QuaternaryInternational,Vol. 31, pp. 13-18, 1996.
Copyright© 1995INQUA/EisevierScienceLtd
Printed in Great Britain.All rights reserved.
1040-6182196$29.00
Pergamon
1040-6182(95)00017-8
DYNAMICS OF THE MONSOON IN THE EQUATORIAL INDIAN OCEAN OVER THE
LAST 260,000 YEARS
Luc Beaufort
Laboratoire de G~ologie du Quaternaire, CNRS-CEREGE, 13549 Aix-en-Provence, France
The coccolithophorids are excellent indicators of the variations in ocean productivity. As an example, the relative abundance of Florisphaera
profunda, a species which lives exclusively in the deepest part of the photic zone, constitutes a reliable monitor of the depth of the nutricline
as a function of climatic variability. Productivity in the Indian Ocean is closely linked with the monsoon. Its intensity affects the depth of the
mixing zone which deepens as the wind stress increases. During the monsoon season, the nutricline is shallow resulting in a higher
productivity. The variations in composition of eoccolith assemblages in well-dated oceanic sediments allows to describe the dynamics of
productivity and therefore the fluctuation of the monsoon intensity. Core MD900963 was retrieved in the equatorial Indian Ocean, East of
the Maldives at a water depth of 2400 m. A precise chronology was established by fine-tuning the high resolution 81sO record with the
SPECMAP Stack. Coccoliths were counted from samples taken at 10 cm intervals in the core, providing a resolution of 2000 years for the
last 260,000 years. The productivity estimates made from coccolith counts show that productivity varied greatly in the Maldives area during
this time. The variations of the coceolith productivity index match the variations of the organic carbon content in the sediment. Spectral
analysis reveals strong precessional cycles, and weak obliquity and eccentricity cycles. This implies that the solar radiation has a dominant
effect on the monsoon variability South of India. The productivity maxima occur during even-numbered 81sO stages, This may indicate that
the productivity events are related to increase of the winter monsoon. However the analysis of the phase between the palaeowoductivity and
seasonal solar radiation curves calculated for the past 260,000 years suggest an alternative hypothesis: the westerlies which blow in April in
the Maldives area (there, the onset of the Monsoon is in advance on the Arabian Sea) would have a major effect on productivity, since the
insolation curve of the end of March at 5"N is in perfect phase with the paleoproductivity record for the last 260,000 years.
INTRODUCTION
The Asian Monsoon climate, driven by differential
land-ocean sensible heating, produces seasonal reversals in
the wind direction which affect strongly the Northern Indian
Ocean. Palaeoelimate studies have shown that the intensity
of the monsoon has varied through time, during the Last
Glacial Maximum the summer monsoon was weaker (Prell
et al., 1980; Duplessy, 1982), and the winter monsoon was
stronger (Duplessy, 1982; Fontugne and Duplessy, 1986;
Sarkar et al. 1990). Clemens et al. (1991) suggested that the
variations of the summer monsoon intensity are associated
with changes in the seasonal pattern of the solar radiation,
whereas variability in global ice volume is not a primary
factor controlling this intensity, which is not in agreement
with general circulation model. The summer monsoon
variability is recorded particularly well in the Arabian Sea
where the strongest winds appears during summer (e.g.
Knox, 1987). The seasonal variations of the wind stress over
the surface water induce dramatic changes on oceanic
productivity (e.g. Ittekkot et al., 1992; Curry et al., 1992)
which can be identified in the fossil records from
palaeoproductivity proxies (e.g. Cullen, 1981; Prell and
Curry, 1981; Clemens and Prell, 1990; Canlet et al., 1992).
The effects produced by the summer monsoon are
particularly well observed in the areas of the Somalian and
Arabian margins where seasonal upwelling are induced by
the SW (summer) monsoon while it is stopped by the NE
(winter) monsoon. The record of palaeoproductivity
variations in these upwelling systems corresponds
essentially to the dynamic of the summer monsoon. At the
opposite, records of the variability of the winter monsoon
were found in the Bay of Bengal and Andaman Sea
(Fontugne and Duplessy, 1986) where this monsoon season
is predominant. Today, very little is known about the
variability of the seasonal winds over the rest of the Indian
Ocean, where seasonal contrasts are less pronounced
(regions where both monsoon seasons are recorded). The
giant piston core MD900963 retrieved during the
SEYMAMA expedition of the French R/V Marion Dufresne
in 1990, at the junction between the Bay of Bengal and the
Arabian Sea, provides an excellent record for investigating
the dynamics of the monsoon in the equatorial Indian Ocean.
COCCOLITHS AS PALAEOPRODUCTIVITY
INDICATORS
The coccolithophores (Primnesiophyceae) constitute one
of the major phytoplanktonic group. As primary producers
their growth depend on the amount of light and nutrients
available in the photic zone. The photic zone in a stratified
ocean presents strong contrasts of these two parameters on a
vertical scale: the upper photic zone (0 to -60 m) is depleted
in nutrients, whereas light is strongly limited in the lower
photic zone (-60 to -180 m). The phytoplanktonic
communities are adapted to these physicochemical
differences and in consequence, their composition differs
with depth (Venriek, 1982, 1990). At low latitude, the lowerphotic-zone coccolithophore communities are dominated by
Florisphaera profunda and Thorosphaera flabellata while
most of the other species live in the upper part of the photic
zone (Okada and Honjo, 1973). That aspect of the coccolith
vertical zonation has been used successfully for
palaeoproductivity studies by Molfino and McIntyre (1990),
13
14
L. Beaufort
who use the abundance of F. profunda in fossil assemblages
to monitor the climatic control of the depth of the nutrieline.
F. profunda increases in abundance in coccolith assemblages
when the upper photic zone is impoverished in nutrients,
hence, when productivity decreases. In contrast, the
abundance of species such as Emiliania huxleyi and
Gephyrocapsa oceanica are high when productivity
increases in a given oceanic area. Productivity in the Indian
ocean is closely linked with the monsoon intensity. The wind
stress affects the depth of the mixing zone (deeper when
wind stress is high). In other words, during the monsoon
season, the nutrieline is shallower, resulting in a higher
productivity. The coecolith index based on the relative
abundance of F. profunda is a good tool for reconstructing
the past variations of the windstress and thus of the monsoon.
zo.
lo.
o..
10Z0•
MATERIAL AND M E T H O D
40"
Core MD900963 was retrieved on the eastern slope of the
Maldives Islands at 5*03'30 N and 73"52'60 E in 2446 m of
water depth (Fig. 1). More than 50 m of calcareous
nannofossil ooze with foraminifera were recovered. Oxygen
stable isotope measurements were performed on the
planktonic foraminifer Globigerinoides ruber (white)
(Bassinot et al., 1994) providing a detailed chronological
framework when compared to the 81sO SPECMAP stack
(Fig. 2). The average sedimentation rate is 5 cm/1000 year.
The core was sampled at a 10 cm interval which corresponds
to a chronologic resolution of about 2000 years. The upper
13 m of the core studied here provides a continuous record
for the last 260,000 years. A smear slide was prepared from
each sample; 300 coccoliths per slide were counted and the
percentage of F. profunda was established from this count.
The 95% confidence interval varies between ±2 and ±6%
depending on the percentage (Mosimann, 1965). Settling
slides were prepared from 56 levels which permit to count
the absolute abundance of coccoliths (Beaufort, 1992). The
number of view fields needed to obtain a total number of 300
coccoliths (F. profunda excluded) were counted. The
number of coccoliths per gram of sediment is calculated
using the equation in Beaufort (1992).
RESULTS
Large variations of the F. profunda index are observed,
which range between 5 and 85% (Fig. 3), much above the
confidence interval. All peaks and troughs which shape the
curve are well defined, being delineated by several
successive points. The sampling interval in thus sufficient
enough to characterize the variations is the core. Based on
our knowledge, of the ecology of F. profunda (Okada and
Honjo, 1973), an increase of the relative abundance of this
species in coccolith assemblages reflects a deepening of the
nutricline equivalent to a decrease of primary productivity in
the photic zone (Molfino and Melntyre, 1990). The curve is
presented on a reverse scale, in order to have higher
productivity in the upper part of the graph (Fig. 3).
Rostek et al. (1993) measured the Co~ content in the upper
10 m of the Core MD900963. An increase in the organic
carbon content in the sediment is often interpreted as a result
$0"
60"
70 °
80"
90"
FIG I Location of Core M D 900963
-2
-3-
-1.5
-2.5 -
-2-
-0.5
-1.5-
0
0.5
-1-
1
-0.5 -
i/
1.5
:"
I
I
I
I
I
I
I
I
I
I
I
2
FIG. 2. Comparison of the 61sO record of the planktonic foraminifera G.
ruber obtained from Core MD 900963 (Bassinot et al., 1994) and the
SPECMAP Stack after correlation.
of an increase of productivity in the surface water (e.g.
MUller and Suess, 1979). Similar variations in productivity,
are inferred independently from percentage variations of F.
profunda and Con (Fig. 3). The good match between the two
series conforts t h e interpretation of F. profunda as a
palaeoproductivity indicator. Higher productivity is
observed in both series around 18, 70, 90, 110, 130, and 160
kyr during even-numbered 5nO stages.
Bassinot et al. (1994) found that the planktonic
foraminifera fragmentation in Core MD900963 was more
important at levels of high organic content, indicating that
the dissolution of the foraminifera was higher during high
productivity events due to a higher oxidation of the organic
matter incorporated into the sediments. The F. profunda
index cannot reflect dissolution for at least two reasons: (1)
no evidence of dissolution was seen qualitatively on the
coccolith as seen in the light microscope, (2) the total
abundance of the coccoliths increases at level richer in
organic carbon and foraminifera fragments, levels where the
dissolution should have been greater. The decrease of the
relative abundance of F. profunda results therefore from an
increase of the absolute number of coceoliths due to higher
productivity and not from increased dissolution.
MonsoonDynamicsin the IndianOcean
15
0 ~
j,o.!
.|
~
80-
700-
-
/
100
0.9
I
BI
0"8
t
0.7
0.6
0.5-
0.4=
0.3.
0.2
70t
.m 60=
E
P'
&
4,P
=u
,E
50-
40-
4~
|
0
!
!
4O
I
SO
)
'i~C
)
I
!
160
!
I
200
!
!
24O
I
Aoe Oqrr P )
FIG. 3. Variationsof the percentageof F. profunda in the coccolithassemblageof Core MD 900963 (uppergraph)in comparisonwith
other proxies: absoluteabundancein placoliths,organic carbon content (from Bassinotet al., 1994) and percentageof foraminifera
fragmentsversusentireforaminifera(fromBassinotet al., 1994).The shadedareascorrespondto periodof higherproductivity.
SPECTRAL ANALYSIS
The good chronological framework and the high
resolution in the series are favourable for spectral analysis.
The statistical techniques used are standard procedures based
on the Fourier transform, which permits an estimation of the
power density spectrum of series, of the coherency arid phase
between two series, and of filtering of the series at given
frequencies (band pass filtering). The programmes used
belong to a package written by D. Paillard (CFR, GiffYvette,
France) in a large part inspired from routines of SPECMAP
programs and Numerical Recipes (Press et al., 1986).
16
L. Beaufort
Spectral estimates of these records are made using both
1.5
Blackman-Tukey and maximum entropy algorithms (e.g.
:25 ky~
Jenkins and Watts, 1968; Press et al., 1986). The maximum
! :\
entropy method gives very high resolution spectra which
help to precisely estimate frequencies peaks. With this latter
1
ew
method, however, there is no test of evaluating the
significance of the peaks. Also, an abundance of spurious
peaks may be found when too high an order of resolution is
chosen (Press et al., 1986). These authors recommend the
o.~ I'~' ky,
use of this algorithm in conjunction with more conservative
methods. The Blackman-Tukey technique is used for this
. .,,
:
:
.
purpose with the maximum entropy because it is usually
considered statistically robust, allowing an estimation of the
Ol
I
I
I
I
I
significance of peaks; it is often used by palaeoo
0.oa
0.o4
o.oe
o.o8
o.1
Frequency (Cycle kyr "I)
climatologists. The phases and coherencies between series
are calculated by Cross-Blackman-Tukey.
Spectral analysis of the F. profunda series shows FIG. 4. Periodograms of the F. profunda index. Solid line:
Blackman-Tukcy; Dashed line: maximum ena~apy; BW is bandwidth, and
significant power rising at 23 kyr, periods which closely the small horizontal bars within peaks are the lower bounds of one-sided
correspond to precession cycle (Berger and Loutre, 1991)
95% confidence intervals attached to estimates of B.-T. peaks heights.
(Fig. 4). This indicates that it is the Earth' precession cycles
which have the dominant effect on the monsoon climate.
200
Smaller peaks correspond to other known orbital cycles: 111
kyr correspond to the eccentricity and 40 kyr cycle is the
0.8
150
period of the obliquity. The secondary importance of the
obliquity signal is not surprising, since this parameter does
I 0.6
not affect dramatically the solar radiation at low latitude. The
eccentricity has a strong effect on the monsoon strength,
i 0.4
because the climatic conditions (e.g. sea-surface
temperature, sea level, land albedo and CO2 concentration)
0.2
o
during glacial periods were different enough from those in
interglacial times to affect the timing and strength of the
i
I
I
I
-50
monsoon (Prell and Kutzbach, 1987) (the glacial cycles in
0
0.02
0.04
0.06
0.08
0.1
the late Pleistocene are closely linked to the eccentricity
Ft=Iu~cy (o/de kyr "l)
periods, see Imbrie et al., 1993). A small (but significant) FIG. 5. Cross spectral analysis (Blackman-Tukey) of the F. profunda index
peak in the eccentricity band of the F. profunda series is in (inverted) versus the SPECMAP stack (inverted). The horizontal line
represents the 80~ confidence interval level.
this aspect surprising. Clemens et al. (1991) reached to the
same results in their study of the monsoon in the Western
Arabian Sea. This study provides additional evidence in
favour of a dominant effect of the solar radiations on the Arabian Sea from results of Clemens et al. ( 1991) have been
long-term variability of the monsoon. Cross-spectral also plotted (Fig. 6).
analysis between the F. profunda series (the curve is inverted
The simplest explanation which may be given for this
to give the increase of productivity in a positive way) and the phase opposition, is that the variations of intensity of the
8'sO SPECMAP Stack shows significant coherencies in the winter monsoon induced the fluctuations of productivity
Milankovitch frequency bands (Fig. 5). The series are almost observed in Core MD900963. From studies of cores
in phase opposition (173°). This corresponds to the fact that retrieved in the eastern Bay of Bengal and in the Andaman
the productivity events were observed during even- Sea, the winter monsoon appears to have been stronger
numbered isotopic stages. Cross-spectral analysis between during the Last Glacial Maximum (Fontugne and Duplessy,
the F. profunda series and the solar radiation received 21 1986; Sarkar et al., 1990). This could correspond to the
June at 65"N (which has the greatest influence on present case. However, these studies are based on variations
glaciations; Berger, 1978) indicates a significant coherency of salinity, which are related to the amount of precipitation
(0.94) at 23 kyr with a phase of 83" (not shown).
over India and the Indian Ocean, and thus to the SW
monsoon or to productivity variations in the area of winter
DISCUSSION
upwelling. These studies address a seasonal problem, the
effect of the summer monsoon being not recorded, whereas
It is not the summer monsoon which influenced past the present study describes an index of the annual windstress.
changes in productivity in the Maldives. There, the situation The effect of an increase of the winter monsoon may be
is opposite to what is observed in the western Arabian Sea easily balanced by the decrease of the summer monsoon. It
where the summer monsoon is well recorded. This is therefore complex to imagine that the winter monsoon has
opposition is visualized on the precession phase wheel, in varied, while the summer monsoon intensity remained
which the phase of palaeoproductivity index of the western constant. Moreover, it is difficult to predict a strong effect of
i
,i
11oo,,o,
17
Monsoon Dynamics in the Indian Ocean
Precession cycle
I m U l ~ June21
6SON
F.~
/InNl~bn
8! 8~coAtu.)
F,pedu~b /SPECMAP
\
a/-~
"-
0mln~e
FIG. 6. Coherence and phase for the precession cycle compared with the results of Clemens et al. (1991). The zero phase is set up to the
June perihelion. Position of the vectors indicate the lag in degree existing between the different proxies and the June perihelion. The lengths
of the vectors represent the coherency with the June perihelion. The vectors of Ba flux, Opal flux, G. bulloides, and Grain size are from the
study of Clemens et al. ( 1991) on Western Arabian Sea cores.
may be due to the lag existing between the timing of the
maximum production season and June 21. Looking at the
phase wheel this way, the maximum of production is in phase
480
with a maximum of insolation of March 27, which is the time
of the onset of the monsoon season on this area. This is
visualized by comparing the precession component of the F.
440
profunda series (filtered series at 23 kyr) and the insolation
received March 27 at 5°N (Fig. 7). The correspondence
42O
between the two curves is striking. If this explanation is
4O0
correct, it would imply that the records of past activity of the
monsoon should be diachronous in the Indian Ocean,
i
380
I
I
occurring earlier in the south-east and progressing towards
260
3OO
0
ao
lOO
160
200
the north-west. The seasonal position of the low-Jets over the
Indian Ocean would be the key for reconstructing the timing
FIG. 7. Solid line: filtered series of the inverted (to have productivity events
with positive values) percentage of F. profunda with a band pass filter of these maximum production events.
8
500
i
centered at 20 kyr with cutting frequency of 17 and 24 kyr. Dashed line:
Variations of the solar radiation at 5"N on March 27 (using the equation of
Berger, 1978 in a package written by D. Paillard (CFR, Gif/Yvette,
France)).
the winter monsoon on productivity east of the Maldives. If
a local upwelling were to be created seasonally in this zone,
it would be with a SW wind, therefore during summer. It
would be logical that the summer monsoon have the
dominant effect on the long-term variation of the annual and
wind stress in this area.
An alternative explanation of this opposition between the
western Arabian Sea and the Maldives on the phase wheel
may be given. The summer monsoon has an abrupt onset at
the end of May in the Arabian Sea. But further south, the
onset occurs earlier, and corresponds to the position of the
low-level Jets. For example, windstress observations made at
Gan in the Maldives during several years indicate that the
onset occurs between the end of March and the beginning of
April (Knox, 1987). The 'windy season' is usually over by
the end of May. The difference of phase obtained between
the F. profunda series and the insolation curve for June 21,
ACKNOWLEDGEMENTS
I thank M.-P. Aubry and E. Vincent for constructive review of the
manuscript. I am grateful to F. Bassinot, F. Rostek, E. Bard, Y. Lancelot and
L. Labeyrie for stimulating discussions. I am thankful to N. Buchet, for her
help in technical preparations. The French program FNEDC supported this
study. Those results were presented at the IUGS-Unesco CLIP program
meeting, August 1994. This is an LGQ contribution number 95002.
REFERENCES
Bassinot, F.C., Beaufort, L., Vincent, E., Labeyrie, L.D., Rostek, F., MUller,
P.J., Quidelleur, X. and Lancelot, Y. (1994). Coarse fraction
fluctuations in pelagic carbonate sediments from the tropical Indian
Ocean: a 1500 kyr record of carbonate dissolution. Paleoceanography,
9, 579--600.
Beaufort, L. (1992). Adaptation of the random settling method for
quantitative studies of calcareous nannofossils. Micropaleontology, 37,
415-418.
Berger, A.L. (1978). Long-term variations of caloric insolation resulting
from Earth's Orbital elements. Quaternary Research, 9, 1978.
Berger, A. and Loutre, M.F. (1991). Insolation values for the climate of the
last 10 million years. Quaternary Science Reviews, 10, 297-317.
Caulet, J.-., V6nec-Peyr~, M.-T., Vergnand-Grazzini, C. and Nigrini, C.
18
L. Beaufort
(1992). Variation of South Somalian upwelling during the last 160 ka:
radiolarian and foraminifera records in core MD85674. In:
Summerhayes, C.P., Prell, W.L., Eraeis, K.C. (eds), UpwellingSystems:
Evolution Since the Early Miocene, pp. 379-389, Geological Society
Special Publication, No. 64, London.
Clemens, S.C. and Preil, W.L. (1990). Late Pleistocene variability of
Arabian sea summer monsoon winds and continental aridity: eolian
records from the lithogenic component of deap-sea sediments.
Paleoceanography, 5, 109-145.
Clemens, S., Prell, W., Murray, D., Shimmield, G. and Graham, W. ( 1991).
Forcing mechanisms of the Indian Ocean monsoon. Nature, 353,
720-725.
Cuilen, J.L. (1981). Microfossil evidence for changing salinity patterns in
the Bay of Bengal over the last 20,000 years. Palaeogeography
Palaeoclimatology, Palaeoccology,35, 315-356.
Curry, W.B., Ostermann, D.R., Guptha, V.S. and Ittekkot, V. (1992).
Foraminiferal production and monsoonal upwellin8 in the Arabian Sea:
evidence from sediments traps. In: S ~ h a y e s , C.P., Prell, W.L.,
Emeis, K.C. (eds), Upwelling Systems: Evolution since the early
Miocene, pp. 93-106, Geological Society Special Publication, No. 64,
London.
Duplessy, J.-C. (1982). Glacial to interglacial contrasts in the northern
Indian Ocean. Nature, 295, 494-.498.
Fontugne, M.R. and Duplessy, J.-C. (1986). Variations of the monsoon
regime during the upper quaternary: evidence from carbon isotopic
record of organic matter in North Indian sediment cores.
PalaeogeographyPalaeoclimatologyPalaeoccoiogy,$6, 69-88.
Imbrie, I., Berger, A., Boyle, E.A., Clemens, S.C., Duffy, A., Howard,
W.R., Kukla, G., Kutzbach, J., Martinson, D.G., Mclntyre, A., Mix,
A.C., Molfino, B., Morley, J.J., Peterson, L.C., Pisias, N.G., Prell, W.L.,
Raymo, M.E., Shacldeton, N.J. and Toggweier, J.R. (1993). On the
structure and origin of major Glaciation cycles 2. the 10,000-year cycle.
Paleoceanography,8, 699-736.
Ittekkot, V., Haake, B., Bartsch, M., Hair, R.R. and Ramaswamy, V. (1992).
Organic carbon removal in the sea: the continental connection. In:
Summerhayes,C.P., Prell, W.L., Emeis, K.C. (ede), UpwelltngSystems:
Evolution since the early Miocene, pp. 167-176, Geological society
Special Publication No. 64, London.
Jenkins, G.M. and Watts, D.G. (1968). Spectral analysis and its
applications, Holden-Day, San Francisco.
Knox, R.A. (1987). The Indian Ocean: Interaction with the monsoon. In:
Fein, J.S., Stephens, P.L. (eds), Monsoons, 632, Wiley, New York.
Molfino, B. and McIntyre, A. (1990). Precessional forcing of the nutricline
dynamics in the F_.quatodalAtlantic. Science, 249, 766-769.
Mosimann, J.E. (1965). Statistical methods for the pollen analyst:
multinomial and negative multinomial techniques. In: Kummel, B.a.R.,
D. (ed.), Handbookof Paleontological Techniques, pp. 636--673, W.H.
Freeman, San Francisco.
MUller, PJ. and Suess, E. (1979). Productivity, sedimentation rate, and
sedimentary organic matter in the oceans, I. Organic carbon
preservation. Deep Sea Research, 26, 1347-1362.
Okada, H. and Honjo, S. (1973). The distribution of oceanic
coccolithophorids in the Pacific. 20, 355-374.
Prell, W.L. and Curry, W.B. (1981). Faunal and isotopic indices of
monsoonal upweiling: western Arabian Sea. Oceanologica Acta, 4,
91-98.
Prell, W.L., Hutson, W.H., Williams, D.F., B~, A.W.H., Geitzenauer, K.
and Moifino, B. (1980). Surface circulation of the Indian Ocean during
the last Glacial Maximum, approximately 18,000 BP. Quaternary
Research, 14, 309-336.
Prell, W.L. and Kutzbach, J.E. (1987). Monsoon variability over the past
150,000 years. Journal of Geophysical Research, 92, 8411-8425.
Press, W.H., Flannery, B.P., Teukolsky, S.A. and Vetterling (1986).
Numerical Recipes: The Art of Scientific Computing, Cambridge
University Press, Cambridge.
Rostek, F., Ruhland, G., Bassinot, F.C., MUller, P.J., Labeyrie, L.D.,
Lancelot, Y. and Bard, E. (1993). Reconstructing sea surface
temperature and salinity using 8~aOand alkenone records. Nature, 364,
319-321.
Sarkar, A., Ramesh, R., S.K., B. and Rajagopalan, G. (1990). Oxygen
isotope evidence for a stronger winter monsoon current during the last
glaciation. Nature, 343, 549-551.
Venrick, E.L. (1982). Phytoplankton in an oligotrophic ocean: observations
and question. Ecological Monographs, Vol. 52, pp. 129-154.
Venrick, E.L. (1990). Phytoplankton in an ologotruphic ocean: species
structure and interannual variability. Ecology, 71, 1574-1563.