux in the core GeoB 7101-1 in order to see more (6 – 27 dpm cm2 kyr1) (Figure 6). Unlike opal and
productivity proxies. organic carbon rain rates, the authigenic U flux do not
[20] During the MIS 4 – 5, accumulation rates of authi- reveal a cyclicity. Variations of authigenic U fluxes are
genic uranium are low (7 dpm cm2 kyr1) and during the erratic and do not show any climatic signals. How can we

8 of 15
PA3012 DEZILEAU ET AL.: PAST PRODUCTIVITY OFF CHILE PA3012

Figure 6. Comparison between authigenic uranium fluxes (dpm cm2 kyr1) and total and vertical
organic carbon fluxes (mg cm2 kyr1) for the last 100,000 years derived from deep-sea sediment core
GeoB 7101-1.

reconcile authigenic U flux with opal and organic carbon? authigenic U. Thus variation in authigenic U in our cores
The recent discovery that authigenic U fluxes may also be may also be explained by a change in the contribution of
influenced by lateral transport of organic carbon, or a PNU due to a change in the water column oxygen concen-
change of oxygen concentration in water column [François tration. In conclusion, we suggest that the authigenic U flux
et al., 1997; Anderson et al., 1998; Frank et al., 2000; cannot be used as a direct proxy of palaeoproductivity in the
Dezileau et al., 2002], helps reconcile the authigenic U flux eastern South Pacific.
record with those of the other proxies.
[21] The flux of organic carbon to the seabed may be
influenced by sediment focusing. That is, focused depo- 6. Discussion
sition of sinking particles by deep-sea currents may 6.1. Regional Paleoceanographic and Paleoclimatic
enhance significantly the local flux of organic matter to Implications
the seabed relative to the regional average flux of organic [23] Biogenic opal and organic carbon vertical rain rates
matter exported from surface waters [Anderson et al., and the opal/carbonate ratio (Figures 5a – 5c) reveal a
1998; François et al., 1993]. Accumulation of sediments strong cyclicity, which is dominated by precession (a
in the continental slope off northern Chile is strongly cycle of
20,000 years). High and low values are in
enhanced by sediment focusing. Lateral transport estima- phase with summertime (January) insolation maxima and
tions for the last 100 kyr suggest that sediment redistri- minima, calculated for 25
S latitude from the Earth’s
bution is higher during the LGM and Holocene periods. known orbital variations [Berger and Loutre, 1991]. These
In that case, we suggest that a significant part of the variations were in phase during the last 100,000 years.
authigenic U was attributed to intense sediment focusing Iron vertical rain rates also show a strong precessional
during these periods (Figure 6). cyclicity. The patterns of iron vertical rain rate and of
[22] In the water column, particulate nonlithogenic U iron content itself (Figure 5d) are remarkably similar to
(PNU: excess U above detrital background levels) is found the down-core records of the proxies for biogenic pro-
in marine particulate matter [Anderson, 1982]. Zheng et al. ductivity (Figures 5a – 5c), suggesting iron availability to
[2002] have recently shown that PNU formed in surface be a factor responsible for past changes in productivity.
waters may, under certain circumstances (i.e., low concen- Although Fe/Al ratios could be affected by the presence
tration of oxygen in the water column), contribute signifi- of authigenic iron, particularly in low oxygen environ-
cantly to the total burial of authigenic U (between 20% and ments, variations in the La/Yb ratios (Figure 5c), which
80% below the oxygen minimum zone (OMZ) off Califor- are not altered by changes in the redox conditions, show
nia). Our two gravity cores were raised below the OMZ, the same pattern (see section 5.2). Thus our results
thus, the PNU must certainly contribute significantly to the indicate that the supply of iron and consequently biolog-
total burial of authigenic U in this area. Moreover, Zheng et ical productivity has varied in response to past changes in
al. [2002] showed that the preservation of PNU as particles orbital forcing.
sink through the water column varies with the water column [24] The potential for fertilization of iron-limited waters
oxygen content. Evidence for past changes in oxygen depends on the source of the iron. Iron supplied in the
concentration in the water column off Peru have been form of riverine or atmospheric dust is more likely
shown by Ganeshram et al. [2000]. This means that the biologically available at the surface than iron from other
oxygen content of the water column off northern Chile have sources (e.g., lithogenic material transported by deep-
also probably changed between glacial/interglacial periods ocean currents, for example). Recent studies [Lamy et
affecting the contribution of PNU to the total burial of al., 1998, 2000] on core GeoB 3375-1 have clearly

9 of 15
PA3012 DEZILEAU ET AL.: PAST PRODUCTIVITY OFF CHILE PA3012

productivity, a higher utilization of nutrient through iron

fertilization would be a more possible explanation. This is
illustrated by results obtained by Lyle et al. [1992] in the
central equatorial Pacific where upwelling strength varia-
tions were unable to explain organic carbon burial with-
out taking into account terrestrial input (aeolian) and
probably iron fertilization.
[26] We propose that variations in marine productivity
at the precessional timescale are due to changes in
terrestrial weathering regimes. At present, maximum pre-
cipitation in Central Chile [Aceituno and Vidal, 1990] is
during the austral winter. The increasing amounts of
rainfall are directly related to the annual frequency of
atmospheric frontal systems arriving from the west, which
Figure 7. Comparison between the d13C of N. pachyderma also increase with latitude (cf. Figure 1). In our
(thick line) and biogenic opal vertical rain rates for the last study region, the Norte Chico, most of the precipitation
100,000 years derived from deep-sea sediment cores GeoB is deposited in the form of snow [Aceituno and
3375-1 and GeoB 7101-1. The arrows indicate ‘‘stronger’’ or Vidal, 1990] during the winter season. The few rivers
‘‘lower’’ upwelling intensity using d13C of N. pachyderma here display a clear ‘‘snow’’ regime, with a maximum
as proxy. fluvial runoff during the period of melting (austral sum-
mer). During precessional maxima (winter insolation
shown that during summer insolation maxima the litho- minima), a latitudinal shift toward the north in the
genic material in this region is mainly of fluvial origin patterns of onshore precipitation (cf. Figure 1), for
coming from the Andes (Figure 5e), a result that is example due to a
5
northward displacement of the
supported by our data on rare Earth elements (i.e., the southern westerlies and thus of the climatic zones in
La/Yb ratio). We conclude that during these periods, iron Chile [Lamy et al., 1998], should have occurred. Snow
fluxes were dominantly of fluvial origin and therefore precipitation was probably much higher during the winter
passed through the surface ocean en route to the seabed. and the fluvial runoff more significant during the summer.
During precessional maxima, biologically available iron These periods of higher river runoff should have trans-
associated with this fluvial material should have some- ported significant amounts of iron-rich terrigenous mate-
what relieved iron limitation in this present-day coastal rial from the Andes (high La/Yb and Fe/Al ratio) into the
HNLC upwelling regime. ocean. The high river discharges could then have caused
[25] Another hypothesis could be that biological pro- an increase in iron concentrations, as well as an increase
ductivity is driven by upwelling. In other words, it was in silicate concentrations, in the ocean surface waters and
the delivery of large amount of nutrients including iron induced an increase in biological productivity in the
via upwelling, which stimulated higher biogenic opal and region, particularly through diatom blooms. During pre-
organic carbon vertical rain rates indicative of higher cessional minima, aridity was probably even stronger than
productivity. We here investigate this hypothesis using today, with low precipitation during the winter and
the d13C of N. pachyderma. The d13C of N. pachyderma absence of rivers during the summer. Hence the sediment
varies between 0.2 to 1.7% (Figure 7). An increase of load of the ocean surface waters supplied by river
upwelling intensity would make go up deeper and colder discharges must have been low and iron was not suffi-
water with lower d13C of dissolved inorganic carbon ciently available in the ocean surface waters, limiting
(DIC). Thus, if biological productivity had been only primary productivity in the region.
driven by upwelling intensity, we should have observed [27] An analogous situation in the precipitation and
a correlation between low values of d13C of N. pachy- river runoff occurs at the interannual timescale. During
derma and high opal vertical rain rates (Figure 7). This El Niño events above average precipitation and river
correlation is not observed. The d13C variations are runoff are observed in central Chile [Aceituno and Vidal,
obviously independent of those of the productivity prox- 1990; Montecinos and Aceituno, 2003]; the opposite is
ies and do not reflect the periodicity recorded for opal observed during La Niña periods. Positive rainfall anoma-
fluxes. Moreover, the period of increased productivity lies during El Niño winters in the region are associated
observed in our cores do not correlate with the maximal with a northern shift in the storm tracks and enhanced
upwelling activity (early MIS 1, MIS 3 and upper MIS 5) blocking activity over the Amundsen-Bellinghausen Seas
in the Peru-Chile current off the Peru margin [Oberhänsli [Montecinos and Aceituno, 2003]. Hence an El Niño-
et al., 1990]. In addition, Schrader and Sorknes [1990] Southern Oscillation-like mechanism may have modulated
consider that increased upwelling strength was very slight the iron supply by rivers at the precessional timescale. An
during these periods. Whatever the upwelling strength ENSO modeling experiment over the past glacial-inter-
was, the Peru Chile current area seemed to have remained glacial cycles has shown that changes in low-latitude
a HNLC zone. Within these zones, the physical supply of insolation induce variations in the frequency and intensity
nutrients exceed the biological demand, consequently an of ENSO events [Clement et al., 2000]. However, we
increase of upwelling strength is not needed to increase found that variations in our iron vertical rain rates are not

10 of 15
PA3012 DEZILEAU ET AL.: PAST PRODUCTIVITY OFF CHILE PA3012

[31] Presently, nutrients and productivity in the EEP south

of the equator reflect the upwelling of deeper Equatorial
Undercurrent Water along the margin of Peru. This water
has its ultimate source in the SW Pacific subantarctic
[Toggweiler et al., 1991], an area where waters are rich in
nutrients [Fitzwater et al., 1996]. Loubere et al. [2003]
proposed that during the LGM the supply of Equatorial
Undercurrent Water from the subantarctic was reduced and
that there was a proportional increase in water from sub-
tropical sources (poor in nutrients) that upwelled along the
Peru margin. The implication of these results is that pro-
ductivity in the EEP has been strongly affected by ocean
circulation rather than just the tradewinds and upwelling
alone.
Figure 8. Comparison between iron vertical rain rates
[32] Investigation of surface sediments from the Chilean
(thin line) based on point-by-point normalization to 230Thex


continental slope points to the Antarctic Circumpolar Cur-
and sea surface temperature anomalies (thick line) in the
rent (ACC) as the principal nutrient source that sustains the
region Niño 3 (5
N – 5
S, 150
– 90
W) modeled by
high productivity in the Humboldt Current, where, besides
Clement et al. [2000].
being slowly consumed, the nutrients are continuously
recycled in the coastal upwelling system while moving to
the north [Hebbeln et al., 2000]. These nutrients are
in phase with the predictive ENSO model [Clement et al., effectively consumed by marine microorganisms in the
2000] at such a timescale (Figure 8), an almost constant coastal upwelling system due to a supplement of iron
shift of about 10 kyr occurs between the model and our derived from nearby hinterland and resuspended sediments
iron fluxes. Further studies are thus required to clearly from the continental shelf. During the LGM, a northward
establish the links between rainfall in the region and displacement of the ACC (rich in major nutrients, i.e., Si, N
anomalies in the tropical Pacific at orbital timescales. and P), in line of the northward movement of the southern
westerlies, would bring the main nutrient source closer to
6.2. Comparison With Equatorial Upwelling (Peru) our core sites [Hebbeln et al., 2000]. This is, however, not
During the Last Glacial Maximum sufficient. With Si (and other major nutrients transported by
[28] It has been most commonly thought that produc- the ACC) but no Fe being supplied to waters, primary
tivity on the glacial-interglacial timescale in the eastern producers in this region would have been probably limited
equatorial Pacific (EEP) was controlled by upwelling and (HNLC conditions). The increase of iron-rich terrigenous
the strength of the tradewinds. The expectation was that material from the Andes during the LGM has permitted an
stronger winds during glacials would increase the rate of increase of productivity. The implication of this result is that
upwelling and raise productivity through increased supply productivity off Chile has also been strongly affected by
of nutrients. other processes than just upwelling alone.
[29] Several tools have been used to infer past productiv- [33] In conclusion, productivity in the Chilean upwelling
ity of Peru and Chile Margins. Some were based on does not vary in the same way than the Peruvian
accumulation rates of biogenic components (organic carbon, upwelling during the LGM. It seems that two systems
opal and barium accumulation rates) and others based on work both independently. This result may only be
proxies independent of accumulation rate calculations explained by a different origin of nutrients and micro-
(abundance of planktic and benthic foraminifera, geochem- nutrients.
ical proxies). Since there is a risk that the accumulation rates
are affected by redistribution of sediment by bottom cur- 6.3. Global Paleoceanographic Implications
rents we have decided to take into account only works based [34] Iron limitation of phytoplankton primary productiv-
on proxies independent of accumulation rate calculations. ity in HNLC regions has become a central paradigm in
[30] Regional reconstruction of productivity patterns by biological oceanography [Hutchins and Bruland, 1998;
Loubere [2000] indicates that in the area downstream of the Martin and Fitzwater, 1988; De Baar et al., 1990; Coale
Peru margin, export productivity was lower during the Last et al., 1996, 2004; Hutchins et al., 2002] and a main
Glacial Maximum (LGM). This interpretation is coincident hypothesis [Martin, 2000] to explain glacial/interglacial
with results from independent geochemical tracers based on changes in paleoproductivity and atmospheric pCO 2.
ratios of productivity sensitive elements [Loubere and However, the conventional view has been that variations
Murray, 2001; Ganeshram et al., 2000]. On the contrary, in productivity in coastal upwelling systems (and their
regional reconstruction of productivity off Chile [study on response to orbital forcing) are mainly driven by changes
species composition of planktic foraminifera [Hebbeln et in the meridional surface wind field, and consequently in
al., 2002; this work] shows that export productivity was the input of major nutrients from subsurface to surface
higher during the LGM. How can we reconcile the higher waters. Moreover, although fertilization of the photic zone
productivity off Chile with the lower productivity down- by the upwelling process promotes high primary produc-
stream of the Peru margin? tion and uptake of CO2 from the atmosphere and the

11 of 15
PA3012 DEZILEAU ET AL.: PAST PRODUCTIVITY OFF CHILE PA3012

corresponding high rates of export of carbon to the depth, ciently available in the ocean surface waters, limiting
upwelling promotes also a strong outgassing of CO2 to primary productivity in the region.
the atmosphere. Hence increased upwelling and primary
production do not necessarily mean a significant net
increase in uptake of atmospheric CO2. We have shown Appendix A: Calculating Age-Corrected Excess
here that past changes in productivity in the upwelling 230Th Activity at the Time of Deposition
system of the eastern South Pacific may have been
controlled mainly by the availability of Andean iron, [37] There are two prerequisites for utilization of 230Th
through changes in the hydrological conditions, and that data to determine vertical rain rates. First, the measured
this availability has been tightly linked to orbital forcing concentrations of 230Th have to be corrected both for
during the last 100,000 years. Therefore variations in detrital and authigenic contributions produced from decay
paleoproductivity in major coastal upwelling systems of 234U in the sediment in order to calculate the amount of
230
due to iron fertilization may have had a significant Th originating from scavenging in the oceanic water
influence on past variations in atmospheric pCO2. How- column (excess concentrations). Second, an independent
ever, we also show that the Chilean upwelling region timescale for the core is necessary to calculate the
never really has been highly productive in comparison corresponding value at the time of deposition, (230Th
ex),
with other upwelling regions or the Southern Ocean in for each measured 230Thex content.
view of the very low biogenic opal deposition and low [38] The total 238U measured in the samples consists of
Corg contents. Thus the potential importance of this two components: uranium present in detrital minerals
region as an area of glacial CO2 drawdown has most (238Ud) and authigenic uranium (238U auth) derived from
likely been negligible. This is, however, an important seawater.
result because it has been argued that the middle and low [39] Detrital uranium (238Ud) is estimated as:
latitude HNLC areas of the oceans such as the coastal
upwelling regions may have been an important sink for

glacial atmospheric CO2. 238
Ud ¼ 238
U=232 Th  232
Th; ðA1Þ
d

7. Conclusions where (238U/232Th)d is the detrital U/Th ratio and (232Th)s is

[35] We have analyzed variations of ocean productivity the measured content of 232Th in the sediment samples.
232
through the last 100,000 years in the upwelling system of Th is considered to be exclusively present in detrital
the eastern South Pacific, as recorded in sediments from minerals [Anderson, 1982].
two gravity cores retrieved at the same location from the [40] The authigenic uranium (238U auth) contents of the
continental slope off northern Chile. We demonstrated samples were then calculated using the relation:
that strong changes in accumulation rates recorded in
these cores were due to changes in lateral transport by
bottom currents rather than reflecting changes in particle 238
Uauth ¼ 238 Umeas  238 Ud ;
settling rates from the overlying water column. In this
case, opal and organic carbon accumulation rates cannot
be used to assess past changes in productivity. This
information can be obtained, however, by normalization where 238Umeas is the measured total uranium content.
to 230Th which corrects for syndepositional sediment [41] In order to calculate the activities of detrital and
redistribution. authigenic uranium, François et al. [1993] assumed a
[36] Biogenic opal and organic carbon vertical rain (238U/232Th)d activity ratio of the detrital sediment phase
rates and the opal/carbonate ratio reveal a strong cy- in the range of 0.5 to 1. We use 232Th  0.75, which
clicity, which is dominated by the precessional period represents the estimate of the mean activity of detrital 238U
(
20,000 years). Iron vertical rain rates also show a in dpm/g [cf. Wedepohl, 1995].
strong precessional cyclicity. The patterns of iron and [42] The following equations (equation (A1)) are applied
terrigenous material vertical rain rates are remarkably to calculate the age-corrected excess 230Th activity (noted
230
similar to the down-core records of the proxies for
):
Thex
biogenic productivity, suggesting iron availability to be
a main factor responsible for past changes in productivity.
During precessional maxima increased precipitation in the
region, for example due to a northward shift of the 230 230
 
Thex ¼ Thmeas:  232 Th  0:75
southern westerlies or an ENSO-like mechanism, may
238 h

have brought high inputs of iron from the Andes to the  Uauth:  1  el230 t
coastal ocean enhancing primary productivity there. Dur-



ing precessional minima, aridity was probably even þ l230 = l230  l234  1  eðl230 l234 Þt
stronger than today. Hence the sediment load of the

 i
ocean surface waters supplied by river discharges must  234 U =238 U 1 ðA2Þ
have been low and iron may thus not have been suffi- i

12 of 15
PA3012 DEZILEAU ET AL.: PAST PRODUCTIVITY OFF CHILE PA3012

with in oligotrophic regions to an maximum excess of circa

50% in high flux areas. In the extreme case, if we
238
232  assume a strong upwelling and high mass flux in our
Uauth: ¼ 238 Umeas:  Th  0:75 ;
study region, the 230Thex flux may be increased by a
factor of 1.5 by boundary scavenging. In that case, only
where (232Th  0.75) represents the estimate of the mean ratios between sediment accumulation rates and vertical
activity of detrital 238U in dpm g1; l230 and l234 are the sediment rain rates (called focusing factor by Suman and
corresponding decay constants of the 230Th and 234U; Bacon [1989]) above 1.5 may be interpreted as sediment
(234U/238U) refers to the initial activity ratio of 234U and redistribution. For the above reasons, the observed
238
U at the time of sediment deposition; and t is the age in 230
Th
ex fluxes (Fs) which are up to fivefold the 230Th
years estimated from the previously determined stratigra- production flux (Fw) cannot be due to increased bound-
phy. The 238Uauth. term represents the equilibrium activity ary scavenging but must have been caused by an increase
of 230Th produced from authigenic uranium, assuming of the focusing intensity.
uranium precipitation occurred at the time of sediment [46] The second assumption for estimation of the vertical
deposition. sediment rain rate is that the 230Th concentration in the
[43] After the stratigraphy is established, the excess redistributed sediment is equal to that of the sediment
activities of 230Thex are decay-corrected to the time of raining down from the overlying water column (i.e., the
deposition of the sediment (230Thex
): redeposited sediment came from an area of similar water
depth). As suggested by François et al. [1993], it is also
230
Thex ¼ 230 Thex  el230 t : possible that the redistributed sediment was initially depos-
ited at a shallower depth prior to being transported to its
final site of deposition. In this case, it can be shown that the
Appendix B focusing factor would underestimate the fraction of sedi-
[44] There are several assumptions and problems that may ment brought by lateral transport and vertical paleofluxes
lead to uncertainties in calculating the vertical rain rates: would overestimate true rain rates. However, several lines
[45] The first assumption is that lateral isopycnal trans- of evidence suggest that this problem has not been impor-
port of dissolved 230Th within the water column (bound- tant for core GeoB7101-1. If lateral transport of sediment
ary scavenging) from high to low particle flux areas, is initially deposited in shallower areas significantly affected
negligible. Boundary scavenging occurs when marginal paleoflux estimates, we would expect a correlation between
sediments receive much greater fluxes of particle-reactive the two, i.e., the higher the focusing factor, the higher the
elements than open ocean sediments. An increased lateral vertical particle rain rate calculated by the 230Th method
transport of isotopes along isopycnals should result in would be. Since this correlation is not observed (Figures 3a
increased radionuclide fluxes to the sediment and should and 3b), the redistributed sediment must have come from an
coincide with increased sedimentation rates. In this study, area of similar depth. Moreover, visual examination of the
core GeoB7101-1 has higher 230Thex
fluxes (Fs) than the sediment core did not reveal any hints for turbidites
230
Th production flux (Fw) (Figure 3b). So, an increased throughout the core. Pathways of turbidity currents are
amount of ‘‘boundary scavenging’’ due to a higher channeled and restricted to rare submarine canyons on the
particle rain rate could explain the high 230Th fluxes. continental slope off northern Chile [Thornburg and Kulm,
However, several lines of evidence suggest that this effect 1987].
is unimportant for our core. Indeed, in different oceanic
regions, François et al. [1990] and Yu et al. [2001] [47] Acknowledgments. We thank A. Montecinos, C. Lange, and
showed that while total mass flux collected by sediment T. Platt for useful comments and discussions and R. De Pol and G. Yuras
traps varied 15-fold between low and high productivity for help with the figures. This work was supported by the Chilean National
Commission for Scientific and Technological Research (CONICYT)
regions, the accompanying flux of 230Th varied from a through the FONDAP Programme. LD was supported by a postdoctoral
deficit (compared to 230Th production rate) of circa 40% fellowship from ECOS Sud (France) and Fundación Andes (Chile).

References
Aceituno, P., and F. Vidal (1990), Variabilidad in productivity of the Southern Ocean, J. Mar. Bentaleb, I., C. Caratini, M. Fontugne, M. T.
interanual en el caudal de rı́os andinos Syst., 17, 497 – 514. Morzadec-Kerfourn, J. P. Pascal, and C. Tissot
de Chile Central en relación con la tem- Anderson, R. F., Z. Chase, M. Q. Fleisher, and (1997), Monsoon regime variations during the
peratura de la superficie del mar en el J. Sachs (2002), The Southern Ocean’s biolo- late Holocene in the southwestern India, in
Pacı́fico Central, Rev. Soc. Ing. Hidraul., 5, gical pump during the Last Glacial Maximum, Third Millennium BC Social Collapse and Cli-
9 – 17. Deep Sea Res. Part II, 49, 1909 – 1938. mate Change, edited by N. Dalfes, G. Kukla,
Anderson, R. F. (1982), Concentration, vertical Arrhenius, G. (1952), Sediment cores from the and H. Weiss, pp. 475 – 488, Springer-Verlag,
flux, and remineralization of particulate ura- east Pacific, Deep Sea Exped., 5, 189 – 201. New York.
nium in seawater, Geochim. Cosmochim. Acta, Bacon, M. P. (1984), Glacial to interglacial Berger, A., and P. J. Loutre (1991), Insolation
46, 1293 – 1299. changes in carbonate and clay sedimenta- values for the climate of the last 10 million
Anderson, R. F., N. Kumar, M. R. A. Mortlock, tion in the Atlantic ocean estimated from years, Quat. Sci. Rev., 10, 297 – 317.
230
P. N. Froelich, P. Kubik, B. Dittrich-Hannen, Th measurements, Isot. Geosci., 2, 97 – Berger, W. H., V. S. Smetacek, and G. Wefer
and M. Sutter (1998), Late Quaternary changes 111. (1989), Ocean productivity and paleoproduc-

13 of 15
PA3012 DEZILEAU ET AL.: PAST PRODUCTIVITY OFF CHILE PA3012

tivity—An overview, in Productivity of the stratification to low atmospheric CO2 concen- radiometer data, J. Plankton Res., 17, 1245 –
Ocean: Present and Past, edited by W. H. trations during the last glacial period, Nature, 1271.
Berger, V. S. Smetacek, and G. Wefer, pp. 1 – 389, 929 – 935. Loubere, P. (2000), Marine control of biological
34, John Wiley, Hoboken, N. J. François, R., M. Frank, M. M. Rutgers van der production in the eastern equatorial Pacific
Brink, K. H., and T. J. Cowles (1991), The Loeff, and M. P. Bacon (2004), 230Th normal- Ocean, Nature, 406, 497 – 500.
coastal transition zone program, J. Geophys. ization: An essential tool for interpreting sedi- Loubere, P., and R. Murray (2001), Late Pleisto-
Res., 96, 14,637 – 14,648. mentary fluxes during the late Quaternary, cene bioproductivity in the eastern equatorial
Broecker, W. S. (1982), Glacial to interglacial Paleoceanography, 19, PA1018, doi:10.1029/ Pacific: Patterns and comparison of proxies,
changes in ocean chemistry, Prog. Oceanogr., 2003PA000939. Eos Trans. AGU, 82(20), Abstract PP12A-
11, 151 – 197. Frank, M., A. Eisenhauer, W. J. Bonn, P. Walter, 0478.
Chase, Z., R. F. A. Anderson, M. Q. F. Fleisher, H. Grobe, P. W. Kubik, B. Dittrich-Hannen, Loubere, P., M. Fariduddin, and R. W. Murray
and P. W. Kubik (2003), Accumulation of bio- and A. Mangini (1995), Sediment redistribu- (2003), Patterns of export production in the
genic and lithogenic material in the Pacific tion versus paleproductivity change: Weddell eastern equatorial Pacific over the past
sector of the Southern Ocean during the past Sea margin sediment stratigraphy and biogenic 130,000 years, Paleoceanography, 18(2),
40,000 years, Deep Sea Res. Part II, 50, 799 – particle flux of the last 250.000 years deduced 1028, doi:10.1029/2001PA000658.
832. from 230Thex, 10Be and biogenic barium pro- Lyle, M., D. W. Murray, B. P. Finney,
Clement, A. C., R. Seager, and M. A. Cane files, Earth Planet. Sci. Lett., 136, 559 – 573. J. Dymond, J. M. Robbins, and K. Brooksforce
(2000), Suppression of El Niño during the Frank, M., R. Gersonde, M. Rutgers Van der (1988), The record of late Pleistocene biogenic
mid-Holocene by changes in the Earth’s orbit, Loeff, G. Bohrmann, C. C. Nurnberg, P. W. sedimentation in the eastern tropical Pacific
Paleoceanography, 15, 731 – 737. Kubik, M. Suter, and A. Mangini (2000), Ocean, Paleoceanography, 3, 39 – 59.
Coale, K. H., et al. (1996), A massive phyto- Similar glacial and interglacial export biopro- Lyle, M. W., F. G. Prahl, and M. A. Sparrow
plankton bloom induced by an ecosystem-scale ductivity in the Atlantic sector of the Southern (1992), Upwelling and productivity changes
iron fertilization experiment in the equatorial Ocean: Multiproxy evidence and implications inferred from a temperature record in the cen-
Pacific Ocean, Nature, 383, 495 – 501. for glacial atmospheric CO2, Paleoceanogra- tral equatorial Pacific, Nature, 355, 812 – 815.
Coale, K. H., et al. (2004), Southern Ocean iron phy, 15(6), 642 – 658. Mariotti, A., F. Gadel, P. Giresse, and X. Kinga-
enrichment experiment: Carbon cycling in Ganeshram, R., T. Pedersen, S. Calvert, Mouzeo (1991), Carbon isotope compositions
High and low Si waters, Science, 304, 408 – G. McNeill, and M. Fontugne (2000), Glacial- and geochemistry of particulate organic matter
414. interglacial variability in denitrification in the in the Congo river: Application to the study of
Daneri, G., et al. (2000), Primary production and world’s ocean: Causes and consequences, quaternary sediments of the mouth river,
community respiration in the Humboldt Cur- Paleoceanography, 15, 361 – 376. Chem. Geol., 86, 345 – 357.
rent system off Chile and associated oceanic Hebbeln, D., M. Marchant, T. Freudenthal, and Martin, J. H. (2000), Glacial-interglacial CO2
areas, Mar. Ecol. Prog. Ser., 197, 41 – 49. G. Wefer (2000), Surface sediment distribution change: The iron hypothesis, Paleoceanogra-
De Baar, H. J. W., et al. (1990), On iron limita- along the Chilean continental slope related to phy, 5, 1 – 13.
tion of the Southern Ocean: Experimental upwelling and productivity, Mar. Geol., 164, Martin, J. H., and S. E. Fitzwater (1988), Iron
observations in the Weddell and Scotia Seas, 119 – 137. deficiency limits phytoplankton growth in the
Mar. Ecol. Prog. Ser., 65, 105 – 122. Hebbeln, D., M. Marchant, and G. Wefer (2002), northeast Pacific subarctic, Nature, 331, 341 –
Dezileau, L., G. Bareille, J. L. Reyss, and Paleoproductivity in the southern Peru-Chile 343.
F. Lemoine (2000), Evidence for strong sedi- Current through the last 33,000 years, Mar. Montecinos, A., and P. Aceituno (2003), Season-
ment redistribution by bottom currents along Geol., 186, 487 – 504. ality of the ENSO-related rainfall variability in
the Southeast Indian Ridge, Deep Sea Res. Henderson, G. M., C. Heinze, R. F. Anderson, central Chile and associated circulation anoma-
Part I, 47, 1899 – 1936. and A. M. E. Winguth (1999), Global distribu- lies, J. Clim., 16, 281 – 296.
Dezileau, L., G. Bareille, and J. L. Reyss (2002), tion of the 230Th flux to ocean sediments con- Müller, P. J., and R. Schneider (1993), An auto-
Enrichments in authigenic uranium in glacial strained by GCM modelling, Deep Sea Res. mated leaching method for the determination
sediments of the Southern Ocean, C. R. Part, 46, 1861 – 1893. of opal in sediments and particulate matter,
Geosci., 334, 1039 – 1046. Hutchins, D. A., and K. W. Bruland (1998), Iron- Deep Sea Res., 40, 425 – 444.
Dezileau, L., J. L. Reyss, and F. Lemoine (2003), limited diatom growth and Si:N uptake ratios Oberhänsli, H., P. Heinze, L. Diester-Haass, and
Late Quaternary changes in biogenic opal in a coastal upwelling regime, Nature, 393, G. Wefer (1990), Upwelling off Peru during
fluxes in the Southern Indian Ocean, Mar. 561 – 564. the last 430,000 yr and its relationship to the
Geol., 202, 143 – 158. Hutchins, D. A., et al. (2002), Phytoplankton bottom water environment, as deduced from
Emerson, S., and J. I. Hedges (1988), Processes iron limitation in the Humboldt Current and coarse grain size distributions and analyses of
controlling the organic carbon content of open Peru upwelling, Limnol. Oceanogr., 47, 997 – benthic foraminifers at hole 679D and 681B,
ocean sediments, Paleoceanography, 3, 621 – 1011. Leg 112, Proc. Ocean Drill. Project, Sci.
634. Kay, S. M., C. Mpodozis, V. A. Ramos, and Results, 112, 369 – 390.
Fitzwater, S. E., K. H. Coale, R. M. Gordon, F. Munizaga (1991), Magma source variations Paillard, D., L. Labeyrie, and P. Yiou (1996),
K. S. Johnson, and M. E. Ondrusek (1996), for mid-late Tertiary magmatic rock associated Macintosh program performs time-series ana-
Iron deficiency and phytoplankton growth in with a shallowing subduction zone and a thick- lysis, Eos Trans. AGU, 77, 379.
the equatorial Pacific, Deep Sea Res. Part II, ening crust in the central Andes (28 – 33 S), in Pedersen, T. F., and S. E. Calvert (1990),
43, 995 – 1015. Andean Magmatism and Its Tectonic Setting, Anoxia versus productivity: What controls
Fontugne, M. R., and J. C. Duplessy (1981), Spec. Pap., vol. 265, edited by R. S. Harmon the formation of organic carbon rich sedi-
Organic carbon isotopes fractionation by mar- and C. W. Rapela, pp. 113 – 137, Geol. Soc. of ments and sedimentary rocks?, AAPG Bull.,
ine plankton in the temperate range 1 to Am., Boulder, Colo. 74, 454 – 466.
31
C, Oceanol. Acta, 4, 85 – 90. Kumar, N., R. F. Anderson, R. A. Mortlock, P. N. Pichowiak, S. (1994), Early Jurassic to early
François, R., M. P. Bacon, and D. O. Suman Froelich, P. Kubik, P. Dittrich-Hannen, and Cretaceous magmatism in the coastal cordil-
(1990), Thorium 230 profiling in deep-sea M. Suter (1995), Increased biological produc- lera and the central depression of north
sediments: High-resolution records of flux tivity and export production in glacial South- Chile, in Tectonics of the Southern Central
and dissolution of carbonate in the equatorial ern Ocean, Nature, 378, 675 – 680. Andes, edited by Reutter, Scheuber, and
Atlantic during the last 24,000 years, Paleo- Lamy, F., D. Hebbeln, and G. Wefer (1998), Late Wigger, pp. 203 – 217, Springer-Verlag, New
ceanography, 5(5), 761 – 787. Quaternary precessional cycles of terrigenous York.
François, R., M. P. Bacon, M. A. Altabet, and sediment input off the Norte Chico, Chile Quay, P. D., D. O. Wilbur, J. E. Richey, J. I.
L. D. Labeyrie (1993), Glacial/interglacial (27.5
S) and paleoclimatic implications, Hedges, A. H. Devol, and R. Victoria
changes in sediment rain rate in the SW Indian Palaeogeogr. Palaeoclimatol. Palaeoecol., (1992), Carbon cycling in the Amazon River:
sector of Subantarctic waters as recorded by 141, 233 – 251. Implication from the 13C compositions of
230
Th, 231Pa, U, and d15N, Paleoceanography, Lamy, F., J. Klump, D. Hebbeln, and G. Wefer particles and solutes, Limnol. Oceanogr.,
8(5), 611 – 629. (2000), Late Quaternary rapid climate change 37, 857 – 871.
François, R., M. A. Altabet, E.-F. Yu, D. M. in northern Chile, Terra Nova, 12, 8 – 13. Sarnthein, M., K. Winn, J. C. Duplessy, and
Sigman, M. Frank, M. P. Bacon, G. Bohrmann, Longhurst, A., S. Sathyendranath, T. Platt, and M. R. Fontugne (1988), Global variations of
G. Bareille, and L. D. Labeyrie (1997), C. Caverhill (1995), An estimate of global pri- surface ocean productivity in low and mid
Contribution of southern ocean surface-water mary production in the ocean from satellite latitudes: Influence on CO2 reservoirs of the

14 of 15
PA3012 DEZILEAU ET AL.: PAST PRODUCTIVITY OFF CHILE PA3012

deep ocean and atmosphere during the last cords, edited by C. P. Summerhayes pp. 1 – sediments, Geochim. Cosmochim. Acta,
21,000 years, Paleoceanography, 3, 361 – 399. 37, John Wiley, Hoboken, N. J. 66(17), 3085 – 3092.
Schneider, E. K., Z. Zhu, B. S. Giese, B. Huang, Thornburg, T. M., and L. D. Kulm (1987), Sedi-
B. P. Kirtman, J. Shukla, and J. A. Carton mentation in the Chile Trench: Depositional
(1997), Annual cycle and ENSO in a coupled morphologies, lithofacies, and stratigraphy,
ocean-atmosphere general circulation model, Geol. Soc. Am. Bull., 98, 33 – 52. 
Mon. Weather Rev., 125, 680 – 702. Toggweiler, J., D. Dixon, and W. Broecker L. Dezileau, Laboratoire de Dynamique de la
Schrader, H., and R. Sorknes (1990), Spatial and (1991), The Peru upwelling and the ventilation Lithosphère, Université de Montpellier 2, Mont-
temporal variation of Peruvian coastal of the South Pacific thermocline, J. Geophys. pellier F-34095, France. ([email protected]
upwelling during the latest Quaternary, Leg Res., 96, 20,467 – 20,497. montp2.fr)
112, Proc. Ocean Drill. Project, Sci. Results, Wedepohl, K. H. (1995), The composition of the M. Fontugne and J.-L. Reyss, Laboratoire des
112, 391 – 406. continental crust, Geochim. Cosmochim. Acta, Sciences du Climat et de l’Environnement,
Suman, D. O., and M. P. Bacon (1989), Varia- 59, 1217 – 1232. CNRS/Commissariat à l’Energie Atomique,
tions in Holocene sedimentation in the North Yu, E. F., R. Francois, M. P. Bacon, and A. P. F-91198 Gif-sur-Yvette, France.
American basin determined from 230Th mea- Fleer (2001), Fluxes of 230Th and 231Pa to the D. Hebbeln and F. Lamy, Fachbereich Geo-
surements, Deep Sea Res., 36(6), 869 – 878. deep sea: Implications for the interpretation of wissenschaften, Universität Bremen, Postfach
Summerhayes, C. P., K.-C. Emeis, M. V. Angel, excess 230Th and 231Pa/230Th profiles in sedi- 330440, D-28334 Bremen, Germany.
R. L. Smith, and B. Zeitzschel (1995), ments, Earth. Planet. Sci. Lett., 191, 219 – 230. O. Ulloa, Centro de Investigación Oceanográ-
Upwelling in the ocean: Modern processes Zheng, Y., R. F. Anderson, A. Van Geen, fica, Programa Regional de Oceanografia Fisica
and ancient records, in Upwelling in the and M. Q. Fleisher (2002), Preservation of y Clima, Departamento de Oceanografı́a, Uni-
Ocean: Modern Processes and Ancient Re- particulate non-lithogenic uranium in marine versidad de Concepción, Concepción, Chile.

15 of 15
PA3012 DEZILEAU ET AL.: PAST PRODUCTIVITY OFF CHILE PA3012

Figure 1. Composite of annual average of phytoplankton pigment concentration (Chl) of the eastern
South Pacific and annual maximum normalized difference vegetation index (NDVI) of the central and
southern Andean region (both for year 2000). These data were derived from the SeaWiFS sensor
(SeaWiFS Project) Distributed Active Archive Center. Top left inset: map of the study area with location
of the two cores GeoB 3375-1 and GeoB 7101-1 (27
280S, 71
150W). Bottom left inset: chlorophyll and
nutrient profiles of the upper 100 m depth at station GeoB 3377-1 (27
280S; 71
31.50W). Right inset:
average of annual continental rainfall data in Chile (http://www.meteochile.cl).

2 of 15